Add files using upload-large-folder tool

bigscience/evaluation/README.md

@@ -0,0 +1,7 @@
+# Evaluation
+
+This folder contains scripts and results for intermediate evaluation, mostly based on zero-shot prompting performance. Most are performed with Eleuther AI's [LM eval harness](https://github.com/EleutherAI/lm-evaluation-harness).
+
+Evaluated models:
+- BLOOM (tr11 / The `bigscience/bloom` model in 176B / 6B3 / 2B5 / 1B3 / 750M / 350M variants)
+- [13B](https://github.com/bigscience-workshop/bigscience/blob/master/evaluation/Tr1-13B-harness-eval.json)

bigscience/evaluation/results/tr1/Tr1-13B-harness-eval.json

@@ -0,0 +1,165 @@
+{
+  "results": {
+    "lambada": {
+      "ppl": 5.020137688328123,
+      "ppl_stderr": 0.11575351197990837,
+      "acc": 0.634193673588201,
+      "acc_stderr": 0.006710403442216892
+    },
+    "winogrande": {
+      "acc": 0.6471981057616417,
+      "acc_stderr": 0.013429728101788954
+    },
+    "hellaswag": {
+      "acc": 0.5416251742680741,
+      "acc_stderr": 0.004972460206842306,
+      "acc_norm": 0.7162915753833897,
+      "acc_norm_stderr": 0.004498757194493409
+    },
+    "piqa": {
+      "acc": 0.7769314472252449,
+      "acc_stderr": 0.009713057213018522,
+      "acc_norm": 0.7878128400435256,
+      "acc_norm_stderr": 0.009539299828174046
+    },
+    "cola": {
+      "mcc": 0.05586916675965605,
+      "mcc_stderr": 0.034250689348891604
+    },
+    "mnli": {
+      "acc": 0.3959246051961284,
+      "acc_stderr": 0.004936609703575665
+    },
+    "mnli_mismatched": {
+      "acc": 0.3984947111472742,
+      "acc_stderr": 0.004937784794740595
+    },
+    "mrpc": {
+      "acc": 0.6764705882352942,
+      "acc_stderr": 0.023189113109403536,
+      "f1": 0.8058823529411765,
+      "f1_stderr": 0.016598529068410604
+    },
+    "rte": {
+      "acc": 0.5234657039711191,
+      "acc_stderr": 0.03006330041190266
+    },
+    "qnli": {
+      "acc": 0.5171151382024529,
+      "acc_stderr": 0.006761445834294947
+    },
+    "qqp": {
+      "acc": 0.36772198862231015,
+      "acc_stderr": 0.0023981002797098354,
+      "f1": 0.532523819102829,
+      "f1_stderr": 0.0025759259415034795
+    },
+    "sst": {
+      "acc": 0.5137614678899083,
+      "acc_stderr": 0.01693543564494107
+    },
+    "wnli": {
+      "acc": 0.18309859154929578,
+      "acc_stderr": 0.046225147349214284
+    },
+    "boolq": {
+      "acc": 0.5868501529051988,
+      "acc_stderr": 0.008612117547803569
+    },
+    "copa": {
+      "acc": 0.88,
+      "acc_stderr": 0.03265986323710906
+    },
+    "multirc": {
+      "acc": 0.017838405036726127,
+      "acc_stderr": 0.00428993794671089
+    },
+    "record": {
+      "f1": 0.885354285714286,
+      "f1_stderr": 0.00314773987203575,
+      "em": 0.8783,
+      "em_stderr": 0.003269553486028481
+    },
+    "wic": {
+      "acc": 0.49843260188087773,
+      "acc_stderr": 0.019810623954060382
+    },
+    "wsc": {
+      "acc": 0.5,
+      "acc_stderr": 0.04926646390821466
+    },
+    "prost": {
+      "acc": 0.28047608881298036,
+      "acc_stderr": 0.003282038627279345,
+      "acc_norm": 0.2830380017079419,
+      "acc_norm_stderr": 0.003291119066155946
+    },
+    "mc_taco": {
+      "em": 0.12612612612612611,
+      "f1": 0.4965489467730623
+    },
+    "pubmedqa": {
+      "acc": 0.615,
+      "acc_stderr": 0.015395194445410805
+    },
+    "sciq": {
+      "acc": 0.895,
+      "acc_stderr": 0.009698921026024957,
+      "acc_norm": 0.815,
+      "acc_norm_stderr": 0.012285191326386693
+    },
+    "triviaqa": {
+      "acc": 0.13294440024750287,
+      "acc_stderr": 0.0031921904944669202
+    },
+    "arc_easy": {
+      "acc": 0.6813973063973064,
+      "acc_stderr": 0.009560775507673364,
+      "acc_norm": 0.6001683501683501,
+      "acc_norm_stderr": 0.010051788039412911
+    },
+    "arc_challenge": {
+      "acc": 0.3216723549488055,
+      "acc_stderr": 0.013650488084494164,
+      "acc_norm": 0.34215017064846415,
+      "acc_norm_stderr": 0.013864152159177275
+    },
+    "logiqa": {
+      "acc": 0.23195084485407066,
+      "acc_stderr": 0.0165552524979259,
+      "acc_norm": 0.2749615975422427,
+      "acc_norm_stderr": 0.01751297178222522
+    },
+    "openbookqa": {
+      "acc": 0.294,
+      "acc_stderr": 0.020395095484936603,
+      "acc_norm": 0.412,
+      "acc_norm_stderr": 0.022033677993740865
+    },
+    "race": {
+      "acc": 0.3741626794258373,
+      "acc_stderr": 0.014976513181619648
+    },
+    "headqa": {
+      "acc": 0.22283005105762219,
+      "acc_stderr": 0.007948594863726302,
+      "acc_norm": 0.26258205689277897,
+      "acc_norm_stderr": 0.00840494460823324
+    },
+    "mathqa": {
+      "acc": 0.2375209380234506,
+      "acc_stderr": 0.0077905030438074,
+      "acc_norm": 0.23450586264656617,
+      "acc_norm_stderr": 0.007756188894243557
+    },
+    "webqs": {
+      "acc": 0.0265748031496063,
+      "acc_stderr": 0.003568875174120304
+    },
+    "wikitext": {
+      "word_perplexity": 12.921754196505068,
+      "byte_perplexity": 1.6136995247803747,
+      "bits_per_byte": 0.4785293844744369
+    }
+  }
+}

bigscience/evaluation/results/tr11/README.md

@@ -0,0 +1,11 @@
+# BigScience BLOOM Evaluation Results
+
+This folder contains evaluation results of the BLOOM model family.
+
+## Evaluation Procedure
+
+- bslmeval files were created using the below:
+    - https://github.com/bigscience-workshop/Megatron-DeepSpeed/pull/291
+    - https://github.com/bigscience-workshop/lm-evaluation-harness
+- humaneval files were created using the HumanEval code dataset with the below:
+    - https://github.com/loubnabnl/bloom-code-evaluation

bigscience/evaluation/results/tr11/bloom1b3/bslmeval.json

@@ -0,0 +1,2938 @@
+{
+  "results": {
+    "arc_challenge": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.23464163822525597,
+        "acc_norm": 0.26791808873720135,
+        "acc_norm_stderr": 0.012942030195136423,
+        "acc_stderr": 0.012383873560768673
+      }
+    },
+    "arc_easy": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.5631313131313131,
+        "acc_norm": 0.4810606060606061,
+        "acc_norm_stderr": 0.010252420496894487,
+        "acc_stderr": 0.010177672928157678
+      }
+    },
+    "axb+GPT-3 style": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4855072463768116,
+        "acc_norm": 0.5878623188405797,
+        "acc_norm_stderr": 0.014820785339690506,
+        "acc_stderr": 0.015048725939283577,
+        "prompt_name": "GPT-3 style",
+        "task_name": "axb"
+      }
+    },
+    "axb+MNLI crowdsource": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.447463768115942,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.0149717153798021,
+        "prompt_name": "MNLI crowdsource",
+        "task_name": "axb"
+      }
+    },
+    "axb+based on the previous passage": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4846014492753623,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.015047910329698355,
+        "prompt_name": "based on the previous passage",
+        "task_name": "axb"
+      }
+    },
+    "axb+can we infer": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.421195652173913,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.014866888213508284,
+        "prompt_name": "can we infer",
+        "task_name": "axb"
+      }
+    },
+    "axb+does it follow that": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4375,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.014936970932375573,
+        "prompt_name": "does it follow that",
+        "task_name": "axb"
+      }
+    },
+    "axb+does this imply": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5353260869565217,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.015017429208641943,
+        "prompt_name": "does this imply",
+        "task_name": "axb"
+      }
+    },
+    "axb+guaranteed true": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.44655797101449274,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.014968808595500557,
+        "prompt_name": "guaranteed true",
+        "task_name": "axb"
+      }
+    },
+    "axb+justified in saying": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4365942028985507,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.014933509475434285,
+        "prompt_name": "justified in saying",
+        "task_name": "axb"
+      }
+    },
+    "axb+must be true": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4266304347826087,
+        "acc_norm": 0.4166666666666667,
+        "acc_norm_stderr": 0.014844481058991162,
+        "acc_stderr": 0.014892084059444173,
+        "prompt_name": "must be true",
+        "task_name": "axb"
+      }
+    },
+    "axb+should assume": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5163043478260869,
+        "acc_norm": 0.4157608695652174,
+        "acc_norm_stderr": 0.014839845193003246,
+        "acc_stderr": 0.015047045240919796,
+        "prompt_name": "should assume",
+        "task_name": "axb"
+      }
+    },
+    "axg+GPT-3 style": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4803370786516854,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.02651671646679541,
+        "parity": 0.9606741573033708,
+        "parity_stderr": 0.01460967112412074,
+        "prompt_name": "GPT-3 style",
+        "task_name": "axg"
+      }
+    },
+    "axg+MNLI crowdsource": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5140449438202247,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026526773058212952,
+        "parity": 0.9719101123595506,
+        "parity_stderr": 0.012419422972302346,
+        "prompt_name": "MNLI crowdsource",
+        "task_name": "axg"
+      }
+    },
+    "axg+based on the previous passage": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026537244621713762,
+        "parity": 0.9662921348314607,
+        "parity_stderr": 0.013565419020002358,
+        "prompt_name": "based on the previous passage",
+        "task_name": "axg"
+      }
+    },
+    "axg+can we infer": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026537244621713762,
+        "parity": 0.9887640449438202,
+        "parity_stderr": 0.007922544664164389,
+        "prompt_name": "can we infer",
+        "task_name": "axg"
+      }
+    },
+    "axg+does it follow that": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026537244621713762,
+        "parity": 1.0,
+        "parity_stderr": 0.0,
+        "prompt_name": "does it follow that",
+        "task_name": "axg"
+      }
+    },
+    "axg+does this imply": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.49719101123595505,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026536825838510643,
+        "parity": 0.9831460674157303,
+        "parity_stderr": 0.009675491064988365,
+        "prompt_name": "does this imply",
+        "task_name": "axg"
+      }
+    },
+    "axg+guaranteed true": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.48314606741573035,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026522164260489825,
+        "parity": 0.9887640449438202,
+        "parity_stderr": 0.007922544664164387,
+        "prompt_name": "guaranteed true",
+        "task_name": "axg"
+      }
+    },
+    "axg+justified in saying": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026537244621713762,
+        "parity": 0.9887640449438202,
+        "parity_stderr": 0.007922544664164385,
+        "prompt_name": "justified in saying",
+        "task_name": "axg"
+      }
+    },
+    "axg+must be true": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4803370786516854,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026516716466795417,
+        "parity": 0.9719101123595506,
+        "parity_stderr": 0.012419422972302347,
+        "prompt_name": "must be true",
+        "task_name": "axg"
+      }
+    },
+    "axg+should assume": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.49719101123595505,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.026537244621713762,
+        "acc_stderr": 0.026536825838510643,
+        "parity": 0.949438202247191,
+        "parity_stderr": 0.016468648928151884,
+        "prompt_name": "should assume",
+        "task_name": "axg"
+      }
+    },
+    "boolq": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.617737003058104,
+        "acc_stderr": 0.008499149690449272
+      }
+    },
+    "boolq+GPT-3 Style": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5896024464831804,
+        "acc_norm": 0.6211009174311927,
+        "acc_norm_stderr": 0.008484678718565017,
+        "acc_stderr": 0.008603488048617526,
+        "prompt_name": "GPT-3 Style",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+I wonder\u2026": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.563914373088685,
+        "acc_norm": 0.6217125382262997,
+        "acc_norm_stderr": 0.00848200113393099,
+        "acc_stderr": 0.008673312776324934,
+        "prompt_name": "I wonder\u2026",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+after_reading": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6217125382262997,
+        "acc_norm": 0.3804281345565749,
+        "acc_norm_stderr": 0.008491310027059626,
+        "acc_stderr": 0.00848200113393099,
+        "prompt_name": "after_reading",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+based on the following passage": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.3798165137614679,
+        "acc_norm": 0.6012232415902141,
+        "acc_norm_stderr": 0.008563973987729906,
+        "acc_stderr": 0.008488668235778644,
+        "prompt_name": "based on the following passage",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+based on the previous passage": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6146788990825688,
+        "acc_norm": 0.6217125382262997,
+        "acc_norm_stderr": 0.00848200113393099,
+        "acc_stderr": 0.008511930879680635,
+        "prompt_name": "based on the previous passage",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+could you tell me\u2026": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5840978593272171,
+        "acc_norm": 0.6217125382262997,
+        "acc_norm_stderr": 0.00848200113393099,
+        "acc_stderr": 0.008620469604001,
+        "prompt_name": "could you tell me\u2026",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+exam": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6220183486238532,
+        "acc_norm": 0.6217125382262997,
+        "acc_norm_stderr": 0.00848200113393099,
+        "acc_stderr": 0.008480656964585267,
+        "prompt_name": "exam",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+exercise": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6217125382262997,
+        "acc_norm": 0.46788990825688076,
+        "acc_norm_stderr": 0.0087270030269178,
+        "acc_stderr": 0.00848200113393099,
+        "prompt_name": "exercise",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+valid_binary": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.491131498470948,
+        "acc_norm": 0.37370030581039754,
+        "acc_norm_stderr": 0.008461461177104003,
+        "acc_stderr": 0.008743679265456042,
+        "prompt_name": "valid_binary",
+        "task_name": "boolq"
+      }
+    },
+    "boolq+yes_no_question": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.5951070336391437,
+        "acc_norm": 0.6217125382262997,
+        "acc_norm_stderr": 0.00848200113393099,
+        "acc_stderr": 0.008585393347962319,
+        "prompt_name": "yes_no_question",
+        "task_name": "boolq"
+      }
+    },
+    "cb+GPT-3 style": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.42857142857142855,
+        "acc_stderr": 0.06672848092813057,
+        "f1": 0.21956970232832299,
+        "prompt_name": "GPT-3 style",
+        "task_name": "cb"
+      }
+    },
+    "cb+MNLI crowdsource": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.42857142857142855,
+        "acc_stderr": 0.06672848092813057,
+        "f1": 0.21956970232832299,
+        "prompt_name": "MNLI crowdsource",
+        "task_name": "cb"
+      }
+    },
+    "cb+always/sometimes/never": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.08928571428571429,
+        "acc_stderr": 0.038450387280282494,
+        "f1": 0.054644808743169404,
+        "prompt_name": "always/sometimes/never",
+        "task_name": "cb"
+      }
+    },
+    "cb+based on the previous passage": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.35714285714285715,
+        "acc_stderr": 0.06460957383809221,
+        "f1": 0.2094181249110827,
+        "prompt_name": "based on the previous passage",
+        "task_name": "cb"
+      }
+    },
+    "cb+can we infer": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.25,
+        "acc_stderr": 0.058387420812114225,
+        "f1": 0.15483870967741933,
+        "prompt_name": "can we infer",
+        "task_name": "cb"
+      }
+    },
+    "cb+claim true/false/inconclusive": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.42857142857142855,
+        "acc_stderr": 0.06672848092813057,
+        "f1": 0.21956970232832299,
+        "prompt_name": "claim true/false/inconclusive",
+        "task_name": "cb"
+      }
+    },
+    "cb+consider always/sometimes/never": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.08928571428571429,
+        "acc_stderr": 0.038450387280282494,
+        "f1": 0.054644808743169404,
+        "prompt_name": "consider always/sometimes/never",
+        "task_name": "cb"
+      }
+    },
+    "cb+does it follow that": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.30357142857142855,
+        "acc_stderr": 0.06199938655510754,
+        "f1": 0.2613574165298303,
+        "prompt_name": "does it follow that",
+        "task_name": "cb"
+      }
+    },
+    "cb+does this imply": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.10714285714285714,
+        "acc_stderr": 0.0417053005800816,
+        "f1": 0.11222753854332802,
+        "prompt_name": "does this imply",
+        "task_name": "cb"
+      }
+    },
+    "cb+guaranteed true": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.21428571428571427,
+        "acc_stderr": 0.055328333517248834,
+        "f1": 0.15883777239709443,
+        "prompt_name": "guaranteed true",
+        "task_name": "cb"
+      }
+    },
+    "cb+guaranteed/possible/impossible": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.10714285714285714,
+        "acc_stderr": 0.0417053005800816,
+        "f1": 0.07871939736346516,
+        "prompt_name": "guaranteed/possible/impossible",
+        "task_name": "cb"
+      }
+    },
+    "cb+justified in saying": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.21428571428571427,
+        "acc_stderr": 0.055328333517248834,
+        "f1": 0.1623009758602979,
+        "prompt_name": "justified in saying",
+        "task_name": "cb"
+      }
+    },
+    "cb+must be true": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.19642857142857142,
+        "acc_stderr": 0.05357142857142859,
+        "f1": 0.1384656508954825,
+        "prompt_name": "must be true",
+        "task_name": "cb"
+      }
+    },
+    "cb+should assume": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.19642857142857142,
+        "acc_stderr": 0.05357142857142858,
+        "f1": 0.14613935969868175,
+        "prompt_name": "should assume",
+        "task_name": "cb"
+      }
+    },
+    "cb+take the following as truth": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.4107142857142857,
+        "acc_stderr": 0.06633634150359538,
+        "f1": 0.1940928270042194,
+        "prompt_name": "take the following as truth",
+        "task_name": "cb"
+      }
+    },
+    "cola+Following sentence acceptable": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6625119846596357,
+        "acc_norm": 0.31064237775647174,
+        "acc_norm_stderr": 0.014335695984672221,
+        "acc_stderr": 0.014648467353878477,
+        "prompt_name": "Following sentence acceptable",
+        "task_name": "cola"
+      }
+    },
+    "cola+Make sense yes no": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.3710450623202301,
+        "acc_norm": 0.6903163950143816,
+        "acc_norm_stderr": 0.014323506235950028,
+        "acc_stderr": 0.01496543118537874,
+        "prompt_name": "Make sense yes no",
+        "task_name": "cola"
+      }
+    },
+    "cola+Previous sentence acceptable": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6864813039309684,
+        "acc_norm": 0.6912751677852349,
+        "acc_norm_stderr": 0.014311244461311299,
+        "acc_stderr": 0.014371834902632595,
+        "prompt_name": "Previous sentence acceptable",
+        "task_name": "cola"
+      }
+    },
+    "cola+editing": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.46596356663470756,
+        "acc_norm": 0.6912751677852349,
+        "acc_norm_stderr": 0.014311244461311299,
+        "acc_stderr": 0.015453525186655532,
+        "prompt_name": "editing",
+        "task_name": "cola"
+      }
+    },
+    "cola+is_this_correct": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.6893576222435283,
+        "acc_norm": 0.6912751677852349,
+        "acc_norm_stderr": 0.014311244461311299,
+        "acc_stderr": 0.014335695984672223,
+        "prompt_name": "is_this_correct",
+        "task_name": "cola"
+      }
+    },
+    "copa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.7,
+        "acc_stderr": 0.046056618647183814
+      }
+    },
+    "copa+C1 or C2? premise, so/because\u2026": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.65,
+        "acc_norm": 0.57,
+        "acc_norm_stderr": 0.049756985195624284,
+        "acc_stderr": 0.047937248544110196,
+        "prompt_name": "C1 or C2? premise, so/because\u2026",
+        "task_name": "copa"
+      }
+    },
+    "copa+best_option": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.52,
+        "acc_norm": 0.49,
+        "acc_norm_stderr": 0.05024183937956911,
+        "acc_stderr": 0.050211673156867795,
+        "prompt_name": "best_option",
+        "task_name": "copa"
+      }
+    },
+    "copa+cause_effect": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.56,
+        "acc_norm": 0.45,
+        "acc_norm_stderr": 0.05,
+        "acc_stderr": 0.04988876515698589,
+        "prompt_name": "cause_effect",
+        "task_name": "copa"
+      }
+    },
+    "copa+choose": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.53,
+        "acc_norm": 0.46,
+        "acc_norm_stderr": 0.05009082659620333,
+        "acc_stderr": 0.05016135580465919,
+        "prompt_name": "choose",
+        "task_name": "copa"
+      }
+    },
+    "copa+exercise": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.54,
+        "acc_norm": 0.48,
+        "acc_norm_stderr": 0.050211673156867795,
+        "acc_stderr": 0.05009082659620332,
+        "prompt_name": "exercise",
+        "task_name": "copa"
+      }
+    },
+    "copa+i_am_hesitating": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.56,
+        "acc_norm": 0.48,
+        "acc_norm_stderr": 0.050211673156867795,
+        "acc_stderr": 0.04988876515698589,
+        "prompt_name": "i_am_hesitating",
+        "task_name": "copa"
+      }
+    },
+    "copa+more likely": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.53,
+        "acc_norm": 0.49,
+        "acc_norm_stderr": 0.05024183937956911,
+        "acc_stderr": 0.05016135580465919,
+        "prompt_name": "more likely",
+        "task_name": "copa"
+      }
+    },
+    "copa+plausible_alternatives": {
+      "2022-07-15-11-47-34": {
+        "acc": 0.56,
+        "acc_norm": 0.53,
+        "acc_norm_stderr": 0.05016135580465919,
+        "acc_stderr": 0.04988876515698589,
+        "prompt_name": "plausible_alternatives",
+        "task_name": "copa"
+      }
+    },
+    "crows_pairs_english+1": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.49552772808586765,
+        "acc_norm": 0.49552772808586765,
+        "acc_norm_stderr": 0.012212810647205384,
+        "acc_stderr": 0.012212810647205384,
+        "prompt_name": "1",
+        "task_name": "crows_pairs_english"
+      }
+    },
+    "crows_pairs_english+2": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4883720930232558,
+        "acc_norm": 0.4883720930232558,
+        "acc_norm_stderr": 0.012209996095069646,
+        "acc_stderr": 0.012209996095069646,
+        "prompt_name": "2",
+        "task_name": "crows_pairs_english"
+      }
+    },
+    "crows_pairs_english+3": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5163983303518187,
+        "acc_norm": 0.4836016696481813,
+        "acc_norm_stderr": 0.012206729011137944,
+        "acc_stderr": 0.012206729011137944,
+        "prompt_name": "3",
+        "task_name": "crows_pairs_english"
+      }
+    },
+    "crows_pairs_english+4": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4919499105545617,
+        "acc_norm": 0.4919499105545617,
+        "acc_norm_stderr": 0.01221171617623539,
+        "acc_stderr": 0.01221171617623539,
+        "prompt_name": "4",
+        "task_name": "crows_pairs_english"
+      }
+    },
+    "crows_pairs_english+A_preference": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5104353011329755,
+        "acc_norm": 0.5104353011329755,
+        "acc_norm_stderr": 0.012210638982043397,
+        "acc_stderr": 0.012210638982043397,
+        "prompt_name": "A_preference",
+        "task_name": "crows_pairs_english"
+      }
+    },
+    "crows_pairs_english+A_stereotype_true": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4907573047107931,
+        "acc_norm": 0.5062611806797853,
+        "acc_norm_stderr": 0.012212341600228735,
+        "acc_stderr": 0.012211212339167695,
+        "prompt_name": "A_stereotype_true",
+        "task_name": "crows_pairs_english"
+      }
+    },
+    "crows_pairs_french+1_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.48598688133571855,
+        "acc_norm": 0.48598688133571855,
+        "acc_norm_stderr": 0.012208501686447066,
+        "acc_stderr": 0.012208501686447066,
+        "prompt_name": "1_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "crows_pairs_french+2_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.49850924269528923,
+        "acc_norm": 0.49850924269528923,
+        "acc_norm_stderr": 0.01221324493389968,
+        "acc_stderr": 0.01221324493389968,
+        "prompt_name": "2_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "crows_pairs_french+3_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.49612403100775193,
+        "acc_norm": 0.49612403100775193,
+        "acc_norm_stderr": 0.012212932249036454,
+        "acc_stderr": 0.012212932249036454,
+        "prompt_name": "3_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "crows_pairs_french+4_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5313059033989267,
+        "acc_norm": 0.5313059033989267,
+        "acc_norm_stderr": 0.012189336188399829,
+        "acc_stderr": 0.012189336188399829,
+        "prompt_name": "4_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "crows_pairs_french+A_preference_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4847942754919499,
+        "acc_norm": 0.4847942754919499,
+        "acc_norm_stderr": 0.01220765013925874,
+        "acc_stderr": 0.01220765013925874,
+        "prompt_name": "A_preference_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "crows_pairs_french+A_reality_check_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.505664877757901,
+        "acc_norm": 0.505664877757901,
+        "acc_norm_stderr": 0.012212515323431717,
+        "acc_stderr": 0.012212515323431717,
+        "prompt_name": "A_reality_check_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "crows_pairs_french+A_stereotype_true_fr": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5020870602265951,
+        "acc_norm": 0.5020870602265951,
+        "acc_norm_stderr": 0.012213192820312026,
+        "acc_stderr": 0.012213192820312026,
+        "prompt_name": "A_stereotype_true_fr",
+        "task_name": "crows_pairs_french"
+      }
+    },
+    "diabla+Is the error present? (same lang)": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.07741823242867084,
+        "acc_norm": 0.07741823242867084,
+        "acc_norm_stderr": 0.0035253599064790993,
+        "acc_stderr": 0.0035253599064790993,
+        "prompt_name": "Is the error present? (same lang)",
+        "task_name": "diabla"
+      }
+    },
+    "diabla+Which is automatic?": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4966945024356298,
+        "acc_norm": 0.4966945024356298,
+        "acc_norm_stderr": 0.0065953813991735995,
+        "acc_stderr": 0.0065953813991735995,
+        "prompt_name": "Which is automatic?",
+        "task_name": "diabla"
+      }
+    },
+    "gsarti/flores_101_afr+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.8175051369933213,
+        "byte_perplexity": 7.049422805555328,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_afr",
+        "word_perplexity": 139324.0466654445
+      }
+    },
+    "gsarti/flores_101_amh+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.0608666814101815,
+        "byte_perplexity": 4.172368790188039,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_amh",
+        "word_perplexity": 105036774.30501972
+      }
+    },
+    "gsarti/flores_101_ara+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 0.8797352167688847,
+        "byte_perplexity": 1.8400375612633983,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ara",
+        "word_perplexity": 674.8640314665696
+      }
+    },
+    "gsarti/flores_101_asm+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.458711333673663,
+        "byte_perplexity": 5.497254736157445,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_asm",
+        "word_perplexity": 6763188828222.085
+      }
+    },
+    "gsarti/flores_101_ast+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.0909386784329675,
+        "byte_perplexity": 4.260251728273795,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ast",
+        "word_perplexity": 10657.272913539553
+      }
+    },
+    "gsarti/flores_101_azj+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.9432455349850195,
+        "byte_perplexity": 7.691396328945705,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_azj",
+        "word_perplexity": 45923924.18878753
+      }
+    },
+    "gsarti/flores_101_bel+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.914816732584341,
+        "byte_perplexity": 3.7706591215465943,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_bel",
+        "word_perplexity": 23935692.781315073
+      }
+    },
+    "gsarti/flores_101_ben+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.3432036318231058,
+        "byte_perplexity": 5.074281765515423,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ben",
+        "word_perplexity": 2480418685142.412
+      }
+    },
+    "gsarti/flores_101_bos+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.665248069942796,
+        "byte_perplexity": 6.343363734045183,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_bos",
+        "word_perplexity": 229622.13691086147
+      }
+    },
+    "gsarti/flores_101_bul+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.5136770683283687,
+        "byte_perplexity": 2.8553687444403257,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_bul",
+        "word_perplexity": 194851.13344620814
+      }
+    },
+    "gsarti/flores_101_cat+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.2376904653775254,
+        "byte_perplexity": 2.358207169698056,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_cat",
+        "word_perplexity": 179.13123174533087
+      }
+    },
+    "gsarti/flores_101_ceb+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.7858604115174295,
+        "byte_perplexity": 6.896481056329736,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ceb",
+        "word_perplexity": 113330.67154113152
+      }
+    },
+    "gsarti/flores_101_ces+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.6150694333085327,
+        "byte_perplexity": 6.126526835715164,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ces",
+        "word_perplexity": 625101.1441414964
+      }
+    },
+    "gsarti/flores_101_ckb+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.9421776126623524,
+        "byte_perplexity": 3.842852526862475,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ckb",
+        "word_perplexity": 11104497.438038943
+      }
+    },
+    "gsarti/flores_101_cym+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.8470317241534553,
+        "byte_perplexity": 14.390369428021707,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_cym",
+        "word_perplexity": 5900331.966242436
+      }
+    },
+    "gsarti/flores_101_dan+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.5307665257708245,
+        "byte_perplexity": 5.778786323448377,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_dan",
+        "word_perplexity": 71695.50336412797
+      }
+    },
+    "gsarti/flores_101_deu+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.7492158999678582,
+        "byte_perplexity": 3.361758059911202,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_deu",
+        "word_perplexity": 5647.282599404732
+      }
+    },
+    "gsarti/flores_101_ell+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.3862374641150543,
+        "byte_perplexity": 2.6139607239932805,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ell",
+        "word_perplexity": 102751.5248402687
+      }
+    },
+    "gsarti/flores_101_eng+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.0435427545613876,
+        "byte_perplexity": 2.061283234268159,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_eng",
+        "word_perplexity": 75.56480997823662
+      }
+    },
+    "gsarti/flores_101_est+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.340809503762674,
+        "byte_perplexity": 10.131736127467489,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_est",
+        "word_perplexity": 92602633.82439691
+      }
+    },
+    "gsarti/flores_101_fas+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.6586730625582675,
+        "byte_perplexity": 3.1572599808371367,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_fas",
+        "word_perplexity": 59965.98383842629
+      }
+    },
+    "gsarti/flores_101_fin+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.9093822743068216,
+        "byte_perplexity": 7.5129644427067355,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_fin",
+        "word_perplexity": 91621886.60145952
+      }
+    },
+    "gsarti/flores_101_fra+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.0127395726746855,
+        "byte_perplexity": 2.0177390037335385,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_fra",
+        "word_perplexity": 89.45884576931464
+      }
+    },
+    "gsarti/flores_101_ful+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.561969238361191,
+        "byte_perplexity": 11.810263420287875,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ful",
+        "word_perplexity": 908715.1423017589
+      }
+    },
+    "gsarti/flores_101_gle+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.2798070331865063,
+        "byte_perplexity": 9.712259930753122,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_gle",
+        "word_perplexity": 1548851.5929806433
+      }
+    },
+    "gsarti/flores_101_glg+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.6877168009728167,
+        "byte_perplexity": 3.2214647330840154,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_glg",
+        "word_perplexity": 1537.3193913761668
+      }
+    },
+    "gsarti/flores_101_guj+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.357806609400009,
+        "byte_perplexity": 5.125904532570054,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_guj",
+        "word_perplexity": 133216198508.6925
+      }
+    },
+    "gsarti/flores_101_hau+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.4659038057537184,
+        "byte_perplexity": 11.049458818357667,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_hau",
+        "word_perplexity": 730749.6449046461
+      }
+    },
+    "gsarti/flores_101_heb+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.8889611054621571,
+        "byte_perplexity": 3.7036842387723694,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_heb",
+        "word_perplexity": 880255.4148832298
+      }
+    },
+    "gsarti/flores_101_hin+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.195760704215568,
+        "byte_perplexity": 4.581311639568996,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_hin",
+        "word_perplexity": 453226793.5348556
+      }
+    },
+    "gsarti/flores_101_hrv+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.7016816564307984,
+        "byte_perplexity": 6.50559790827845,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_hrv",
+        "word_perplexity": 307789.1462790266
+      }
+    },
+    "gsarti/flores_101_hun+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.8470581600253615,
+        "byte_perplexity": 7.19531655942431,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_hun",
+        "word_perplexity": 8545882.19823639
+      }
+    },
+    "gsarti/flores_101_hye+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.906169044483402,
+        "byte_perplexity": 3.7481249397064547,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_hye",
+        "word_perplexity": 99262887.01092263
+      }
+    },
+    "gsarti/flores_101_ibo+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.6012385649422316,
+        "byte_perplexity": 6.06807351892086,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ibo",
+        "word_perplexity": 99576.38125028457
+      }
+    },
+    "gsarti/flores_101_ind+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.1501325666473412,
+        "byte_perplexity": 2.2193428661828962,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ind",
+        "word_perplexity": 299.41864562936706
+      }
+    },
+    "gsarti/flores_101_isl+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.1394769822824644,
+        "byte_perplexity": 8.812045732299993,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_isl",
+        "word_perplexity": 3947458.536983725
+      }
+    },
+    "gsarti/flores_101_ita+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.695253347487448,
+        "byte_perplexity": 3.238337491305615,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ita",
+        "word_perplexity": 1951.0663459405935
+      }
+    },
+    "gsarti/flores_101_jav+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.899297993680408,
+        "byte_perplexity": 7.460632752007581,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_jav",
+        "word_perplexity": 956961.3940329206
+      }
+    },
+    "gsarti/flores_101_jpn+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.539549942005635,
+        "byte_perplexity": 2.907038023970581,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_jpn",
+        "word_perplexity": 6.0024027118732196e+69
+      }
+    },
+    "gsarti/flores_101_kam+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.515626316920499,
+        "byte_perplexity": 11.436917146974627,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kam",
+        "word_perplexity": 4288601.196402131
+      }
+    },
+    "gsarti/flores_101_kan+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.3991591199422513,
+        "byte_perplexity": 5.274956219477929,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kan",
+        "word_perplexity": 5.3861539364992216e+16
+      }
+    },
+    "gsarti/flores_101_kat+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.3325401608568794,
+        "byte_perplexity": 2.5184571084900518,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kat",
+        "word_perplexity": 1133105340.614723
+      }
+    },
+    "gsarti/flores_101_kaz+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.845791322405974,
+        "byte_perplexity": 3.5945005448756477,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kaz",
+        "word_perplexity": 89537342.10068764
+      }
+    },
+    "gsarti/flores_101_kea+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.2143692668645976,
+        "byte_perplexity": 9.281572608888562,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kea",
+        "word_perplexity": 438558.0012817139
+      }
+    },
+    "gsarti/flores_101_kir+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.9511242166700078,
+        "byte_perplexity": 3.8667573034119127,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kir",
+        "word_perplexity": 235337758.18519488
+      }
+    },
+    "gsarti/flores_101_kor+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.023500324792833,
+        "byte_perplexity": 4.065690303705374,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_kor",
+        "word_perplexity": 1684949.6449262113
+      }
+    },
+    "gsarti/flores_101_lao+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.6376750107826055,
+        "byte_perplexity": 3.1116396826339545,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_lao",
+        "word_perplexity": 3.0817754157127624e+28
+      }
+    },
+    "gsarti/flores_101_lav+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.075865182775687,
+        "byte_perplexity": 8.431943399753028,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_lav",
+        "word_perplexity": 20692036.880855087
+      }
+    },
+    "gsarti/flores_101_lin+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.018221991102226,
+        "byte_perplexity": 8.10168498947524,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_lin",
+        "word_perplexity": 259077.7174090486
+      }
+    },
+    "gsarti/flores_101_lit+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.0526165270213905,
+        "byte_perplexity": 8.297153789252596,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_lit",
+        "word_perplexity": 22011900.13997282
+      }
+    },
+    "gsarti/flores_101_ltz+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.2407955989852377,
+        "byte_perplexity": 9.453152958003827,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ltz",
+        "word_perplexity": 6731220.931729273
+      }
+    },
+    "gsarti/flores_101_lug+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 3.2150119431528754,
+        "byte_perplexity": 9.285708185212261,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_lug",
+        "word_perplexity": 32046806.791237485
+      }
+    },
+    "gsarti/flores_101_luo+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.609093857404177,
+        "byte_perplexity": 12.202407052163576,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_luo",
+        "word_perplexity": 1485111.1306447538
+      }
+    },
+    "gsarti/flores_101_mal+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.1588237245178132,
+        "byte_perplexity": 4.465506197375413,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mal",
+        "word_perplexity": 4.8990954217696134e+17
+      }
+    },
+    "gsarti/flores_101_mar+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.454064685835334,
+        "byte_perplexity": 5.479577601103449,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mar",
+        "word_perplexity": 53348101396468.1
+      }
+    },
+    "gsarti/flores_101_mkd+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.6388651004482695,
+        "byte_perplexity": 3.11420755589491,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mkd",
+        "word_perplexity": 513306.31562258815
+      }
+    },
+    "gsarti/flores_101_mlt+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 4.014730236310589,
+        "byte_perplexity": 16.164200382975334,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mlt",
+        "word_perplexity": 3271065298.9525104
+      }
+    },
+    "gsarti/flores_101_mon+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.8368760183021453,
+        "byte_perplexity": 3.5723563966116956,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mon",
+        "word_perplexity": 11967156.496346941
+      }
+    },
+    "gsarti/flores_101_mri+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.037967287223778,
+        "byte_perplexity": 8.213330128288407,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mri",
+        "word_perplexity": 42667.84366725716
+      }
+    },
+    "gsarti/flores_101_msa+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.4109363519680242,
+        "byte_perplexity": 2.659096901190639,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_msa",
+        "word_perplexity": 1188.7251531670374
+      }
+    },
+    "gsarti/flores_101_mya+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.4101030557435918,
+        "byte_perplexity": 2.657561458464019,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_mya",
+        "word_perplexity": 5.887577237013639e+18
+      }
+    },
+    "gsarti/flores_101_nld+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.1813098607926804,
+        "byte_perplexity": 4.535651709856251,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_nld",
+        "word_perplexity": 13951.877058430618
+      }
+    },
+    "gsarti/flores_101_nob+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.561165630453858,
+        "byte_perplexity": 5.901843358131797,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_nob",
+        "word_perplexity": 64134.3587194621
+      }
+    },
+    "gsarti/flores_101_npi+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.361386302448311,
+        "byte_perplexity": 5.138638996619111,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_npi",
+        "word_perplexity": 7452421298650.788
+      }
+    },
+    "gsarti/flores_101_nso+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.150046187635368,
+        "byte_perplexity": 8.876839962509171,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_nso",
+        "word_perplexity": 133251.3907730927
+      }
+    },
+    "gsarti/flores_101_nya+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.166160871838487,
+        "byte_perplexity": 8.97654874419086,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_nya",
+        "word_perplexity": 13237249.320560299
+      }
+    },
+    "gsarti/flores_101_oci+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.3544826611123932,
+        "byte_perplexity": 5.114108118049416,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_oci",
+        "word_perplexity": 29786.57326210068
+      }
+    },
+    "gsarti/flores_101_orm+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.7457001993717243,
+        "byte_perplexity": 13.414303089263644,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_orm",
+        "word_perplexity": 1286222337.8393624
+      }
+    },
+    "gsarti/flores_101_ory+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.3466784891528936,
+        "byte_perplexity": 5.086518347981296,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ory",
+        "word_perplexity": 8232620282886.167
+      }
+    },
+    "gsarti/flores_101_pan+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.3255600077385723,
+        "byte_perplexity": 5.012603107956229,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_pan",
+        "word_perplexity": 2003582065.835696
+      }
+    },
+    "gsarti/flores_101_pol+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.3688414865658434,
+        "byte_perplexity": 5.165261846492578,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_pol",
+        "word_perplexity": 239703.75452947227
+      }
+    },
+    "gsarti/flores_101_por+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.0087385096181816,
+        "byte_perplexity": 2.012150908931838,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_por",
+        "word_perplexity": 78.66129921108659
+      }
+    },
+    "gsarti/flores_101_pus+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.2173729850313615,
+        "byte_perplexity": 4.650458574106675,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_pus",
+        "word_perplexity": 200303.57214724104
+      }
+    },
+    "gsarti/flores_101_ron+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.486356022105963,
+        "byte_perplexity": 5.603607947317877,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ron",
+        "word_perplexity": 80490.92705368399
+      }
+    },
+    "gsarti/flores_101_rus+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.095728414417906,
+        "byte_perplexity": 2.1372096174466697,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_rus",
+        "word_perplexity": 22038.65288574451
+      }
+    },
+    "gsarti/flores_101_slk+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.8667803584469502,
+        "byte_perplexity": 7.294354718439043,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_slk",
+        "word_perplexity": 1873211.2703176092
+      }
+    },
+    "gsarti/flores_101_slv+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.894935550489075,
+        "byte_perplexity": 7.438107250941839,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_slv",
+        "word_perplexity": 609965.8362492598
+      }
+    },
+    "gsarti/flores_101_sna+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.226698783453375,
+        "byte_perplexity": 9.361234419948593,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_sna",
+        "word_perplexity": 151658287.08006003
+      }
+    },
+    "gsarti/flores_101_snd+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.505484320885354,
+        "byte_perplexity": 5.678399375652783,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_snd",
+        "word_perplexity": 2195879.0537875695
+      }
+    },
+    "gsarti/flores_101_som+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.6579492747174616,
+        "byte_perplexity": 12.622705630414286,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_som",
+        "word_perplexity": 12921970.127169678
+      }
+    },
+    "gsarti/flores_101_spa+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 0.9441289779054047,
+        "byte_perplexity": 1.9240269109386998,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_spa",
+        "word_perplexity": 55.14408503293887
+      }
+    },
+    "gsarti/flores_101_srp+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.6091583939601046,
+        "byte_perplexity": 3.050738229673983,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_srp",
+        "word_perplexity": 359037.4163692842
+      }
+    },
+    "gsarti/flores_101_swe+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.4943222333483153,
+        "byte_perplexity": 5.634635291846611,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_swe",
+        "word_perplexity": 104567.9891705103
+      }
+    },
+    "gsarti/flores_101_swh+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.9721156771582438,
+        "byte_perplexity": 3.923430589092355,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_swh",
+        "word_perplexity": 6985.646204087442
+      }
+    },
+    "gsarti/flores_101_tam+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.0999329236632325,
+        "byte_perplexity": 4.286894531607389,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_tam",
+        "word_perplexity": 4220234444737767.0
+      }
+    },
+    "gsarti/flores_101_tel+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.549014618212334,
+        "byte_perplexity": 5.852344181819556,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_tel",
+        "word_perplexity": 7315913985648022.0
+      }
+    },
+    "gsarti/flores_101_tgk+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.9399053923480125,
+        "byte_perplexity": 3.836804862794101,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_tgk",
+        "word_perplexity": 10003619.893239152
+      }
+    },
+    "gsarti/flores_101_tgl+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 2.645461413001105,
+        "byte_perplexity": 6.256957969905079,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_tgl",
+        "word_perplexity": 87554.31770184237
+      }
+    },
+    "gsarti/flores_101_tha+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.2979178211163922,
+        "byte_perplexity": 2.458737675753546,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_tha",
+        "word_perplexity": 6.85384626099906e+32
+      }
+    },
+    "gsarti/flores_101_tur+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.4123830232149,
+        "byte_perplexity": 5.323529328304652,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_tur",
+        "word_perplexity": 1230000.8194755162
+      }
+    },
+    "gsarti/flores_101_ukr+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.5282644195953918,
+        "byte_perplexity": 2.8843863497020608,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_ukr",
+        "word_perplexity": 780615.9486315987
+      }
+    },
+    "gsarti/flores_101_umb+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.710219475046473,
+        "byte_perplexity": 13.088423907901921,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_umb",
+        "word_perplexity": 346118506.64866126
+      }
+    },
+    "gsarti/flores_101_urd+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 1.0075988539165108,
+        "byte_perplexity": 2.010562039704537,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_urd",
+        "word_perplexity": 335.1943886252716
+      }
+    },
+    "gsarti/flores_101_uzb+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.69831120498359,
+        "byte_perplexity": 12.980834294137205,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_uzb",
+        "word_perplexity": 1248263505.2751954
+      }
+    },
+    "gsarti/flores_101_vie+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 0.8461114961807352,
+        "byte_perplexity": 1.7976491760484148,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_vie",
+        "word_perplexity": 33.51752264232948
+      }
+    },
+    "gsarti/flores_101_wol+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.332383415073327,
+        "byte_perplexity": 10.072733993132132,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_wol",
+        "word_perplexity": 199684.7010180392
+      }
+    },
+    "gsarti/flores_101_xho+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.0428982143908727,
+        "byte_perplexity": 8.241450154294917,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_xho",
+        "word_perplexity": 141017733.33017766
+      }
+    },
+    "gsarti/flores_101_yor+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 2.62429549091613,
+        "byte_perplexity": 6.165831615133067,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_yor",
+        "word_perplexity": 171980.641422536
+      }
+    },
+    "gsarti/flores_101_zho_simpl+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.2156521449449949,
+        "byte_perplexity": 2.322457417595381,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_zho_simpl",
+        "word_perplexity": 1.0554528210220222e+21
+      }
+    },
+    "gsarti/flores_101_zho_trad+null": {
+      "2022-07-14-10-03-25": {
+        "bits_per_byte": 1.3622834584784203,
+        "byte_perplexity": 2.5709177552415134,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_zho_trad",
+        "word_perplexity": 4.787781515987923e+24
+      }
+    },
+    "gsarti/flores_101_zul+null": {
+      "2022-07-14-12-00-55": {
+        "bits_per_byte": 3.2020451216662975,
+        "byte_perplexity": 9.202622963132773,
+        "prompt_name": "null",
+        "task_name": "gsarti/flores_101_zul",
+        "word_perplexity": 998742068.9481835
+      }
+    },
+    "headqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.25419401896425964,
+        "acc_norm": 0.29576951130561635,
+        "acc_norm_stderr": 0.008717251898361422,
+        "acc_stderr": 0.008316509290190668
+      }
+    },
+    "hellaswag": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.37621987651862177,
+        "acc_norm": 0.46564429396534557,
+        "acc_norm_stderr": 0.004977988452502641,
+        "acc_stderr": 0.004834461997944872
+      }
+    },
+    "lambada": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.46322530564719583,
+        "acc_stderr": 0.006947110835634445,
+        "ppl": 12.583447597222621,
+        "ppl_stderr": 0.4021518609838198
+      }
+    },
+    "logiqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.21658986175115208,
+        "acc_norm": 0.28110599078341014,
+        "acc_norm_stderr": 0.017632374626460005,
+        "acc_stderr": 0.016156860583178303
+      }
+    },
+    "mathqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.2489112227805695,
+        "acc_norm": 0.2422110552763819,
+        "acc_norm_stderr": 0.007842810183504986,
+        "acc_stderr": 0.007915319798861361
+      }
+    },
+    "mc_taco": {
+      "2022-07-13-11-29-13": {
+        "em": 0.12537537537537538,
+        "f1": 0.4747075325110886
+      }
+    },
+    "mnli+GPT-3 style": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3564951604686704,
+        "acc_norm": 0.335303107488538,
+        "acc_norm_stderr": 0.004765490263584639,
+        "acc_stderr": 0.004834813222301984,
+        "prompt_name": "GPT-3 style",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+MNLI crowdsource": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3548650025471218,
+        "acc_norm": 0.37982679572083544,
+        "acc_norm_stderr": 0.004899212442097964,
+        "acc_stderr": 0.004829852406948984,
+        "prompt_name": "MNLI crowdsource",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+always/sometimes/never": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.31920529801324504,
+        "acc_norm": 0.31818644931227713,
+        "acc_norm_stderr": 0.004701653585969693,
+        "acc_stderr": 0.004705655206722177,
+        "prompt_name": "always/sometimes/never",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+based on the previous passage": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.34070300560366784,
+        "acc_norm": 0.33245033112582784,
+        "acc_norm_stderr": 0.004755346314564714,
+        "acc_stderr": 0.004784157883834768,
+        "prompt_name": "based on the previous passage",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+can we infer": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.36271013754457465,
+        "acc_norm": 0.3392766174223128,
+        "acc_norm_stderr": 0.004779294320017342,
+        "acc_stderr": 0.004853167998709484,
+        "prompt_name": "can we infer",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+claim true/false/inconclusive": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.35384615384615387,
+        "acc_norm": 0.3169638308711156,
+        "acc_norm_stderr": 0.004696817414398099,
+        "acc_stderr": 0.004826720820135633,
+        "prompt_name": "claim true/false/inconclusive",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+consider always/sometimes/never": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3183902190524707,
+        "acc_norm": 0.31818644931227713,
+        "acc_norm_stderr": 0.004701653585969693,
+        "acc_stderr": 0.004702455981984395,
+        "prompt_name": "consider always/sometimes/never",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+does it follow that": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3784004075394804,
+        "acc_norm": 0.3499745287824758,
+        "acc_norm_stderr": 0.004814601860231488,
+        "acc_stderr": 0.00489562485968904,
+        "prompt_name": "does it follow that",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+does this imply": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.33224656138563424,
+        "acc_norm": 0.31920529801324504,
+        "acc_norm_stderr": 0.004705655206722178,
+        "acc_stderr": 0.004754614244749308,
+        "prompt_name": "does this imply",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+guaranteed true": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.35731023942944473,
+        "acc_norm": 0.3398879266428935,
+        "acc_norm_stderr": 0.004781384619510542,
+        "acc_stderr": 0.004837270730680468,
+        "prompt_name": "guaranteed true",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+guaranteed/possible/impossible": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.32317880794701986,
+        "acc_norm": 0.3390728476821192,
+        "acc_norm_stderr": 0.004778595579555236,
+        "acc_stderr": 0.004721015048648592,
+        "prompt_name": "guaranteed/possible/impossible",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+justified in saying": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3611818644931228,
+        "acc_norm": 0.33438614365766683,
+        "acc_norm_stderr": 0.004762251055102503,
+        "acc_stderr": 0.0048487362318538965,
+        "prompt_name": "justified in saying",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+must be true": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3532348446255731,
+        "acc_norm": 0.3400916963830871,
+        "acc_norm_stderr": 0.004782079413482068,
+        "acc_stderr": 0.004824830369595005,
+        "prompt_name": "must be true",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+should assume": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3532348446255731,
+        "acc_norm": 0.32236372898624555,
+        "acc_norm_stderr": 0.004717896188851781,
+        "acc_stderr": 0.004824830369595005,
+        "prompt_name": "should assume",
+        "task_name": "mnli"
+      }
+    },
+    "mnli+take the following as truth": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3540499235863474,
+        "acc_norm": 0.32654100866021396,
+        "acc_norm_stderr": 0.004733707466562015,
+        "acc_stderr": 0.004827349052909375,
+        "prompt_name": "take the following as truth",
+        "task_name": "mnli"
+      }
+    },
+    "mnli_mismatched+GPT-3 style": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3558787632221318,
+        "acc_norm": 0.3365541090317331,
+        "acc_norm_stderr": 0.0047657510794410825,
+        "acc_stderr": 0.004828764189286043,
+        "prompt_name": "GPT-3 style",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+MNLI crowdsource": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3524206672091131,
+        "acc_norm": 0.3876118795768918,
+        "acc_norm_stderr": 0.004913750149712027,
+        "acc_stderr": 0.004818127922877737,
+        "prompt_name": "MNLI crowdsource",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+always/sometimes/never": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3187550854353133,
+        "acc_norm": 0.318246541903987,
+        "acc_norm_stderr": 0.004697823254367764,
+        "acc_stderr": 0.004699821349212815,
+        "prompt_name": "always/sometimes/never",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+based on the previous passage": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3442839707078926,
+        "acc_norm": 0.3240439381611066,
+        "acc_norm_stderr": 0.00472022103875238,
+        "acc_stderr": 0.004792007109263922,
+        "prompt_name": "based on the previous passage",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+can we infer": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3628966639544345,
+        "acc_norm": 0.33909682668836455,
+        "acc_norm_stderr": 0.0047745443668395,
+        "acc_stderr": 0.004849506876045877,
+        "prompt_name": "can we infer",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+claim true/false/inconclusive": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3517087062652563,
+        "acc_norm": 0.31550040683482505,
+        "acc_norm_stderr": 0.004686921836958016,
+        "acc_stderr": 0.004815903833418159,
+        "prompt_name": "claim true/false/inconclusive",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+consider always/sometimes/never": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.318246541903987,
+        "acc_norm": 0.318246541903987,
+        "acc_norm_stderr": 0.004697823254367764,
+        "acc_stderr": 0.004697823254367764,
+        "prompt_name": "consider always/sometimes/never",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+does it follow that": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.38923921887713586,
+        "acc_norm": 0.34926769731489016,
+        "acc_norm_stderr": 0.004808189163919754,
+        "acc_stderr": 0.004917507365149974,
+        "prompt_name": "does it follow that",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+does this imply": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3233319772172498,
+        "acc_norm": 0.3184499593165175,
+        "acc_norm_stderr": 0.0046986232661144,
+        "acc_stderr": 0.0047175151956513625,
+        "prompt_name": "does this imply",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+guaranteed true": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.36208299430431246,
+        "acc_norm": 0.3303498779495525,
+        "acc_norm_stderr": 0.004743645253038162,
+        "acc_stderr": 0.00484715944530685,
+        "prompt_name": "guaranteed true",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+guaranteed/possible/impossible": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.32048413344182264,
+        "acc_norm": 0.33848657445077296,
+        "acc_norm_stderr": 0.004772448023078353,
+        "acc_stderr": 0.004706566719294992,
+        "prompt_name": "guaranteed/possible/impossible",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+justified in saying": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.371033360455655,
+        "acc_norm": 0.32648494711147275,
+        "acc_norm_stderr": 0.004729403696523803,
+        "acc_stderr": 0.004872158826748743,
+        "prompt_name": "justified in saying",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+must be true": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3565907241659886,
+        "acc_norm": 0.3373677786818552,
+        "acc_norm_stderr": 0.004768581700693004,
+        "acc_stderr": 0.004830919845456573,
+        "prompt_name": "must be true",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+should assume": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.35740439381611067,
+        "acc_norm": 0.32231489015459724,
+        "acc_norm_stderr": 0.0047136280360736155,
+        "acc_stderr": 0.0048333692129862065,
+        "prompt_name": "should assume",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mnli_mismatched+take the following as truth": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3522172497965826,
+        "acc_norm": 0.3263832384052075,
+        "acc_norm_stderr": 0.004729024000627127,
+        "acc_stderr": 0.004817493665633715,
+        "prompt_name": "take the following as truth",
+        "task_name": "mnli_mismatched"
+      }
+    },
+    "mrpc": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.6813725490196079,
+        "acc_stderr": 0.023095996571841474,
+        "f1": 0.8104956268221574,
+        "f1_stderr": 0.016329211455484924
+      }
+    },
+    "multirc": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.011542497376705142,
+        "acc_stderr": 0.003461867320927179
+      }
+    },
+    "multirc+I was going to say\u2026": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5082508250825083,
+        "acc_norm": 0.4280115511551155,
+        "acc_norm_stderr": 0.007106976252751536,
+        "acc_stderr": 0.007180825220905937,
+        "prompt_name": "I was going to say\u2026",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+Would it be good to answer\u2026": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.45173267326732675,
+        "acc_norm": 0.4278052805280528,
+        "acc_norm_stderr": 0.007106544557507229,
+        "acc_stderr": 0.007148261386088041,
+        "prompt_name": "Would it be good to answer\u2026",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+confirm": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4280115511551155,
+        "acc_norm": 0.4280115511551155,
+        "acc_norm_stderr": 0.007106976252751536,
+        "acc_stderr": 0.007106976252751536,
+        "prompt_name": "confirm",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+correct": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5532178217821783,
+        "acc_norm": 0.4643151815181518,
+        "acc_norm_stderr": 0.00716348904876326,
+        "acc_stderr": 0.007141007544074806,
+        "prompt_name": "correct",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+decide_valid": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5107260726072608,
+        "acc_norm": 0.4280115511551155,
+        "acc_norm_stderr": 0.007106976252751536,
+        "acc_stderr": 0.007180150402551771,
+        "prompt_name": "decide_valid",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+found_this_answer": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4278052805280528,
+        "acc_norm": 0.4280115511551155,
+        "acc_norm_stderr": 0.007106976252751536,
+        "acc_stderr": 0.007106544557507229,
+        "prompt_name": "found_this_answer",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+grading": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.429042904290429,
+        "acc_norm": 0.4280115511551155,
+        "acc_norm_stderr": 0.007106976252751536,
+        "acc_stderr": 0.007109115814226985,
+        "prompt_name": "grading",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+is the correct answer\u2026": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4498762376237624,
+        "acc_norm": 0.4273927392739274,
+        "acc_norm_stderr": 0.007105677382236137,
+        "acc_stderr": 0.0071456249799065185,
+        "prompt_name": "is the correct answer\u2026",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+is\u2026 a correct answer?": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.4278052805280528,
+        "acc_norm": 0.4280115511551155,
+        "acc_norm_stderr": 0.007106976252751536,
+        "acc_stderr": 0.007106544557507229,
+        "prompt_name": "is\u2026 a correct answer?",
+        "task_name": "multirc"
+      }
+    },
+    "multirc+paragraph\u2026 question\u2026 is it\u2026 ?": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.5030940594059405,
+        "acc_norm": 0.42883663366336633,
+        "acc_norm_stderr": 0.007108690423137722,
+        "acc_stderr": 0.007181665598939583,
+        "prompt_name": "paragraph\u2026 question\u2026 is it\u2026 ?",
+        "task_name": "multirc"
+      }
+    },
+    "openbookqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.214,
+        "acc_norm": 0.298,
+        "acc_norm_stderr": 0.020475118092988978,
+        "acc_stderr": 0.01835979750238702
+      }
+    },
+    "piqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.6871599564744287,
+        "acc_norm": 0.7002176278563657,
+        "acc_norm_stderr": 0.010689686967138092,
+        "acc_stderr": 0.010817714425701112
+      }
+    },
+    "prost": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.23505550811272416,
+        "acc_norm": 0.2670260461144321,
+        "acc_norm_stderr": 0.0032321702981822874,
+        "acc_stderr": 0.0030979423271461875
+      }
+    },
+    "pubmedqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.56,
+        "acc_stderr": 0.015704987954361798
+      }
+    },
+    "qnli": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.4962474830679114,
+        "acc_stderr": 0.006765220016415222
+      }
+    },
+    "qqp": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.3681424684640119,
+        "acc_stderr": 0.0023986729832071816,
+        "f1": 0.5381138352498734,
+        "f1_stderr": 0.002555831569895799
+      }
+    },
+    "qqp+answer": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.40558990848379917,
+        "acc_norm": 0.36816720257234725,
+        "acc_norm_stderr": 0.002398706610614492,
+        "acc_stderr": 0.002441969063495092,
+        "prompt_name": "answer",
+        "task_name": "qqp"
+      }
+    },
+    "qqp+duplicate": {
+      "2022-07-12-22-45-57": {
+        "acc": 0.3788523373732377,
+        "acc_norm": 0.36816720257234725,
+        "acc_norm_stderr": 0.002398706610614492,
+        "acc_stderr": 0.002412603277723025,
+        "prompt_name": "duplicate",
+        "task_name": "qqp"
+      }
+    },
+    "qqp+duplicate or not": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.5761315854563444,
+        "acc_norm": 0.6318327974276527,
+        "acc_norm_stderr": 0.002398706610614492,
+        "acc_stderr": 0.0024577056660753426,
+        "prompt_name": "duplicate or not",
+        "task_name": "qqp"
+      }
+    },
+    "qqp+meaning": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.3681424684640119,
+        "acc_norm": 0.36816720257234725,
+        "acc_norm_stderr": 0.002398706610614492,
+        "acc_stderr": 0.0023986729832071916,
+        "prompt_name": "meaning",
+        "task_name": "qqp"
+      }
+    },
+    "qqp+quora": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.36821667078901804,
+        "acc_norm": 0.36816720257234725,
+        "acc_norm_stderr": 0.002398706610614492,
+        "acc_stderr": 0.0023987738450886556,
+        "prompt_name": "quora",
+        "task_name": "qqp"
+      }
+    },
+    "qqp+same thing": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.5099431115508286,
+        "acc_norm": 0.36816720257234725,
+        "acc_norm_stderr": 0.002398706610614492,
+        "acc_stderr": 0.002486208885430481,
+        "prompt_name": "same thing",
+        "task_name": "qqp"
+      }
+    },
+    "race": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.3320574162679426,
+        "acc_stderr": 0.014575582129545914
+      }
+    },
+    "rte": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.5342960288808665,
+        "acc_stderr": 0.030025579819366426
+      }
+    },
+    "rte+does the claim\u2026 follow the fact\u2026": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.4729241877256318,
+        "acc_norm": 0.5270758122743683,
+        "acc_norm_stderr": 0.0300523034631437,
+        "acc_stderr": 0.030052303463143706,
+        "prompt_name": "does the claim\u2026 follow the fact\u2026",
+        "task_name": "rte"
+      }
+    },
+    "rte+entailment explained": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.49458483754512633,
+        "acc_norm": 0.4729241877256318,
+        "acc_norm_stderr": 0.0300523034631437,
+        "acc_stderr": 0.030094698123239966,
+        "prompt_name": "entailment explained",
+        "task_name": "rte"
+      }
+    },
+    "rte+imply": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.48375451263537905,
+        "acc_norm": 0.5270758122743683,
+        "acc_norm_stderr": 0.0300523034631437,
+        "acc_stderr": 0.030080573208738064,
+        "prompt_name": "imply",
+        "task_name": "rte"
+      }
+    },
+    "rte+imply separated": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.45126353790613716,
+        "acc_norm": 0.5270758122743683,
+        "acc_norm_stderr": 0.0300523034631437,
+        "acc_stderr": 0.029953149241808943,
+        "prompt_name": "imply separated",
+        "task_name": "rte"
+      }
+    },
+    "rte+mean": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.48014440433212996,
+        "acc_norm": 0.5270758122743683,
+        "acc_norm_stderr": 0.0300523034631437,
+        "acc_stderr": 0.030072723167317194,
+        "prompt_name": "mean",
+        "task_name": "rte"
+      }
+    },
+    "sciq": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.853,
+        "acc_norm": 0.771,
+        "acc_norm_stderr": 0.013294199326613609,
+        "acc_stderr": 0.011203415395160335
+      }
+    },
+    "sst": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.6823394495412844,
+        "acc_stderr": 0.015775124845202545
+      }
+    },
+    "sst+following positive negative": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.8061926605504587,
+        "acc_norm": 0.8061926605504587,
+        "acc_norm_stderr": 0.013393542261521812,
+        "acc_stderr": 0.013393542261521812,
+        "prompt_name": "following positive negative",
+        "task_name": "sst"
+      }
+    },
+    "sst+happy or mad": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.5091743119266054,
+        "acc_norm": 0.5091743119266054,
+        "acc_norm_stderr": 0.01693900152535154,
+        "acc_stderr": 0.01693900152535154,
+        "prompt_name": "happy or mad",
+        "task_name": "sst"
+      }
+    },
+    "sst+positive negative after": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.6204128440366973,
+        "acc_norm": 0.6204128440366973,
+        "acc_norm_stderr": 0.016443227556688766,
+        "acc_stderr": 0.016443227556688766,
+        "prompt_name": "positive negative after",
+        "task_name": "sst"
+      }
+    },
+    "sst+review": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.5091743119266054,
+        "acc_norm": 0.5091743119266054,
+        "acc_norm_stderr": 0.01693900152535154,
+        "acc_stderr": 0.01693900152535154,
+        "prompt_name": "review",
+        "task_name": "sst"
+      }
+    },
+    "sst+said": {
+      "2022-07-13-19-23-37": {
+        "acc": 0.4908256880733945,
+        "acc_norm": 0.5091743119266054,
+        "acc_norm_stderr": 0.01693900152535154,
+        "acc_stderr": 0.01693900152535154,
+        "prompt_name": "said",
+        "task_name": "sst"
+      }
+    },
+    "triviaqa": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.0313798285158667,
+        "acc_stderr": 0.0016392014864795154
+      }
+    },
+    "webqs": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.012795275590551181,
+        "acc_stderr": 0.0024938680596856277
+      }
+    },
+    "wic": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.5,
+        "acc_stderr": 0.01981072129375818
+      }
+    },
+    "wic+GPT-3-prompt": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.01981072129375818,
+        "acc_stderr": 0.01981072129375818,
+        "prompt_name": "GPT-3-prompt",
+        "task_name": "wic"
+      }
+    },
+    "wic+GPT-3-prompt-with-label": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.49216300940438873,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.01981072129375818,
+        "acc_stderr": 0.019808287657813832,
+        "prompt_name": "GPT-3-prompt-with-label",
+        "task_name": "wic"
+      }
+    },
+    "wic+affirmation_true_or_false": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5,
+        "acc_norm": 0.5078369905956113,
+        "acc_norm_stderr": 0.019808287657813832,
+        "acc_stderr": 0.01981072129375818,
+        "prompt_name": "affirmation_true_or_false",
+        "task_name": "wic"
+      }
+    },
+    "wic+grammar_homework": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5094043887147336,
+        "acc_norm": 0.49843260188087773,
+        "acc_norm_stderr": 0.019810623954060382,
+        "acc_stderr": 0.019807216763271497,
+        "prompt_name": "grammar_homework",
+        "task_name": "wic"
+      }
+    },
+    "wic+polysemous": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.512539184952978,
+        "acc_norm": 0.49843260188087773,
+        "acc_norm_stderr": 0.019810623954060382,
+        "acc_stderr": 0.019804490588592596,
+        "prompt_name": "polysemous",
+        "task_name": "wic"
+      }
+    },
+    "wic+question-context": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5266457680250783,
+        "acc_norm": 0.5031347962382445,
+        "acc_norm_stderr": 0.019810331932097542,
+        "acc_stderr": 0.019782570188812167,
+        "prompt_name": "question-context",
+        "task_name": "wic"
+      }
+    },
+    "wic+question-context-meaning": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5438871473354232,
+        "acc_norm": 0.5015673981191222,
+        "acc_norm_stderr": 0.019810623954060382,
+        "acc_stderr": 0.019734259601993404,
+        "prompt_name": "question-context-meaning",
+        "task_name": "wic"
+      }
+    },
+    "wic+question-context-meaning-with-label": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5156739811912225,
+        "acc_norm": 0.5015673981191222,
+        "acc_norm_stderr": 0.019810623954060382,
+        "acc_stderr": 0.019800984955347847,
+        "prompt_name": "question-context-meaning-with-label",
+        "task_name": "wic"
+      }
+    },
+    "wic+same_sense": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5047021943573667,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.01981072129375818,
+        "acc_stderr": 0.019809845219259763,
+        "prompt_name": "same_sense",
+        "task_name": "wic"
+      }
+    },
+    "wic+similar-sense": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.542319749216301,
+        "acc_norm": 0.5,
+        "acc_norm_stderr": 0.01981072129375818,
+        "acc_stderr": 0.01973963328373276,
+        "prompt_name": "similar-sense",
+        "task_name": "wic"
+      }
+    },
+    "winogrande": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.5730071033938438,
+        "acc_stderr": 0.013901878072575058
+      }
+    },
+    "wnli": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.43661971830985913,
+        "acc_stderr": 0.0592793555841297
+      }
+    },
+    "wnli+confident": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.43661971830985913,
+        "acc_norm": 0.43661971830985913,
+        "acc_norm_stderr": 0.0592793555841297,
+        "acc_stderr": 0.0592793555841297,
+        "prompt_name": "confident",
+        "task_name": "wnli"
+      }
+    },
+    "wnli+entailment explained": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.39436619718309857,
+        "acc_norm": 0.43661971830985913,
+        "acc_norm_stderr": 0.0592793555841297,
+        "acc_stderr": 0.058412510854444266,
+        "prompt_name": "entailment explained",
+        "task_name": "wnli"
+      }
+    },
+    "wnli+imply": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.4225352112676056,
+        "acc_norm": 0.43661971830985913,
+        "acc_norm_stderr": 0.0592793555841297,
+        "acc_stderr": 0.05903984205682581,
+        "prompt_name": "imply",
+        "task_name": "wnli"
+      }
+    },
+    "wnli+justified": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.43661971830985913,
+        "acc_norm": 0.43661971830985913,
+        "acc_norm_stderr": 0.0592793555841297,
+        "acc_stderr": 0.0592793555841297,
+        "prompt_name": "justified",
+        "task_name": "wnli"
+      }
+    },
+    "wnli+mean": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.6619718309859155,
+        "acc_norm": 0.43661971830985913,
+        "acc_norm_stderr": 0.0592793555841297,
+        "acc_stderr": 0.05653887739133513,
+        "prompt_name": "mean",
+        "task_name": "wnli"
+      }
+    },
+    "wsc": {
+      "2022-07-13-11-29-13": {
+        "acc": 0.36538461538461536,
+        "acc_stderr": 0.0474473339327792
+      }
+    },
+    "wsc+GPT-3 Style": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.36538461538461536,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.0474473339327792,
+        "prompt_name": "GPT-3 Style",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+I think they mean": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.36538461538461536,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.0474473339327792,
+        "prompt_name": "I think they mean",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+Who or what is/are": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.40384615384615385,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.048346889526540184,
+        "prompt_name": "Who or what is/are",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+by p they mean": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.36538461538461536,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.0474473339327792,
+        "prompt_name": "by p they mean",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+does p stand for": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.375,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.04770204856076104,
+        "prompt_name": "does p stand for",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+does the pronoun refer to": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.5480769230769231,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.049038186969314335,
+        "prompt_name": "does the pronoun refer to",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+in other words": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.36538461538461536,
+        "acc_norm": 0.5288461538461539,
+        "acc_norm_stderr": 0.04918440626354964,
+        "acc_stderr": 0.0474473339327792,
+        "prompt_name": "in other words",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+p is/are r": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.36538461538461536,
+        "acc_norm": 0.34615384615384615,
+        "acc_norm_stderr": 0.04687634642174987,
+        "acc_stderr": 0.0474473339327792,
+        "prompt_name": "p is/are r",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+replaced with": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.6153846153846154,
+        "acc_norm": 0.36538461538461536,
+        "acc_norm_stderr": 0.0474473339327792,
+        "acc_stderr": 0.047936688680750406,
+        "prompt_name": "replaced with",
+        "task_name": "wsc"
+      }
+    },
+    "wsc+the pronoun refers to": {
+      "2022-07-14-10-03-25": {
+        "acc": 0.36538461538461536,
+        "acc_norm": 0.5865384615384616,
+        "acc_norm_stderr": 0.04852294969729053,
+        "acc_stderr": 0.0474473339327792,
+        "prompt_name": "the pronoun refers to",
+        "task_name": "wsc"
+      }
+    }
+  },
+  "versions": {
+    "arc_challenge": 0,
+    "arc_easy": 0,
+    "axb+GPT-3 style": 0,
+    "axb+MNLI crowdsource": 0,
+    "axb+based on the previous passage": 0,
+    "axb+can we infer": 0,
+    "axb+does it follow that": 0,
+    "axb+does this imply": 0,
+    "axb+guaranteed true": 0,
+    "axb+justified in saying": 0,
+    "axb+must be true": 0,
+    "axb+should assume": 0,
+    "axg+GPT-3 style": 0,
+    "axg+MNLI crowdsource": 0,
+    "axg+based on the previous passage": 0,
+    "axg+can we infer": 0,
+    "axg+does it follow that": 0,
+    "axg+does this imply": 0,
+    "axg+guaranteed true": 0,
+    "axg+justified in saying": 0,
+    "axg+must be true": 0,
+    "axg+should assume": 0,
+    "boolq": 1,
+    "boolq+GPT-3 Style": 0,
+    "boolq+I wonder\u2026": 0,
+    "boolq+after_reading": 0,
+    "boolq+based on the following passage": 0,
+    "boolq+based on the previous passage": 0,
+    "boolq+could you tell me\u2026": 0,
+    "boolq+exam": 0,
+    "boolq+exercise": 0,
+    "boolq+valid_binary": 0,
+    "boolq+yes_no_question": 0,
+    "cb+GPT-3 style": 0,
+    "cb+MNLI crowdsource": 0,
+    "cb+always/sometimes/never": 0,
+    "cb+based on the previous passage": 0,
+    "cb+can we infer": 0,
+    "cb+claim true/false/inconclusive": 0,
+    "cb+consider always/sometimes/never": 0,
+    "cb+does it follow that": 0,
+    "cb+does this imply": 0,
+    "cb+guaranteed true": 0,
+    "cb+guaranteed/possible/impossible": 0,
+    "cb+justified in saying": 0,
+    "cb+must be true": 0,
+    "cb+should assume": 0,
+    "cb+take the following as truth": 0,
+    "cola+Following sentence acceptable": 0,
+    "cola+Make sense yes no": 0,
+    "cola+Previous sentence acceptable": 0,
+    "cola+editing": 0,
+    "cola+is_this_correct": 0,
+    "copa": 0,
+    "copa+C1 or C2? premise, so/because\u2026": 0,
+    "copa+best_option": 0,
+    "copa+cause_effect": 0,
+    "copa+choose": 0,
+    "copa+exercise": 0,
+    "copa+i_am_hesitating": 0,
+    "copa+more likely": 0,
+    "copa+plausible_alternatives": 0,
+    "crows_pairs_english+1": 0,
+    "crows_pairs_english+2": 0,
+    "crows_pairs_english+3": 0,
+    "crows_pairs_english+4": 0,
+    "crows_pairs_english+A_preference": 0,
+    "crows_pairs_english+A_reality_check": 0,
+    "crows_pairs_english+A_stereotype_true": 0,
+    "crows_pairs_french+1_fr": 0,
+    "crows_pairs_french+2_fr": 0,
+    "crows_pairs_french+3_fr": 0,
+    "crows_pairs_french+4_fr": 0,
+    "crows_pairs_french+A_preference_fr": 0,
+    "crows_pairs_french+A_reality_check_fr": 0,
+    "crows_pairs_french+A_stereotype_true_fr": 0,
+    "diabla+Is the error present? (same lang)": 0,
+    "diabla+Which is automatic?": 0,
+    "gsarti/flores_101_afr+null": 0,
+    "gsarti/flores_101_amh+null": 0,
+    "gsarti/flores_101_ara+null": 0,
+    "gsarti/flores_101_asm+null": 0,
+    "gsarti/flores_101_ast+null": 0,
+    "gsarti/flores_101_azj+null": 0,
+    "gsarti/flores_101_bel+null": 0,
+    "gsarti/flores_101_ben+null": 0,
+    "gsarti/flores_101_bos+null": 0,
+    "gsarti/flores_101_bul+null": 0,
+    "gsarti/flores_101_cat+null": 0,
+    "gsarti/flores_101_ceb+null": 0,
+    "gsarti/flores_101_ces+null": 0,
+    "gsarti/flores_101_ckb+null": 0,
+    "gsarti/flores_101_cym+null": 0,
+    "gsarti/flores_101_dan+null": 0,
+    "gsarti/flores_101_deu+null": 0,
+    "gsarti/flores_101_ell+null": 0,
+    "gsarti/flores_101_eng+null": 0,
+    "gsarti/flores_101_est+null": 0,
+    "gsarti/flores_101_fas+null": 0,
+    "gsarti/flores_101_fin+null": 0,
+    "gsarti/flores_101_fra+null": 0,
+    "gsarti/flores_101_ful+null": 0,
+    "gsarti/flores_101_gle+null": 0,
+    "gsarti/flores_101_glg+null": 0,
+    "gsarti/flores_101_guj+null": 0,
+    "gsarti/flores_101_hau+null": 0,
+    "gsarti/flores_101_heb+null": 0,
+    "gsarti/flores_101_hin+null": 0,
+    "gsarti/flores_101_hrv+null": 0,
+    "gsarti/flores_101_hun+null": 0,
+    "gsarti/flores_101_hye+null": 0,
+    "gsarti/flores_101_ibo+null": 0,
+    "gsarti/flores_101_ind+null": 0,
+    "gsarti/flores_101_isl+null": 0,
+    "gsarti/flores_101_ita+null": 0,
+    "gsarti/flores_101_jav+null": 0,
+    "gsarti/flores_101_jpn+null": 0,
+    "gsarti/flores_101_kam+null": 0,
+    "gsarti/flores_101_kan+null": 0,
+    "gsarti/flores_101_kat+null": 0,
+    "gsarti/flores_101_kaz+null": 0,
+    "gsarti/flores_101_kea+null": 0,
+    "gsarti/flores_101_kir+null": 0,
+    "gsarti/flores_101_kor+null": 0,
+    "gsarti/flores_101_lao+null": 0,
+    "gsarti/flores_101_lav+null": 0,
+    "gsarti/flores_101_lin+null": 0,
+    "gsarti/flores_101_lit+null": 0,
+    "gsarti/flores_101_ltz+null": 0,
+    "gsarti/flores_101_lug+null": 0,
+    "gsarti/flores_101_luo+null": 0,
+    "gsarti/flores_101_mal+null": 0,
+    "gsarti/flores_101_mar+null": 0,
+    "gsarti/flores_101_mkd+null": 0,
+    "gsarti/flores_101_mlt+null": 0,
+    "gsarti/flores_101_mon+null": 0,
+    "gsarti/flores_101_mri+null": 0,
+    "gsarti/flores_101_msa+null": 0,
+    "gsarti/flores_101_mya+null": 0,
+    "gsarti/flores_101_nld+null": 0,
+    "gsarti/flores_101_nob+null": 0,
+    "gsarti/flores_101_npi+null": 0,
+    "gsarti/flores_101_nso+null": 0,
+    "gsarti/flores_101_nya+null": 0,
+    "gsarti/flores_101_oci+null": 0,
+    "gsarti/flores_101_orm+null": 0,
+    "gsarti/flores_101_ory+null": 0,
+    "gsarti/flores_101_pan+null": 0,
+    "gsarti/flores_101_pol+null": 0,
+    "gsarti/flores_101_por+null": 0,
+    "gsarti/flores_101_pus+null": 0,
+    "gsarti/flores_101_ron+null": 0,
+    "gsarti/flores_101_rus+null": 0,
+    "gsarti/flores_101_slk+null": 0,
+    "gsarti/flores_101_slv+null": 0,
+    "gsarti/flores_101_sna+null": 0,
+    "gsarti/flores_101_snd+null": 0,
+    "gsarti/flores_101_som+null": 0,
+    "gsarti/flores_101_spa+null": 0,
+    "gsarti/flores_101_srp+null": 0,
+    "gsarti/flores_101_swe+null": 0,
+    "gsarti/flores_101_swh+null": 0,
+    "gsarti/flores_101_tam+null": 0,
+    "gsarti/flores_101_tel+null": 0,
+    "gsarti/flores_101_tgk+null": 0,
+    "gsarti/flores_101_tgl+null": 0,
+    "gsarti/flores_101_tha+null": 0,
+    "gsarti/flores_101_tur+null": 0,
+    "gsarti/flores_101_ukr+null": 0,
+    "gsarti/flores_101_umb+null": 0,
+    "gsarti/flores_101_urd+null": 0,
+    "gsarti/flores_101_uzb+null": 0,
+    "gsarti/flores_101_vie+null": 0,
+    "gsarti/flores_101_wol+null": 0,
+    "gsarti/flores_101_xho+null": 0,
+    "gsarti/flores_101_yor+null": 0,
+    "gsarti/flores_101_zho_simpl+null": 0,
+    "gsarti/flores_101_zho_trad+null": 0,
+    "gsarti/flores_101_zul+null": 0,
+    "headqa": 0,
+    "hellaswag": 0,
+    "lambada": 0,
+    "logiqa": 0,
+    "mathqa": 0,
+    "mc_taco": 0,
+    "mnli+GPT-3 style": 0,
+    "mnli+MNLI crowdsource": 0,
+    "mnli+always/sometimes/never": 0,
+    "mnli+based on the previous passage": 0,
+    "mnli+can we infer": 0,
+    "mnli+claim true/false/inconclusive": 0,
+    "mnli+consider always/sometimes/never": 0,
+    "mnli+does it follow that": 0,
+    "mnli+does this imply": 0,
+    "mnli+guaranteed true": 0,
+    "mnli+guaranteed/possible/impossible": 0,
+    "mnli+justified in saying": 0,
+    "mnli+must be true": 0,
+    "mnli+should assume": 0,
+    "mnli+take the following as truth": 0,
+    "mnli_mismatched+GPT-3 style": 0,
+    "mnli_mismatched+MNLI crowdsource": 0,
+    "mnli_mismatched+always/sometimes/never": 0,
+    "mnli_mismatched+based on the previous passage": 0,
+    "mnli_mismatched+can we infer": 0,
+    "mnli_mismatched+claim true/false/inconclusive": 0,
+    "mnli_mismatched+consider always/sometimes/never": 0,
+    "mnli_mismatched+does it follow that": 0,
+    "mnli_mismatched+does this imply": 0,
+    "mnli_mismatched+guaranteed true": 0,
+    "mnli_mismatched+guaranteed/possible/impossible": 0,
+    "mnli_mismatched+justified in saying": 0,
+    "mnli_mismatched+must be true": 0,
+    "mnli_mismatched+should assume": 0,
+    "mnli_mismatched+take the following as truth": 0,
+    "mrpc": 0,
+    "multirc": 1,
+    "multirc+I was going to say\u2026": 0,
+    "multirc+Would it be good to answer\u2026": 0,
+    "multirc+confirm": 0,
+    "multirc+correct": 0,
+    "multirc+decide_valid": 0,
+    "multirc+found_this_answer": 0,
+    "multirc+grading": 0,
+    "multirc+is the correct answer\u2026": 0,
+    "multirc+is\u2026 a correct answer?": 0,
+    "multirc+paragraph\u2026 question\u2026 is it\u2026 ?": 0,
+    "openbookqa": 0,
+    "piqa": 0,
+    "prost": 0,
+    "pubmedqa": 0,
+    "qnli": 0,
+    "qqp": 0,
+    "qqp+answer": 0,
+    "qqp+duplicate": 0,
+    "qqp+duplicate or not": 0,
+    "qqp+meaning": 0,
+    "qqp+quora": 0,
+    "qqp+same thing": 0,
+    "race": 1,
+    "rte": 0,
+    "rte+does the claim\u2026 follow the fact\u2026": 0,
+    "rte+entailment explained": 0,
+    "rte+imply": 0,
+    "rte+imply separated": 0,
+    "rte+mean": 0,
+    "sciq": 0,
+    "sst": 0,
+    "sst+following positive negative": 0,
+    "sst+happy or mad": 0,
+    "sst+positive negative after": 0,
+    "sst+review": 0,
+    "sst+said": 0,
+    "triviaqa": 0,
+    "webqs": 0,
+    "wic": 0,
+    "wic+GPT-3-prompt": 0,
+    "wic+GPT-3-prompt-with-label": 0,
+    "wic+affirmation_true_or_false": 0,
+    "wic+grammar_homework": 0,
+    "wic+polysemous": 0,
+    "wic+question-context": 0,
+    "wic+question-context-meaning": 0,
+    "wic+question-context-meaning-with-label": 0,
+    "wic+same_sense": 0,
+    "wic+similar-sense": 0,
+    "winogrande": 0,
+    "wnli": 1,
+    "wnli+confident": 1,
+    "wnli+entailment explained": 1,
+    "wnli+imply": 1,
+    "wnli+justified": 1,
+    "wnli+mean": 1,
+    "wsc": 0,
+    "wsc+GPT-3 Style": 0,
+    "wsc+I think they mean": 0,
+    "wsc+Who or what is/are": 0,
+    "wsc+by p they mean": 0,
+    "wsc+does p stand for": 0,
+    "wsc+does the pronoun refer to": 0,
+    "wsc+in other words": 0,
+    "wsc+p is/are r": 0,
+    "wsc+replaced with": 0,
+    "wsc+the pronoun refers to": 0
+  }
+}

bigscience/evaluation/results/tr11/bloom1b3/bslmevalfiles/tr11b-1b3-ml-bsevalharness-results_lm-eval_global_step340500_2022-07-13-19-23-37.json

@@ -0,0 +1,701 @@
+{
+    "results": [
+        {
+            "task_name": "qqp",
+            "prompt_name": "answer",
+            "acc": 0.40558990848379917,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "c0182cd1-c7ac-4abe-829f-4651536af951",
+            "prompt_jinja": "Can an answer to \"{{question1}}\" also be used to answer \"{{question2}}\"? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": false,
+            "comment": "",
+            "acc_stderr": 0.002441969063495092
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "answer",
+            "acc_norm": 0.36816720257234725,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "c0182cd1-c7ac-4abe-829f-4651536af951",
+            "prompt_jinja": "Can an answer to \"{{question1}}\" also be used to answer \"{{question2}}\"? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": false,
+            "comment": "",
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "duplicate",
+            "acc": 0.3788523373732377,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "fd244bd3-ca3b-4e4f-9722-fd006c50e157",
+            "prompt_jinja": "I received the questions \"{{question1}}\" and \"{{question2}}\". Are they duplicates? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.002412603277723025
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "duplicate",
+            "acc_norm": 0.36816720257234725,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "fd244bd3-ca3b-4e4f-9722-fd006c50e157",
+            "prompt_jinja": "I received the questions \"{{question1}}\" and \"{{question2}}\". Are they duplicates? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "duplicate or not",
+            "acc": 0.5761315854563444,
+            "fixed_answer_choice_list": [
+                "not duplicates",
+                "duplicates"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "94972071-a726-42a3-a726-13f414b65e67",
+            "prompt_jinja": "{{question1}}\n{{question2}}\nPick one: These questions are \"{{\"duplicates\"}}\" or \"{{\"not duplicates\"}}\".\n|||\n{{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0024577056660753426
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "duplicate or not",
+            "acc_norm": 0.6318327974276527,
+            "fixed_answer_choice_list": [
+                "not duplicates",
+                "duplicates"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "94972071-a726-42a3-a726-13f414b65e67",
+            "prompt_jinja": "{{question1}}\n{{question2}}\nPick one: These questions are \"{{\"duplicates\"}}\" or \"{{\"not duplicates\"}}\".\n|||\n{{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "meaning",
+            "acc": 0.3681424684640119,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "c0724198-97e7-44a1-89d8-c51e97ce0b04",
+            "prompt_jinja": "Question 1: {{question1}}\nQuestion 2: {{question2}}\n\nDo these two questions convey the same meaning? Yes or no? ||| {{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0023986729832071916
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "meaning",
+            "acc_norm": 0.36816720257234725,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "c0724198-97e7-44a1-89d8-c51e97ce0b04",
+            "prompt_jinja": "Question 1: {{question1}}\nQuestion 2: {{question2}}\n\nDo these two questions convey the same meaning? Yes or no? ||| {{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "quora",
+            "acc": 0.36821667078901804,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "8e711799-a57c-4941-833b-466bedfb80ad",
+            "prompt_jinja": "I'm an administrator on the website Quora. There are two posts, one that asks \"{{question1}}\" and another that asks \"{{question2}}\". I can merge questions if they are asking the same thing. Can I merge these two questions? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0023987738450886556
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "quora",
+            "acc_norm": 0.36816720257234725,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "8e711799-a57c-4941-833b-466bedfb80ad",
+            "prompt_jinja": "I'm an administrator on the website Quora. There are two posts, one that asks \"{{question1}}\" and another that asks \"{{question2}}\". I can merge questions if they are asking the same thing. Can I merge these two questions? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "same thing",
+            "acc": 0.5099431115508286,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "a45ad5cd-a3ba-4ab2-a728-a9ea0f27102b",
+            "prompt_jinja": "Are the questions \"{{question1}}\" and \"{{question2}}\" asking the same thing? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.002486208885430481
+        },
+        {
+            "task_name": "qqp",
+            "prompt_name": "same thing",
+            "acc_norm": 0.36816720257234725,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "qqp",
+            "subset": null,
+            "prompt_id": "a45ad5cd-a3ba-4ab2-a728-a9ea0f27102b",
+            "prompt_jinja": "Are the questions \"{{question1}}\" and \"{{question2}}\" asking the same thing? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "does the claim\u2026 follow the fact\u2026",
+            "acc": 0.4729241877256318,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "4ee6ff27-de63-4e7b-a9d4-82a17eba407a",
+            "prompt_jinja": "Does the claim \"{{sentence2}}\" follow from the fact that \"{{sentence1}}\"? Please answer either {{\"yes\"}} or {{\"no\"}}.\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.030052303463143706
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "does the claim\u2026 follow the fact\u2026",
+            "acc_norm": 0.5270758122743683,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "4ee6ff27-de63-4e7b-a9d4-82a17eba407a",
+            "prompt_jinja": "Does the claim \"{{sentence2}}\" follow from the fact that \"{{sentence1}}\"? Please answer either {{\"yes\"}} or {{\"no\"}}.\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "entailment explained",
+            "acc": 0.49458483754512633,
+            "fixed_answer_choice_list": [
+                "entailment",
+                "not entailment"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "9e2b4267-ec23-44c8-b82a-107e2c890fec",
+            "prompt_jinja": "We say that one sentence \"{{\"entails\"}}\" another sentence when the first sentence implies the second sentence. Consider the following two sentences:\n{{sentence1}}\n{{sentence2}}\nIs the relationship from the first to the second sentence \"{{\"entailment\"}}\" or \"{{\"not entailment\"}}\"?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.030094698123239966
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "entailment explained",
+            "acc_norm": 0.4729241877256318,
+            "fixed_answer_choice_list": [
+                "entailment",
+                "not entailment"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "9e2b4267-ec23-44c8-b82a-107e2c890fec",
+            "prompt_jinja": "We say that one sentence \"{{\"entails\"}}\" another sentence when the first sentence implies the second sentence. Consider the following two sentences:\n{{sentence1}}\n{{sentence2}}\nIs the relationship from the first to the second sentence \"{{\"entailment\"}}\" or \"{{\"not entailment\"}}\"?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "imply",
+            "acc": 0.48375451263537905,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "c8dfc879-40f2-412d-be1e-4cd70107f6e6",
+            "prompt_jinja": "Does \"{{sentence1}}\" imply that \"{{sentence2}}\"? Please answer either {{\"yes\"}} or {{\"no\"}}.\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.030080573208738064
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "imply",
+            "acc_norm": 0.5270758122743683,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "c8dfc879-40f2-412d-be1e-4cd70107f6e6",
+            "prompt_jinja": "Does \"{{sentence1}}\" imply that \"{{sentence2}}\"? Please answer either {{\"yes\"}} or {{\"no\"}}.\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "imply separated",
+            "acc": 0.45126353790613716,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "f56ffced-9b16-431a-8a17-501e63cddf73",
+            "prompt_jinja": "{{sentence1}}\nDoes this imply\n{{sentence2}}\nPlease answer {{\"A) yes or B) no.\"}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.029953149241808943
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "imply separated",
+            "acc_norm": 0.5270758122743683,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "f56ffced-9b16-431a-8a17-501e63cddf73",
+            "prompt_jinja": "{{sentence1}}\nDoes this imply\n{{sentence2}}\nPlease answer {{\"A) yes or B) no.\"}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "mean",
+            "acc": 0.48014440433212996,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "03a7ae07-5ddd-46c4-92f3-2152223d44ec",
+            "prompt_jinja": "{{sentence1}}\nDoes this mean that \"{{sentence2}}\" is true? {{\"A) yes or B) no.\"}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.030072723167317194
+        },
+        {
+            "task_name": "rte",
+            "prompt_name": "mean",
+            "acc_norm": 0.5270758122743683,
+            "fixed_answer_choice_list": [
+                "yes",
+                "no"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "rte",
+            "subset": null,
+            "prompt_id": "03a7ae07-5ddd-46c4-92f3-2152223d44ec",
+            "prompt_jinja": "{{sentence1}}\nDoes this mean that \"{{sentence2}}\" is true? {{\"A) yes or B) no.\"}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "following positive negative",
+            "acc": 0.8061926605504587,
+            "fixed_answer_choice_list": [
+                "negative",
+                "positive"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "63c6b2be-8ecd-42ad-88c7-0d1dc1a8323a",
+            "prompt_jinja": "Does the following sentence have a {{\"positive\"}} or {{\"negative\"}} sentiment?\n{{sentence}}\n|||\n{{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.013393542261521812
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "following positive negative",
+            "acc_norm": 0.8061926605504587,
+            "fixed_answer_choice_list": [
+                "negative",
+                "positive"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "63c6b2be-8ecd-42ad-88c7-0d1dc1a8323a",
+            "prompt_jinja": "Does the following sentence have a {{\"positive\"}} or {{\"negative\"}} sentiment?\n{{sentence}}\n|||\n{{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.013393542261521812
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "happy or mad",
+            "acc": 0.5091743119266054,
+            "fixed_answer_choice_list": [
+                "bad",
+                "good"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "6dd74cd5-e074-4612-9e96-c17ca88c3bc4",
+            "prompt_jinja": "Someone sent me an email with the sentence \"{{sentence}}\". Do you think they are feeling {{\"good\"}} or {{\"bad\"}}? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.01693900152535154
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "happy or mad",
+            "acc_norm": 0.5091743119266054,
+            "fixed_answer_choice_list": [
+                "bad",
+                "good"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "6dd74cd5-e074-4612-9e96-c17ca88c3bc4",
+            "prompt_jinja": "Someone sent me an email with the sentence \"{{sentence}}\". Do you think they are feeling {{\"good\"}} or {{\"bad\"}}? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01693900152535154
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "positive negative after",
+            "acc": 0.6204128440366973,
+            "fixed_answer_choice_list": [
+                "negative",
+                "positive"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "11d1c505-9232-4c35-82a4-4c3642843e2e",
+            "prompt_jinja": "{{sentence}}\nQuestion: Was that sentence {{\"positive\"}} or {{\"negative\"}}? Answer: ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.016443227556688766
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "positive negative after",
+            "acc_norm": 0.6204128440366973,
+            "fixed_answer_choice_list": [
+                "negative",
+                "positive"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "11d1c505-9232-4c35-82a4-4c3642843e2e",
+            "prompt_jinja": "{{sentence}}\nQuestion: Was that sentence {{\"positive\"}} or {{\"negative\"}}? Answer: ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.016443227556688766
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "review",
+            "acc": 0.5091743119266054,
+            "fixed_answer_choice_list": [
+                "negative",
+                "positive"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "228fcae7-7f4c-4e3c-9ac4-e49b26bc103d",
+            "prompt_jinja": "I'm reading a review that says \"{{sentence}}\".\n\nDo you think the review is {{\"positive\"}} or {{\"negative\"}}? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.01693900152535154
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "review",
+            "acc_norm": 0.5091743119266054,
+            "fixed_answer_choice_list": [
+                "negative",
+                "positive"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "228fcae7-7f4c-4e3c-9ac4-e49b26bc103d",
+            "prompt_jinja": "I'm reading a review that says \"{{sentence}}\".\n\nDo you think the review is {{\"positive\"}} or {{\"negative\"}}? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01693900152535154
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "said",
+            "acc": 0.4908256880733945,
+            "fixed_answer_choice_list": [
+                "sad",
+                "happy"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "5aa0cea9-0f8d-454d-b25b-b0d4cda273b8",
+            "prompt_jinja": "Someone just said to me \"{{sentence}}\".\n\nDo you think they are {{\"sad\"}} or {{\"happy\"}}? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.01693900152535154
+        },
+        {
+            "task_name": "sst",
+            "prompt_name": "said",
+            "acc_norm": 0.5091743119266054,
+            "fixed_answer_choice_list": [
+                "sad",
+                "happy"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "sst2",
+            "subset": null,
+            "prompt_id": "5aa0cea9-0f8d-454d-b25b-b0d4cda273b8",
+            "prompt_jinja": "Someone just said to me \"{{sentence}}\".\n\nDo you think they are {{\"sad\"}} or {{\"happy\"}}? ||| {{ answer_choices[label] }}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01693900152535154
+        }
+    ],
+    "versions": {
+        "qqp+answer": 0,
+        "qqp+duplicate": 0,
+        "qqp+duplicate or not": 0,
+        "qqp+meaning": 0,
+        "qqp+quora": 0,
+        "qqp+same thing": 0,
+        "rte+does the claim\u2026 follow the fact\u2026": 0,
+        "rte+entailment explained": 0,
+        "rte+imply": 0,
+        "rte+imply separated": 0,
+        "rte+mean": 0,
+        "sst+following positive negative": 0,
+        "sst+happy or mad": 0,
+        "sst+positive negative after": 0,
+        "sst+review": 0,
+        "sst+said": 0
+    },
+    "table_results": {
+        "qqp+answer": {
+            "task_name": "qqp",
+            "prompt_name": "answer",
+            "acc": 0.40558990848379917,
+            "acc_stderr": 0.002441969063495092,
+            "acc_norm": 0.36816720257234725,
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        "qqp+duplicate": {
+            "task_name": "qqp",
+            "prompt_name": "duplicate",
+            "acc": 0.3788523373732377,
+            "acc_stderr": 0.002412603277723025,
+            "acc_norm": 0.36816720257234725,
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        "qqp+duplicate or not": {
+            "task_name": "qqp",
+            "prompt_name": "duplicate or not",
+            "acc": 0.5761315854563444,
+            "acc_stderr": 0.0024577056660753426,
+            "acc_norm": 0.6318327974276527,
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        "qqp+meaning": {
+            "task_name": "qqp",
+            "prompt_name": "meaning",
+            "acc": 0.3681424684640119,
+            "acc_stderr": 0.0023986729832071916,
+            "acc_norm": 0.36816720257234725,
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        "qqp+quora": {
+            "task_name": "qqp",
+            "prompt_name": "quora",
+            "acc": 0.36821667078901804,
+            "acc_stderr": 0.0023987738450886556,
+            "acc_norm": 0.36816720257234725,
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        "qqp+same thing": {
+            "task_name": "qqp",
+            "prompt_name": "same thing",
+            "acc": 0.5099431115508286,
+            "acc_stderr": 0.002486208885430481,
+            "acc_norm": 0.36816720257234725,
+            "acc_norm_stderr": 0.002398706610614492
+        },
+        "rte+does the claim\u2026 follow the fact\u2026": {
+            "task_name": "rte",
+            "prompt_name": "does the claim\u2026 follow the fact\u2026",
+            "acc": 0.4729241877256318,
+            "acc_stderr": 0.030052303463143706,
+            "acc_norm": 0.5270758122743683,
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        "rte+entailment explained": {
+            "task_name": "rte",
+            "prompt_name": "entailment explained",
+            "acc": 0.49458483754512633,
+            "acc_stderr": 0.030094698123239966,
+            "acc_norm": 0.4729241877256318,
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        "rte+imply": {
+            "task_name": "rte",
+            "prompt_name": "imply",
+            "acc": 0.48375451263537905,
+            "acc_stderr": 0.030080573208738064,
+            "acc_norm": 0.5270758122743683,
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        "rte+imply separated": {
+            "task_name": "rte",
+            "prompt_name": "imply separated",
+            "acc": 0.45126353790613716,
+            "acc_stderr": 0.029953149241808943,
+            "acc_norm": 0.5270758122743683,
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        "rte+mean": {
+            "task_name": "rte",
+            "prompt_name": "mean",
+            "acc": 0.48014440433212996,
+            "acc_stderr": 0.030072723167317194,
+            "acc_norm": 0.5270758122743683,
+            "acc_norm_stderr": 0.0300523034631437
+        },
+        "sst+following positive negative": {
+            "task_name": "sst",
+            "prompt_name": "following positive negative",
+            "acc": 0.8061926605504587,
+            "acc_stderr": 0.013393542261521812,
+            "acc_norm": 0.8061926605504587,
+            "acc_norm_stderr": 0.013393542261521812
+        },
+        "sst+happy or mad": {
+            "task_name": "sst",
+            "prompt_name": "happy or mad",
+            "acc": 0.5091743119266054,
+            "acc_stderr": 0.01693900152535154,
+            "acc_norm": 0.5091743119266054,
+            "acc_norm_stderr": 0.01693900152535154
+        },
+        "sst+positive negative after": {
+            "task_name": "sst",
+            "prompt_name": "positive negative after",
+            "acc": 0.6204128440366973,
+            "acc_stderr": 0.016443227556688766,
+            "acc_norm": 0.6204128440366973,
+            "acc_norm_stderr": 0.016443227556688766
+        },
+        "sst+review": {
+            "task_name": "sst",
+            "prompt_name": "review",
+            "acc": 0.5091743119266054,
+            "acc_stderr": 0.01693900152535154,
+            "acc_norm": 0.5091743119266054,
+            "acc_norm_stderr": 0.01693900152535154
+        },
+        "sst+said": {
+            "task_name": "sst",
+            "prompt_name": "said",
+            "acc": 0.4908256880733945,
+            "acc_stderr": 0.01693900152535154,
+            "acc_norm": 0.5091743119266054,
+            "acc_norm_stderr": 0.01693900152535154
+        }
+    },
+    "config": {
+        "adaptive_seq_len": true,
+        "num_fewshot": 0,
+        "bootstrap_iters": 100000
+    }
+}

bigscience/evaluation/results/tr11/bloom1b3/humaneval_temp06.json

@@ -0,0 +1 @@
+{"pass@1": 0.031249999999999993, "pass@10": 0.07447701667197712, "pass@100": 0.1253791767704454}

bigscience/evaluation/results/tr11/bloom1b3/humaneval_temp08.json

@@ -0,0 +1 @@
+{"pass@1": 0.023475609756097564, "pass@10": 0.06591235746713595, "pass@100": 0.12748827115496364}

bigscience/evaluation/results/tr11/bloom2b5/bslmevalfiles/concat.py

@@ -0,0 +1,103 @@
+import argparse
+import json
+import re
+from pathlib import Path
+from re import Pattern
+from typing import List, Dict
+
+
+def get_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--results-dir", required=True, type=Path, help="Path to the list of results")
+    parser.add_argument("--concatenate-output-file", required=True, type=Path, help="Path to store the final output file")
+    return parser.parse_args()
+
+MODEL = "tr11b-1b3-ml-bsevalharness-results_lm-eval_global_step337250"
+# MODEL = "global_step95000"
+RESULTS_REGEX = re.compile(rf"(eai|bs)_results_lm-eval_{MODEL}_(\d{4}-\d{2}-\d{2}-\d{2}-\d{2}-\d{2})_backup\.json")
+RESULTS_REGEX = re.compile(rf"{MODEL}_*.json")
+#tr11b-1b3-ml-bsevalharness-results_lm-eval_global_step340500_2022-07-14-10-03-25.json
+def get_all_files_that_match_results_in_folder(root_folder: Path) -> List[Path]:
+    json_files = []
+    for folder in root_folder.iterdir():
+        if folder.is_dir():
+            json_files += get_all_files_that_match_results_in_folder(folder)
+        else:
+            # it's actually a file
+            file = folder
+
+            #match = RESULTS_REGEX.match(file.name)
+
+            if not str(file.name).endswith("json"):
+                continue
+            else:
+                json_files.append(file)
+    return json_files
+
+def sort_dict(dictionary: Dict) -> Dict:
+    results = {}
+
+    for key, value in sorted(dictionary.items()):
+        new_value = value
+
+        if isinstance(value, dict):
+            new_value = sort_dict(new_value)
+        elif isinstance(value, list):
+            new_value = sorted(value)
+
+        results[key] = new_value
+
+    return results
+
+def main():
+    args = get_args()
+
+    # Get all json files
+    json_files = get_all_files_that_match_results_in_folder(args.results_dir)
+    print("GOT", json_files)
+    # Merge all json files
+    final_result = {
+        "results": {},
+        "versions": {}
+    }
+    for file in json_files:
+        with open(file, "r") as fi:
+            task_result = json.load(fi)
+
+        #match = RESULTS_REGEX.match(file.name)
+        #assert match is not None
+        prefix = "bs" if "bs" in file.name else "eai"#match.group(1)
+        datetime_string = file.name[file.name.index("global_step337250_") + len("global_step337250_"):].replace(".json", "")#match.group(2)
+
+        if prefix == "eai":
+            results_key = "results"
+        elif prefix == "bs":
+            results_key = "table_results"
+        else:
+            raise ValueError(f"Unsupported key: {prefix}")
+
+        for key, value in task_result[results_key].items():
+            if key not in final_result["results"]:
+                final_result["results"][key] = {
+                    datetime_string: value
+                }
+            #else:
+            #    assert datetime_string not in final_result["results"][key]
+            #    final_result["results"][key][datetime_string] = value
+
+        for key, value in task_result["versions"].items():
+            final_result["versions"][key] = value
+
+    # We sort dict, better for serialization
+    print(final_result)
+    final_result = sort_dict(final_result)
+
+    # Save result
+    with open(args.concatenate_output_file, "w") as fo:
+        json.dump(final_result, fo, indent=2)
+
+    pass
+
+if __name__ == "__main__":
+    main()
+

bigscience/evaluation/results/tr11/bloom2b5/bslmevalfiles/tr11c-2b5-ml-bsevalharness-results_lm-eval_global_step337250_2022-07-14-20-09-16.json

@@ -0,0 +1,1255 @@
+{
+    "results": [
+        {
+            "task_name": "gsarti/flores_101_kor",
+            "prompt_name": null,
+            "word_perplexity": 1199924.6918920355
+        },
+        {
+            "task_name": "gsarti/flores_101_kor",
+            "prompt_name": null,
+            "byte_perplexity": 3.932884847226212
+        },
+        {
+            "task_name": "gsarti/flores_101_kor",
+            "prompt_name": null,
+            "bits_per_byte": 1.9755879455567535
+        },
+        {
+            "task_name": "gsarti/flores_101_kir",
+            "prompt_name": null,
+            "word_perplexity": 140474672.36703426
+        },
+        {
+            "task_name": "gsarti/flores_101_kir",
+            "prompt_name": null,
+            "byte_perplexity": 3.729278369847201
+        },
+        {
+            "task_name": "gsarti/flores_101_kir",
+            "prompt_name": null,
+            "bits_per_byte": 1.8988964902756764
+        },
+        {
+            "task_name": "gsarti/flores_101_lao",
+            "prompt_name": null,
+            "word_perplexity": 6.1350041352351446e+26
+        },
+        {
+            "task_name": "gsarti/flores_101_lao",
+            "prompt_name": null,
+            "byte_perplexity": 2.9077314760849924
+        },
+        {
+            "task_name": "gsarti/flores_101_lao",
+            "prompt_name": null,
+            "bits_per_byte": 1.5398940450457603
+        },
+        {
+            "task_name": "gsarti/flores_101_lav",
+            "prompt_name": null,
+            "word_perplexity": 10925745.685132286
+        },
+        {
+            "task_name": "gsarti/flores_101_lav",
+            "prompt_name": null,
+            "byte_perplexity": 7.777221919194806
+        },
+        {
+            "task_name": "gsarti/flores_101_lav",
+            "prompt_name": null,
+            "bits_per_byte": 2.959254905963978
+        },
+        {
+            "task_name": "gsarti/flores_101_lin",
+            "prompt_name": null,
+            "word_perplexity": 166841.83897098716
+        },
+        {
+            "task_name": "gsarti/flores_101_lin",
+            "prompt_name": null,
+            "byte_perplexity": 7.524842908050988
+        },
+        {
+            "task_name": "gsarti/flores_101_lin",
+            "prompt_name": null,
+            "bits_per_byte": 2.9116614638468965
+        },
+        {
+            "task_name": "gsarti/flores_101_lit",
+            "prompt_name": null,
+            "word_perplexity": 8532364.031813102
+        },
+        {
+            "task_name": "gsarti/flores_101_lit",
+            "prompt_name": null,
+            "byte_perplexity": 7.369179434621725
+        },
+        {
+            "task_name": "gsarti/flores_101_lit",
+            "prompt_name": null,
+            "bits_per_byte": 2.88150398275188
+        },
+        {
+            "task_name": "gsarti/flores_101_luo",
+            "prompt_name": null,
+            "word_perplexity": 1335199.656768974
+        },
+        {
+            "task_name": "gsarti/flores_101_luo",
+            "prompt_name": null,
+            "byte_perplexity": 11.975963093623681
+        },
+        {
+            "task_name": "gsarti/flores_101_luo",
+            "prompt_name": null,
+            "bits_per_byte": 3.5820697754437467
+        },
+        {
+            "task_name": "gsarti/flores_101_ltz",
+            "prompt_name": null,
+            "word_perplexity": 4081613.1281958995
+        },
+        {
+            "task_name": "gsarti/flores_101_ltz",
+            "prompt_name": null,
+            "byte_perplexity": 8.801059747949214
+        },
+        {
+            "task_name": "gsarti/flores_101_ltz",
+            "prompt_name": null,
+            "bits_per_byte": 3.1376772511430198
+        },
+        {
+            "task_name": "gsarti/flores_101_mkd",
+            "prompt_name": null,
+            "word_perplexity": 291548.6603872499
+        },
+        {
+            "task_name": "gsarti/flores_101_mkd",
+            "prompt_name": null,
+            "byte_perplexity": 2.9656732291754087
+        },
+        {
+            "task_name": "gsarti/flores_101_mkd",
+            "prompt_name": null,
+            "bits_per_byte": 1.5683596441110415
+        },
+        {
+            "task_name": "gsarti/flores_101_msa",
+            "prompt_name": null,
+            "word_perplexity": 931.4191160965655
+        },
+        {
+            "task_name": "gsarti/flores_101_msa",
+            "prompt_name": null,
+            "byte_perplexity": 2.5710001772665634
+        },
+        {
+            "task_name": "gsarti/flores_101_msa",
+            "prompt_name": null,
+            "bits_per_byte": 1.3623297096432079
+        },
+        {
+            "task_name": "gsarti/flores_101_mal",
+            "prompt_name": null,
+            "word_perplexity": 1.207348615509252e+18
+        },
+        {
+            "task_name": "gsarti/flores_101_mal",
+            "prompt_name": null,
+            "byte_perplexity": 4.615948455160037
+        },
+        {
+            "task_name": "gsarti/flores_101_mal",
+            "prompt_name": null,
+            "bits_per_byte": 2.2066271139530245
+        },
+        {
+            "task_name": "gsarti/flores_101_mlt",
+            "prompt_name": null,
+            "word_perplexity": 1820552051.5260184
+        },
+        {
+            "task_name": "gsarti/flores_101_mlt",
+            "prompt_name": null,
+            "byte_perplexity": 15.004773437665275
+        },
+        {
+            "task_name": "gsarti/flores_101_mlt",
+            "prompt_name": null,
+            "bits_per_byte": 3.9073496302297994
+        },
+        {
+            "task_name": "gsarti/flores_101_mri",
+            "prompt_name": null,
+            "word_perplexity": 26466.98082941409
+        },
+        {
+            "task_name": "gsarti/flores_101_mri",
+            "prompt_name": null,
+            "byte_perplexity": 7.474035895661322
+        },
+        {
+            "task_name": "gsarti/flores_101_mri",
+            "prompt_name": null,
+            "bits_per_byte": 2.9018874925878335
+        },
+        {
+            "task_name": "gsarti/flores_101_mar",
+            "prompt_name": null,
+            "word_perplexity": 54017030487867.64
+        },
+        {
+            "task_name": "gsarti/flores_101_mar",
+            "prompt_name": null,
+            "byte_perplexity": 5.483253482821379
+        },
+        {
+            "task_name": "gsarti/flores_101_mar",
+            "prompt_name": null,
+            "bits_per_byte": 2.4550321688665875
+        },
+        {
+            "task_name": "gsarti/flores_101_mon",
+            "prompt_name": null,
+            "word_perplexity": 6612951.176601774
+        },
+        {
+            "task_name": "gsarti/flores_101_mon",
+            "prompt_name": null,
+            "byte_perplexity": 3.410598542315402
+        },
+        {
+            "task_name": "gsarti/flores_101_mon",
+            "prompt_name": null,
+            "bits_per_byte": 1.7700249469487581
+        },
+        {
+            "task_name": "gsarti/flores_101_npi",
+            "prompt_name": null,
+            "word_perplexity": 9218412485042.457
+        },
+        {
+            "task_name": "gsarti/flores_101_npi",
+            "prompt_name": null,
+            "byte_perplexity": 5.199342701937889
+        },
+        {
+            "task_name": "gsarti/flores_101_npi",
+            "prompt_name": null,
+            "bits_per_byte": 2.3783292500628397
+        },
+        {
+            "task_name": "gsarti/flores_101_nso",
+            "prompt_name": null,
+            "word_perplexity": 84236.45826211123
+        },
+        {
+            "task_name": "gsarti/flores_101_nso",
+            "prompt_name": null,
+            "byte_perplexity": 8.154626800955667
+        },
+        {
+            "task_name": "gsarti/flores_101_nso",
+            "prompt_name": null,
+            "bits_per_byte": 3.027618853058479
+        },
+        {
+            "task_name": "gsarti/flores_101_nob",
+            "prompt_name": null,
+            "word_perplexity": 36969.51682419191
+        },
+        {
+            "task_name": "gsarti/flores_101_nob",
+            "prompt_name": null,
+            "byte_perplexity": 5.402763169129877
+        },
+        {
+            "task_name": "gsarti/flores_101_nob",
+            "prompt_name": null,
+            "bits_per_byte": 2.4336974426149056
+        },
+        {
+            "task_name": "gsarti/flores_101_nya",
+            "prompt_name": null,
+            "word_perplexity": 6609896.030066139
+        },
+        {
+            "task_name": "gsarti/flores_101_nya",
+            "prompt_name": null,
+            "byte_perplexity": 8.179860208369393
+        },
+        {
+            "task_name": "gsarti/flores_101_nya",
+            "prompt_name": null,
+            "bits_per_byte": 3.0320761881040017
+        },
+        {
+            "task_name": "gsarti/flores_101_oci",
+            "prompt_name": null,
+            "word_perplexity": 21641.316763505896
+        },
+        {
+            "task_name": "gsarti/flores_101_oci",
+            "prompt_name": null,
+            "byte_perplexity": 4.8617357393685845
+        },
+        {
+            "task_name": "gsarti/flores_101_oci",
+            "prompt_name": null,
+            "bits_per_byte": 2.2814714775164466
+        },
+        {
+            "task_name": "gsarti/flores_101_ory",
+            "prompt_name": null,
+            "word_perplexity": 11873283711992.748
+        },
+        {
+            "task_name": "gsarti/flores_101_ory",
+            "prompt_name": null,
+            "byte_perplexity": 5.189421861225964
+        },
+        {
+            "task_name": "gsarti/flores_101_ory",
+            "prompt_name": null,
+            "bits_per_byte": 2.375573820972048
+        },
+        {
+            "task_name": "gsarti/flores_101_orm",
+            "prompt_name": null,
+            "word_perplexity": 944722910.1683049
+        },
+        {
+            "task_name": "gsarti/flores_101_orm",
+            "prompt_name": null,
+            "byte_perplexity": 12.911595421079408
+        },
+        {
+            "task_name": "gsarti/flores_101_orm",
+            "prompt_name": null,
+            "bits_per_byte": 3.690595373136525
+        },
+        {
+            "task_name": "gsarti/flores_101_pus",
+            "prompt_name": null,
+            "word_perplexity": 153261.38659736273
+        },
+        {
+            "task_name": "gsarti/flores_101_pus",
+            "prompt_name": null,
+            "byte_perplexity": 4.4963371422771585
+        },
+        {
+            "task_name": "gsarti/flores_101_pus",
+            "prompt_name": null,
+            "bits_per_byte": 2.1687502151085742
+        },
+        {
+            "task_name": "gsarti/flores_101_fas",
+            "prompt_name": null,
+            "word_perplexity": 44174.10652942002
+        },
+        {
+            "task_name": "gsarti/flores_101_fas",
+            "prompt_name": null,
+            "byte_perplexity": 3.058009097116482
+        },
+        {
+            "task_name": "gsarti/flores_101_fas",
+            "prompt_name": null,
+            "bits_per_byte": 1.6125926985055565
+        },
+        {
+            "task_name": "gsarti/flores_101_pol",
+            "prompt_name": null,
+            "word_perplexity": 104253.80848720921
+        },
+        {
+            "task_name": "gsarti/flores_101_pol",
+            "prompt_name": null,
+            "byte_perplexity": 4.625550458479643
+        },
+        {
+            "task_name": "gsarti/flores_101_pol",
+            "prompt_name": null,
+            "bits_per_byte": 2.2096250621616695
+        },
+        {
+            "task_name": "gsarti/flores_101_por",
+            "prompt_name": null,
+            "word_perplexity": 70.12185258792593
+        },
+        {
+            "task_name": "gsarti/flores_101_por",
+            "prompt_name": null,
+            "byte_perplexity": 1.9754515986213523
+        },
+        {
+            "task_name": "gsarti/flores_101_por",
+            "prompt_name": null,
+            "bits_per_byte": 0.9821824986646657
+        },
+        {
+            "task_name": "gsarti/flores_101_pan",
+            "prompt_name": null,
+            "word_perplexity": 847925284.3968099
+        },
+        {
+            "task_name": "gsarti/flores_101_pan",
+            "prompt_name": null,
+            "byte_perplexity": 4.698477289331806
+        },
+        {
+            "task_name": "gsarti/flores_101_pan",
+            "prompt_name": null,
+            "bits_per_byte": 2.2321932752863454
+        },
+        {
+            "task_name": "gsarti/flores_101_ron",
+            "prompt_name": null,
+            "word_perplexity": 36440.61611845943
+        },
+        {
+            "task_name": "gsarti/flores_101_ron",
+            "prompt_name": null,
+            "byte_perplexity": 4.965456830031304
+        },
+        {
+            "task_name": "gsarti/flores_101_ron",
+            "prompt_name": null,
+            "bits_per_byte": 2.31192645412871
+        },
+        {
+            "task_name": "gsarti/flores_101_rus",
+            "prompt_name": null,
+            "word_perplexity": 12717.27557342625
+        },
+        {
+            "task_name": "gsarti/flores_101_rus",
+            "prompt_name": null,
+            "byte_perplexity": 2.0498020542445303
+        },
+        {
+            "task_name": "gsarti/flores_101_rus",
+            "prompt_name": null,
+            "bits_per_byte": 1.0354845979511649
+        },
+        {
+            "task_name": "gsarti/flores_101_srp",
+            "prompt_name": null,
+            "word_perplexity": 179094.36755355867
+        },
+        {
+            "task_name": "gsarti/flores_101_srp",
+            "prompt_name": null,
+            "byte_perplexity": 2.871214785885079
+        },
+        {
+            "task_name": "gsarti/flores_101_srp",
+            "prompt_name": null,
+            "bits_per_byte": 1.5216612577275341
+        },
+        {
+            "task_name": "gsarti/flores_101_sna",
+            "prompt_name": null,
+            "word_perplexity": 64794029.630749054
+        },
+        {
+            "task_name": "gsarti/flores_101_sna",
+            "prompt_name": null,
+            "byte_perplexity": 8.462166771382726
+        },
+        {
+            "task_name": "gsarti/flores_101_sna",
+            "prompt_name": null,
+            "bits_per_byte": 3.0810271184378166
+        },
+        {
+            "task_name": "gsarti/flores_101_snd",
+            "prompt_name": null,
+            "word_perplexity": 1593844.7987764536
+        },
+        {
+            "task_name": "gsarti/flores_101_snd",
+            "prompt_name": null,
+            "byte_perplexity": 5.466066951221973
+        },
+        {
+            "task_name": "gsarti/flores_101_snd",
+            "prompt_name": null,
+            "bits_per_byte": 2.450503130846187
+        },
+        {
+            "task_name": "gsarti/flores_101_slk",
+            "prompt_name": null,
+            "word_perplexity": 766753.5771631876
+        },
+        {
+            "task_name": "gsarti/flores_101_slk",
+            "prompt_name": null,
+            "byte_perplexity": 6.450822127057479
+        },
+        {
+            "task_name": "gsarti/flores_101_slk",
+            "prompt_name": null,
+            "bits_per_byte": 2.6894830369770566
+        },
+        {
+            "task_name": "gsarti/flores_101_slv",
+            "prompt_name": null,
+            "word_perplexity": 281495.6973621906
+        },
+        {
+            "task_name": "gsarti/flores_101_slv",
+            "prompt_name": null,
+            "byte_perplexity": 6.620252120186232
+        },
+        {
+            "task_name": "gsarti/flores_101_slv",
+            "prompt_name": null,
+            "bits_per_byte": 2.726886160479057
+        },
+        {
+            "task_name": "gsarti/flores_101_som",
+            "prompt_name": null,
+            "word_perplexity": 9117591.536991648
+        },
+        {
+            "task_name": "gsarti/flores_101_som",
+            "prompt_name": null,
+            "byte_perplexity": 11.95918054093392
+        },
+        {
+            "task_name": "gsarti/flores_101_som",
+            "prompt_name": null,
+            "bits_per_byte": 3.5800466324138576
+        },
+        {
+            "task_name": "gsarti/flores_101_ckb",
+            "prompt_name": null,
+            "word_perplexity": 7641937.513844287
+        },
+        {
+            "task_name": "gsarti/flores_101_ckb",
+            "prompt_name": null,
+            "byte_perplexity": 3.7255124939234765
+        },
+        {
+            "task_name": "gsarti/flores_101_ckb",
+            "prompt_name": null,
+            "bits_per_byte": 1.8974389011678956
+        },
+        {
+            "task_name": "gsarti/flores_101_spa",
+            "prompt_name": null,
+            "word_perplexity": 50.48600403475257
+        },
+        {
+            "task_name": "gsarti/flores_101_spa",
+            "prompt_name": null,
+            "byte_perplexity": 1.8965140104323535
+        },
+        {
+            "task_name": "gsarti/flores_101_spa",
+            "prompt_name": null,
+            "bits_per_byte": 0.9233500295317635
+        },
+        {
+            "task_name": "gsarti/flores_101_swh",
+            "prompt_name": null,
+            "word_perplexity": 4756.310957867697
+        },
+        {
+            "task_name": "gsarti/flores_101_swh",
+            "prompt_name": null,
+            "byte_perplexity": 3.6973091886730676
+        },
+        {
+            "task_name": "gsarti/flores_101_swh",
+            "prompt_name": null,
+            "bits_per_byte": 1.8864756944079395
+        },
+        {
+            "task_name": "gsarti/flores_101_swe",
+            "prompt_name": null,
+            "word_perplexity": 50609.194691403645
+        },
+        {
+            "task_name": "gsarti/flores_101_swe",
+            "prompt_name": null,
+            "byte_perplexity": 5.054972008155866
+        },
+        {
+            "task_name": "gsarti/flores_101_swe",
+            "prompt_name": null,
+            "bits_per_byte": 2.3377031032447033
+        },
+        {
+            "task_name": "gsarti/flores_101_tgk",
+            "prompt_name": null,
+            "word_perplexity": 4653242.643384356
+        },
+        {
+            "task_name": "gsarti/flores_101_tgk",
+            "prompt_name": null,
+            "byte_perplexity": 3.5994818827380426
+        },
+        {
+            "task_name": "gsarti/flores_101_tgk",
+            "prompt_name": null,
+            "bits_per_byte": 1.847789256832959
+        },
+        {
+            "task_name": "gsarti/flores_101_tam",
+            "prompt_name": null,
+            "word_perplexity": 1.7375636861561886e+16
+        },
+        {
+            "task_name": "gsarti/flores_101_tam",
+            "prompt_name": null,
+            "byte_perplexity": 4.539493400469833
+        },
+        {
+            "task_name": "gsarti/flores_101_tam",
+            "prompt_name": null,
+            "bits_per_byte": 2.182531304254031
+        },
+        {
+            "task_name": "gsarti/flores_101_tel",
+            "prompt_name": null,
+            "word_perplexity": 6240250468604343.0
+        },
+        {
+            "task_name": "gsarti/flores_101_tel",
+            "prompt_name": null,
+            "byte_perplexity": 5.807499987508966
+        },
+        {
+            "task_name": "gsarti/flores_101_tel",
+            "prompt_name": null,
+            "bits_per_byte": 2.537917245931069
+        },
+        {
+            "task_name": "gsarti/flores_101_tha",
+            "prompt_name": null,
+            "word_perplexity": 2.7023221906004898e+31
+        },
+        {
+            "task_name": "gsarti/flores_101_tha",
+            "prompt_name": null,
+            "byte_perplexity": 2.365940201944242
+        },
+        {
+            "task_name": "gsarti/flores_101_tha",
+            "prompt_name": null,
+            "bits_per_byte": 1.242413610681628
+        },
+        {
+            "task_name": "gsarti/flores_101_tur",
+            "prompt_name": null,
+            "word_perplexity": 598170.0194818947
+        },
+        {
+            "task_name": "gsarti/flores_101_tur",
+            "prompt_name": null,
+            "byte_perplexity": 4.885014749844601
+        },
+        {
+            "task_name": "gsarti/flores_101_tur",
+            "prompt_name": null,
+            "bits_per_byte": 2.288362918282818
+        },
+        {
+            "task_name": "gsarti/flores_101_ukr",
+            "prompt_name": null,
+            "word_perplexity": 375312.1511987307
+        },
+        {
+            "task_name": "gsarti/flores_101_ukr",
+            "prompt_name": null,
+            "byte_perplexity": 2.7240934990288483
+        },
+        {
+            "task_name": "gsarti/flores_101_ukr",
+            "prompt_name": null,
+            "bits_per_byte": 1.445776221804572
+        },
+        {
+            "task_name": "gsarti/flores_101_umb",
+            "prompt_name": null,
+            "word_perplexity": 286182026.84727985
+        },
+        {
+            "task_name": "gsarti/flores_101_umb",
+            "prompt_name": null,
+            "byte_perplexity": 12.766915508610673
+        },
+        {
+            "task_name": "gsarti/flores_101_umb",
+            "prompt_name": null,
+            "bits_per_byte": 3.6743381063848357
+        },
+        {
+            "task_name": "gsarti/flores_101_urd",
+            "prompt_name": null,
+            "word_perplexity": 294.7473718166965
+        },
+        {
+            "task_name": "gsarti/flores_101_urd",
+            "prompt_name": null,
+            "byte_perplexity": 1.9797467071381232
+        },
+        {
+            "task_name": "gsarti/flores_101_urd",
+            "prompt_name": null,
+            "bits_per_byte": 0.9853158607436239
+        },
+        {
+            "task_name": "gsarti/flores_101_uzb",
+            "prompt_name": null,
+            "word_perplexity": 657971096.5030558
+        },
+        {
+            "task_name": "gsarti/flores_101_uzb",
+            "prompt_name": null,
+            "byte_perplexity": 12.002337637722146
+        },
+        {
+            "task_name": "gsarti/flores_101_uzb",
+            "prompt_name": null,
+            "bits_per_byte": 3.5852435148799184
+        },
+        {
+            "task_name": "gsarti/flores_101_vie",
+            "prompt_name": null,
+            "word_perplexity": 30.113286809710246
+        },
+        {
+            "task_name": "gsarti/flores_101_vie",
+            "prompt_name": null,
+            "byte_perplexity": 1.76578415476397
+        },
+        {
+            "task_name": "gsarti/flores_101_vie",
+            "prompt_name": null,
+            "bits_per_byte": 0.8203090021691818
+        },
+        {
+            "task_name": "gsarti/flores_101_cym",
+            "prompt_name": null,
+            "word_perplexity": 2638019.4579179045
+        },
+        {
+            "task_name": "gsarti/flores_101_cym",
+            "prompt_name": null,
+            "byte_perplexity": 12.539424151448149
+        },
+        {
+            "task_name": "gsarti/flores_101_cym",
+            "prompt_name": null,
+            "bits_per_byte": 3.6483991915978407
+        },
+        {
+            "task_name": "gsarti/flores_101_wol",
+            "prompt_name": null,
+            "word_perplexity": 119795.78671768666
+        },
+        {
+            "task_name": "gsarti/flores_101_wol",
+            "prompt_name": null,
+            "byte_perplexity": 9.144285650306488
+        },
+        {
+            "task_name": "gsarti/flores_101_wol",
+            "prompt_name": null,
+            "bits_per_byte": 3.1928704713393357
+        },
+        {
+            "task_name": "gsarti/flores_101_xho",
+            "prompt_name": null,
+            "word_perplexity": 54307092.21333007
+        },
+        {
+            "task_name": "gsarti/flores_101_xho",
+            "prompt_name": null,
+            "byte_perplexity": 7.403240538286952
+        },
+        {
+            "task_name": "gsarti/flores_101_xho",
+            "prompt_name": null,
+            "bits_per_byte": 2.8881569038733983
+        },
+        {
+            "task_name": "gsarti/flores_101_yor",
+            "prompt_name": null,
+            "word_perplexity": 130267.12232132205
+        },
+        {
+            "task_name": "gsarti/flores_101_yor",
+            "prompt_name": null,
+            "byte_perplexity": 5.91272037551173
+        },
+        {
+            "task_name": "gsarti/flores_101_yor",
+            "prompt_name": null,
+            "bits_per_byte": 2.5638220507535796
+        },
+        {
+            "task_name": "gsarti/flores_101_zul",
+            "prompt_name": null,
+            "word_perplexity": 493606524.8156374
+        },
+        {
+            "task_name": "gsarti/flores_101_zul",
+            "prompt_name": null,
+            "byte_perplexity": 8.53353320693145
+        },
+        {
+            "task_name": "gsarti/flores_101_zul",
+            "prompt_name": null,
+            "bits_per_byte": 3.0931431957905224
+        }
+    ],
+    "versions": {
+        "gsarti/flores_101_kor+null": 0,
+        "gsarti/flores_101_kir+null": 0,
+        "gsarti/flores_101_lao+null": 0,
+        "gsarti/flores_101_lav+null": 0,
+        "gsarti/flores_101_lin+null": 0,
+        "gsarti/flores_101_lit+null": 0,
+        "gsarti/flores_101_luo+null": 0,
+        "gsarti/flores_101_ltz+null": 0,
+        "gsarti/flores_101_mkd+null": 0,
+        "gsarti/flores_101_msa+null": 0,
+        "gsarti/flores_101_mal+null": 0,
+        "gsarti/flores_101_mlt+null": 0,
+        "gsarti/flores_101_mri+null": 0,
+        "gsarti/flores_101_mar+null": 0,
+        "gsarti/flores_101_mon+null": 0,
+        "gsarti/flores_101_npi+null": 0,
+        "gsarti/flores_101_nso+null": 0,
+        "gsarti/flores_101_nob+null": 0,
+        "gsarti/flores_101_nya+null": 0,
+        "gsarti/flores_101_oci+null": 0,
+        "gsarti/flores_101_ory+null": 0,
+        "gsarti/flores_101_orm+null": 0,
+        "gsarti/flores_101_pus+null": 0,
+        "gsarti/flores_101_fas+null": 0,
+        "gsarti/flores_101_pol+null": 0,
+        "gsarti/flores_101_por+null": 0,
+        "gsarti/flores_101_pan+null": 0,
+        "gsarti/flores_101_ron+null": 0,
+        "gsarti/flores_101_rus+null": 0,
+        "gsarti/flores_101_srp+null": 0,
+        "gsarti/flores_101_sna+null": 0,
+        "gsarti/flores_101_snd+null": 0,
+        "gsarti/flores_101_slk+null": 0,
+        "gsarti/flores_101_slv+null": 0,
+        "gsarti/flores_101_som+null": 0,
+        "gsarti/flores_101_ckb+null": 0,
+        "gsarti/flores_101_spa+null": 0,
+        "gsarti/flores_101_swh+null": 0,
+        "gsarti/flores_101_swe+null": 0,
+        "gsarti/flores_101_tgk+null": 0,
+        "gsarti/flores_101_tam+null": 0,
+        "gsarti/flores_101_tel+null": 0,
+        "gsarti/flores_101_tha+null": 0,
+        "gsarti/flores_101_tur+null": 0,
+        "gsarti/flores_101_ukr+null": 0,
+        "gsarti/flores_101_umb+null": 0,
+        "gsarti/flores_101_urd+null": 0,
+        "gsarti/flores_101_uzb+null": 0,
+        "gsarti/flores_101_vie+null": 0,
+        "gsarti/flores_101_cym+null": 0,
+        "gsarti/flores_101_wol+null": 0,
+        "gsarti/flores_101_xho+null": 0,
+        "gsarti/flores_101_yor+null": 0,
+        "gsarti/flores_101_zul+null": 0
+    },
+    "table_results": {
+        "gsarti/flores_101_kor+null": {
+            "task_name": "gsarti/flores_101_kor",
+            "prompt_name": "null",
+            "word_perplexity": 1199924.6918920355,
+            "byte_perplexity": 3.932884847226212,
+            "bits_per_byte": 1.9755879455567535
+        },
+        "gsarti/flores_101_kir+null": {
+            "task_name": "gsarti/flores_101_kir",
+            "prompt_name": "null",
+            "word_perplexity": 140474672.36703426,
+            "byte_perplexity": 3.729278369847201,
+            "bits_per_byte": 1.8988964902756764
+        },
+        "gsarti/flores_101_lao+null": {
+            "task_name": "gsarti/flores_101_lao",
+            "prompt_name": "null",
+            "word_perplexity": 6.1350041352351446e+26,
+            "byte_perplexity": 2.9077314760849924,
+            "bits_per_byte": 1.5398940450457603
+        },
+        "gsarti/flores_101_lav+null": {
+            "task_name": "gsarti/flores_101_lav",
+            "prompt_name": "null",
+            "word_perplexity": 10925745.685132286,
+            "byte_perplexity": 7.777221919194806,
+            "bits_per_byte": 2.959254905963978
+        },
+        "gsarti/flores_101_lin+null": {
+            "task_name": "gsarti/flores_101_lin",
+            "prompt_name": "null",
+            "word_perplexity": 166841.83897098716,
+            "byte_perplexity": 7.524842908050988,
+            "bits_per_byte": 2.9116614638468965
+        },
+        "gsarti/flores_101_lit+null": {
+            "task_name": "gsarti/flores_101_lit",
+            "prompt_name": "null",
+            "word_perplexity": 8532364.031813102,
+            "byte_perplexity": 7.369179434621725,
+            "bits_per_byte": 2.88150398275188
+        },
+        "gsarti/flores_101_luo+null": {
+            "task_name": "gsarti/flores_101_luo",
+            "prompt_name": "null",
+            "word_perplexity": 1335199.656768974,
+            "byte_perplexity": 11.975963093623681,
+            "bits_per_byte": 3.5820697754437467
+        },
+        "gsarti/flores_101_ltz+null": {
+            "task_name": "gsarti/flores_101_ltz",
+            "prompt_name": "null",
+            "word_perplexity": 4081613.1281958995,
+            "byte_perplexity": 8.801059747949214,
+            "bits_per_byte": 3.1376772511430198
+        },
+        "gsarti/flores_101_mkd+null": {
+            "task_name": "gsarti/flores_101_mkd",
+            "prompt_name": "null",
+            "word_perplexity": 291548.6603872499,
+            "byte_perplexity": 2.9656732291754087,
+            "bits_per_byte": 1.5683596441110415
+        },
+        "gsarti/flores_101_msa+null": {
+            "task_name": "gsarti/flores_101_msa",
+            "prompt_name": "null",
+            "word_perplexity": 931.4191160965655,
+            "byte_perplexity": 2.5710001772665634,
+            "bits_per_byte": 1.3623297096432079
+        },
+        "gsarti/flores_101_mal+null": {
+            "task_name": "gsarti/flores_101_mal",
+            "prompt_name": "null",
+            "word_perplexity": 1.207348615509252e+18,
+            "byte_perplexity": 4.615948455160037,
+            "bits_per_byte": 2.2066271139530245
+        },
+        "gsarti/flores_101_mlt+null": {
+            "task_name": "gsarti/flores_101_mlt",
+            "prompt_name": "null",
+            "word_perplexity": 1820552051.5260184,
+            "byte_perplexity": 15.004773437665275,
+            "bits_per_byte": 3.9073496302297994
+        },
+        "gsarti/flores_101_mri+null": {
+            "task_name": "gsarti/flores_101_mri",
+            "prompt_name": "null",
+            "word_perplexity": 26466.98082941409,
+            "byte_perplexity": 7.474035895661322,
+            "bits_per_byte": 2.9018874925878335
+        },
+        "gsarti/flores_101_mar+null": {
+            "task_name": "gsarti/flores_101_mar",
+            "prompt_name": "null",
+            "word_perplexity": 54017030487867.64,
+            "byte_perplexity": 5.483253482821379,
+            "bits_per_byte": 2.4550321688665875
+        },
+        "gsarti/flores_101_mon+null": {
+            "task_name": "gsarti/flores_101_mon",
+            "prompt_name": "null",
+            "word_perplexity": 6612951.176601774,
+            "byte_perplexity": 3.410598542315402,
+            "bits_per_byte": 1.7700249469487581
+        },
+        "gsarti/flores_101_npi+null": {
+            "task_name": "gsarti/flores_101_npi",
+            "prompt_name": "null",
+            "word_perplexity": 9218412485042.457,
+            "byte_perplexity": 5.199342701937889,
+            "bits_per_byte": 2.3783292500628397
+        },
+        "gsarti/flores_101_nso+null": {
+            "task_name": "gsarti/flores_101_nso",
+            "prompt_name": "null",
+            "word_perplexity": 84236.45826211123,
+            "byte_perplexity": 8.154626800955667,
+            "bits_per_byte": 3.027618853058479
+        },
+        "gsarti/flores_101_nob+null": {
+            "task_name": "gsarti/flores_101_nob",
+            "prompt_name": "null",
+            "word_perplexity": 36969.51682419191,
+            "byte_perplexity": 5.402763169129877,
+            "bits_per_byte": 2.4336974426149056
+        },
+        "gsarti/flores_101_nya+null": {
+            "task_name": "gsarti/flores_101_nya",
+            "prompt_name": "null",
+            "word_perplexity": 6609896.030066139,
+            "byte_perplexity": 8.179860208369393,
+            "bits_per_byte": 3.0320761881040017
+        },
+        "gsarti/flores_101_oci+null": {
+            "task_name": "gsarti/flores_101_oci",
+            "prompt_name": "null",
+            "word_perplexity": 21641.316763505896,
+            "byte_perplexity": 4.8617357393685845,
+            "bits_per_byte": 2.2814714775164466
+        },
+        "gsarti/flores_101_ory+null": {
+            "task_name": "gsarti/flores_101_ory",
+            "prompt_name": "null",
+            "word_perplexity": 11873283711992.748,
+            "byte_perplexity": 5.189421861225964,
+            "bits_per_byte": 2.375573820972048
+        },
+        "gsarti/flores_101_orm+null": {
+            "task_name": "gsarti/flores_101_orm",
+            "prompt_name": "null",
+            "word_perplexity": 944722910.1683049,
+            "byte_perplexity": 12.911595421079408,
+            "bits_per_byte": 3.690595373136525
+        },
+        "gsarti/flores_101_pus+null": {
+            "task_name": "gsarti/flores_101_pus",
+            "prompt_name": "null",
+            "word_perplexity": 153261.38659736273,
+            "byte_perplexity": 4.4963371422771585,
+            "bits_per_byte": 2.1687502151085742
+        },
+        "gsarti/flores_101_fas+null": {
+            "task_name": "gsarti/flores_101_fas",
+            "prompt_name": "null",
+            "word_perplexity": 44174.10652942002,
+            "byte_perplexity": 3.058009097116482,
+            "bits_per_byte": 1.6125926985055565
+        },
+        "gsarti/flores_101_pol+null": {
+            "task_name": "gsarti/flores_101_pol",
+            "prompt_name": "null",
+            "word_perplexity": 104253.80848720921,
+            "byte_perplexity": 4.625550458479643,
+            "bits_per_byte": 2.2096250621616695
+        },
+        "gsarti/flores_101_por+null": {
+            "task_name": "gsarti/flores_101_por",
+            "prompt_name": "null",
+            "word_perplexity": 70.12185258792593,
+            "byte_perplexity": 1.9754515986213523,
+            "bits_per_byte": 0.9821824986646657
+        },
+        "gsarti/flores_101_pan+null": {
+            "task_name": "gsarti/flores_101_pan",
+            "prompt_name": "null",
+            "word_perplexity": 847925284.3968099,
+            "byte_perplexity": 4.698477289331806,
+            "bits_per_byte": 2.2321932752863454
+        },
+        "gsarti/flores_101_ron+null": {
+            "task_name": "gsarti/flores_101_ron",
+            "prompt_name": "null",
+            "word_perplexity": 36440.61611845943,
+            "byte_perplexity": 4.965456830031304,
+            "bits_per_byte": 2.31192645412871
+        },
+        "gsarti/flores_101_rus+null": {
+            "task_name": "gsarti/flores_101_rus",
+            "prompt_name": "null",
+            "word_perplexity": 12717.27557342625,
+            "byte_perplexity": 2.0498020542445303,
+            "bits_per_byte": 1.0354845979511649
+        },
+        "gsarti/flores_101_srp+null": {
+            "task_name": "gsarti/flores_101_srp",
+            "prompt_name": "null",
+            "word_perplexity": 179094.36755355867,
+            "byte_perplexity": 2.871214785885079,
+            "bits_per_byte": 1.5216612577275341
+        },
+        "gsarti/flores_101_sna+null": {
+            "task_name": "gsarti/flores_101_sna",
+            "prompt_name": "null",
+            "word_perplexity": 64794029.630749054,
+            "byte_perplexity": 8.462166771382726,
+            "bits_per_byte": 3.0810271184378166
+        },
+        "gsarti/flores_101_snd+null": {
+            "task_name": "gsarti/flores_101_snd",
+            "prompt_name": "null",
+            "word_perplexity": 1593844.7987764536,
+            "byte_perplexity": 5.466066951221973,
+            "bits_per_byte": 2.450503130846187
+        },
+        "gsarti/flores_101_slk+null": {
+            "task_name": "gsarti/flores_101_slk",
+            "prompt_name": "null",
+            "word_perplexity": 766753.5771631876,
+            "byte_perplexity": 6.450822127057479,
+            "bits_per_byte": 2.6894830369770566
+        },
+        "gsarti/flores_101_slv+null": {
+            "task_name": "gsarti/flores_101_slv",
+            "prompt_name": "null",
+            "word_perplexity": 281495.6973621906,
+            "byte_perplexity": 6.620252120186232,
+            "bits_per_byte": 2.726886160479057
+        },
+        "gsarti/flores_101_som+null": {
+            "task_name": "gsarti/flores_101_som",
+            "prompt_name": "null",
+            "word_perplexity": 9117591.536991648,
+            "byte_perplexity": 11.95918054093392,
+            "bits_per_byte": 3.5800466324138576
+        },
+        "gsarti/flores_101_ckb+null": {
+            "task_name": "gsarti/flores_101_ckb",
+            "prompt_name": "null",
+            "word_perplexity": 7641937.513844287,
+            "byte_perplexity": 3.7255124939234765,
+            "bits_per_byte": 1.8974389011678956
+        },
+        "gsarti/flores_101_spa+null": {
+            "task_name": "gsarti/flores_101_spa",
+            "prompt_name": "null",
+            "word_perplexity": 50.48600403475257,
+            "byte_perplexity": 1.8965140104323535,
+            "bits_per_byte": 0.9233500295317635
+        },
+        "gsarti/flores_101_swh+null": {
+            "task_name": "gsarti/flores_101_swh",
+            "prompt_name": "null",
+            "word_perplexity": 4756.310957867697,
+            "byte_perplexity": 3.6973091886730676,
+            "bits_per_byte": 1.8864756944079395
+        },
+        "gsarti/flores_101_swe+null": {
+            "task_name": "gsarti/flores_101_swe",
+            "prompt_name": "null",
+            "word_perplexity": 50609.194691403645,
+            "byte_perplexity": 5.054972008155866,
+            "bits_per_byte": 2.3377031032447033
+        },
+        "gsarti/flores_101_tgk+null": {
+            "task_name": "gsarti/flores_101_tgk",
+            "prompt_name": "null",
+            "word_perplexity": 4653242.643384356,
+            "byte_perplexity": 3.5994818827380426,
+            "bits_per_byte": 1.847789256832959
+        },
+        "gsarti/flores_101_tam+null": {
+            "task_name": "gsarti/flores_101_tam",
+            "prompt_name": "null",
+            "word_perplexity": 1.7375636861561886e+16,
+            "byte_perplexity": 4.539493400469833,
+            "bits_per_byte": 2.182531304254031
+        },
+        "gsarti/flores_101_tel+null": {
+            "task_name": "gsarti/flores_101_tel",
+            "prompt_name": "null",
+            "word_perplexity": 6240250468604343.0,
+            "byte_perplexity": 5.807499987508966,
+            "bits_per_byte": 2.537917245931069
+        },
+        "gsarti/flores_101_tha+null": {
+            "task_name": "gsarti/flores_101_tha",
+            "prompt_name": "null",
+            "word_perplexity": 2.7023221906004898e+31,
+            "byte_perplexity": 2.365940201944242,
+            "bits_per_byte": 1.242413610681628
+        },
+        "gsarti/flores_101_tur+null": {
+            "task_name": "gsarti/flores_101_tur",
+            "prompt_name": "null",
+            "word_perplexity": 598170.0194818947,
+            "byte_perplexity": 4.885014749844601,
+            "bits_per_byte": 2.288362918282818
+        },
+        "gsarti/flores_101_ukr+null": {
+            "task_name": "gsarti/flores_101_ukr",
+            "prompt_name": "null",
+            "word_perplexity": 375312.1511987307,
+            "byte_perplexity": 2.7240934990288483,
+            "bits_per_byte": 1.445776221804572
+        },
+        "gsarti/flores_101_umb+null": {
+            "task_name": "gsarti/flores_101_umb",
+            "prompt_name": "null",
+            "word_perplexity": 286182026.84727985,
+            "byte_perplexity": 12.766915508610673,
+            "bits_per_byte": 3.6743381063848357
+        },
+        "gsarti/flores_101_urd+null": {
+            "task_name": "gsarti/flores_101_urd",
+            "prompt_name": "null",
+            "word_perplexity": 294.7473718166965,
+            "byte_perplexity": 1.9797467071381232,
+            "bits_per_byte": 0.9853158607436239
+        },
+        "gsarti/flores_101_uzb+null": {
+            "task_name": "gsarti/flores_101_uzb",
+            "prompt_name": "null",
+            "word_perplexity": 657971096.5030558,
+            "byte_perplexity": 12.002337637722146,
+            "bits_per_byte": 3.5852435148799184
+        },
+        "gsarti/flores_101_vie+null": {
+            "task_name": "gsarti/flores_101_vie",
+            "prompt_name": "null",
+            "word_perplexity": 30.113286809710246,
+            "byte_perplexity": 1.76578415476397,
+            "bits_per_byte": 0.8203090021691818
+        },
+        "gsarti/flores_101_cym+null": {
+            "task_name": "gsarti/flores_101_cym",
+            "prompt_name": "null",
+            "word_perplexity": 2638019.4579179045,
+            "byte_perplexity": 12.539424151448149,
+            "bits_per_byte": 3.6483991915978407
+        },
+        "gsarti/flores_101_wol+null": {
+            "task_name": "gsarti/flores_101_wol",
+            "prompt_name": "null",
+            "word_perplexity": 119795.78671768666,
+            "byte_perplexity": 9.144285650306488,
+            "bits_per_byte": 3.1928704713393357
+        },
+        "gsarti/flores_101_xho+null": {
+            "task_name": "gsarti/flores_101_xho",
+            "prompt_name": "null",
+            "word_perplexity": 54307092.21333007,
+            "byte_perplexity": 7.403240538286952,
+            "bits_per_byte": 2.8881569038733983
+        },
+        "gsarti/flores_101_yor+null": {
+            "task_name": "gsarti/flores_101_yor",
+            "prompt_name": "null",
+            "word_perplexity": 130267.12232132205,
+            "byte_perplexity": 5.91272037551173,
+            "bits_per_byte": 2.5638220507535796
+        },
+        "gsarti/flores_101_zul+null": {
+            "task_name": "gsarti/flores_101_zul",
+            "prompt_name": "null",
+            "word_perplexity": 493606524.8156374,
+            "byte_perplexity": 8.53353320693145,
+            "bits_per_byte": 3.0931431957905224
+        }
+    },
+    "config": {
+        "adaptive_seq_len": true,
+        "num_fewshot": 0,
+        "bootstrap_iters": 100000
+    }
+}

bigscience/evaluation/results/tr11/bloom2b5/bslmevalfiles/tr11c-2b5-ml-evalharness-results_lm-eval_global_step337250_2022-07-13-09-55-04.json

@@ -0,0 +1,172 @@
+{
+    "results": {
+        "arc_challenge": {
+            "acc": 0.27986348122866894,
+            "acc_stderr": 0.013119040897725922,
+            "acc_norm": 0.3054607508532423,
+            "acc_norm_stderr": 0.013460080478002498
+        },
+        "arc_easy": {
+            "acc": 0.5946969696969697,
+            "acc_stderr": 0.010074093589739182,
+            "acc_norm": 0.5324074074074074,
+            "acc_norm_stderr": 0.010238210368801902
+        },
+        "boolq": {
+            "acc": 0.6165137614678899,
+            "acc_stderr": 0.008504304838837027
+        },
+        "copa": {
+            "acc": 0.74,
+            "acc_stderr": 0.04408440022768078
+        },
+        "headqa": {
+            "acc": 0.26440554339897887,
+            "acc_stderr": 0.008423643607316284,
+            "acc_norm": 0.3099927060539752,
+            "acc_norm_stderr": 0.008833810133604958
+        },
+        "hellaswag": {
+            "acc": 0.41236805417247563,
+            "acc_stderr": 0.004912547040132878,
+            "acc_norm": 0.527185819557857,
+            "acc_norm_stderr": 0.0049824003689396615
+        },
+        "lambada": {
+            "ppl": 9.094305394880015,
+            "ppl_stderr": 0.2651922806718523,
+            "acc": 0.5181447700368718,
+            "acc_stderr": 0.0069613892910728266
+        },
+        "logiqa": {
+            "acc": 0.2073732718894009,
+            "acc_stderr": 0.015902084913876333,
+            "acc_norm": 0.29185867895545314,
+            "acc_norm_stderr": 0.017831570553971925
+        },
+        "mathqa": {
+            "acc": 0.24958123953098826,
+            "acc_stderr": 0.007922429819042544,
+            "acc_norm": 0.2492462311557789,
+            "acc_norm_stderr": 0.007918877981680667
+        },
+        "mc_taco": {
+            "em": 0.11936936936936937,
+            "f1": 0.4957122298258418
+        },
+        "mrpc": {
+            "acc": 0.5857843137254902,
+            "acc_stderr": 0.02441658575130785,
+            "f1": 0.6998223801065719,
+            "f1_stderr": 0.021967079752819446
+        },
+        "multirc": {
+            "acc": 0.012591815320041973,
+            "acc_stderr": 0.0036138827653638874
+        },
+        "openbookqa": {
+            "acc": 0.216,
+            "acc_stderr": 0.01842190906141194,
+            "acc_norm": 0.322,
+            "acc_norm_stderr": 0.020916668330019882
+        },
+        "piqa": {
+            "acc": 0.7078346028291621,
+            "acc_stderr": 0.010610252174513661,
+            "acc_norm": 0.705114254624592,
+            "acc_norm_stderr": 0.010639030620156982
+        },
+        "prost": {
+            "acc": 0.22683603757472245,
+            "acc_stderr": 0.003059602302050251,
+            "acc_norm": 0.26371690862510677,
+            "acc_norm_stderr": 0.003219323004106053
+        },
+        "pubmedqa": {
+            "acc": 0.616,
+            "acc_stderr": 0.01538768276189707
+        },
+        "qnli": {
+            "acc": 0.5072304594545122,
+            "acc_stderr": 0.006764703129634549
+        },
+        "qqp": {
+            "acc": 0.38211723967350975,
+            "acc_stderr": 0.0024166004681771985,
+            "f1": 0.5301408768597062,
+            "f1_stderr": 0.002619199330934276
+        },
+        "race": {
+            "acc": 0.3521531100478469,
+            "acc_stderr": 0.014782629897202264
+        },
+        "rte": {
+            "acc": 0.5631768953068592,
+            "acc_stderr": 0.029855247390314945
+        },
+        "sciq": {
+            "acc": 0.892,
+            "acc_stderr": 0.009820001651345703,
+            "acc_norm": 0.817,
+            "acc_norm_stderr": 0.012233587399477823
+        },
+        "sst": {
+            "acc": 0.49426605504587157,
+            "acc_stderr": 0.01694073961990489
+        },
+        "triviaqa": {
+            "acc": 0.041633518960487934,
+            "acc_stderr": 0.0018780954895624524
+        },
+        "webqs": {
+            "acc": 0.01673228346456693,
+            "acc_stderr": 0.0028461549169432184
+        },
+        "wic": {
+            "acc": 0.49843260188087773,
+            "acc_stderr": 0.019810623954060382
+        },
+        "winogrande": {
+            "acc": 0.5864246250986582,
+            "acc_stderr": 0.013840971763195303
+        },
+        "wnli": {
+            "acc": 0.4507042253521127,
+            "acc_stderr": 0.05947027187737998
+        },
+        "wsc": {
+            "acc": 0.375,
+            "acc_stderr": 0.04770204856076104
+        }
+    },
+    "versions": {
+        "arc_challenge": 0,
+        "arc_easy": 0,
+        "boolq": 1,
+        "copa": 0,
+        "headqa": 0,
+        "hellaswag": 0,
+        "lambada": 0,
+        "logiqa": 0,
+        "mathqa": 0,
+        "mc_taco": 0,
+        "mrpc": 0,
+        "multirc": 1,
+        "openbookqa": 0,
+        "piqa": 0,
+        "prost": 0,
+        "pubmedqa": 0,
+        "qnli": 0,
+        "qqp": 0,
+        "race": 1,
+        "rte": 0,
+        "sciq": 0,
+        "sst": 0,
+        "triviaqa": 0,
+        "webqs": 0,
+        "wic": 0,
+        "winogrande": 0,
+        "wnli": 1,
+        "wsc": 0
+    }
+}

bigscience/evaluation/results/tr11/bloom2b5/humaneval_temp06.json

@@ -0,0 +1 @@
+{"pass@1": 0.04460365853658537, "pass@10": 0.11354616672373204, "pass@100": 0.1866822927112951}

bigscience/evaluation/results/tr11/bloom2b5/mdtable.txt

@@ -0,0 +1,143 @@
+| Task | Language | Metric | BLOOM-2B5 |
+|:----|:----|:----|:----:|
+| arc_challenge | eng | acc ↑ | 0.28 |
+| arc_easy | eng | acc ↑ | 0.595 |
+| axb (Median of 10 prompts) | eng | acc ↑ | 0.443 |
+| axg (Median of 10 prompts) | eng | acc ↑ | 0.5 |
+| boolq (Median of 11 prompts) | eng | acc ↑ | 0.617 |
+| cb (Median of 15 prompts) | eng | acc ↑ | 0.304 |
+| cola (Median of 5 prompts) | eng | acc ↑ | 0.611 |
+| copa (Median of 9 prompts) | eng | acc ↑ | 0.63 |
+| crows_pairs_english (Median of 6 prompts) | eng | acc ↑ | 0.497 |
+| crows_pairs_french (Median of 7 prompts) | fra | acc ↑ | 0.503 |
+| diabla (Median of 2 prompts) | eng | acc ↑ | 0.289 |
+| gsarti/flores_101_afr | afr | byte_perplexity ↓ | 6.501 |
+| gsarti/flores_101_amh | amh | byte_perplexity ↓ | 3.973 |
+| gsarti/flores_101_ara | ara | byte_perplexity ↓ | 1.808 |
+| gsarti/flores_101_asm | asm | byte_perplexity ↓ | 5.699 |
+| gsarti/flores_101_ast | ast | byte_perplexity ↓ | 3.925 |
+| gsarti/flores_101_azj | azj | byte_perplexity ↓ | 6.943 |
+| gsarti/flores_101_bel | bel | byte_perplexity ↓ | 3.614 |
+| gsarti/flores_101_ben | ben | byte_perplexity ↓ | 5.121 |
+| gsarti/flores_101_bos | bos | byte_perplexity ↓ | 5.653 |
+| gsarti/flores_101_bul | bul | byte_perplexity ↓ | 2.701 |
+| gsarti/flores_101_cat | cat | byte_perplexity ↓ | 2.305 |
+| gsarti/flores_101_ceb | ceb | byte_perplexity ↓ | 6.291 |
+| gsarti/flores_101_ces | ces | byte_perplexity ↓ | 5.447 |
+| gsarti/flores_101_ckb | ckb | byte_perplexity ↓ | 3.726 |
+| gsarti/flores_101_cym | cym | byte_perplexity ↓ | 12.539 |
+| gsarti/flores_101_dan | dan | byte_perplexity ↓ | 5.183 |
+| gsarti/flores_101_deu | deu | byte_perplexity ↓ | 3.118 |
+| gsarti/flores_101_ell | ell | byte_perplexity ↓ | 2.468 |
+| gsarti/flores_101_eng | eng | byte_perplexity ↓ | 2.019 |
+| gsarti/flores_101_est | est | byte_perplexity ↓ | 9.117 |
+| gsarti/flores_101_fas | fas | byte_perplexity ↓ | 3.058 |
+| gsarti/flores_101_fin | fin | byte_perplexity ↓ | 6.847 |
+| gsarti/flores_101_fra | fra | byte_perplexity ↓ | 1.998 |
+| gsarti/flores_101_ful | ful | byte_perplexity ↓ | 11.466 |
+| gsarti/flores_101_gle | gle | byte_perplexity ↓ | 8.681 |
+| gsarti/flores_101_glg | glg | byte_perplexity ↓ | 3.03 |
+| gsarti/flores_101_guj | guj | byte_perplexity ↓ | 4.955 |
+| gsarti/flores_101_hau | hau | byte_perplexity ↓ | 10.758 |
+| gsarti/flores_101_heb | heb | byte_perplexity ↓ | 3.6 |
+| gsarti/flores_101_hin | hin | byte_perplexity ↓ | 4.713 |
+| gsarti/flores_101_hrv | hrv | byte_perplexity ↓ | 5.822 |
+| gsarti/flores_101_hun | hun | byte_perplexity ↓ | 6.44 |
+| gsarti/flores_101_hye | hye | byte_perplexity ↓ | 3.658 |
+| gsarti/flores_101_ibo | ibo | byte_perplexity ↓ | 5.565 |
+| gsarti/flores_101_ind | ind | byte_perplexity ↓ | 2.16 |
+| gsarti/flores_101_isl | isl | byte_perplexity ↓ | 8.082 |
+| gsarti/flores_101_ita | ita | byte_perplexity ↓ | 2.969 |
+| gsarti/flores_101_jav | jav | byte_perplexity ↓ | 7.057 |
+| gsarti/flores_101_jpn | jpn | byte_perplexity ↓ | 2.776 |
+| gsarti/flores_101_kam | kam | byte_perplexity ↓ | 11.073 |
+| gsarti/flores_101_kan | kan | byte_perplexity ↓ | 5.552 |
+| gsarti/flores_101_kat | kat | byte_perplexity ↓ | 2.523 |
+| gsarti/flores_101_kaz | kaz | byte_perplexity ↓ | 3.39 |
+| gsarti/flores_101_kea | kea | byte_perplexity ↓ | 8.919 |
+| gsarti/flores_101_kir | kir | byte_perplexity ↓ | 3.729 |
+| gsarti/flores_101_kor | kor | byte_perplexity ↓ | 3.933 |
+| gsarti/flores_101_lao | lao | byte_perplexity ↓ | 2.908 |
+| gsarti/flores_101_lav | lav | byte_perplexity ↓ | 7.777 |
+| gsarti/flores_101_lin | lin | byte_perplexity ↓ | 7.525 |
+| gsarti/flores_101_lit | lit | byte_perplexity ↓ | 7.369 |
+| gsarti/flores_101_ltz | ltz | byte_perplexity ↓ | 8.801 |
+| gsarti/flores_101_lug | lug | byte_perplexity ↓ | 8.483 |
+| gsarti/flores_101_luo | luo | byte_perplexity ↓ | 11.976 |
+| gsarti/flores_101_mal | mal | byte_perplexity ↓ | 4.616 |
+| gsarti/flores_101_mar | mar | byte_perplexity ↓ | 5.483 |
+| gsarti/flores_101_mkd | mkd | byte_perplexity ↓ | 2.966 |
+| gsarti/flores_101_mlt | mlt | byte_perplexity ↓ | 15.005 |
+| gsarti/flores_101_mon | mon | byte_perplexity ↓ | 3.411 |
+| gsarti/flores_101_mri | mri | byte_perplexity ↓ | 7.474 |
+| gsarti/flores_101_msa | msa | byte_perplexity ↓ | 2.571 |
+| gsarti/flores_101_mya | mya | byte_perplexity ↓ | 2.414 |
+| gsarti/flores_101_nld | nld | byte_perplexity ↓ | 4.128 |
+| gsarti/flores_101_nob | nob | byte_perplexity ↓ | 5.403 |
+| gsarti/flores_101_npi | npi | byte_perplexity ↓ | 5.199 |
+| gsarti/flores_101_nso | nso | byte_perplexity ↓ | 8.155 |
+| gsarti/flores_101_nya | nya | byte_perplexity ↓ | 8.18 |
+| gsarti/flores_101_oci | oci | byte_perplexity ↓ | 4.862 |
+| gsarti/flores_101_orm | orm | byte_perplexity ↓ | 12.912 |
+| gsarti/flores_101_ory | ory | byte_perplexity ↓ | 5.189 |
+| gsarti/flores_101_pan | pan | byte_perplexity ↓ | 4.698 |
+| gsarti/flores_101_pol | pol | byte_perplexity ↓ | 4.626 |
+| gsarti/flores_101_por | por | byte_perplexity ↓ | 1.975 |
+| gsarti/flores_101_pus | pus | byte_perplexity ↓ | 4.496 |
+| gsarti/flores_101_ron | ron | byte_perplexity ↓ | 4.965 |
+| gsarti/flores_101_rus | rus | byte_perplexity ↓ | 2.05 |
+| gsarti/flores_101_slk | slk | byte_perplexity ↓ | 6.451 |
+| gsarti/flores_101_slv | slv | byte_perplexity ↓ | 6.62 |
+| gsarti/flores_101_sna | sna | byte_perplexity ↓ | 8.462 |
+| gsarti/flores_101_snd | snd | byte_perplexity ↓ | 5.466 |
+| gsarti/flores_101_som | som | byte_perplexity ↓ | 11.959 |
+| gsarti/flores_101_spa | spa | byte_perplexity ↓ | 1.897 |
+| gsarti/flores_101_srp | srp | byte_perplexity ↓ | 2.871 |
+| gsarti/flores_101_swe | swe | byte_perplexity ↓ | 5.055 |
+| gsarti/flores_101_swh | swh | byte_perplexity ↓ | 3.697 |
+| gsarti/flores_101_tam | tam | byte_perplexity ↓ | 4.539 |
+| gsarti/flores_101_tel | tel | byte_perplexity ↓ | 5.807 |
+| gsarti/flores_101_tgk | tgk | byte_perplexity ↓ | 3.599 |
+| gsarti/flores_101_tgl | tgl | byte_perplexity ↓ | 5.667 |
+| gsarti/flores_101_tha | tha | byte_perplexity ↓ | 2.366 |
+| gsarti/flores_101_tur | tur | byte_perplexity ↓ | 4.885 |
+| gsarti/flores_101_ukr | ukr | byte_perplexity ↓ | 2.724 |
+| gsarti/flores_101_umb | umb | byte_perplexity ↓ | 12.767 |
+| gsarti/flores_101_urd | urd | byte_perplexity ↓ | 1.98 |
+| gsarti/flores_101_uzb | uzb | byte_perplexity ↓ | 12.002 |
+| gsarti/flores_101_vie | vie | byte_perplexity ↓ | 1.766 |
+| gsarti/flores_101_wol | wol | byte_perplexity ↓ | 9.144 |
+| gsarti/flores_101_xho | xho | byte_perplexity ↓ | 7.403 |
+| gsarti/flores_101_yor | yor | byte_perplexity ↓ | 5.913 |
+| gsarti/flores_101_zho_simpl | zho_simpl | byte_perplexity ↓ | 2.277 |
+| gsarti/flores_101_zho_trad | zho_trad | byte_perplexity ↓ | 2.518 |
+| gsarti/flores_101_zul | zul | byte_perplexity ↓ | 8.534 |
+| headqa | esp | acc ↑ | 0.264 |
+| hellaswag | eng | acc ↑ | 0.412 |
+| logiqa | eng | acc ↑ | 0.207 |
+| mathqa | eng | acc ↑ | 0.25 |
+| mc_taco | eng | em ↑ | 0.119 |
+| mnli (Median of 15 prompts) | eng | acc ↑ | 0.355 |
+| mnli_mismatched (Median of 15 prompts) | eng | acc ↑ | 0.352 |
+| mrpc | eng | acc ↑ | 0.586 |
+| multirc (Median of 11 prompts) | eng | acc ↑ | 0.538 |
+| openbookqa | eng | acc ↑ | 0.216 |
+| piqa | eng | acc ↑ | 0.708 |
+| prost | eng | acc ↑ | 0.227 |
+| pubmedqa | eng | acc ↑ | 0.616 |
+| qnli | eng | acc ↑ | 0.507 |
+| qqp (Median of 7 prompts) | eng | acc ↑ | 0.384 |
+| race | eng | acc ↑ | 0.352 |
+| rte (Median of 6 prompts) | eng | acc ↑ | 0.477 |
+| sciq | eng | acc ↑ | 0.892 |
+| sst (Median of 6 prompts) | eng | acc ↑ | 0.518 |
+| triviaqa | eng | acc ↑ | 0.042 |
+| tydiqa_primary (Median of 24 prompts) | eng | acc ↑ | 0.301 |
+| webqs | eng | acc ↑ | 0.017 |
+| wic (Median of 11 prompts) | eng | acc ↑ | 0.502 |
+| winogrande | eng | acc ↑ | 0.586 |
+| wnli (Median of 6 prompts) | eng | acc ↑ | 0.472 |
+| wsc (Median of 11 prompts) | eng | acc ↑ | 0.442 |
+| humaneval | python | pass@1 ↑ | 0.155 |
+| humaneval | python | pass@10 ↑ | 0.322 |
+| humaneval | python | pass@100 ↑ | 0.555 |

bigscience/evaluation/results/tr11/get_templates.sh

@@ -0,0 +1,27 @@
+DATASETS_AND_CONFIGS=(
+piaf,None,None
+GEM/wiki_lingua,ar,ar
+GEM/wiki_lingua,en,en
+GEM/wiki_lingua,es,es
+GEM/wiki_lingua,fr,fr
+GEM/wiki_lingua,hi,hi
+GEM/wiki_lingua,id,id
+GEM/wiki_lingua,pt,pt
+GEM/wiki_lingua,vi,vi
+GEM/wiki_lingua,zh,zh
+GEM/web_nlg,en,en
+GEM/web_nlg,ru,ru
+wmt14,fr-en,fr-en
+)
+
+# Unique ones: 0 1 2 5 6 7 8 9 10 11
+for val in {0..12}; do
+    DATASET_AND_CONFIG=${DATASETS_AND_CONFIGS[$val]}
+    IFS=',' read dataset_name dataset_config_name template_config_name <<< "${DATASET_AND_CONFIG}"
+    echo $dataset_config_name
+    python evaluation/results/tr13/tzeroeval/get_templates.py \
+            --dataset_name $dataset_name \
+            --dataset_config_name $dataset_config_name \
+            --template_config_name $template_config_name
+done
+

bigscience/evaluation/results/tr13/download_bslmeval.slurm

@@ -0,0 +1,37 @@
+#!/bin/bash
+#SBATCH --job-name=download-bslmeval
+#SBATCH --partition=prepost
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=10           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@cpu
+
+set -x -e
+
+echo "START TIME: $(date)"
+
+source $six_ALL_CCFRWORK/start-py38-pt111
+conda activate muennighofflmeval
+
+#export HF_DATASETS_OFFLINE=1
+#export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasetseval
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export TOKENIZERS_PARALLELISM=false
+
+cd /gpfsscratch/rech/six/commun/experiments/muennighoff/bloomckpt/lm-evaluation-harness
+
+#  GEM/web_nlg_en,GEM/web_nlg_en_challenge_test_numbers,GEM/web_nlg_en_challenge_test_scramble,GEM/web_nlg_en_challenge_validation_sample,GEM/web_nlg_ru,GEM/web_nlg_ru_challenge_test_scramble,GEM/web_nlg_ru_challenge_validation_sample,GEM/wiki_auto_asset_turk_challenge_test_asset_backtranslation,GEM/wiki_auto_asset_turk_challenge_test_asset_bfp02,GEM/wiki_auto_asset_turk_challenge_test_asset_bfp05,GEM/wiki_auto_asset_turk_challenge_test_asset_nopunc,GEM/wiki_auto_asset_turk_challenge_test_turk_backtranslation,GEM/wiki_auto_asset_turk_challenge_test_turk_bfp02,GEM/wiki_auto_asset_turk_challenge_test_turk_bfp05,GEM/wiki_auto_asset_turk_challenge_test_turk_nopunc,GEM/wiki_auto_asset_turk_test_asset,GEM/wiki_auto_asset_turk_test_turk,GEM/wiki_lingua_ar,GEM/wiki_lingua_cs,GEM/wiki_lingua_de,GEM/wiki_lingua_en,GEM/wiki_lingua_es,GEM/wiki_lingua_fr,GEM/wiki_lingua_hi,GEM/wiki_lingua_id,GEM/wiki_lingua_it,GEM/wiki_lingua_ja,GEM/wiki_lingua_ko,GEM/wiki_lingua_nl,GEM/wiki_lingua_pt,GEM/wiki_lingua_ru,GEM/wiki_lingua_th,GEM/wiki_lingua_tr,GEM/wiki_lingua_vi,GEM/wiki_lingua_zh,gem_xsum,gem_xsum_challenge_sample,gem_xsum_challenge_test_backtranslation,gem_xsum_challenge_test_bfp_02,gem_xsum_challenge_test_bfp_05,gem_xsum_challenge_test_covid,gem_xsum_challenge_test_nopunc \
+python3 main.py --model hf-causal \
+    --model_args pretrained=hf-internal-testing/tiny-random-gpt2,use_accelerate=True,tokenizer=hf-internal-testing/tiny-random-gpt2,dtype=float16 \
+    --tasks wmt14_fr_en,wmt19_ru_en,wmt19_zh_en \
+    --device cuda \
+    --limit 1 \
+    --no_cache \
+    --num_fewshot 0

bigscience/evaluation/results/tr13/results_to_csv.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+
+# this script converts results.json:
+#
+#   "results": {
+#     "arc_challenge": {
+#       "acc": 0.24232081911262798,
+#       "acc_stderr": 0.01252159329580012,
+#       "acc_norm": 0.2764505119453925,
+#       "acc_norm_stderr": 0.013069662474252425
+#     },
+#
+# into a format expected by a spreadsheet, which is:
+#
+#   task          metric   value    err
+#   arc_challenge acc      xxx      yyy
+#   arc_challenge acc_norm xxx      yyy
+#   arc_challenge f1       xxx      yyy
+#
+# usage:
+# report-to-csv.py results.json
+
+
+import sys
+import statistics
+import json
+import io
+import csv
+
+results_file = sys.argv[1]
+
+csv_file = results_file.replace("json", "csv")
+
+print(f"Converting {results_file} to {csv_file}")
+
+with io.open(results_file, 'r', encoding='utf-8') as f:
+    raw_results = json.load(f)
+
+results = {}
+for ds_name, v in sorted(raw_results.items()):
+    results[ds_name.split("/")[-1]] = v
+
+with io.open(csv_file, 'w', encoding='utf-8') as f:
+
+    writer = csv.writer(f)
+    writer.writerow(["dataset", "prompt", "metric", "value"])
+    medians = []
+    for ds_name, v in sorted(results.items()):
+        acc_scores, bleu_scores, rouge2_fmeasure = [], [], []
+        for prompt_name, res in sorted(v.items()):
+            # T0 Eval
+            if "evaluation" in res:
+                for metric, value in sorted(res["evaluation"].items()):
+                    writer.writerow([ds_name, prompt_name, metric, value])
+                    if metric == "accuracy":
+                        acc_scores.append(value)
+            # LM Eval Harness Generation
+            elif "bleu" in res:
+                # Make sure BLEU is 0-1 not 0-100
+                writer.writerow([ds_name, prompt_name, "bleu", res["bleu"] / 100])
+                bleu_scores.append(res["bleu"] / 100)
+
+        if acc_scores:
+            median = statistics.median(acc_scores)
+            medians.append(median)
+            writer.writerow([ds_name, "median", "accuracy", median])
+        elif bleu_scores:
+            median = statistics.median(bleu_scores)
+            medians.append(median)
+            writer.writerow([ds_name, "median", "bleu", median])
+    if medians:
+        writer.writerow(["multiple", "average", "multiple", statistics.mean(medians)])

bigscience/evaluation/results/tr3/README.md

@@ -0,0 +1 @@
+We're interested in understanding when zero shot capabilities appear.

bigscience/evaluation/results/tr3/switch_tokenizer_to_t5_for_tr3e.sh

@@ -0,0 +1,6 @@
+export GIT_LFS_SKIP_SMUDGE=1
+git clone https://huggingface.co/bigscience/tr3e-1B3-c4-checkpoints
+cd tr3e-1B3-c4-checkpoints
+$six_ALL_CCFRWORK/code/bigscience/tools/hub-sync.py --repo-path . --patterns '*bogus*'
+git branch -a | sort -V | perl -lne 'm|(global_step\d+)| && print qx[git checkout $1; perl -pi -e "s|\\"tokenizer_class\\": null|\\"tokenizer_class\\": \\"T5Tokenizer\\"|" config.json; git commit -m "Fix tokenizer_class to use T5 tokenizer" .; git push --set-upstream origin $1]'
+export GIT_LFS_SKIP_SMUDGE=0

bigscience/evaluation/utilities/download_all_models.py

@@ -0,0 +1,47 @@
+from argparse import ArgumentParser
+from multiprocessing import Pool
+
+from requests import HTTPError
+from transformers import AutoModel, AutoTokenizer
+
+def get_args():
+    parser = ArgumentParser()
+    # --experiments bigscience/tr3d-1B3-oscar-checkpoints,bigscience/tr3e-1B3-c4-checkpoints,bigscience/tr3m-1B3-pile-checkpoints
+    parser.add_argument('--experiments', type=lambda s: s.split(','), required=True, help='Experiments we want to download.')
+    # --steps 19500,28500,37500,48000,57000,66000,76500,85500,94500,105000,114000
+    parser.add_argument('--steps', type=lambda s: [int(item) for item in s.split(',')], required=True, help='Steps we should download the model checkpoints')
+    return parser.parse_args()
+
+def _load_model(pretrain:str, revision: str):
+    try:
+        AutoModel.from_pretrained(pretrain, revision=revision)
+        AutoTokenizer.from_pretrained(pretrain, revision=revision)
+        return f"Loaded: {{pretrain:{pretrain}, revision:{revision}}}"
+    except HTTPError:
+        return f"Failed to load: {{pretrain:{pretrain}, revision:{revision}}}"
+
+def load_model(kwargs):
+    return _load_model(**kwargs)
+
+def main():
+    args = get_args()
+    pretrains = args.experiments
+    steps = args.steps
+    revisions = [f"global_step{step}" for step in steps]
+
+    # with Pool(10) as pool:
+    #     results = pool.imap(
+    #         load_model,
+    #         [{"pretrain": pretrain, "revision": revision} for pretrain in pretrains for revision in revisions],
+    #         chunksize=1
+    #     )
+    #
+    #     for result in results:
+    #         print(result)
+
+
+    for kwargs in [{"pretrain": pretrain, "revision": revision} for pretrain in pretrains for revision in revisions]:
+        print(load_model(kwargs))
+
+if __name__ == "__main__":
+    main()

bigscience/evaluation/utilities/download_all_models.slurm

@@ -0,0 +1,26 @@
+#!/bin/bash
+#SBATCH --job-name=download_all_models
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=10           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --time 10:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=logs/%x.out           # output file name
+#SBATCH --account=six@gpu
+#SBATCH --partition=compil
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-prod
+conda activate thomas_lm_eval
+
+# TODO: replace with local fork of bigscience
+BIGSCIENCE_REPO=$WORK/code/big_science/bigscience/evaluation/results/tr3
+
+pushd $BIGSCIENCE_REPO
+
+# TODO: replace with experiment / steps
+EXPERIMENTS=bigscience/tr3d-1B3-oscar-checkpoints,bigscience/tr3e-1B3-c4-checkpoints,bigscience/tr3m-1B3-pile-checkpoints
+STEPS=$(python -c "print(\",\".join([str(i) for i in range(19500, 118500, 1500)]))")
+
+python download_all_models.py --experiments $EXPERIMENTS --steps $STEPS

bigscience/evaluation/utilities/export_results_through_training_to_wandb.py

@@ -0,0 +1,86 @@
+import os
+
+import numpy as np
+import wandb
+import json
+import argparse
+
+RANDOM_BASELINE={
+    "arc_challenge": 0.2502, # Source: https://arxiv.org/pdf/1803.05457.pdf table 6
+    "arc_easy": 0.2502, # Source: https://arxiv.org/pdf/1803.05457.pdf table 6
+    "boolq": 0.5,
+    "copa": 0.5,
+    "headqa_en": 0.25,
+    "hellaswag": 0.25,
+    "lambada": 0., # Safe to say that random models won't perform well at all.
+    "logiqa": 0.25,
+    "mathqa": (4360 * 1/ 5 - (4475 - 4360) * 1/ 4) / 4475,
+    "mrpc": 0.5,
+    "multirc": 0., # TODO: I couldn't figure it out
+    "openbookqa": 0.25,
+    "piqa": 0.5,
+    "prost": 0.25,
+    "pubmedqa": 1/3,
+    "qnli": 0.5,
+    "qqp": 0.5,
+    "race": 0.25, # Source: https://arxiv.org/pdf/1704.04683.pdf table 5
+    "rte": 0.5,
+    "sciq": 0.25,
+    "sst": 0.5,
+    "triviaqa": 0.,
+    "webqs": 0.,
+    "wic": 0.5,
+    "winogrande": 0.5,
+    "wnli": 0.5,
+    "wsc": 0.5
+}
+
+def normalise(score, task):
+    return (score - RANDOM_BASELINE[task]) / (1. - RANDOM_BASELINE[task])
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--input_files", type=lambda s: s.split(','), required=True)
+    parser.add_argument("--all_tasks", action="store_true")
+    parser.add_argument("--naive_average", action="store_true")
+    parser.add_argument("--acc_average", action="store_true")
+    parser.add_argument("--normalised_acc_average", action="store_true")
+    return parser.parse_args()
+
+def main():
+    args = parse_args()
+    for input_file in args.input_files:
+        assert os.path.basename(input_file).endswith("_agg.json")
+        experiment_name = os.path.basename(input_file).split("_agg.json")[0]
+        with open(input_file, "r") as fi:
+            experiment = json.load(fi)
+
+        results = experiment["results"]
+        tokens = experiment["tokens"]
+        run = wandb.init(project="bigscience-tr3-evaluation-through-training", entity="timerobber", name=experiment_name,
+                         reinit=True)
+        for i, n_tokens in enumerate(tokens):
+            all_values = []
+            acc_average = []
+            normalised_acc_average = []
+            for task, task_results in results.items():
+                values = None
+                for metric, values in task_results.items():
+                    if args.all_tasks:
+                        wandb.log({f"{task}_{metric}": values[i], "tokens": tokens[i]})
+                    if "stderr" not in metric and "ppl" not in metric:
+                        all_values.append(values[i])
+                        if metric == "acc":
+                            acc_average.append(values[i])
+                            normalised_acc_average.append(normalise(values[i], task))
+            if args.naive_average:
+                wandb.log({f"naive_average": np.mean(all_values), "tokens": tokens[i]})
+            if args.acc_average:
+                wandb.log({f"acc_average": np.mean(acc_average), "tokens": tokens[i]})
+            if args.normalised_acc_average:
+                wandb.log({f"normalised_acc_average": np.mean(normalised_acc_average), "tokens": tokens[i]})
+
+        run.finish()
+
+if __name__ == "__main__":
+    main()

bigscience/evaluation/utilities/find_checkpoints_at_token_intervals.py

@@ -0,0 +1,27 @@
+import datasets
+import json
+
+steps_vs_samples = datasets.load_dataset("csv", data_files="run-.-tag-steps-vs-samples_y=steps,x=samples.csv")["train"]
+
+slope = (steps_vs_samples[-1]["Step"] - steps_vs_samples[-2]["Step"]) / (
+        steps_vs_samples[-1]["Value"] - steps_vs_samples[-2]["Value"])
+offset = steps_vs_samples[-1]["Step"] - steps_vs_samples[-1]["Value"] * slope
+
+token_interval = 1e10
+step_interval = 1500
+tokens_per_sample = 2048
+token_count = token_interval
+
+output_checkpoints = []
+
+for item in steps_vs_samples:
+    if item["Step"] * tokens_per_sample > token_count:
+        token_count += token_interval
+        step = step_interval * (item['Value'] // step_interval)
+        tokens = tokens_per_sample * (slope * (step_interval * (item['Value'] // step_interval)) + offset)
+        print(f"step: {step}")
+        print(f"tokens at that step: {tokens}")
+        output_checkpoints.append({"step": step, "tokens": tokens})
+
+
+json.dump(output_checkpoints, open("steps_to_evaluate_with_tokens.json", "w"))

bigscience/evaluation/utilities/plot_all_eval.py

@@ -0,0 +1,45 @@
+import json
+import os
+from argparse import ArgumentParser
+
+from matplotlib import pyplot as plt
+
+
+def get_args():
+    parser = ArgumentParser()
+    parser.add_argument('--input-files', type=lambda s: s.split(','), required=True, help='Input files that hold all evaluation metrics')
+    return parser.parse_args()
+
+def main():
+    args = get_args()
+
+    plots = {} # {"{EVALUATION}_{METRIC}": plt.figure}
+    for input_file in args.input_files:
+        assert os.path.basename(input_file).endswith("_agg.json")
+        experiment_name = os.path.basename(input_file).split("_agg.json")[0]
+        with open(input_file, "r") as fi:
+            experiment = json.load(fi)
+
+        tokens = experiment["tokens"]
+        for evaluation_name, evaluation in experiment["results"].items():
+            for metric_name, metric in evaluation.items():
+                key = f"{evaluation_name}_{metric_name}"
+                if key[-7:] == "_stderr":
+                    continue
+
+                if key not in plots:
+                    plot = plt.figure(len(plots))
+                    plot = plot.add_subplot(1,1,1)
+                    plot.set_title(key)
+                    plots[key] = plot
+
+                plot = plots[key]
+
+                plot.plot(tokens, metric, label=experiment_name)
+
+    for plot in plots.values():
+        plot.legend()
+    plt.show()
+
+if __name__ == "__main__":
+    main()

bigscience/experiments/bandwidth/README.md

@@ -0,0 +1,117 @@
+# Bandwidth tests
+
+## Deepspeed benchmark
+
+https://gist.github.com/stas00/ec5e197b15e2e7aea0153f54d2f97c15
+
+Probably need to adjust `TRIALS` to a higher number to get the more realistic results (after the interconnects is saturated).
+
+Note: tried a larger number but got the same results.
+
+## Single node V100
+
+ssh into a desired node and then:
+```
+export NCCL_DEBUG=info
+python -m torch.distributed.launch --nproc_per_node=4 all_reduce_bench.py 2>&1 | tee n1_32gb_all_reduce_bench.txt
+```
+
+Results:
+- [16gb](./n1_16gb_all_reduce_bench.txt) - `algo throughput: 1329.4242 Gbps`
+- [32gb](./n1_32gb_all_reduce_bench.txt) - `algo throughput: 1323.6244 Gbps`
+
+Here we have NVLink gen 2
+
+https://en.wikipedia.org/wiki/NVLink
+```
+Nvidia GV100 | V100 SXM2 | NVLink 2.0 | 25 GT/s | 300 GByte/s
+```
+So the total is 300GB/s => 2400 Gb/s
+
+and the benchmark clocks 1360 Gb/s - slightly more than half of the max total.
+
+if this test is run a bit longer, it drops to 600 Gbps.
+
+
+
+## 16 nodes V100 32GB
+
+```
+export NCCL_DEBUG=info
+export MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+srun --jobid $SLURM_JOBID bash -c 'python -m torch.distributed.launch --nnodes 16 --nproc_per_node=4 --node_rank $SLURM_PROCID --master_addr $MASTER_ADDR --master_port 12345 all_reduce_bench.py'  2>&1 | tee n16_32gb_all_reduce_bench.txt
+```
+Results:
+
+- [32gp](./n16_32gb_all_reduce_bench.txt) - `algo throughput: 23.2939 to 55.0766 Gbps`
+
+If the test is run much longer it fluctuates between 44 and 57 Gbps.
+
+Currently we have an issue with nccl that doesn't fully utilize Intel OPA full bandwidth. Which is supposed to be 400Gbps max.
+
+
+## 4 nodes V100 32GB
+
+Here is a recent re-run - jan 2022:
+
+script: [all_reduce_bench-32gb-n4.slurm](./all_reduce_bench-32gb-n4.slurm)
+
+
+```
+sbatch all_reduce_bench-32gb-n4.slurm
+```
+
+Results:
+
+[all_reduce_bench-32gb-n4.txt](./all_reduce_bench-32gb-n4.txt) - `algo throughput: 30 to 90 Gbps`
+
+
+
+
+## 4 nodes A100 80GB
+
+
+script: [all_reduce_bench-a100-n4.slurm](./all_reduce_bench-a100-n4.slurm)
+
+
+```
+sbatch all_reduce_bench-a100-n4.slurm
+```
+
+Results:
+
+[all_reduce_bench-a100-n4.txt](./all_reduce_bench-a100-n4.txt) - `algo throughput: 15 to 42 Gbps`
+
+As a reference Azure has [ND A100 v4-series](https://docs.microsoft.com/en-us/azure/virtual-machines/nda100-v4-series) w/ 1.6 Tb/s of interconnect bandwidth per VM. And Jeff Rasley clocked ~1.5Tb/s with this `all_reduce_bench` script.
+
+
+## NCCL tests
+
+https://github.com/nvidia/nccl-tests
+
+The details are explained here:
+
+https://github.com/NVIDIA/nccl-tests/blob/master/doc/PERFORMANCE.md
+
+```
+git clone https://github.com/nvidia/nccl-tests
+cd nccl-tests
+make
+```
+
+
+## Single node
+
+ssh into a desired node and then:
+```
+./build/all_reduce_perf -b 8 -e 128M -f 2 -g 4
+```
+
+
+## 16 nodes
+
+from master node:
+```
+srun --jobid $SLURM_JOBID ./build/all_reduce_perf -b 8 -e 128M -f 2 -g 4
+```
+(not sure if I did it right - didn't have time to read the docs)

bigscience/experiments/bandwidth/all_reduce_bench-32gb-n4.slurm

@@ -0,0 +1,20 @@
+#!/bin/bash
+#SBATCH --job-name=all_reduce_bench-32gb-n4
+#SBATCH --constraint=v100-32g
+#SBATCH --nodes=4
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=40           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:4                 # number of gpus
+#SBATCH --time 0:05:00               # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@gpu
+
+export LOG_FILE=all_reduce_bench-32gb-n4.txt
+export NNODES=4
+export GPUS_PER_NODE=4
+export NCCL_DEBUG=info
+
+export MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+
+srun --jobid $SLURM_JOBID bash -c 'python -m torch.distributed.launch --nnodes $NNODES --nproc_per_node $GPUS_PER_NODE --node_rank $SLURM_PROCID --master_addr $MASTER_ADDR --master_port 12345 all_reduce_bench.py'  2>&1 | tee $LOG_FILE

bigscience/experiments/bandwidth/all_reduce_bench-32gb-n4.txt

@@ -0,0 +1,850 @@
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+local_rank: 0
+local_rank: 0
+local_rank: 2
+local_rank: 1
+local_rank: 1
+local_rank: 3
+local_rank: 3
+local_rank: 2
+local_rank: 1
+local_rank: 3
+local_rank: 2
+local_rank: 0
+local_rank: 0
+local_rank: 1
+local_rank: 3
+local_rank: 2
+0 data size: 4.0 GB
+4 data size: 4.0 GB
+6 data size: 4.0 GB
+11 data size: 4.0 GB
+3 data size: 4.0 GB
+10 data size: 4.0 GB
+7 data size: 4.0 GB
+5 data size: 4.0 GB
+1 data size: 4.0 GB
+2 data size: 4.0 GB
+8 data size: 4.0 GB
+15 data size: 4.0 GB
+13 data size: 4.0 GB
+12 data size: 4.0 GB
+14 data size: 4.0 GB
+9 data size: 4.0 GB
+r6i6n4:257714:257714 [0] NCCL INFO Bootstrap : Using ib0:10.148.7.175<0>
+r6i6n4:257714:257714 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n4:257714:257714 [0] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.175<0>
+r6i6n4:257714:257714 [0] NCCL INFO Using network IB
+NCCL version 2.10.3+cuda11.3
+r6i6n4:257715:257715 [1] NCCL INFO Bootstrap : Using ib0:10.148.7.175<0>
+r6i6n5:378203:378203 [3] NCCL INFO Bootstrap : Using ib0:10.148.7.176<0>
+r6i6n5:378202:378202 [2] NCCL INFO Bootstrap : Using ib0:10.148.7.176<0>
+r6i6n5:378201:378201 [1] NCCL INFO Bootstrap : Using ib0:10.148.7.176<0>
+r6i6n4:257715:257715 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n5:378202:378202 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n5:378203:378203 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n5:378201:378201 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n4:257715:257715 [1] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.175<0>
+r6i6n4:257715:257715 [1] NCCL INFO Using network IB
+r6i6n5:378203:378203 [3] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.176<0>
+r6i6n5:378202:378202 [2] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.176<0>
+r6i6n5:378203:378203 [3] NCCL INFO Using network IB
+r6i6n5:378202:378202 [2] NCCL INFO Using network IB
+r6i6n5:378201:378201 [1] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.176<0>
+r6i6n5:378201:378201 [1] NCCL INFO Using network IB
+r6i6n4:257717:257717 [3] NCCL INFO Bootstrap : Using ib0:10.148.7.175<0>
+r6i6n4:257717:257717 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n4:257717:257717 [3] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.175<0>
+r6i6n4:257717:257717 [3] NCCL INFO Using network IB
+r6i6n4:257716:257716 [2] NCCL INFO Bootstrap : Using ib0:10.148.7.175<0>
+r6i6n4:257716:257716 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n4:257716:257716 [2] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.175<0>
+r6i6n4:257716:257716 [2] NCCL INFO Using network IB
+r6i6n5:378200:378200 [0] NCCL INFO Bootstrap : Using ib0:10.148.7.176<0>
+r6i6n5:378200:378200 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r6i6n5:378200:378200 [0] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.7.176<0>
+r6i6n5:378200:378200 [0] NCCL INFO Using network IB
+r7i6n2:1370349:1370349 [2] NCCL INFO Bootstrap : Using ib0:10.148.0.95<0>
+r7i6n2:1370348:1370348 [1] NCCL INFO Bootstrap : Using ib0:10.148.0.95<0>
+r7i6n2:1370348:1370348 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n2:1370349:1370349 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n2:1370349:1370349 [2] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.95<0>
+r7i6n2:1370349:1370349 [2] NCCL INFO Using network IB
+r7i6n2:1370348:1370348 [1] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.95<0>
+r7i6n2:1370348:1370348 [1] NCCL INFO Using network IB
+r7i6n3:610063:610063 [0] NCCL INFO Bootstrap : Using ib0:10.148.0.96<0>
+r7i6n3:610066:610066 [3] NCCL INFO Bootstrap : Using ib0:10.148.0.96<0>
+r7i6n3:610063:610063 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n3:610066:610066 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n3:610066:610066 [3] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.96<0>
+r7i6n3:610066:610066 [3] NCCL INFO Using network IB
+r7i6n3:610063:610063 [0] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.96<0>
+r7i6n3:610063:610063 [0] NCCL INFO Using network IB
+r7i6n2:1370347:1370347 [0] NCCL INFO Bootstrap : Using ib0:10.148.0.95<0>
+r7i6n2:1370347:1370347 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n2:1370347:1370347 [0] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.95<0>
+r7i6n2:1370347:1370347 [0] NCCL INFO Using network IB
+r7i6n3:610065:610065 [2] NCCL INFO Bootstrap : Using ib0:10.148.0.96<0>
+r7i6n3:610065:610065 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n3:610064:610064 [1] NCCL INFO Bootstrap : Using ib0:10.148.0.96<0>
+r7i6n3:610065:610065 [2] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.96<0>
+r7i6n3:610065:610065 [2] NCCL INFO Using network IB
+r7i6n3:610064:610064 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n3:610064:610064 [1] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.96<0>
+r7i6n3:610064:610064 [1] NCCL INFO Using network IB
+r7i6n2:1370350:1370350 [3] NCCL INFO Bootstrap : Using ib0:10.148.0.95<0>
+r7i6n2:1370350:1370350 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i6n2:1370350:1370350 [3] NCCL INFO NET/IB : Using [0]hfi1_0:1/IB [1]hfi1_1:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.0.95<0>
+r7i6n2:1370350:1370350 [3] NCCL INFO Using network IB
+r7i6n3:610064:610133 [1] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] -1/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] -1/-1/-1->13->12
+r7i6n3:610065:610128 [2] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->10 [2] 15/-1/-1->14->13 [3] 15/6/-1->14->-1
+r7i6n3:610066:610123 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 12/-1/-1->15->14 [2] -1/-1/-1->15->14 [3] 12/-1/-1->15->14
+r6i6n4:257714:257762 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+r6i6n4:257714:257762 [0] NCCL INFO Channel 01/04 :    0   3   2   1   4   7   6   5   8  11  10   9  12  15  14  13
+r6i6n4:257714:257762 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+r6i6n4:257714:257762 [0] NCCL INFO Channel 03/04 :    0   3   2   1   4   7   6   5   8  11  10   9  12  15  14  13
+r6i6n4:257714:257762 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/-1/-1->0->3 [2] 1/-1/-1->0->4 [3] 1/-1/-1->0->3
+r6i6n4:257716:257777 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/10/-1->2->-1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->6
+r6i6n4:257715:257767 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] -1/-1/-1->1->0
+r7i6n3:610063:610122 [0] NCCL INFO Trees [0] 13/-1/-1->12->8 [1] 13/-1/-1->12->15 [2] 13/4/-1->12->-1 [3] 13/-1/-1->12->15
+r6i6n4:257714:257762 [0] NCCL INFO Setting affinity for GPU 0 to 0fffff
+r6i6n4:257715:257767 [1] NCCL INFO Setting affinity for GPU 1 to 0fffff
+r6i6n4:257717:257772 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] 0/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] 0/-1/-1->3->2
+r6i6n4:257716:257777 [2] NCCL INFO Setting affinity for GPU 2 to ff,fff00000
+r7i6n2:1370350:1370452 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] 8/6/-1->11->10 [2] -1/-1/-1->11->10 [3] 8/-1/-1->11->10
+r7i6n2:1370349:1370433 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/14/-1->10->2 [2] 11/-1/-1->10->9 [3] 11/-1/-1->10->7
+r6i6n4:257717:257772 [3] NCCL INFO Setting affinity for GPU 3 to ff,fff00000
+r7i6n2:1370347:1370447 [0] NCCL INFO Trees [0] 9/12/-1->8->0 [1] 9/-1/-1->8->11 [2] 9/-1/-1->8->5 [3] 9/-1/-1->8->11
+r7i6n2:1370348:1370434 [1] NCCL INFO Trees [0] 10/4/-1->9->8 [1] -1/-1/-1->9->8 [2] 10/-1/-1->9->8 [3] -1/-1/-1->9->8
+r6i6n5:378202:378256 [2] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->11 [2] 7/-1/-1->6->5 [3] 7/2/-1->6->14
+r6i6n5:378203:378255 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 4/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] 4/10/-1->7->6
+r6i6n5:378201:378257 [1] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] -1/-1/-1->5->4 [2] 6/8/-1->5->4 [3] -1/-1/-1->5->4
+r6i6n5:378200:378262 [0] NCCL INFO Trees [0] 5/-1/-1->4->9 [1] 5/-1/-1->4->7 [2] 5/0/-1->4->12 [3] 5/-1/-1->4->7
+r6i6n5:378202:378256 [2] NCCL INFO Setting affinity for GPU 2 to ff,fff00000
+r6i6n5:378203:378255 [3] NCCL INFO Setting affinity for GPU 3 to ff,fff00000
+r6i6n5:378201:378257 [1] NCCL INFO Setting affinity for GPU 1 to 0fffff
+r6i6n5:378200:378262 [0] NCCL INFO Setting affinity for GPU 0 to 0fffff
+r7i6n3:610064:610133 [1] NCCL INFO Setting affinity for GPU 1 to 0fffff
+r7i6n3:610065:610128 [2] NCCL INFO Setting affinity for GPU 2 to ff,fff00000
+r7i6n3:610066:610123 [3] NCCL INFO Setting affinity for GPU 3 to ff,fff00000
+r7i6n3:610063:610122 [0] NCCL INFO Setting affinity for GPU 0 to 0fffff
+r7i6n2:1370350:1370452 [3] NCCL INFO Setting affinity for GPU 3 to ff,fff00000
+r7i6n2:1370349:1370433 [2] NCCL INFO Setting affinity for GPU 2 to ff,fff00000
+r7i6n2:1370347:1370447 [0] NCCL INFO Setting affinity for GPU 0 to 0fffff
+r7i6n2:1370348:1370434 [1] NCCL INFO Setting affinity for GPU 1 to 0fffff
+r7i6n3:610064:610133 [1] NCCL INFO Channel 00 : 13[1c000] -> 14[88000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Channel 00 : 5[1c000] -> 6[88000] via P2P/IPC
+r7i6n3:610064:610133 [1] NCCL INFO Channel 02 : 13[1c000] -> 14[88000] via P2P/IPC
+r6i6n4:257715:257767 [1] NCCL INFO Channel 00 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 00 : 14[88000] -> 15[8a000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 00 : 2[88000] -> 3[8a000] via P2P/IPC
+r6i6n4:257715:257767 [1] NCCL INFO Channel 02 : 1[1c000] -> 2[88000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Channel 02 : 5[1c000] -> 6[88000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 00 : 6[88000] -> 7[8a000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 02 : 14[88000] -> 15[8a000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 02 : 2[88000] -> 3[8a000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 02 : 6[88000] -> 7[8a000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Channel 00 : 15[8a000] -> 0[1a000] [receive] via NET/IB/1
+r6i6n5:378200:378262 [0] NCCL INFO Channel 00 : 3[8a000] -> 4[1a000] [receive] via NET/IB/1
+r7i6n3:610063:610122 [0] NCCL INFO Channel 00 : 11[8a000] -> 12[1a000] [receive] via NET/IB/1
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 00 : 9[1c000] -> 10[88000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 00 : 10[88000] -> 11[8a000] via P2P/IPC
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 02 : 9[1c000] -> 10[88000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 02 : 10[88000] -> 11[8a000] via P2P/IPC
+r6i6n4:257717:257772 [3] NCCL INFO Channel 00 : 3[8a000] -> 4[1a000] [send] via NET/IB/3
+r7i6n3:610066:610123 [3] NCCL INFO Channel 00 : 15[8a000] -> 0[1a000] [send] via NET/IB/3
+r6i6n5:378203:378255 [3] NCCL INFO Channel 00 : 7[8a000] -> 8[1a000] [send] via NET/IB/3
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 00 : 7[8a000] -> 8[1a000] [receive] via NET/IB/1
+r6i6n4:257714:257762 [0] NCCL INFO Channel 02 : 15[8a000] -> 0[1a000] [receive] via NET/IB/1
+r6i6n5:378200:378262 [0] NCCL INFO Channel 02 : 3[8a000] -> 4[1a000] [receive] via NET/IB/1
+r7i6n3:610063:610122 [0] NCCL INFO Channel 02 : 11[8a000] -> 12[1a000] [receive] via NET/IB/1
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 00 : 11[8a000] -> 12[1a000] [send] via NET/IB/3
+r6i6n4:257717:257772 [3] NCCL INFO Channel 02 : 3[8a000] -> 4[1a000] [send] via NET/IB/3
+r7i6n3:610066:610123 [3] NCCL INFO Channel 02 : 15[8a000] -> 0[1a000] [send] via NET/IB/3
+r6i6n4:257716:257777 [2] NCCL INFO Channel 01 : 2[88000] -> 1[1c000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 03 : 2[88000] -> 1[1c000] via P2P/IPC
+r6i6n5:378203:378255 [3] NCCL INFO Channel 02 : 7[8a000] -> 8[1a000] [send] via NET/IB/3
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 02 : 7[8a000] -> 8[1a000] [receive] via NET/IB/1
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 02 : 11[8a000] -> 12[1a000] [send] via NET/IB/3
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 01 : 10[88000] -> 9[1c000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 03 : 10[88000] -> 9[1c000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 01 : 6[88000] -> 5[1c000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 03 : 6[88000] -> 5[1c000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Channel 00 : 12[1a000] -> 13[1c000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Channel 02 : 12[1a000] -> 13[1c000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 01 : 14[88000] -> 13[1c000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 03 : 14[88000] -> 13[1c000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 00 : 4[1a000] -> 5[1c000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 02 : 4[1a000] -> 5[1c000] via P2P/IPC
+r7i6n3:610064:610133 [1] NCCL INFO Channel 01 : 13[1c000] -> 0[1a000] [send] via NET/IB/1
+r6i6n4:257714:257762 [0] NCCL INFO Channel 00 : 0[1a000] -> 1[1c000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Channel 02 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Channel 01 : 9[1c000] -> 12[1a000] [receive] via NET/IB/1
+r7i6n3:610064:610133 [1] NCCL INFO Channel 03 : 13[1c000] -> 0[1a000] [send] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 00 : 8[1a000] -> 9[1c000] via P2P/IPC
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 02 : 8[1a000] -> 9[1c000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 01 : 1[1c000] -> 4[1a000] [receive] via NET/IB/1
+r6i6n4:257714:257762 [0] NCCL INFO Channel 01 : 13[1c000] -> 0[1a000] [receive] via NET/IB/1
+r6i6n4:257715:257767 [1] NCCL INFO Channel 01 : 1[1c000] -> 4[1a000] [send] via NET/IB/1
+r7i6n3:610063:610122 [0] NCCL INFO Channel 03 : 9[1c000] -> 12[1a000] [receive] via NET/IB/1
+r7i6n3:610063:610122 [0] NCCL INFO Channel 01 : 12[1a000] -> 15[8a000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Channel 03 : 12[1a000] -> 15[8a000] via P2P/IPC
+r7i6n3:610066:610123 [3] NCCL INFO Channel 01 : 15[8a000] -> 14[88000] via P2P/IPC
+r7i6n3:610066:610123 [3] NCCL INFO Channel 03 : 15[8a000] -> 14[88000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 03 : 1[1c000] -> 4[1a000] [receive] via NET/IB/1
+r7i6n3:610066:610123 [3] NCCL INFO Connected all rings
+r6i6n5:378200:378262 [0] NCCL INFO Channel 01 : 4[1a000] -> 7[8a000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 03 : 4[1a000] -> 7[8a000] via P2P/IPC
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 01 : 5[1c000] -> 8[1a000] [receive] via NET/IB/1
+r6i6n4:257714:257762 [0] NCCL INFO Channel 03 : 13[1c000] -> 0[1a000] [receive] via NET/IB/1
+r6i6n5:378203:378255 [3] NCCL INFO Channel 01 : 7[8a000] -> 6[88000] via P2P/IPC
+r6i6n4:257715:257767 [1] NCCL INFO Channel 03 : 1[1c000] -> 4[1a000] [send] via NET/IB/1
+r6i6n4:257714:257762 [0] NCCL INFO Channel 01 : 0[1a000] -> 3[8a000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Channel 03 : 0[1a000] -> 3[8a000] via P2P/IPC
+r6i6n5:378203:378255 [3] NCCL INFO Channel 03 : 7[8a000] -> 6[88000] via P2P/IPC
+r6i6n5:378203:378255 [3] NCCL INFO Connected all rings
+r6i6n4:257717:257772 [3] NCCL INFO Channel 01 : 3[8a000] -> 2[88000] via P2P/IPC
+r6i6n4:257717:257772 [3] NCCL INFO Channel 03 : 3[8a000] -> 2[88000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Connected all rings
+r6i6n5:378200:378262 [0] NCCL INFO Channel 01 : 4[1a000] -> 5[1c000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 03 : 4[1a000] -> 5[1c000] via P2P/IPC
+r6i6n4:257715:257767 [1] NCCL INFO Connected all rings
+r6i6n4:257717:257772 [3] NCCL INFO Connected all rings
+r6i6n4:257716:257777 [2] NCCL INFO Connected all rings
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 03 : 5[1c000] -> 8[1a000] [receive] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 01 : 8[1a000] -> 11[8a000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 01 : 2[88000] -> 3[8a000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 03 : 2[88000] -> 3[8a000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Channel 01 : 5[1c000] -> 8[1a000] [send] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 03 : 8[1a000] -> 11[8a000] via P2P/IPC
+r6i6n4:257717:257772 [3] NCCL INFO Channel 01 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 01 : 9[1c000] -> 12[1a000] [send] via NET/IB/1
+r6i6n4:257717:257772 [3] NCCL INFO Channel 03 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 01 : 11[8a000] -> 10[88000] via P2P/IPC
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 03 : 11[8a000] -> 10[88000] via P2P/IPC
+r7i6n2:1370350:1370452 [3] NCCL INFO Connected all rings
+r6i6n5:378201:378257 [1] NCCL INFO Channel 03 : 5[1c000] -> 8[1a000] [send] via NET/IB/1
+r6i6n4:257716:257777 [2] NCCL INFO Channel 03 : 2[88000] -> 6[88000] [send] via NET/IB/3
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 03 : 9[1c000] -> 12[1a000] [send] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Connected all rings
+r6i6n4:257714:257762 [0] NCCL INFO Connected all rings
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 01 : 8[1a000] -> 9[1c000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Channel 01 : 0[1a000] -> 1[1c000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Connected all rings
+r6i6n4:257714:257762 [0] NCCL INFO Channel 03 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Connected all rings
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 03 : 8[1a000] -> 9[1c000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Connected all rings
+r7i6n3:610064:610133 [1] NCCL INFO Connected all rings
+r6i6n5:378202:378256 [2] NCCL INFO Channel 01 : 6[88000] -> 7[8a000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Channel 01 : 12[1a000] -> 13[1c000] via P2P/IPC
+r7i6n2:1370348:1370434 [1] NCCL INFO Connected all rings
+r6i6n4:257715:257767 [1] NCCL INFO Channel 00 : 1[1c000] -> 0[1a000] via P2P/IPC
+r6i6n4:257715:257767 [1] NCCL INFO Channel 01 : 1[1c000] -> 0[1a000] via P2P/IPC
+r6i6n4:257715:257767 [1] NCCL INFO Channel 02 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i6n3:610063:610122 [0] NCCL INFO Channel 03 : 12[1a000] -> 13[1c000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 03 : 6[88000] -> 7[8a000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Connected all rings
+r7i6n2:1370349:1370433 [2] NCCL INFO Connected all rings
+r6i6n4:257715:257767 [1] NCCL INFO Channel 03 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 01 : 14[88000] -> 15[8a000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 01 : 10[88000] -> 11[8a000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 03 : 14[88000] -> 15[8a000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 03 : 10[88000] -> 11[8a000] via P2P/IPC
+r7i6n3:610066:610123 [3] NCCL INFO Channel 01 : 15[8a000] -> 12[1a000] via P2P/IPC
+r7i6n3:610064:610133 [1] NCCL INFO Channel 00 : 13[1c000] -> 12[1a000] via P2P/IPC
+r7i6n3:610066:610123 [3] NCCL INFO Channel 03 : 15[8a000] -> 12[1a000] via P2P/IPC
+r7i6n3:610064:610133 [1] NCCL INFO Channel 01 : 13[1c000] -> 12[1a000] via P2P/IPC
+r7i6n3:610064:610133 [1] NCCL INFO Channel 02 : 13[1c000] -> 12[1a000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Channel 02 : 0[1a000] -> 4[1a000] [send] via NET/IB/1
+r7i6n3:610064:610133 [1] NCCL INFO Channel 03 : 13[1c000] -> 12[1a000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 02 : 0[1a000] -> 4[1a000] [receive] via NET/IB/1
+r6i6n5:378201:378257 [1] NCCL INFO Channel 02 : 5[1c000] -> 8[1a000] [send] via NET/IB/1
+r6i6n5:378203:378255 [3] NCCL INFO Channel 03 : 7[8a000] -> 10[88000] [send] via NET/IB/3
+r7i6n3:610063:610122 [0] NCCL INFO Channel 00 : 8[1a000] -> 12[1a000] [receive] via NET/IB/1
+r6i6n5:378202:378256 [2] NCCL INFO Channel 03 : 2[88000] -> 6[88000] [receive] via NET/IB/3
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 02 : 5[1c000] -> 8[1a000] [receive] via NET/IB/1
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 00 : 4[1a000] -> 9[1c000] [receive] via NET/IB/1
+r7i6n3:610065:610128 [2] NCCL INFO Channel 01 : 10[88000] -> 14[88000] [receive] via NET/IB/3
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 01 : 6[88000] -> 11[8a000] [receive] via NET/IB/3
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 03 : 7[8a000] -> 10[88000] [receive] via NET/IB/3
+r6i6n4:257714:257762 [0] NCCL INFO Channel 00 : 8[1a000] -> 0[1a000] [receive] via NET/IB/1
+r6i6n5:378200:378262 [0] NCCL INFO Channel 00 : 4[1a000] -> 9[1c000] [send] via NET/IB/1
+r6i6n4:257716:257777 [2] NCCL INFO Channel 01 : 10[88000] -> 2[88000] [receive] via NET/IB/3
+r6i6n5:378202:378256 [2] NCCL INFO Channel 01 : 6[88000] -> 11[8a000] [send] via NET/IB/3
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 00 : 8[1a000] -> 12[1a000] [send] via NET/IB/1
+r6i6n5:378201:378257 [1] NCCL INFO Channel 02 : 8[1a000] -> 5[1c000] [receive] via NET/IB/1
+r6i6n5:378203:378255 [3] NCCL INFO Channel 03 : 10[88000] -> 7[8a000] [receive] via NET/IB/3
+r6i6n5:378203:378255 [3] NCCL INFO Channel 01 : 7[8a000] -> 4[1a000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 01 : 10[88000] -> 14[88000] [send] via NET/IB/3
+r6i6n5:378203:378255 [3] NCCL INFO Channel 03 : 7[8a000] -> 4[1a000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Channel 00 : 0[1a000] -> 8[1a000] [send] via NET/IB/1
+r6i6n4:257716:257777 [2] NCCL INFO Channel 01 : 2[88000] -> 10[88000] [send] via NET/IB/3
+r6i6n5:378200:378262 [0] NCCL INFO Channel 02 : 12[1a000] -> 4[1a000] [receive] via NET/IB/1
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 00 : 9[1c000] -> 4[1a000] [send] via NET/IB/1
+r6i6n5:378202:378256 [2] NCCL INFO Channel 03 : 14[88000] -> 6[88000] [receive] via NET/IB/3
+r7i6n3:610063:610122 [0] NCCL INFO Channel 02 : 4[1a000] -> 12[1a000] [receive] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 00 : 0[1a000] -> 8[1a000] [receive] via NET/IB/1
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 01 : 11[8a000] -> 6[88000] [send] via NET/IB/3
+r7i6n3:610065:610128 [2] NCCL INFO Channel 03 : 6[88000] -> 14[88000] [receive] via NET/IB/3
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 01 : 2[88000] -> 10[88000] [receive] via NET/IB/3
+r6i6n5:378200:378262 [0] NCCL INFO Channel 02 : 4[1a000] -> 12[1a000] [send] via NET/IB/1
+r6i6n5:378202:378256 [2] NCCL INFO Channel 03 : 6[88000] -> 14[88000] [send] via NET/IB/3
+r7i6n3:610063:610122 [0] NCCL INFO Channel 02 : 12[1a000] -> 4[1a000] [send] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 00 : 8[1a000] -> 0[1a000] [send] via NET/IB/1
+r7i6n3:610065:610128 [2] NCCL INFO Channel 03 : 14[88000] -> 6[88000] [send] via NET/IB/3
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 01 : 10[88000] -> 2[88000] [send] via NET/IB/3
+r6i6n5:378200:378262 [0] NCCL INFO Channel 00 : 9[1c000] -> 4[1a000] [receive] via NET/IB/1
+r7i6n3:610063:610122 [0] NCCL INFO Channel 00 : 12[1a000] -> 8[1a000] [send] via NET/IB/1
+r6i6n4:257714:257762 [0] NCCL INFO Channel 02 : 4[1a000] -> 0[1a000] [receive] via NET/IB/1
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 00 : 12[1a000] -> 8[1a000] [receive] via NET/IB/1
+r7i6n3:610065:610128 [2] NCCL INFO Channel 01 : 14[88000] -> 10[88000] [send] via NET/IB/3
+r6i6n5:378202:378256 [2] NCCL INFO Channel 01 : 11[8a000] -> 6[88000] [receive] via NET/IB/3
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 00 : 9[1c000] -> 8[1a000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 03 : 6[88000] -> 2[88000] [receive] via NET/IB/3
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 01 : 9[1c000] -> 8[1a000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 01 : 14[88000] -> 10[88000] [receive] via NET/IB/3
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 01 : 11[8a000] -> 8[1a000] via P2P/IPC
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 02 : 9[1c000] -> 8[1a000] via P2P/IPC
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 03 : 11[8a000] -> 8[1a000] via P2P/IPC
+r7i6n3:610066:610123 [3] NCCL INFO Channel 00 : 15[8a000] -> 14[88000] via P2P/IPC
+r7i6n2:1370348:1370434 [1] NCCL INFO Channel 03 : 9[1c000] -> 8[1a000] via P2P/IPC
+r7i6n3:610066:610123 [3] NCCL INFO Channel 02 : 15[8a000] -> 14[88000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 00 : 14[88000] -> 13[1c000] via P2P/IPC
+r7i6n3:610065:610128 [2] NCCL INFO Channel 02 : 14[88000] -> 13[1c000] via P2P/IPC
+r6i6n5:378200:378262 [0] NCCL INFO Channel 02 : 4[1a000] -> 0[1a000] [send] via NET/IB/1
+r7i6n3:610063:610122 [0] NCCL INFO Connected all trees
+r7i6n3:610063:610122 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n3:610063:610122 [0] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n3:610066:610123 [3] NCCL INFO Connected all trees
+r7i6n3:610066:610123 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n3:610066:610123 [3] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n5:378202:378256 [2] NCCL INFO Channel 03 : 6[88000] -> 2[88000] [send] via NET/IB/3
+r7i6n3:610065:610128 [2] NCCL INFO Connected all trees
+r7i6n3:610065:610128 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n3:610065:610128 [2] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n2:1370347:1370447 [0] NCCL INFO Channel 02 : 8[1a000] -> 5[1c000] [send] via NET/IB/1
+r7i6n3:610064:610133 [1] NCCL INFO Connected all trees
+r7i6n3:610064:610133 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n3:610064:610133 [1] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n3:610064:610133 [1] NCCL INFO comm 0x1471e8002fb0 rank 13 nranks 16 cudaDev 1 busId 1c000 - Init COMPLETE
+r7i6n3:610063:610122 [0] NCCL INFO comm 0x148058002fb0 rank 12 nranks 16 cudaDev 0 busId 1a000 - Init COMPLETE
+r7i6n3:610066:610123 [3] NCCL INFO comm 0x155220002fb0 rank 15 nranks 16 cudaDev 3 busId 8a000 - Init COMPLETE
+r7i6n3:610065:610128 [2] NCCL INFO comm 0x1521c8002fb0 rank 14 nranks 16 cudaDev 2 busId 88000 - Init COMPLETE
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 03 : 10[88000] -> 7[8a000] [send] via NET/IB/3
+r6i6n4:257717:257772 [3] NCCL INFO Channel 00 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 00 : 11[8a000] -> 10[88000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 00 : 6[88000] -> 5[1c000] via P2P/IPC
+r6i6n4:257717:257772 [3] NCCL INFO Channel 02 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i6n2:1370350:1370452 [3] NCCL INFO Channel 02 : 11[8a000] -> 10[88000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 00 : 2[88000] -> 1[1c000] via P2P/IPC
+r6i6n4:257716:257777 [2] NCCL INFO Channel 02 : 2[88000] -> 1[1c000] via P2P/IPC
+r6i6n5:378202:378256 [2] NCCL INFO Channel 02 : 6[88000] -> 5[1c000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Channel 00 : 5[1c000] -> 4[1a000] via P2P/IPC
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 00 : 10[88000] -> 9[1c000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Channel 01 : 5[1c000] -> 4[1a000] via P2P/IPC
+r6i6n5:378203:378255 [3] NCCL INFO Channel 00 : 7[8a000] -> 6[88000] via P2P/IPC
+r6i6n4:257717:257772 [3] NCCL INFO Connected all trees
+r7i6n2:1370349:1370433 [2] NCCL INFO Channel 02 : 10[88000] -> 9[1c000] via P2P/IPC
+r6i6n4:257717:257772 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n4:257717:257772 [3] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n5:378201:378257 [1] NCCL INFO Channel 02 : 5[1c000] -> 4[1a000] via P2P/IPC
+r6i6n5:378203:378255 [3] NCCL INFO Channel 02 : 7[8a000] -> 6[88000] via P2P/IPC
+r6i6n5:378201:378257 [1] NCCL INFO Channel 03 : 5[1c000] -> 4[1a000] via P2P/IPC
+r6i6n4:257714:257762 [0] NCCL INFO Connected all trees
+r7i6n2:1370350:1370452 [3] NCCL INFO Connected all trees
+r6i6n4:257714:257762 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n4:257714:257762 [0] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n4:257716:257777 [2] NCCL INFO Connected all trees
+r6i6n4:257716:257777 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n4:257716:257777 [2] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n2:1370350:1370452 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n2:1370350:1370452 [3] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n5:378203:378255 [3] NCCL INFO Connected all trees
+r6i6n5:378203:378255 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n5:378203:378255 [3] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n4:257715:257767 [1] NCCL INFO Connected all trees
+r6i6n4:257715:257767 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n4:257715:257767 [1] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n4:257714:257762 [0] NCCL INFO comm 0x145844002fb0 rank 0 nranks 16 cudaDev 0 busId 1a000 - Init COMPLETE
+r6i6n4:257715:257767 [1] NCCL INFO comm 0x14c6f8002fb0 rank 1 nranks 16 cudaDev 1 busId 1c000 - Init COMPLETE
+r6i6n4:257717:257772 [3] NCCL INFO comm 0x149830002fb0 rank 3 nranks 16 cudaDev 3 busId 8a000 - Init COMPLETE
+r6i6n4:257716:257777 [2] NCCL INFO comm 0x151a88002fb0 rank 2 nranks 16 cudaDev 2 busId 88000 - Init COMPLETE
+r7i6n2:1370349:1370433 [2] NCCL INFO Connected all trees
+r7i6n2:1370349:1370433 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n2:1370349:1370433 [2] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n2:1370347:1370447 [0] NCCL INFO Connected all trees
+r7i6n2:1370347:1370447 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r7i6n2:1370347:1370447 [0] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n4:257714:257714 [0] NCCL INFO Launch mode Parallel
+r6i6n5:378202:378256 [2] NCCL INFO Connected all trees
+r6i6n5:378202:378256 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n5:378202:378256 [2] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n2:1370348:1370434 [1] NCCL INFO Connected all trees
+r7i6n2:1370348:1370434 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n5:378200:378262 [0] NCCL INFO Connected all trees
+r7i6n2:1370348:1370434 [1] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r6i6n5:378200:378262 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n5:378200:378262 [0] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n2:1370347:1370447 [0] NCCL INFO comm 0x151020002fb0 rank 8 nranks 16 cudaDev 0 busId 1a000 - Init COMPLETE
+r6i6n5:378201:378257 [1] NCCL INFO Connected all trees
+r7i6n2:1370348:1370434 [1] NCCL INFO comm 0x14d418002fb0 rank 9 nranks 16 cudaDev 1 busId 1c000 - Init COMPLETE
+r7i6n2:1370350:1370452 [3] NCCL INFO comm 0x154f28002fb0 rank 11 nranks 16 cudaDev 3 busId 8a000 - Init COMPLETE
+r6i6n5:378201:378257 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 8/8/512
+r6i6n5:378201:378257 [1] NCCL INFO 4 coll channels, 4 p2p channels, 1 p2p channels per peer
+r7i6n2:1370349:1370433 [2] NCCL INFO comm 0x154d48002fb0 rank 10 nranks 16 cudaDev 2 busId 88000 - Init COMPLETE
+r6i6n5:378200:378262 [0] NCCL INFO comm 0x153408002fb0 rank 4 nranks 16 cudaDev 0 busId 1a000 - Init COMPLETE
+r6i6n5:378202:378256 [2] NCCL INFO comm 0x154188002fb0 rank 6 nranks 16 cudaDev 2 busId 88000 - Init COMPLETE
+r6i6n5:378201:378257 [1] NCCL INFO comm 0x14c900002fb0 rank 5 nranks 16 cudaDev 1 busId 1c000 - Init COMPLETE
+r6i6n5:378203:378255 [3] NCCL INFO comm 0x14ef58002fb0 rank 7 nranks 16 cudaDev 3 busId 8a000 - Init COMPLETE
+ignore me 6
+14:
+ duration: 1.1593 sec
+ algo throughput: 55204675592.2273 bps, 55.2047 Gbps
+ busbw: 51.7544  Gbps
+ignore me 6
+ignore me 6
+15:
+ duration: 1.2942 sec
+ algo throughput: 49451976290.1993 bps, 49.4520 Gbps
+ busbw: 46.3612  Gbps
+13:
+ duration: 1.1545 sec
+ algo throughput: 55435153048.8659 bps, 55.4352 Gbps
+ busbw: 51.9705  Gbps
+ignore me 6
+12:
+ duration: 1.2946 sec
+ algo throughput: 49434318117.6515 bps, 49.4343 Gbps
+ busbw: 46.3447  Gbps
+ignore me 6
+ignore me 6
+9:
+ duration: 1.4402 sec
+ algo throughput: 44438492090.8732 bps, 44.4385 Gbps
+ busbw: 41.6611  Gbps
+10:
+ duration: 1.4738 sec
+ algo throughput: 43424520166.0441 bps, 43.4245 Gbps
+ignore me 6
+ busbw: 40.7105  Gbps
+ignore me 6
+ignore me 6
+3:
+ duration: 1.7691 sec
+ algo throughput: 36177572497.7145 bps, 36.1776 Gbps
+ busbw: 33.9165  Gbps
+0:
+11:
+ duration: 1.0927 sec
+ duration: 1.8093 sec
+ algo throughput: 35371927695.6812 bps, 35.3719 Gbps
+ busbw: 33.1612  Gbps
+ algo throughput: 58569704243.7844 bps, 58.5697 Gbps
+ busbw: 54.9091  Gbps
+ignore me 6
+ignore me 6
+5:
+ duration: 2.0802 sec
+ algo throughput: 30765780785.6832 bps, 30.7658 Gbps
+ busbw: 28.8429  Gbps
+ignore me 6
+6:
+ duration: 2.1418 sec
+ algo throughput: 29880845367.0138 bps, 29.8808 Gbps
+ busbw: 28.0133  Gbps
+ignore me 6
+8:
+ duration: 1.2561 sec
+ algo throughput: 50951080615.8564 bps, 50.9511 Gbps
+ busbw: 47.7666  Gbps
+7:
+ duration: 1.8124 sec
+ algo throughput: 35312957596.3833 bps, 35.3130 Gbps
+ busbw: 33.1059  Gbps
+ignore me 6
+4:
+ duration: 1.7526 sec
+ algo throughput: 36517122206.3803 bps, 36.5171 Gbps
+ busbw: 34.2348  Gbps
+ignore me 6
+1:
+ duration: 1.8395 sec
+ algo throughput: 34792737240.4271 bps, 34.7927 Gbps
+ busbw: 32.6182  Gbps
+ignore me 6
+2:
+ duration: 1.7637 sec
+ algo throughput: 36287170944.4988 bps, 36.2872 Gbps
+ busbw: 34.0192  Gbps
+ignore me 109
+14:
+ duration: 0.7080 sec
+ algo throughput: 90399491760.9001 bps, 90.3995 Gbps
+ busbw: 84.7495  Gbps
+ignore me 109
+15:
+ duration: 0.7080 sec
+ algo throughput: 90395163203.6951 bps, 90.3952 Gbps
+ busbw: 84.7455  Gbps
+ignore me 109
+13:
+ duration: 0.7081 sec
+ algo throughput: 90382326783.5510 bps, 90.3823 Gbps
+ busbw: 84.7334  Gbps
+ignore me 109
+12:
+ duration: 0.7080 sec
+ algo throughput: 90401745663.7657 bps, 90.4017 Gbps
+ busbw: 84.7516  Gbps
+ignore me 109
+9:
+ duration: 0.7080 sec
+ algo throughput: 90395783074.5905 bps, 90.3958 Gbps
+ busbw: 84.7460  Gbps
+ignore me 109
+10:
+ duration: 0.7082 sec
+ algo throughput: 90374224799.5715 bps, 90.3742 Gbps
+ busbw: 84.7258  Gbps
+ignore me 109
+0:
+ duration: 0.7083 sec
+ algo throughput: 90354374863.7591 bps, 90.3544 Gbps
+ busbw: 84.7072  Gbps
+ignore me 109
+11:
+ duration: 0.7084 sec
+ algo throughput: 90343336684.2220 bps, 90.3433 Gbps
+ busbw: 84.6969  Gbps
+ignore me 109
+3:
+ duration: 0.7087 sec
+ algo throughput: 90311896434.2268 bps, 90.3119 Gbps
+ busbw: 84.6674  Gbps
+ignore me 109
+8:
+ duration: 0.7085 sec
+ algo throughput: 90330088518.1323 bps, 90.3301 Gbps
+ busbw: 84.6845  Gbps
+ignore me 109
+ignore me 109
+2:
+ duration: 0.7085 sec
+ algo throughput: 90337030385.0629 bps, 90.3370 Gbps
+ busbw: 84.6910  Gbps
+5:
+ duration: 0.7088 sec
+ algo throughput: 90287308758.8899 bps, 90.2873 Gbps
+ busbw: 84.6444  Gbps
+ignore me 109
+ignore me 109
+1:
+ duration: 0.7089 sec
+ algo throughput: 90280901515.7927 bps, 90.2809 Gbps
+ busbw: 84.6383  Gbps
+6:
+ duration: 0.7090 sec
+ algo throughput: 90270047942.0345 bps, 90.2700 Gbps
+ busbw: 84.6282  Gbps
+ignore me 109
+7:
+ duration: 0.7090 sec
+ algo throughput: 90272586091.4933 bps, 90.2726 Gbps
+ busbw: 84.6305  Gbps
+ignore me 109
+4:
+ duration: 0.7085 sec
+ algo throughput: 90337161208.6908 bps, 90.3372 Gbps
+ busbw: 84.6911  Gbps
+ignore me 1749
+14:
+ duration: 0.7107 sec
+ algo throughput: 90058256584.7650 bps, 90.0583 Gbps
+ busbw: 84.4296  Gbps
+ignore me 1749
+ignore me 1749
+15:
+ duration: 0.7107 sec
+ algo throughput: 90057464420.3045 bps, 90.0575 Gbps
+ busbw: 84.4289  Gbps
+13:
+ duration: 0.7106 sec
+ algo throughput: 90070702828.5613 bps, 90.0707 Gbps
+ busbw: 84.4413  Gbps
+ignore me 1749
+ignore me 1749
+12:
+ duration: 0.7106 sec
+ algo throughput: 90059933061.1509 bps, 90.0599 Gbps
+ busbw: 84.4312  Gbps
+9:
+ duration: 0.7105 sec
+ algo throughput: 90071340053.9053 bps, 90.0713 Gbps
+ busbw: 84.4419  Gbps
+ignore me 1749
+10:
+ duration: 0.7106 sec
+ algo throughput: 90063253431.3530 bps, 90.0633 Gbps
+ busbw: 84.4343  Gbps
+ignore me 1749
+ignore me 1749
+11:
+ duration: 0.7106 sec
+ algo throughput: 90065670303.2662 bps, 90.0657 Gbps
+ busbw: 84.4366  Gbps
+0:
+ duration: 0.7107 sec
+ algo throughput: 90053334417.7426 bps, 90.0533 Gbps
+ busbw: 84.4250  Gbps
+ignore me 1749
+3:
+ duration: 0.7106 sec
+ algo throughput: 90068692693.3661 bps, 90.0687 Gbps
+ busbw: 84.4394  Gbps
+ignore me 1749
+ignore me 1749
+ignore me 1749
+8:
+ duration: 0.7105 sec
+2:
+ duration: 0.7104 sec
+ algo throughput: 90072894085.7098 bps, 90.0729 Gbps
+ busbw: 84.4433  Gbps
+ algo throughput: 90091360420.7079 bps, 90.0914 Gbps
+ busbw: 84.4607  Gbps
+ignore me 1749
+ignore me 1749
+5:
+ duration: 0.7104 sec
+ algo throughput: 90091316675.7603 bps, 90.0913 Gbps
+ busbw: 84.4606  Gbps
+1:
+ duration: 0.7103 sec
+ algo throughput: 90101456511.8536 bps, 90.1015 Gbps
+ busbw: 84.4701  Gbps
+6:
+ duration: 0.7103 sec
+ algo throughput: 90107024226.3038 bps, 90.1070 Gbps
+ busbw: 84.4753  Gbps
+ignore me 1749
+7:
+ duration: 0.7103 sec
+ algo throughput: 90107799997.7677 bps, 90.1078 Gbps
+ busbw: 84.4761  Gbps
+ignore me 1749
+4:
+ duration: 0.7103 sec
+ algo throughput: 90102477650.2766 bps, 90.1025 Gbps
+ busbw: 84.4711  Gbps
+ignore me 27986
+14:
+ duration: 0.7092 sec
+ algo throughput: 90242129271.5844 bps, 90.2421 Gbps
+ busbw: 84.6020  Gbps
+ignore me 27986
+ignore me 27986
+15:
+ duration: 0.7093 sec
+ algo throughput: 90233065038.0259 bps, 90.2331 Gbps
+ busbw: 84.5935  Gbps
+13:
+ duration: 0.7093 sec
+ algo throughput: 90226024022.6829 bps, 90.2260 Gbps
+ busbw: 84.5869  Gbps
+ignore me 27986
+12:
+ duration: 0.7092 sec
+ algo throughput: 90236901241.3211 bps, 90.2369 Gbps
+ busbw: 84.5971  Gbps
+ignore me 27986
+9:
+ duration: 0.7093 sec
+ algo throughput: 90231794012.9985 bps, 90.2318 Gbps
+ busbw: 84.5923  Gbps
+ignore me 27986
+10:
+ duration: 0.7093 sec
+ algo throughput: 90224093186.3902 bps, 90.2241 Gbps
+ busbw: 84.5851  Gbps
+ignore me 27986
+ignore me 27986
+0:
+ duration: 0.7092 sec
+11:
+ duration: 0.7092 sec
+ algo throughput: 90246123531.0302 bps, 90.2461 Gbps
+ busbw: 84.6057  Gbps
+ algo throughput: 90237670852.4900 bps, 90.2377 Gbps
+ busbw: 84.5978  Gbps
+ignore me 27986
+3:
+ duration: 0.7093 sec
+ algo throughput: 90235789890.2677 bps, 90.2358 Gbps
+ busbw: 84.5961  Gbps
+ignore me 27986
+8:
+ duration: 0.7092 sec
+ algo throughput: 90238335770.9699 bps, 90.2383 Gbps
+ busbw: 84.5984  Gbps
+ignore me 27986
+ignore me 27986
+2:
+ duration: 0.7093 sec
+ algo throughput: 90223737057.9605 bps, 90.2237 Gbps
+ busbw: 84.5848  Gbps
+ignore me 27986
+ignore me 27986
+5:
+ duration: 0.7093 sec
+ algo throughput: 90226816489.8323 bps, 90.2268 Gbps
+ busbw: 84.5876  Gbps
+6:
+ duration: 0.7093 sec
+ algo throughput: 90227312447.8407 bps, 90.2273 Gbps
+ busbw: 84.5881  Gbps
+1:
+ duration: 0.7094 sec
+ algo throughput: 90222924803.6610 bps, 90.2229 Gbps
+ busbw: 84.5840  Gbps
+ignore me 27986
+7:
+ duration: 0.7093 sec
+ algo throughput: 90229254099.1920 bps, 90.2293 Gbps
+ busbw: 84.5899  Gbps
+ignore me 27986
+4:
+ duration: 0.7094 sec
+ algo throughput: 90217548148.5392 bps, 90.2175 Gbps
+ busbw: 84.5790  Gbps
+ignore me 447779
+14:
+ duration: 0.7079 sec
+ algo throughput: 90401898007.1683 bps, 90.4019 Gbps
+ busbw: 84.7518  Gbps
+ignore me 447779
+13:
+ duration: 0.7078 sec
+ algo throughput: 90422510545.5320 bps, 90.4225 Gbps
+ busbw: 84.7711  Gbps
+ignore me 447779
+15:
+ duration: 0.7080 sec
+ algo throughput: 90397684358.3370 bps, 90.3977 Gbps
+ busbw: 84.7478  Gbps
+ignore me 447779
+12:
+ duration: 0.7080 sec
+ algo throughput: 90398934791.1951 bps, 90.3989 Gbps
+ busbw: 84.7490  Gbps
+ignore me 447779
+10:
+ duration: 0.7079 sec
+ algo throughput: 90404439072.1211 bps, 90.4044 Gbps
+ busbw: 84.7542  Gbps
+ignore me 447779
+11:
+ duration: 0.7078 sec
+ algo throughput: 90415260229.4886 bps, 90.4153 Gbps
+ busbw: 84.7643  Gbps
+ignore me 447779
+ignore me 447779
+9:
+ duration: 0.7086 sec
+ algo throughput: 90317814308.9687 bps, 90.3178 Gbps
+ busbw: 84.6730  Gbps
+0:
+ duration: 0.7081 sec
+ algo throughput: 90384670565.8098 bps, 90.3847 Gbps
+ busbw: 84.7356  Gbps
+ignore me 447779
+8:
+ duration: 0.7080 sec
+ algo throughput: 90401729311.5575 bps, 90.4017 Gbps
+ busbw: 84.7516  Gbps
+ignore me 447779
+ignore me 447779
+2:
+ duration: 0.7081 sec
+ algo throughput: 90388659575.5084 bps, 90.3887 Gbps
+ busbw: 84.7394  Gbps
+1:
+ duration: 0.7080 sec
+ algo throughput: 90397847806.5952 bps, 90.3978 Gbps
+ busbw: 84.7480  Gbps
+ignore me 447779
+ignore me 447779
+3:
+ duration: 0.7086 sec
+6:
+ duration: 0.7083 sec
+ algo throughput: 90320233896.8830 bps, 90.3202 Gbps
+ busbw: 84.6752  Gbps
+ algo throughput: 90360979559.3395 bps, 90.3610 Gbps
+ busbw: 84.7134  Gbps
+ignore me 447779
+7:
+ duration: 0.7083 sec
+ algo throughput: 90360919482.0588 bps, 90.3609 Gbps
+ busbw: 84.7134  Gbps
+ignore me 447779
+5:
+ duration: 0.7087 sec
+ algo throughput: 90307672345.7347 bps, 90.3077 Gbps
+ busbw: 84.6634  Gbps
+ignore me 447779
+4:
+ duration: 0.7085 sec
+ algo throughput: 90328680753.0585 bps, 90.3287 Gbps
+ busbw: 84.6831  Gbps

bigscience/experiments/bandwidth/all_reduce_bench-a100-n4.slurm

@@ -0,0 +1,24 @@
+#!/bin/bash
+#SBATCH --job-name=all_reduce_bench-a100-n4
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --nodes=4
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=64           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:8                 # number of gpus
+#SBATCH --time 0:05:00               # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+
+source $six_ALL_CCFRWORK/code/tr11-176B-ml/bigscience/train/tr11-176B-ml/start-tr11-176B-ml
+
+export NNODES=$SLURM_NNODES
+export GPUS_PER_NODE=8
+export NCCL_DEBUG=info
+
+export LOG_FILE=all_reduce_bench-a100-$NNODES.txt
+
+export MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+
+srun --jobid $SLURM_JOBID bash -c 'python -m torch.distributed.launch --nnodes $NNODES --nproc_per_node $GPUS_PER_NODE --node_rank $SLURM_PROCID --master_addr $MASTER_ADDR --master_port 12345 all_reduce_bench.py'  2>&1 | tee $LOG_FILE

bigscience/experiments/bandwidth/all_reduce_bench-a100-n4.txt

@@ -0,0 +1,1424 @@
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+/gpfswork/rech/six/commun/conda/hf-prod/lib/python3.8/site-packages/torch/distributed/launch.py:178: FutureWarning: The module torch.distributed.launch is deprecated
+and will be removed in future. Use torchrun.
+Note that --use_env is set by default in torchrun.
+If your script expects `--local_rank` argument to be set, please
+change it to read from `os.environ['LOCAL_RANK']` instead. See 
+https://pytorch.org/docs/stable/distributed.html#launch-utility for 
+further instructions
+
+  warnings.warn(
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+WARNING:torch.distributed.run:
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+local_rank: 1
+local_rank: 3
+local_rank: 5
+local_rank: 6
+local_rank: 7
+local_rank: 2
+local_rank: 2
+local_rank: 3
+local_rank: 2
+local_rank: 1
+local_rank: 0
+local_rank: 2
+local_rank: 0
+local_rank: 0
+local_rank: 5
+local_rank: 1
+local_rank: 4
+local_rank: 3
+local_rank: 7
+local_rank: 7
+local_rank: 6
+local_rank: 6
+local_rank: 4
+local_rank: 5
+local_rank: 5
+local_rank: 1
+local_rank: 3
+local_rank: 4
+local_rank: 6
+local_rank: 7
+local_rank: 0
+local_rank: 4
+0 data size: 4.0 GB
+1 data size: 4.0 GB
+5 data size: 4.0 GB
+20 data size: 4.0 GB
+30 data size: 4.0 GB
+3 data size: 4.0 GB
+12 data size: 4.0 GB
+21 data size: 4.0 GB
+28 data size: 4.0 GB
+17 data size: 4.0 GB
+2 data size: 4.0 GB
+25 data size: 4.0 GB
+19 data size: 4.0 GB
+22 data size: 4.0 GB
+16 data size: 4.0 GB
+15 data size: 4.0 GB
+26 data size: 4.0 GB
+27 data size: 4.0 GB
+6 data size: 4.0 GB
+24 data size: 4.0 GB
+9 data size: 4.0 GB
+29 data size: 4.0 GB
+23 data size: 4.0 GB
+31 data size: 4.0 GB
+14 data size: 4.0 GB
+7 data size: 4.0 GB
+18 data size: 4.0 GB
+8 data size: 4.0 GB
+11 data size: 4.0 GB
+10 data size: 4.0 GB
+13 data size: 4.0 GB
+4 data size: 4.0 GB
+jean-zay-iam37:261379:261379 [0] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261379:261379 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam37:261379:261379 [0] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261379:261379 [0] NCCL INFO Using network IB
+NCCL version 2.10.3+cuda11.3
+jean-zay-iam37:261380:261380 [1] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261383:261383 [4] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261384:261384 [5] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261386:261386 [7] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261381:261381 [2] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261382:261382 [3] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam41:276753:276753 [7] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276748:276748 [2] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276747:276747 [1] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276752:276752 [6] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276750:276750 [4] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276751:276751 [5] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276746:276746 [0] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam41:276749:276749 [3] NCCL INFO Bootstrap : Using ib0:10.148.8.206<0>
+jean-zay-iam37:261384:261384 [5] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263016:263016 [1] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263017:263017 [2] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263018:263018 [3] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263021:263021 [6] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263019:263019 [4] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263015:263015 [0] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263020:263020 [5] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam52:263022:263022 [7] NCCL INFO Bootstrap : Using ib0:10.148.8.217<0>
+jean-zay-iam37:261386:261386 [7] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam37:261381:261381 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam37:261380:261380 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam37:261383:261383 [4] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289973:289973 [6] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289967:289967 [0] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289969:289969 [2] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289971:289971 [4] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289970:289970 [3] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289972:289972 [5] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289968:289968 [1] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam40:289974:289974 [7] NCCL INFO Bootstrap : Using ib0:10.148.8.205<0>
+jean-zay-iam37:261382:261382 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276751:276751 [5] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276749:276749 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276746:276746 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276753:276753 [7] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276748:276748 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276752:276752 [6] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276747:276747 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam41:276750:276750 [4] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263017:263017 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263016:263016 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289968:289968 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289974:289974 [7] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263021:263021 [6] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263015:263015 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263019:263019 [4] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263020:263020 [5] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289973:289973 [6] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263018:263018 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam52:263022:263022 [7] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289972:289972 [5] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289969:289969 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289971:289971 [4] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289967:289967 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam40:289970:289970 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam37:261384:261384 [5] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261384:261384 [5] NCCL INFO Using network IB
+jean-zay-iam37:261380:261380 [1] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261380:261380 [1] NCCL INFO Using network IB
+jean-zay-iam37:261383:261383 [4] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261381:261381 [2] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261386:261386 [7] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261381:261381 [2] NCCL INFO Using network IB
+jean-zay-iam37:261383:261383 [4] NCCL INFO Using network IB
+jean-zay-iam37:261386:261386 [7] NCCL INFO Using network IB
+jean-zay-iam37:261382:261382 [3] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261382:261382 [3] NCCL INFO Using network IB
+jean-zay-iam41:276751:276751 [5] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276751:276751 [5] NCCL INFO Using network IB
+jean-zay-iam41:276748:276748 [2] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276748:276748 [2] NCCL INFO Using network IB
+jean-zay-iam41:276747:276747 [1] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276747:276747 [1] NCCL INFO Using network IB
+jean-zay-iam41:276752:276752 [6] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276746:276746 [0] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276749:276749 [3] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276753:276753 [7] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276749:276749 [3] NCCL INFO Using network IB
+jean-zay-iam41:276750:276750 [4] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.206<0>
+jean-zay-iam41:276746:276746 [0] NCCL INFO Using network IB
+jean-zay-iam41:276753:276753 [7] NCCL INFO Using network IB
+jean-zay-iam41:276752:276752 [6] NCCL INFO Using network IB
+jean-zay-iam41:276750:276750 [4] NCCL INFO Using network IB
+jean-zay-iam52:263022:263022 [7] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263022:263022 [7] NCCL INFO Using network IB
+jean-zay-iam52:263019:263019 [4] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263019:263019 [4] NCCL INFO Using network IB
+jean-zay-iam52:263021:263021 [6] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263021:263021 [6] NCCL INFO Using network IB
+jean-zay-iam52:263017:263017 [2] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263017:263017 [2] NCCL INFO Using network IB
+jean-zay-iam52:263020:263020 [5] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263020:263020 [5] NCCL INFO Using network IB
+jean-zay-iam52:263015:263015 [0] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263018:263018 [3] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263015:263015 [0] NCCL INFO Using network IB
+jean-zay-iam52:263018:263018 [3] NCCL INFO Using network IB
+jean-zay-iam52:263016:263016 [1] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.217<0>
+jean-zay-iam52:263016:263016 [1] NCCL INFO Using network IB
+jean-zay-iam40:289968:289968 [1] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289968:289968 [1] NCCL INFO Using network IB
+jean-zay-iam40:289972:289972 [5] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289972:289972 [5] NCCL INFO Using network IB
+jean-zay-iam40:289969:289969 [2] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289969:289969 [2] NCCL INFO Using network IB
+jean-zay-iam40:289973:289973 [6] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289973:289973 [6] NCCL INFO Using network IB
+jean-zay-iam40:289970:289970 [3] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289970:289970 [3] NCCL INFO Using network IB
+jean-zay-iam40:289967:289967 [0] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289967:289967 [0] NCCL INFO Using network IB
+jean-zay-iam40:289974:289974 [7] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289971:289971 [4] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.205<0>
+jean-zay-iam40:289974:289974 [7] NCCL INFO Using network IB
+jean-zay-iam40:289971:289971 [4] NCCL INFO Using network IB
+jean-zay-iam37:261385:261385 [6] NCCL INFO Bootstrap : Using ib0:10.148.8.203<0>
+jean-zay-iam37:261385:261385 [6] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+jean-zay-iam37:261385:261385 [6] NCCL INFO NET/IB : Using [0]hfi1_1:1/IB [1]hfi1_0:1/IB [2]hfi1_2:1/IB [3]hfi1_3:1/IB ; OOB ib0:10.148.8.203<0>
+jean-zay-iam37:261385:261385 [6] NCCL INFO Using network IB
+jean-zay-iam52:263016:263142 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->24
+jean-zay-iam52:263016:263142 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff
+jean-zay-iam52:263017:263138 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25
+jean-zay-iam52:263017:263138 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff
+jean-zay-iam52:263018:263141 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26
+jean-zay-iam52:263018:263141 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff
+jean-zay-iam52:263021:263137 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29
+jean-zay-iam52:263020:263139 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28
+jean-zay-iam52:263021:263137 [6] NCCL INFO Setting affinity for GPU 6 to ff,00000000
+jean-zay-iam52:263020:263139 [5] NCCL INFO Setting affinity for GPU 5 to ff,00000000
+jean-zay-iam52:263019:263136 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27
+jean-zay-iam52:263022:263135 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] -1/-1/-1->31->30
+jean-zay-iam52:263019:263136 [4] NCCL INFO Setting affinity for GPU 4 to ff,00000000
+jean-zay-iam52:263022:263135 [7] NCCL INFO Setting affinity for GPU 7 to ff,00000000
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 00/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+jean-zay-iam37:261381:261500 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 01/02 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+jean-zay-iam37:261379:261471 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] 1/-1/-1->0->8
+jean-zay-iam37:261380:261497 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0
+jean-zay-iam37:261381:261500 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff
+jean-zay-iam37:261379:261471 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff
+jean-zay-iam37:261380:261497 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff
+jean-zay-iam41:276749:276869 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18
+jean-zay-iam41:276751:276865 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20
+jean-zay-iam41:276749:276869 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff
+jean-zay-iam41:276750:276871 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19
+jean-zay-iam41:276752:276872 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21
+jean-zay-iam52:263015:263140 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] 25/8/-1->24->-1
+jean-zay-iam52:263015:263140 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff
+jean-zay-iam40:289974:290092 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14
+jean-zay-iam41:276751:276865 [5] NCCL INFO Setting affinity for GPU 5 to ff,00000000
+jean-zay-iam41:276750:276871 [4] NCCL INFO Setting affinity for GPU 4 to ff,00000000
+jean-zay-iam41:276752:276872 [6] NCCL INFO Setting affinity for GPU 6 to ff,00000000
+jean-zay-iam41:276753:276870 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] -1/-1/-1->23->22
+jean-zay-iam41:276753:276870 [7] NCCL INFO Setting affinity for GPU 7 to ff,00000000
+jean-zay-iam40:289974:290092 [7] NCCL INFO Setting affinity for GPU 7 to ff,00000000
+jean-zay-iam41:276747:276867 [1] NCCL INFO Trees [0] 18/8/-1->17->16 [1] 18/-1/-1->17->16
+jean-zay-iam41:276748:276866 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17
+jean-zay-iam41:276746:276868 [0] NCCL INFO Trees [0] 17/24/-1->16->0 [1] 17/-1/-1->16->9
+jean-zay-iam41:276748:276866 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff
+jean-zay-iam41:276746:276868 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff
+jean-zay-iam41:276747:276867 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff
+jean-zay-iam37:261382:261501 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2
+jean-zay-iam37:261382:261501 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff
+jean-zay-iam37:261383:261499 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3
+jean-zay-iam37:261383:261499 [4] NCCL INFO Setting affinity for GPU 4 to ff,00000000
+jean-zay-iam37:261385:261506 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5
+jean-zay-iam37:261385:261506 [6] NCCL INFO Setting affinity for GPU 6 to ff,00000000
+jean-zay-iam37:261384:261496 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4
+jean-zay-iam37:261384:261496 [5] NCCL INFO Setting affinity for GPU 5 to ff,00000000
+jean-zay-iam37:261386:261498 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6
+jean-zay-iam37:261386:261498 [7] NCCL INFO Setting affinity for GPU 7 to ff,00000000
+jean-zay-iam40:289967:290091 [0] NCCL INFO Trees [0] 9/-1/-1->8->17 [1] 9/0/-1->8->24
+jean-zay-iam40:289967:290091 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff
+jean-zay-iam40:289969:290087 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9
+jean-zay-iam40:289970:290090 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10
+jean-zay-iam40:289968:290086 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/16/-1->9->8
+jean-zay-iam40:289973:290089 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13
+jean-zay-iam40:289969:290087 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff
+jean-zay-iam40:289970:290090 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff
+jean-zay-iam40:289973:290089 [6] NCCL INFO Setting affinity for GPU 6 to ff,00000000
+jean-zay-iam40:289968:290086 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff
+jean-zay-iam40:289972:290088 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12
+jean-zay-iam40:289971:290093 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11
+jean-zay-iam40:289972:290088 [5] NCCL INFO Setting affinity for GPU 5 to ff,00000000
+jean-zay-iam40:289971:290093 [4] NCCL INFO Setting affinity for GPU 4 to ff,00000000
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 00 : 31[cb000] -> 0[7000] [receive] via NET/IB/1
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 00 : 7[cb000] -> 8[7000] [receive] via NET/IB/1
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 00 : 23[cb000] -> 24[7000] [receive] via NET/IB/1
+jean-zay-iam52:263016:263142 [1] NCCL INFO Channel 00 : 25[b000] -> 26[48000] via P2P/IPC/read
+jean-zay-iam52:263017:263138 [2] NCCL INFO Channel 00 : 26[48000] -> 27[4c000] via P2P/IPC/read
+jean-zay-iam52:263018:263141 [3] NCCL INFO Channel 00 : 27[4c000] -> 28[88000] via P2P/IPC/read
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 00 : 15[cb000] -> 16[7000] [receive] via NET/IB/1
+jean-zay-iam52:263020:263139 [5] NCCL INFO Channel 00 : 29[8b000] -> 30[c8000] via P2P/IPC/read
+jean-zay-iam40:289974:290092 [7] NCCL INFO Channel 00 : 15[cb000] -> 16[7000] [send] via NET/IB/3
+jean-zay-iam41:276752:276872 [6] NCCL INFO Channel 00 : 22[c8000] -> 23[cb000] via P2P/IPC/read
+jean-zay-iam52:263019:263136 [4] NCCL INFO Channel 00 : 28[88000] -> 29[8b000] via P2P/IPC/read
+jean-zay-iam52:263021:263137 [6] NCCL INFO Channel 00 : 30[c8000] -> 31[cb000] via P2P/IPC/read
+jean-zay-iam52:263016:263142 [1] NCCL INFO Channel 01 : 25[b000] -> 26[48000] via P2P/IPC/read
+jean-zay-iam41:276750:276871 [4] NCCL INFO Channel 00 : 20[88000] -> 21[8b000] via P2P/IPC/read
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 01 : 23[cb000] -> 24[7000] [receive] via NET/IB/1
+jean-zay-iam52:263017:263138 [2] NCCL INFO Channel 01 : 26[48000] -> 27[4c000] via P2P/IPC/read
+jean-zay-iam41:276751:276865 [5] NCCL INFO Channel 00 : 21[8b000] -> 22[c8000] via P2P/IPC/read
+jean-zay-iam52:263022:263135 [7] NCCL INFO Channel 00 : 31[cb000] -> 0[7000] [send] via NET/IB/3
+jean-zay-iam41:276749:276869 [3] NCCL INFO Channel 00 : 19[4c000] -> 20[88000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 01 : 31[cb000] -> 0[7000] [receive] via NET/IB/1
+jean-zay-iam37:261381:261500 [2] NCCL INFO Channel 00 : 2[48000] -> 3[4c000] via P2P/IPC/read
+jean-zay-iam52:263018:263141 [3] NCCL INFO Channel 01 : 27[4c000] -> 28[88000] via P2P/IPC/read
+jean-zay-iam41:276747:276867 [1] NCCL INFO Channel 00 : 17[b000] -> 18[48000] via P2P/IPC/read
+jean-zay-iam41:276748:276866 [2] NCCL INFO Channel 00 : 18[48000] -> 19[4c000] via P2P/IPC/read
+jean-zay-iam52:263020:263139 [5] NCCL INFO Channel 01 : 29[8b000] -> 30[c8000] via P2P/IPC/read
+jean-zay-iam37:261380:261497 [1] NCCL INFO Channel 00 : 1[b000] -> 2[48000] via P2P/IPC/read
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 01 : 15[cb000] -> 16[7000] [receive] via NET/IB/1
+jean-zay-iam37:261382:261501 [3] NCCL INFO Channel 00 : 3[4c000] -> 4[88000] via P2P/IPC/read
+jean-zay-iam37:261385:261506 [6] NCCL INFO Channel 00 : 6[c8000] -> 7[cb000] via P2P/IPC/read
+jean-zay-iam41:276752:276872 [6] NCCL INFO Channel 01 : 22[c8000] -> 23[cb000] via P2P/IPC/read
+jean-zay-iam52:263019:263136 [4] NCCL INFO Channel 01 : 28[88000] -> 29[8b000] via P2P/IPC/read
+jean-zay-iam52:263021:263137 [6] NCCL INFO Channel 01 : 30[c8000] -> 31[cb000] via P2P/IPC/read
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 01 : 7[cb000] -> 8[7000] [receive] via NET/IB/1
+jean-zay-iam40:289968:290086 [1] NCCL INFO Channel 00 : 9[b000] -> 10[48000] via P2P/IPC/read
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 00 : 24[7000] -> 25[b000] via P2P/IPC/read
+jean-zay-iam41:276753:276870 [7] NCCL INFO Channel 00 : 23[cb000] -> 24[7000] [send] via NET/IB/3
+jean-zay-iam41:276750:276871 [4] NCCL INFO Channel 01 : 20[88000] -> 21[8b000] via P2P/IPC/read
+jean-zay-iam40:289970:290090 [3] NCCL INFO Channel 00 : 11[4c000] -> 12[88000] via P2P/IPC/read
+jean-zay-iam40:289969:290087 [2] NCCL INFO Channel 00 : 10[48000] -> 11[4c000] via P2P/IPC/read
+jean-zay-iam40:289973:290089 [6] NCCL INFO Channel 00 : 14[c8000] -> 15[cb000] via P2P/IPC/read
+jean-zay-iam40:289972:290088 [5] NCCL INFO Channel 00 : 13[8b000] -> 14[c8000] via P2P/IPC/read
+jean-zay-iam37:261386:261498 [7] NCCL INFO Channel 00 : 7[cb000] -> 8[7000] [send] via NET/IB/3
+jean-zay-iam41:276751:276865 [5] NCCL INFO Channel 01 : 21[8b000] -> 22[c8000] via P2P/IPC/read
+jean-zay-iam41:276749:276869 [3] NCCL INFO Channel 01 : 19[4c000] -> 20[88000] via P2P/IPC/read
+jean-zay-iam40:289974:290092 [7] NCCL INFO Channel 01 : 15[cb000] -> 16[7000] [send] via NET/IB/3
+jean-zay-iam40:289971:290093 [4] NCCL INFO Channel 00 : 12[88000] -> 13[8b000] via P2P/IPC/read
+jean-zay-iam41:276747:276867 [1] NCCL INFO Channel 01 : 17[b000] -> 18[48000] via P2P/IPC/read
+jean-zay-iam41:276748:276866 [2] NCCL INFO Channel 01 : 18[48000] -> 19[4c000] via P2P/IPC/read
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 00 : 16[7000] -> 17[b000] via P2P/IPC/read
+jean-zay-iam37:261383:261499 [4] NCCL INFO Channel 00 : 4[88000] -> 5[8b000] via P2P/IPC/read
+jean-zay-iam52:263022:263135 [7] NCCL INFO Channel 01 : 31[cb000] -> 0[7000] [send] via NET/IB/3
+jean-zay-iam37:261384:261496 [5] NCCL INFO Channel 00 : 5[8b000] -> 6[c8000] via P2P/IPC/read
+jean-zay-iam40:289968:290086 [1] NCCL INFO Channel 01 : 9[b000] -> 10[48000] via P2P/IPC/read
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 00 : 8[7000] -> 9[b000] via P2P/IPC/read
+jean-zay-iam40:289970:290090 [3] NCCL INFO Channel 01 : 11[4c000] -> 12[88000] via P2P/IPC/read
+jean-zay-iam40:289969:290087 [2] NCCL INFO Channel 01 : 10[48000] -> 11[4c000] via P2P/IPC/read
+jean-zay-iam40:289972:290088 [5] NCCL INFO Channel 01 : 13[8b000] -> 14[c8000] via P2P/IPC/read
+jean-zay-iam40:289973:290089 [6] NCCL INFO Channel 01 : 14[c8000] -> 15[cb000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 00 : 0[7000] -> 1[b000] via P2P/IPC/read
+jean-zay-iam37:261381:261500 [2] NCCL INFO Channel 01 : 2[48000] -> 3[4c000] via P2P/IPC/read
+jean-zay-iam40:289971:290093 [4] NCCL INFO Channel 01 : 12[88000] -> 13[8b000] via P2P/IPC/read
+jean-zay-iam37:261380:261497 [1] NCCL INFO Channel 01 : 1[b000] -> 2[48000] via P2P/IPC/read
+jean-zay-iam37:261382:261501 [3] NCCL INFO Channel 01 : 3[4c000] -> 4[88000] via P2P/IPC/read
+jean-zay-iam37:261385:261506 [6] NCCL INFO Channel 01 : 6[c8000] -> 7[cb000] via P2P/IPC/read
+jean-zay-iam37:261386:261498 [7] NCCL INFO Channel 01 : 7[cb000] -> 8[7000] [send] via NET/IB/3
+jean-zay-iam41:276753:276870 [7] NCCL INFO Channel 01 : 23[cb000] -> 24[7000] [send] via NET/IB/3
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 01 : 24[7000] -> 25[b000] via P2P/IPC/read
+jean-zay-iam37:261383:261499 [4] NCCL INFO Channel 01 : 4[88000] -> 5[8b000] via P2P/IPC/read
+jean-zay-iam37:261384:261496 [5] NCCL INFO Channel 01 : 5[8b000] -> 6[c8000] via P2P/IPC/read
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 01 : 8[7000] -> 9[b000] via P2P/IPC/read
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 01 : 16[7000] -> 17[b000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 01 : 0[7000] -> 1[b000] via P2P/IPC/read
+jean-zay-iam52:263017:263138 [2] NCCL INFO Connected all rings
+jean-zay-iam41:276753:276870 [7] NCCL INFO Connected all rings
+jean-zay-iam52:263018:263141 [3] NCCL INFO Connected all rings
+jean-zay-iam52:263015:263140 [0] NCCL INFO Connected all rings
+jean-zay-iam52:263019:263136 [4] NCCL INFO Connected all rings
+jean-zay-iam37:261386:261498 [7] NCCL INFO Connected all rings
+jean-zay-iam41:276751:276865 [5] NCCL INFO Connected all rings
+jean-zay-iam40:289967:290091 [0] NCCL INFO Connected all rings
+jean-zay-iam52:263020:263139 [5] NCCL INFO Connected all rings
+jean-zay-iam40:289974:290092 [7] NCCL INFO Connected all rings
+jean-zay-iam41:276750:276871 [4] NCCL INFO Connected all rings
+jean-zay-iam41:276746:276868 [0] NCCL INFO Connected all rings
+jean-zay-iam52:263022:263135 [7] NCCL INFO Connected all rings
+jean-zay-iam37:261379:261471 [0] NCCL INFO Connected all rings
+jean-zay-iam41:276748:276866 [2] NCCL INFO Connected all rings
+jean-zay-iam41:276749:276869 [3] NCCL INFO Connected all rings
+jean-zay-iam52:263021:263137 [6] NCCL INFO Connected all rings
+jean-zay-iam40:289973:290089 [6] NCCL INFO Connected all rings
+jean-zay-iam40:289969:290087 [2] NCCL INFO Connected all rings
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 00 : 16[7000] -> 24[7000] [receive] via NET/IB/1
+jean-zay-iam41:276753:276870 [7] NCCL INFO Channel 00 : 23[cb000] -> 22[c8000] via P2P/IPC/read
+jean-zay-iam37:261381:261500 [2] NCCL INFO Connected all rings
+jean-zay-iam40:289972:290088 [5] NCCL INFO Connected all rings
+jean-zay-iam52:263016:263142 [1] NCCL INFO Connected all rings
+jean-zay-iam52:263022:263135 [7] NCCL INFO Channel 00 : 31[cb000] -> 30[c8000] via P2P/IPC/read
+jean-zay-iam41:276752:276872 [6] NCCL INFO Connected all rings
+jean-zay-iam40:289970:290090 [3] NCCL INFO Connected all rings
+jean-zay-iam40:289971:290093 [4] NCCL INFO Connected all rings
+jean-zay-iam37:261386:261498 [7] NCCL INFO Channel 00 : 7[cb000] -> 6[c8000] via P2P/IPC/read
+jean-zay-iam37:261382:261501 [3] NCCL INFO Connected all rings
+jean-zay-iam40:289974:290092 [7] NCCL INFO Channel 00 : 15[cb000] -> 14[c8000] via P2P/IPC/read
+jean-zay-iam41:276747:276867 [1] NCCL INFO Connected all rings
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 01 : 0[7000] -> 8[7000] [receive] via NET/IB/1
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 01 : 9[b000] -> 16[7000] [receive] via NET/IB/1
+jean-zay-iam37:261384:261496 [5] NCCL INFO Connected all rings
+jean-zay-iam37:261383:261499 [4] NCCL INFO Connected all rings
+jean-zay-iam40:289968:290086 [1] NCCL INFO Connected all rings
+jean-zay-iam37:261385:261506 [6] NCCL INFO Connected all rings
+jean-zay-iam52:263022:263135 [7] NCCL INFO Channel 01 : 31[cb000] -> 30[c8000] via P2P/IPC/read
+jean-zay-iam41:276753:276870 [7] NCCL INFO Channel 01 : 23[cb000] -> 22[c8000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 01 : 0[7000] -> 8[7000] [send] via NET/IB/1
+jean-zay-iam52:263017:263138 [2] NCCL INFO Channel 00 : 26[48000] -> 25[b000] via P2P/IPC/read
+jean-zay-iam37:261380:261497 [1] NCCL INFO Connected all rings
+jean-zay-iam40:289974:290092 [7] NCCL INFO Channel 01 : 15[cb000] -> 14[c8000] via P2P/IPC/read
+jean-zay-iam52:263018:263141 [3] NCCL INFO Channel 00 : 27[4c000] -> 26[48000] via P2P/IPC/read
+jean-zay-iam52:263019:263136 [4] NCCL INFO Channel 00 : 28[88000] -> 27[4c000] via P2P/IPC/read
+jean-zay-iam52:263020:263139 [5] NCCL INFO Channel 00 : 29[8b000] -> 28[88000] via P2P/IPC/read
+jean-zay-iam41:276751:276865 [5] NCCL INFO Channel 00 : 21[8b000] -> 20[88000] via P2P/IPC/read
+jean-zay-iam41:276747:276867 [1] NCCL INFO Channel 00 : 8[7000] -> 17[b000] [receive] via NET/IB/1
+jean-zay-iam37:261386:261498 [7] NCCL INFO Channel 01 : 7[cb000] -> 6[c8000] via P2P/IPC/read
+jean-zay-iam41:276750:276871 [4] NCCL INFO Channel 00 : 20[88000] -> 19[4c000] via P2P/IPC/read
+jean-zay-iam52:263017:263138 [2] NCCL INFO Channel 01 : 26[48000] -> 25[b000] via P2P/IPC/read
+jean-zay-iam52:263021:263137 [6] NCCL INFO Channel 00 : 30[c8000] -> 29[8b000] via P2P/IPC/read
+jean-zay-iam41:276748:276866 [2] NCCL INFO Channel 00 : 18[48000] -> 17[b000] via P2P/IPC/read
+jean-zay-iam41:276749:276869 [3] NCCL INFO Channel 00 : 19[4c000] -> 18[48000] via P2P/IPC/read
+jean-zay-iam40:289968:290086 [1] NCCL INFO Channel 01 : 9[b000] -> 16[7000] [send] via NET/IB/1
+jean-zay-iam52:263018:263141 [3] NCCL INFO Channel 01 : 27[4c000] -> 26[48000] via P2P/IPC/read
+jean-zay-iam52:263019:263136 [4] NCCL INFO Channel 01 : 28[88000] -> 27[4c000] via P2P/IPC/read
+jean-zay-iam52:263020:263139 [5] NCCL INFO Channel 01 : 29[8b000] -> 28[88000] via P2P/IPC/read
+jean-zay-iam52:263016:263142 [1] NCCL INFO Channel 00 : 25[b000] -> 24[7000] via P2P/IPC/read
+jean-zay-iam41:276751:276865 [5] NCCL INFO Channel 01 : 21[8b000] -> 20[88000] via P2P/IPC/read
+jean-zay-iam41:276750:276871 [4] NCCL INFO Channel 01 : 20[88000] -> 19[4c000] via P2P/IPC/read
+jean-zay-iam41:276752:276872 [6] NCCL INFO Channel 00 : 22[c8000] -> 21[8b000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 00 : 16[7000] -> 0[7000] [receive] via NET/IB/1
+jean-zay-iam41:276748:276866 [2] NCCL INFO Channel 01 : 18[48000] -> 17[b000] via P2P/IPC/read
+jean-zay-iam41:276749:276869 [3] NCCL INFO Channel 01 : 19[4c000] -> 18[48000] via P2P/IPC/read
+jean-zay-iam52:263021:263137 [6] NCCL INFO Channel 01 : 30[c8000] -> 29[8b000] via P2P/IPC/read
+jean-zay-iam40:289973:290089 [6] NCCL INFO Channel 00 : 14[c8000] -> 13[8b000] via P2P/IPC/read
+jean-zay-iam41:276752:276872 [6] NCCL INFO Channel 01 : 22[c8000] -> 21[8b000] via P2P/IPC/read
+jean-zay-iam40:289969:290087 [2] NCCL INFO Channel 00 : 10[48000] -> 9[b000] via P2P/IPC/read
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 00 : 8[7000] -> 17[b000] [send] via NET/IB/1
+jean-zay-iam40:289970:290090 [3] NCCL INFO Channel 00 : 11[4c000] -> 10[48000] via P2P/IPC/read
+jean-zay-iam40:289972:290088 [5] NCCL INFO Channel 00 : 13[8b000] -> 12[88000] via P2P/IPC/read
+jean-zay-iam40:289971:290093 [4] NCCL INFO Channel 00 : 12[88000] -> 11[4c000] via P2P/IPC/read
+jean-zay-iam37:261381:261500 [2] NCCL INFO Channel 00 : 2[48000] -> 1[b000] via P2P/IPC/read
+jean-zay-iam52:263016:263142 [1] NCCL INFO Channel 01 : 25[b000] -> 24[7000] via P2P/IPC/read
+jean-zay-iam37:261382:261501 [3] NCCL INFO Channel 00 : 3[4c000] -> 2[48000] via P2P/IPC/read
+jean-zay-iam37:261385:261506 [6] NCCL INFO Channel 00 : 6[c8000] -> 5[8b000] via P2P/IPC/read
+jean-zay-iam37:261384:261496 [5] NCCL INFO Channel 00 : 5[8b000] -> 4[88000] via P2P/IPC/read
+jean-zay-iam40:289973:290089 [6] NCCL INFO Channel 01 : 14[c8000] -> 13[8b000] via P2P/IPC/read
+jean-zay-iam40:289969:290087 [2] NCCL INFO Channel 01 : 10[48000] -> 9[b000] via P2P/IPC/read
+jean-zay-iam40:289968:290086 [1] NCCL INFO Channel 01 : 16[7000] -> 9[b000] [receive] via NET/IB/1
+jean-zay-iam37:261383:261499 [4] NCCL INFO Channel 00 : 4[88000] -> 3[4c000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 00 : 0[7000] -> 16[7000] [send] via NET/IB/1
+jean-zay-iam40:289970:290090 [3] NCCL INFO Channel 01 : 11[4c000] -> 10[48000] via P2P/IPC/read
+jean-zay-iam40:289972:290088 [5] NCCL INFO Channel 01 : 13[8b000] -> 12[88000] via P2P/IPC/read
+jean-zay-iam40:289971:290093 [4] NCCL INFO Channel 01 : 12[88000] -> 11[4c000] via P2P/IPC/read
+jean-zay-iam37:261380:261497 [1] NCCL INFO Channel 00 : 1[b000] -> 0[7000] via P2P/IPC/read
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 01 : 24[7000] -> 8[7000] [receive] via NET/IB/1
+jean-zay-iam37:261381:261500 [2] NCCL INFO Channel 01 : 2[48000] -> 1[b000] via P2P/IPC/read
+jean-zay-iam37:261382:261501 [3] NCCL INFO Channel 01 : 3[4c000] -> 2[48000] via P2P/IPC/read
+jean-zay-iam37:261385:261506 [6] NCCL INFO Channel 01 : 6[c8000] -> 5[8b000] via P2P/IPC/read
+jean-zay-iam37:261384:261496 [5] NCCL INFO Channel 01 : 5[8b000] -> 4[88000] via P2P/IPC/read
+jean-zay-iam37:261383:261499 [4] NCCL INFO Channel 01 : 4[88000] -> 3[4c000] via P2P/IPC/read
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 00 : 16[7000] -> 24[7000] [send] via NET/IB/1
+jean-zay-iam37:261380:261497 [1] NCCL INFO Channel 01 : 1[b000] -> 0[7000] via P2P/IPC/read
+jean-zay-iam41:276753:276870 [7] NCCL INFO Connected all trees
+jean-zay-iam41:276753:276870 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276753:276870 [7] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263022:263135 [7] NCCL INFO Connected all trees
+jean-zay-iam52:263022:263135 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263022:263135 [7] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276747:276867 [1] NCCL INFO Channel 00 : 17[b000] -> 8[7000] [send] via NET/IB/1
+jean-zay-iam40:289974:290092 [7] NCCL INFO Connected all trees
+jean-zay-iam40:289974:290092 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289974:290092 [7] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263018:263141 [3] NCCL INFO Connected all trees
+jean-zay-iam52:263018:263141 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263018:263141 [3] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 01 : 8[7000] -> 24[7000] [send] via NET/IB/1
+jean-zay-iam52:263019:263136 [4] NCCL INFO Connected all trees
+jean-zay-iam52:263019:263136 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263019:263136 [4] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261386:261498 [7] NCCL INFO Connected all trees
+jean-zay-iam37:261386:261498 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261386:261498 [7] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276750:276871 [4] NCCL INFO Connected all trees
+jean-zay-iam41:276750:276871 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276750:276871 [4] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 00 : 0[7000] -> 16[7000] [receive] via NET/IB/1
+jean-zay-iam41:276749:276869 [3] NCCL INFO Connected all trees
+jean-zay-iam41:276749:276869 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276749:276869 [3] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276752:276872 [6] NCCL INFO Connected all trees
+jean-zay-iam41:276752:276872 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263021:263137 [6] NCCL INFO Connected all trees
+jean-zay-iam41:276752:276872 [6] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263021:263137 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263021:263137 [6] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276751:276865 [5] NCCL INFO Connected all trees
+jean-zay-iam41:276751:276865 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276751:276865 [5] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 01 : 8[7000] -> 24[7000] [receive] via NET/IB/1
+jean-zay-iam52:263020:263139 [5] NCCL INFO Connected all trees
+jean-zay-iam52:263020:263139 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263020:263139 [5] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263017:263138 [2] NCCL INFO Connected all trees
+jean-zay-iam52:263017:263138 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263017:263138 [2] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289973:290089 [6] NCCL INFO Connected all trees
+jean-zay-iam40:289973:290089 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289973:290089 [6] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289970:290090 [3] NCCL INFO Connected all trees
+jean-zay-iam40:289970:290090 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289970:290090 [3] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 00 : 16[7000] -> 0[7000] [send] via NET/IB/1
+jean-zay-iam40:289971:290093 [4] NCCL INFO Connected all trees
+jean-zay-iam40:289971:290093 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289971:290093 [4] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289972:290088 [5] NCCL INFO Connected all trees
+jean-zay-iam40:289972:290088 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289972:290088 [5] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 01 : 24[7000] -> 8[7000] [send] via NET/IB/1
+jean-zay-iam37:261385:261506 [6] NCCL INFO Connected all trees
+jean-zay-iam37:261385:261506 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261385:261506 [6] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261382:261501 [3] NCCL INFO Connected all trees
+jean-zay-iam37:261382:261501 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261382:261501 [3] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261383:261499 [4] NCCL INFO Connected all trees
+jean-zay-iam37:261383:261499 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261383:261499 [4] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261384:261496 [5] NCCL INFO Connected all trees
+jean-zay-iam37:261384:261496 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261384:261496 [5] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 00 : 24[7000] -> 16[7000] [receive] via NET/IB/1
+jean-zay-iam37:261381:261500 [2] NCCL INFO Connected all trees
+jean-zay-iam37:261381:261500 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261381:261500 [2] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261379:261471 [0] NCCL INFO Channel 01 : 8[7000] -> 0[7000] [receive] via NET/IB/1
+jean-zay-iam52:263015:263140 [0] NCCL INFO Channel 00 : 24[7000] -> 16[7000] [send] via NET/IB/1
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 00 : 17[b000] -> 8[7000] [receive] via NET/IB/1
+jean-zay-iam41:276747:276867 [1] NCCL INFO Channel 00 : 17[b000] -> 16[7000] via P2P/IPC/read
+jean-zay-iam41:276747:276867 [1] NCCL INFO Channel 01 : 17[b000] -> 16[7000] via P2P/IPC/read
+jean-zay-iam52:263015:263140 [0] NCCL INFO Connected all trees
+jean-zay-iam52:263015:263140 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263015:263140 [0] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289967:290091 [0] NCCL INFO Channel 01 : 8[7000] -> 0[7000] [send] via NET/IB/1
+jean-zay-iam41:276746:276868 [0] NCCL INFO Channel 01 : 16[7000] -> 9[b000] [send] via NET/IB/1
+jean-zay-iam52:263016:263142 [1] NCCL INFO Connected all trees
+jean-zay-iam52:263016:263142 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam52:263016:263142 [1] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam52:263019:263136 [4] NCCL INFO comm 0x14b1a8002fb0 rank 28 nranks 32 cudaDev 4 busId 88000 - Init COMPLETE
+jean-zay-iam52:263020:263139 [5] NCCL INFO comm 0x151418002fb0 rank 29 nranks 32 cudaDev 5 busId 8b000 - Init COMPLETE
+jean-zay-iam52:263016:263142 [1] NCCL INFO comm 0x145588002fb0 rank 25 nranks 32 cudaDev 1 busId b000 - Init COMPLETE
+jean-zay-iam52:263015:263140 [0] NCCL INFO comm 0x14c858002fb0 rank 24 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+jean-zay-iam52:263017:263138 [2] NCCL INFO comm 0x14e858002fb0 rank 26 nranks 32 cudaDev 2 busId 48000 - Init COMPLETE
+jean-zay-iam52:263018:263141 [3] NCCL INFO comm 0x150208002fb0 rank 27 nranks 32 cudaDev 3 busId 4c000 - Init COMPLETE
+jean-zay-iam52:263021:263137 [6] NCCL INFO comm 0x151df8002fb0 rank 30 nranks 32 cudaDev 6 busId c8000 - Init COMPLETE
+jean-zay-iam52:263022:263135 [7] NCCL INFO comm 0x152728002fb0 rank 31 nranks 32 cudaDev 7 busId cb000 - Init COMPLETE
+jean-zay-iam41:276748:276866 [2] NCCL INFO Connected all trees
+jean-zay-iam41:276748:276866 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276748:276866 [2] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289968:290086 [1] NCCL INFO Channel 00 : 9[b000] -> 8[7000] via P2P/IPC/read
+jean-zay-iam40:289968:290086 [1] NCCL INFO Channel 01 : 9[b000] -> 8[7000] via P2P/IPC/read
+jean-zay-iam37:261379:261471 [0] NCCL INFO Connected all trees
+jean-zay-iam37:261379:261471 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261379:261471 [0] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276746:276868 [0] NCCL INFO Connected all trees
+jean-zay-iam41:276746:276868 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276746:276868 [0] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261380:261497 [1] NCCL INFO Connected all trees
+jean-zay-iam37:261380:261497 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam37:261380:261497 [1] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam37:261380:261497 [1] NCCL INFO comm 0x151790002fb0 rank 1 nranks 32 cudaDev 1 busId b000 - Init COMPLETE
+jean-zay-iam37:261379:261471 [0] NCCL INFO comm 0x151f24002fb0 rank 0 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+jean-zay-iam37:261382:261501 [3] NCCL INFO comm 0x14a538002fb0 rank 3 nranks 32 cudaDev 3 busId 4c000 - Init COMPLETE
+jean-zay-iam37:261381:261500 [2] NCCL INFO comm 0x151028002fb0 rank 2 nranks 32 cudaDev 2 busId 48000 - Init COMPLETE
+jean-zay-iam37:261383:261499 [4] NCCL INFO comm 0x152340002fb0 rank 4 nranks 32 cudaDev 4 busId 88000 - Init COMPLETE
+jean-zay-iam37:261384:261496 [5] NCCL INFO comm 0x14d048002fb0 rank 5 nranks 32 cudaDev 5 busId 8b000 - Init COMPLETE
+jean-zay-iam37:261379:261379 [0] NCCL INFO Launch mode Parallel
+jean-zay-iam37:261386:261498 [7] NCCL INFO comm 0x1519b0002fb0 rank 7 nranks 32 cudaDev 7 busId cb000 - Init COMPLETE
+jean-zay-iam37:261385:261506 [6] NCCL INFO comm 0x14bd98002fb0 rank 6 nranks 32 cudaDev 6 busId c8000 - Init COMPLETE
+jean-zay-iam41:276747:276867 [1] NCCL INFO Connected all trees
+jean-zay-iam41:276747:276867 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam41:276747:276867 [1] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276749:276869 [3] NCCL INFO comm 0x14d508002fb0 rank 19 nranks 32 cudaDev 3 busId 4c000 - Init COMPLETE
+jean-zay-iam40:289967:290091 [0] NCCL INFO Connected all trees
+jean-zay-iam41:276748:276866 [2] NCCL INFO comm 0x14ae78002fb0 rank 18 nranks 32 cudaDev 2 busId 48000 - Init COMPLETE
+jean-zay-iam40:289967:290091 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289967:290091 [0] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam41:276747:276867 [1] NCCL INFO comm 0x14d928002fb0 rank 17 nranks 32 cudaDev 1 busId b000 - Init COMPLETE
+jean-zay-iam41:276750:276871 [4] NCCL INFO comm 0x146d68002fb0 rank 20 nranks 32 cudaDev 4 busId 88000 - Init COMPLETE
+jean-zay-iam41:276753:276870 [7] NCCL INFO comm 0x1523f8002fb0 rank 23 nranks 32 cudaDev 7 busId cb000 - Init COMPLETE
+jean-zay-iam41:276746:276868 [0] NCCL INFO comm 0x152f60002fb0 rank 16 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+jean-zay-iam41:276751:276865 [5] NCCL INFO comm 0x14c788002fb0 rank 21 nranks 32 cudaDev 5 busId 8b000 - Init COMPLETE
+jean-zay-iam41:276752:276872 [6] NCCL INFO comm 0x14e538002fb0 rank 22 nranks 32 cudaDev 6 busId c8000 - Init COMPLETE
+jean-zay-iam40:289968:290086 [1] NCCL INFO Connected all trees
+jean-zay-iam40:289968:290086 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289968:290086 [1] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289969:290087 [2] NCCL INFO Connected all trees
+jean-zay-iam40:289969:290087 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 8/8/512
+jean-zay-iam40:289969:290087 [2] NCCL INFO 2 coll channels, 2 p2p channels, 1 p2p channels per peer
+jean-zay-iam40:289969:290087 [2] NCCL INFO comm 0x154f98002fb0 rank 10 nranks 32 cudaDev 2 busId 48000 - Init COMPLETE
+jean-zay-iam40:289971:290093 [4] NCCL INFO comm 0x1529e8002fb0 rank 12 nranks 32 cudaDev 4 busId 88000 - Init COMPLETE
+jean-zay-iam40:289970:290090 [3] NCCL INFO comm 0x14ee38002fb0 rank 11 nranks 32 cudaDev 3 busId 4c000 - Init COMPLETE
+jean-zay-iam40:289973:290089 [6] NCCL INFO comm 0x145bb0002fb0 rank 14 nranks 32 cudaDev 6 busId c8000 - Init COMPLETE
+jean-zay-iam40:289972:290088 [5] NCCL INFO comm 0x14d508002fb0 rank 13 nranks 32 cudaDev 5 busId 8b000 - Init COMPLETE
+jean-zay-iam40:289968:290086 [1] NCCL INFO comm 0x14d558002fb0 rank 9 nranks 32 cudaDev 1 busId b000 - Init COMPLETE
+jean-zay-iam40:289974:290092 [7] NCCL INFO comm 0x1494b8002fb0 rank 15 nranks 32 cudaDev 7 busId cb000 - Init COMPLETE
+jean-zay-iam40:289967:290091 [0] NCCL INFO comm 0x14aa40002fb0 rank 8 nranks 32 cudaDev 0 busId 7000 - Init COMPLETE
+ignore me 17
+6:
+ duration: 3.9563 sec
+ algo throughput: 16176643777.3540 bps, 16.1766 Gbps
+ busbw: 15.6711  Gbps
+ignore me 17
+7:
+ duration: 4.1011 sec
+ algo throughput: 15605538666.8284 bps, 15.6055 Gbps
+ busbw: 15.1179  Gbps
+ignore me 17
+5:
+ duration: 4.0281 sec
+ algo throughput: 15888388696.7879 bps, 15.8884 Gbps
+ busbw: 15.3919  Gbps
+ignore me 17
+ignore me 17
+27:
+ duration: 4.1446 sec
+ algo throughput: 15441789907.3424 bps, 15.4418 Gbps
+ busbw: 14.9592  Gbps
+4:
+ duration: 4.1584 sec
+ algo throughput: 15390377253.3963 bps, 15.3904 Gbps
+ busbw: 14.9094  Gbps
+ignore me 17
+ignore me 17
+28:
+ duration: 4.0857 sec
+ignore me 17
+ algo throughput: 15664581341.3504 bps, 15.6646 Gbps
+ busbw: 15.1751  Gbps
+26:
+ duration: 4.1296 sec
+ algo throughput: 15497834133.7166 bps, 15.4978 Gbps
+ busbw: 15.0135  Gbps
+3:
+ duration: 4.1508 sec
+ algo throughput: 15418582053.9969 bps, 15.4186 Gbps
+ busbw: 14.9368  Gbps
+ignore me 17
+ignore me 17
+8:
+ duration: 4.2224 sec
+ algo throughput: 15157302718.4214 bps, 15.1573 Gbps
+ busbw: 14.6836  Gbps
+ignore me 17
+29:
+ duration: 4.0621 sec
+ algo throughput: 15755272218.1164 bps, 15.7553 Gbps
+ busbw: 15.2629  Gbps
+25:
+ duration: 4.1516 sec
+ algo throughput: 15415828590.9963 bps, 15.4158 Gbps
+ busbw: 14.9341  Gbps
+ignore me 17
+ignore me 17
+9:
+ duration: 4.0906 sec
+ algo throughput: 15645779547.2488 bps, 15.6458 Gbps
+ busbw: 15.1568  Gbps
+ignore me 17
+ignore me 17
+ignore me 17
+ignore me 17
+23:
+ duration: 4.1569 sec
+30:
+ duration: 4.0722 sec
+ algo throughput: 15716173146.2812 bps, 15.7162 Gbps
+1:
+ duration: 4.0663 sec
+ algo throughput: 15396140153.8145 bps, 15.3961 Gbps
+ busbw: 14.9150  Gbps
+ algo throughput: 15739134214.8659 bps, 15.7391 Gbps
+ busbw: 15.2473  Gbps
+ busbw: 15.2250  Gbps
+22:
+ duration: 4.0428 sec
+ algo throughput: 15830448441.2183 bps, 15.8304 Gbps
+ busbw: 15.3357  Gbps
+ignore me 17
+2:
+ duration: 4.1513 sec
+ algo throughput: 15416737873.4375 bps, 15.4167 Gbps
+ busbw: 14.9350  Gbps
+ignore me 17
+ignore me 17
+10:
+ duration: 4.1135 sec
+24:
+ duration: 4.0613 sec
+ algo throughput: 15758479220.2859 bps, 15.7585 Gbps
+ busbw: 15.2660  Gbps
+ algo throughput: 15558588332.8945 bps, 15.5586 Gbps
+ busbw: 15.0724  Gbps
+ignore me 17
+31:
+ duration: 4.1502 sec
+ algo throughput: 15420839540.9777 bps, 15.4208 Gbps
+ busbw: 14.9389  Gbps
+21:
+ duration: 4.1419 sec
+ algo throughput: 15451690470.9343 bps, 15.4517 Gbps
+ busbw: 14.9688  Gbps
+ignore me 17
+ignore me 17
+ignore me 17
+11:
+ duration: 4.0492 sec
+ algo throughput: 15805693708.4176 bps, 15.8057 Gbps
+20:
+ duration: 4.0993 sec
+ algo throughput: 15612440511.8644 bps, 15.6124 Gbps
+ busbw: 15.1246  Gbps
+0:
+ duration: 4.0120 sec
+ algo throughput: 15952303597.3018 bps, 15.9523 Gbps
+ busbw: 15.3118  Gbps
+ busbw: 15.4538  Gbps
+ignore me 17
+ignore me 17
+12:
+ duration: 4.1850 sec
+ algo throughput: 15292749814.3865 bps, 15.2927 Gbps
+ busbw: 14.8149  Gbps
+19:
+ duration: 4.0412 sec
+ algo throughput: 15836843924.5534 bps, 15.8368 Gbps
+ busbw: 15.3419  Gbps
+ignore me 17
+13:
+ duration: 4.0840 sec
+ algo throughput: 15670769926.9476 bps, 15.6708 Gbps
+ busbw: 15.1811  Gbps
+ignore me 17
+18:
+ duration: 4.1647 sec
+ algo throughput: 15367278261.5983 bps, 15.3673 Gbps
+ busbw: 14.8871  Gbps
+ignore me 17
+14:
+ duration: 4.0438 sec
+ algo throughput: 15826582974.8276 bps, 15.8266 Gbps
+ busbw: 15.3320  Gbps
+ignore me 17
+ignore me 17
+17:
+ duration: 4.1553 sec
+ algo throughput: 15401946302.4121 bps, 15.4019 Gbps
+15:
+ duration: 4.1608 sec
+ algo throughput: 15381558817.4705 bps, 15.3816 Gbps
+ busbw: 14.9206  Gbps
+ busbw: 14.9009  Gbps
+ignore me 17
+16:
+ duration: 4.0474 sec
+ algo throughput: 15812815660.2083 bps, 15.8128 Gbps
+ busbw: 15.3187  Gbps
+ignore me 555
+23:
+ duration: 1.5186 sec
+ algo throughput: 42143980222.5332 bps, 42.1440 Gbps
+ignore me 555
+ busbw: 40.8270  Gbps
+9:
+ duration: 1.5187 sec
+ algo throughput: 42140589448.6002 bps, 42.1406 Gbps
+ busbw: 40.8237  Gbps
+ignore me 555
+22:
+ duration: 1.5187 sec
+ algo throughput: 42140378571.5530 bps, 42.1404 Gbps
+ignore me 555
+ busbw: 40.8235  Gbps
+24:
+ duration: 1.5187 sec
+ algo throughput: 42142240285.3888 bps, 42.1422 Gbps
+ busbw: 40.8253  Gbps
+ignore me 555
+7:
+ duration: 1.5199 sec
+ algo throughput: 42108029847.7049 bps, 42.1080 Gbps
+ busbw: 40.7922  Gbps
+ignore me 555
+ignore me 555
+10:
+ duration: 1.5188 sec
+ algo throughput: 42138916267.0821 bps, 42.1389 Gbps
+ busbw: 40.8221  Gbps
+8:
+ duration: 1.5192 sec
+ algo throughput: 42126338602.2545 bps, 42.1263 Gbps
+ busbw: 40.8099  Gbps
+ignore me 555
+ignore me 555
+ignore me 555
+21:
+ duration: 1.5188 sec
+ algo throughput: 42139898494.4063 bps, 42.1399 Gbps
+ busbw: 40.8230  Gbps
+25:
+ duration: 1.5192 sec
+ algo throughput: 42127092502.8457 bps, 42.1271 Gbps
+ busbw: 40.8106  Gbps
+6:
+ duration: 1.5202 sec
+ algo throughput: 42099423136.7009 bps, 42.0994 Gbps
+ busbw: 40.7838  Gbps
+ignore me 555
+11:
+ duration: 1.5187 sec
+ algo throughput: 42141289163.4721 bps, 42.1413 Gbps
+ignore me 555
+ busbw: 40.8244  Gbps
+20:
+ duration: 1.5188 sec
+ algo throughput: 42139687792.2383 bps, 42.1397 Gbps
+ busbw: 40.8228  Gbps
+ignore me 555
+26:
+ duration: 1.5197 sec
+ algo throughput: 42113294024.4995 bps, 42.1133 Gbps
+ busbw: 40.7973  Gbps
+ignore me 555
+ignore me 555
+5:
+ duration: 1.5202 sec
+ algo throughput: 42100022978.8723 bps, 42.1000 Gbps
+ busbw: 40.7844  Gbps
+12:
+ duration: 1.5187 sec
+ algo throughput: 42141483180.7297 bps, 42.1415 Gbps
+ busbw: 40.8246  Gbps
+ignore me 555
+19:
+ duration: 1.5188 sec
+ algo throughput: 42139070669.3367 bps, 42.1391 Gbps
+ busbw: 40.8222  Gbps
+ignore me 555
+ignore me 555
+13:
+ duration: 1.5187 sec
+ algo throughput: 42140413754.7281 bps, 42.1404 Gbps
+27:
+ duration: 1.5202 sec
+ algo throughput: 42099139976.4359 bps, 42.0991 Gbps
+ busbw: 40.7835  Gbps
+ busbw: 40.8235  Gbps
+ignore me 555
+4:
+ duration: 1.5203 sec
+ algo throughput: 42097969076.0652 bps, 42.0980 Gbps
+ busbw: 40.7824  Gbps
+ignore me 555
+18:
+ duration: 1.5187 sec
+ algo throughput: 42141134996.9228 bps, 42.1411 Gbps
+ busbw: 40.8242  Gbps
+ignore me 555
+28:
+ duration: 1.5203 sec
+ algo throughput: 42097422955.6261 bps, 42.0974 Gbps
+ignore me 555
+ busbw: 40.7819  Gbps
+ignore me 555
+14:
+ duration: 1.5188 sec
+ algo throughput: 42139893361.7641 bps, 42.1399 Gbps
+ busbw: 40.8230  Gbps
+3:
+ duration: 1.5203 sec
+ algo throughput: 42097598433.0412 bps, 42.0976 Gbps
+ busbw: 40.7820  Gbps
+ignore me 555
+17:
+ duration: 1.5188 sec
+ algo throughput: 42139267495.6574 bps, 42.1393 Gbps
+ busbw: 40.8224  Gbps
+ignore me 555
+ignore me 555
+29:
+ duration: 1.5203 sec
+ algo throughput: 42096144082.6273 bps, 42.0961 Gbps
+ignore me 555
+ busbw: 40.7806  Gbps
+15:
+ duration: 1.5188 sec
+ algo throughput: 42137175969.6847 bps, 42.1372 Gbps
+ignore me 555
+ busbw: 40.8204  Gbps
+16:
+ duration: 1.5186 sec
+ algo throughput: 42144770940.2506 bps, 42.1448 Gbps
+ busbw: 40.8277  Gbps
+2:
+ duration: 1.5201 sec
+ algo throughput: 42101391688.1200 bps, 42.1014 Gbps
+ busbw: 40.7857  Gbps
+ignore me 555
+ignore me 555
+30:
+ duration: 1.5203 sec
+ algo throughput: 42096228974.3786 bps, 42.0962 Gbps
+ busbw: 40.7807  Gbps
+1:
+ duration: 1.5204 sec
+ algo throughput: 42095494315.5608 bps, 42.0955 Gbps
+ busbw: 40.7800  Gbps
+ignore me 555
+31:
+ duration: 1.5203 sec
+ algo throughput: 42096577970.2344 bps, 42.0966 Gbps
+ busbw: 40.7811  Gbps
+ignore me 555
+0:
+ duration: 1.5203 sec
+ algo throughput: 42097401467.1174 bps, 42.0974 Gbps
+ busbw: 40.7819  Gbps
+ignore me 17760
+19:
+ duration: 1.5271 sec
+ algo throughput: 41910600634.9022 bps, 41.9106 Gbps
+ busbw: 40.6009  Gbps
+ignore me 17760
+ignore me 17760
+18:
+ duration: 1.5270 sec
+ algo throughput: 41911582289.7142 bps, 41.9116 Gbps
+ busbw: 40.6018  Gbps
+20:
+ duration: 1.5276 sec
+ algo throughput: 41894987422.3905 bps, 41.8950 Gbps
+ busbw: 40.5858  Gbps
+ignore me 17760
+ignore me 17760
+17:
+ duration: 1.5270 sec
+ algo throughput: 41913406576.8859 bps, 41.9134 Gbps
+ busbw: 40.6036  Gbps
+ignore me 17760
+21:
+ duration: 1.5280 sec
+ algo throughput: 41885069299.4918 bps, 41.8851 Gbps
+ busbw: 40.5762  Gbps
+ignore me 17760
+14:
+ duration: 1.5272 sec
+ algo throughput: 41907314947.6113 bps, 41.9073 Gbps
+ busbw: 40.5977  Gbps
+15:
+ duration: 1.5270 sec
+ algo throughput: 41913242272.3447 bps, 41.9132 Gbps
+ busbw: 40.6035  Gbps
+ignore me 17760
+ignore me 17760
+13:
+ duration: 1.5277 sec
+ algo throughput: 41893273876.8880 bps, 41.8933 Gbps
+ busbw: 40.5841  Gbps
+ignore me 17760
+16:
+ duration: 1.5271 sec
+ algo throughput: 41909230280.3461 bps, 41.9092 Gbps
+ busbw: 40.5996  Gbps
+22:
+ duration: 1.5286 sec
+ algo throughput: 41869319488.2197 bps, 41.8693 Gbps
+ busbw: 40.5609  Gbps
+ignore me 17760
+ignore me 17760
+23:
+ duration: 1.5289 sec
+ algo throughput: 41861290350.4216 bps, 41.8613 Gbps
+12:
+ duration: 1.5281 sec
+ algo throughput: 41882850453.1701 bps, 41.8829 Gbps
+ busbw: 40.5740  Gbps
+ busbw: 40.5531  Gbps
+ignore me 17760
+11:
+ duration: 1.5286 sec
+ algo throughput: 41868966830.1641 bps, 41.8690 Gbps
+ busbw: 40.5606  Gbps
+ignore me 17760
+ignore me 17760
+24:
+ duration: 1.5291 sec
+ algo throughput: 41854797523.2289 bps, 41.8548 Gbps
+10:
+ duration: 1.5290 sec
+ algo throughput: 41858049187.4726 bps, 41.8580 Gbps
+ busbw: 40.5468  Gbps
+ busbw: 40.5500  Gbps
+ignore me 17760
+25:
+ duration: 1.5291 sec
+ algo throughput: 41855697296.6685 bps, 41.8557 Gbps
+ busbw: 40.5477  Gbps
+ignore me 17760
+ignore me 17760
+ignore me 17760
+9:
+ duration: 1.5296 sec
+ algo throughput: 41841767653.6339 bps, 41.8418 Gbps
+ busbw: 40.5342  Gbps
+6:
+ duration: 1.5292 sec
+ algo throughput: 41851931325.8954 bps, 41.8519 Gbps
+ busbw: 40.5441  Gbps
+7:
+ duration: 1.5294 sec
+ algo throughput: 41846364025.0241 bps, 41.8464 Gbps
+ busbw: 40.5387  Gbps
+ignore me 17760
+26:
+ duration: 1.5290 sec
+ algo throughput: 41856070811.5191 bps, 41.8561 Gbps
+ busbw: 40.5481  Gbps
+ignore me 17760
+ignore me 17760
+5:
+ duration: 1.5291 sec
+ algo throughput: 41855875143.2076 bps, 41.8559 Gbps
+ busbw: 40.5479  Gbps
+8:
+ duration: 1.5295 sec
+ algo throughput: 41843741534.2125 bps, 41.8437 Gbps
+ busbw: 40.5361  Gbps
+ignore me 17760
+27:
+ duration: 1.5290 sec
+ algo throughput: 41856588048.6577 bps, 41.8566 Gbps
+ busbw: 40.5486  Gbps
+ignore me 17760
+4:
+ duration: 1.5290 sec
+ algo throughput: 41856245346.9914 bps, 41.8562 Gbps
+ busbw: 40.5482  Gbps
+ignore me 17760
+28:
+ duration: 1.5290 sec
+ algo throughput: 41858071525.4799 bps, 41.8581 Gbps
+ busbw: 40.5500  Gbps
+ignore me 17760
+3:
+ duration: 1.5290 sec
+ algo throughput: 41857294677.8322 bps, 41.8573 Gbps
+ busbw: 40.5493  Gbps
+ignore me 17760
+29:
+ duration: 1.5289 sec
+ algo throughput: 41859219678.2562 bps, 41.8592 Gbps
+ busbw: 40.5511  Gbps
+ignore me 17760
+2:
+ duration: 1.5289 sec
+ algo throughput: 41859941759.2278 bps, 41.8599 Gbps
+ busbw: 40.5518  Gbps
+ignore me 17760
+30:
+ duration: 1.5289 sec
+ algo throughput: 41858890268.6218 bps, 41.8589 Gbps
+ busbw: 40.5508  Gbps
+ignore me 17760
+1:
+ duration: 1.5290 sec
+ algo throughput: 41856634528.5093 bps, 41.8566 Gbps
+ busbw: 40.5486  Gbps
+ignore me 17760
+31:
+ duration: 1.5290 sec
+ algo throughput: 41858450586.8372 bps, 41.8585 Gbps
+ busbw: 40.5504  Gbps
+ignore me 17760
+0:
+ duration: 1.5289 sec
+ algo throughput: 41860374323.0033 bps, 41.8604 Gbps
+ busbw: 40.5522  Gbps
+ignore me 568326
+18:
+ duration: 1.5292 sec
+ algo throughput: 41851192689.6061 bps, 41.8512 Gbps
+ busbw: 40.5433  Gbps
+ignore me 568326
+19:
+ duration: 1.5296 sec
+ algo throughput: 41840982602.8527 bps, 41.8410 Gbps
+ busbw: 40.5335  Gbps
+ignore me 568326
+17:
+ duration: 1.5292 sec
+ algo throughput: 41851389273.1359 bps, 41.8514 Gbps
+ busbw: 40.5435  Gbps
+ignore me 568326
+ignore me 568326
+ignore me 568326
+14:
+ duration: 1.5293 sec
+ algo throughput: 41850546358.8408 bps, 41.8505 Gbps
+ busbw: 40.5427  Gbps
+20:
+ duration: 1.5296 sec
+ algo throughput: 41841711605.3523 bps, 41.8417 Gbps
+15:
+ duration: 1.5292 sec
+ algo throughput: 41850900844.4322 bps, 41.8509 Gbps
+ busbw: 40.5342  Gbps
+ busbw: 40.5431  Gbps
+ignore me 568326
+ignore me 568326
+13:
+ duration: 1.5293 sec
+16:
+ duration: 1.5292 sec
+ algo throughput: 41851732548.4344 bps, 41.8517 Gbps
+ busbw: 40.5439  Gbps
+ algo throughput: 41849491619.2404 bps, 41.8495 Gbps
+ busbw: 40.5417  Gbps
+ignore me 568326
+21:
+ duration: 1.5296 sec
+ algo throughput: 41841051125.8787 bps, 41.8411 Gbps
+ busbw: 40.5335  Gbps
+ignore me 568326
+12:
+ duration: 1.5293 sec
+ algo throughput: 41848837733.7002 bps, 41.8488 Gbps
+ busbw: 40.5411  Gbps
+ignore me 568326
+ignore me 568326
+22:
+ duration: 1.5295 sec
+ algo throughput: 41842526390.1754 bps, 41.8425 Gbps
+11:
+ duration: 1.5292 sec
+ algo throughput: 41851402077.7964 bps, 41.8514 Gbps
+ busbw: 40.5349  Gbps
+ busbw: 40.5435  Gbps
+ignore me 568326
+ignore me 568326
+25:
+ duration: 1.5289 sec
+ algo throughput: 41860057899.5817 bps, 41.8601 Gbps
+ busbw: 40.5519  Gbps
+23:
+ duration: 1.5296 sec
+ algo throughput: 41841328471.6004 bps, 41.8413 Gbps
+ busbw: 40.5338  Gbps
+ignore me 568326
+ignore me 568326
+10:
+ duration: 1.5293 sec
+ algo throughput: 41850492064.7668 bps, 41.8505 Gbps
+ignore me 568326
+ busbw: 40.5427  Gbps
+26:
+ duration: 1.5289 sec
+ algo throughput: 41861009756.5066 bps, 41.8610 Gbps
+ busbw: 40.5529  Gbps
+24:
+ duration: 1.5293 sec
+ algo throughput: 41848595317.3039 bps, 41.8486 Gbps
+ busbw: 40.5408  Gbps
+ignore me 568326
+5:
+ duration: 1.5289 sec
+ algo throughput: 41860676073.0211 bps, 41.8607 Gbps
+ignore me 568326
+ busbw: 40.5525  Gbps
+ignore me 568326
+27:
+ duration: 1.5288 sec
+ algo throughput: 41861710376.5379 bps, 41.8617 Gbps
+ busbw: 40.5535  Gbps
+ignore me 568326
+6:
+ duration: 1.5292 sec
+ algo throughput: 41852910485.9393 bps, 41.8529 Gbps
+ busbw: 40.5450  Gbps
+ignore me 568326
+9:
+ duration: 1.5292 sec
+ algo throughput: 41850873996.3972 bps, 41.8509 Gbps
+ignore me 568326
+ busbw: 40.5430  Gbps
+4:
+ duration: 1.5288 sec
+ algo throughput: 41861534698.9598 bps, 41.8615 Gbps
+ busbw: 40.5534  Gbps
+7:
+ duration: 1.5293 sec
+ algo throughput: 41849369678.9657 bps, 41.8494 Gbps
+ busbw: 40.5416  Gbps
+ignore me 568326
+28:
+ duration: 1.5289 sec
+ algo throughput: 41861383911.2504 bps, 41.8614 Gbps
+ busbw: 40.5532  Gbps
+ignore me 568326
+ignore me 568326
+8:
+ duration: 1.5293 sec
+ algo throughput: 41848441035.8316 bps, 41.8484 Gbps
+3:
+ duration: 1.5289 sec
+ algo throughput: 41861481198.7633 bps, 41.8615 Gbps
+ busbw: 40.5533  Gbps
+ busbw: 40.5407  Gbps
+ignore me 568326
+29:
+ duration: 1.5289 sec
+ algo throughput: 41861138665.5933 bps, 41.8611 Gbps
+ busbw: 40.5530  Gbps
+ignore me 568326
+2:
+ duration: 1.5289 sec
+ algo throughput: 41861040340.5475 bps, 41.8610 Gbps
+ busbw: 40.5529  Gbps
+ignore me 568326
+30:
+ duration: 1.5289 sec
+ algo throughput: 41861393521.7231 bps, 41.8614 Gbps
+ignore me 568326
+ busbw: 40.5532  Gbps
+1:
+ duration: 1.5288 sec
+ algo throughput: 41863250360.5825 bps, 41.8633 Gbps
+ busbw: 40.5550  Gbps
+ignore me 568326
+31:
+ duration: 1.5289 sec
+ algo throughput: 41860930490.0206 bps, 41.8609 Gbps
+ busbw: 40.5528  Gbps
+ignore me 568326
+0:
+ duration: 1.5289 sec
+ algo throughput: 41861381313.3954 bps, 41.8614 Gbps
+ busbw: 40.5532  Gbps
+ignore me 18186434
+18:
+ duration: 1.5304 sec
+ algo throughput: 41819308451.5824 bps, 41.8193 Gbps
+ busbw: 40.5125  Gbps
+ignore me 18186434
+19:
+ duration: 1.5304 sec
+ algo throughput: 41819374415.9696 bps, 41.8194 Gbps
+ busbw: 40.5125  Gbps
+ignore me 18186434
+17:
+ duration: 1.5304 sec
+ algo throughput: 41819400154.7344 bps, 41.8194 Gbps
+ busbw: 40.5125  Gbps
+ignore me 18186434
+ignore me 18186434
+15:
+ duration: 1.5303 sec
+ algo throughput: 41821175681.0869 bps, 41.8212 Gbps
+20:
+ duration: 1.5304 sec
+ algo throughput: 41820265560.0101 bps, 41.8203 Gbps
+ busbw: 40.5134  Gbps
+ busbw: 40.5143  Gbps
+ignore me 18186434
+14:
+ duration: 1.5305 sec
+ algo throughput: 41817412474.7738 bps, 41.8174 Gbps
+ busbw: 40.5106  Gbps
+ignore me 18186434
+16:
+ duration: 1.5304 sec
+ algo throughput: 41820405171.5425 bps, 41.8204 Gbps
+ busbw: 40.5135  Gbps
+ignore me 18186434
+ignore me 18186434
+21:
+ duration: 1.5304 sec
+ algo throughput: 41820211341.2948 bps, 41.8202 Gbps
+ busbw: 40.5133  Gbps
+13:
+ duration: 1.5305 sec
+ algo throughput: 41815893542.3173 bps, 41.8159 Gbps
+ busbw: 40.5091  Gbps
+ignore me 18186434
+ignore me 18186434
+22:
+ duration: 1.5304 sec
+ algo throughput: 41819993958.8392 bps, 41.8200 Gbps
+ busbw: 40.5131  Gbps
+12:
+ duration: 1.5305 sec
+ algo throughput: 41816988451.4211 bps, 41.8170 Gbps
+ busbw: 40.5102  Gbps
+ignore me 18186434
+23:
+ duration: 1.5304 sec
+ algo throughput: 41820013685.7934 bps, 41.8200 Gbps
+ busbw: 40.5131  Gbps
+ignore me 18186434
+11:
+ duration: 1.5306 sec
+ algo throughput: 41813631070.6557 bps, 41.8136 Gbps
+ busbw: 40.5070  Gbps
+ignore me 18186434
+10:
+ duration: 1.5306 sec
+ algo throughput: 41813136230.6469 bps, 41.8131 Gbps
+ busbw: 40.5065  Gbps
+ignore me 18186434
+24:
+ duration: 1.5306 sec
+ algo throughput: 41813362805.8615 bps, 41.8134 Gbps
+ busbw: 40.5067  Gbps
+ignore me 18186434
+ignore me 18186434
+9:
+ duration: 1.5306 sec
+ algo throughput: 41814612837.9065 bps, 41.8146 Gbps
+25:
+ duration: 1.5311 sec
+ algo throughput: 41801050732.9013 bps, 41.8011 Gbps
+ busbw: 40.4948  Gbps
+ busbw: 40.5079  Gbps
+ignore me 18186434
+ignore me 18186434
+6:
+ duration: 1.5307 sec
+ algo throughput: 41811611108.9466 bps, 41.8116 Gbps
+ busbw: 40.5050  Gbps
+7:
+ duration: 1.5305 sec
+ algo throughput: 41815091867.5771 bps, 41.8151 Gbps
+ busbw: 40.5084  Gbps
+ignore me 18186434
+8:
+ duration: 1.5304 sec
+ algo throughput: 41818224707.1108 bps, 41.8182 Gbps
+ busbw: 40.5114  Gbps
+ignore me 18186434
+26:
+ duration: 1.5311 sec
+ algo throughput: 41799543931.1436 bps, 41.7995 Gbps
+ busbw: 40.4933  Gbps
+ignore me 18186434
+5:
+ duration: 1.5311 sec
+ algo throughput: 41800540982.4688 bps, 41.8005 Gbps
+ busbw: 40.4943  Gbps
+ignore me 18186434
+27:
+ duration: 1.5311 sec
+ algo throughput: 41798734639.3871 bps, 41.7987 Gbps
+ busbw: 40.4925  Gbps
+ignore me 18186434
+4:
+ duration: 1.5311 sec
+ algo throughput: 41799893567.7921 bps, 41.7999 Gbps
+ busbw: 40.4936  Gbps
+ignore me 18186434
+28:
+ duration: 1.5312 sec
+ algo throughput: 41798021113.2911 bps, 41.7980 Gbps
+ busbw: 40.4918  Gbps
+ignore me 18186434
+3:
+ duration: 1.5311 sec
+ algo throughput: 41799656984.3057 bps, 41.7997 Gbps
+ busbw: 40.4934  Gbps
+ignore me 18186434
+29:
+ duration: 1.5312 sec
+ignore me 18186434
+ algo throughput: 41797483455.9485 bps, 41.7975 Gbps
+ busbw: 40.4913  Gbps
+2:
+ duration: 1.5312 sec
+ algo throughput: 41797889916.8612 bps, 41.7979 Gbps
+ busbw: 40.4917  Gbps
+ignore me 18186434
+30:
+ duration: 1.5312 sec
+ algo throughput: 41797399459.7577 bps, 41.7974 Gbps
+ busbw: 40.4912  Gbps
+ignore me 18186434
+1:
+ duration: 1.5312 sec
+ algo throughput: 41796838922.8479 bps, 41.7968 Gbps
+ busbw: 40.4907  Gbps
+ignore me 18186434
+31:
+ duration: 1.5312 sec
+ algo throughput: 41798535248.2715 bps, 41.7985 Gbps
+ busbw: 40.4923  Gbps
+ignore me 18186434
+0:
+ duration: 1.5312 sec
+ algo throughput: 41797155891.1448 bps, 41.7972 Gbps
+ busbw: 40.4910  Gbps

bigscience/experiments/bandwidth/all_reduce_bench.py

@@ -0,0 +1,66 @@
+# python -m torch.distributed.run --nproc_per_node=2 all_reduce_bench.py
+
+import argparse
+import fcntl
+import os
+import socket
+import time
+import torch
+import torch.distributed as dist
+
+# note: this benchmark doesn't care how many gpus per node one has
+
+TRIALS = 5
+
+N = 500000
+M = 2000
+
+def printflock(*msgs):
+    """ print """
+    with open(__file__, "r") as fh:
+        fcntl.flock(fh, fcntl.LOCK_EX)
+        try:
+            print(*msgs)
+        finally:
+            fcntl.flock(fh, fcntl.LOCK_UN)
+
+def timed_allreduce(mat, id):
+    pre = time.perf_counter()
+    dist.all_reduce(mat)
+    printflock(f"ignore me {int(mat[0][0])}")  # required due to lazy evaluation
+    duration = time.perf_counter() - pre
+    tput = ((M*N*4*2)/duration)*8 # *2 is for send + receive, *8 for gigabits/second
+    size = M * N * 4 # 4 is fp32
+    n = dist.get_world_size()
+    busbw = (size / duration) * (2 * (n - 1) / n) * 8
+    printflock(f"{id}:\n",
+               f"duration: {duration:.4f} sec\n",
+               f"algo throughput: {tput:.4f} bps, {tput/1e9:.4f} Gbps\n",
+               f"busbw: {busbw / 1e9:.4f}  Gbps"
+    )
+
+def run(local_rank):
+    hostname = socket.gethostname()
+    id = f"{hostname}:{local_rank}"
+    global_rank = dist.get_rank()
+
+    printflock(f"{id} data size: {M*N*4/1e9} GB")
+    mat = torch.rand(N, M, dtype=torch.float32).cuda(local_rank)
+
+    for i in range(TRIALS):
+        dist.barrier()
+        if global_rank == 0:
+            print(f"\n\n\n-----------trial-{i}----------------")
+        timed_allreduce(mat, id)
+
+def init_processes(local_rank, fn, backend='nccl'):
+    torch.cuda.set_device(local_rank)
+    dist.init_process_group(backend)
+    fn(local_rank)
+
+
+if __name__ == "__main__":
+    rank = int(os.environ["LOCAL_RANK"])
+    printflock("local_rank: %d" % rank)
+    init_processes(local_rank=rank, fn=run)
+

bigscience/experiments/bandwidth/n1_16gb_all_reduce_bench.txt

@@ -0,0 +1,264 @@
+export NCCL_DEBUG=info
+python -m torch.distributed.launch --nproc_per_node=4 all_reduce_bench.py
+
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+local_rank: 2
+local_rank: 3
+local_rank: 1
+local_rank: 0
+0 data size: 4.0 GB
+2 data size: 4.0 GB
+1 data size: 4.0 GB
+3 data size: 4.0 GB
+r10i4n8:38029:38029 [0] NCCL INFO Bootstrap : Using [0]ib0:10.148.8.71<0> [1]ib1:10.149.8.71<0>
+r10i4n8:38029:38029 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r10i4n8:38029:38029 [0] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.8.71<0>
+r10i4n8:38029:38029 [0] NCCL INFO Using network IB
+NCCL version 2.7.8+cuda10.2
+r10i4n8:38030:38030 [1] NCCL INFO Bootstrap : Using [0]ib0:10.148.8.71<0> [1]ib1:10.149.8.71<0>
+r10i4n8:38030:38030 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r10i4n8:38030:38030 [1] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.8.71<0>
+r10i4n8:38030:38030 [1] NCCL INFO Using network IB
+r10i4n8:38032:38032 [3] NCCL INFO Bootstrap : Using [0]ib0:10.148.8.71<0> [1]ib1:10.149.8.71<0>
+r10i4n8:38032:38032 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r10i4n8:38031:38031 [2] NCCL INFO Bootstrap : Using [0]ib0:10.148.8.71<0> [1]ib1:10.149.8.71<0>
+r10i4n8:38031:38031 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r10i4n8:38032:38032 [3] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.8.71<0>
+r10i4n8:38032:38032 [3] NCCL INFO Using network IB
+r10i4n8:38031:38031 [2] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.8.71<0>
+r10i4n8:38031:38031 [2] NCCL INFO Using network IB
+r10i4n8:38029:38066 [0] NCCL INFO Channel 00/12 :    0   1   2   3
+r10i4n8:38029:38066 [0] NCCL INFO Channel 01/12 :    0   1   3   2
+r10i4n8:38029:38066 [0] NCCL INFO Channel 02/12 :    0   2   3   1
+r10i4n8:38029:38066 [0] NCCL INFO Channel 03/12 :    0   2   1   3
+r10i4n8:38030:38071 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r10i4n8:38032:38077 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r10i4n8:38029:38066 [0] NCCL INFO Channel 04/12 :    0   3   1   2
+r10i4n8:38031:38081 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r10i4n8:38029:38066 [0] NCCL INFO Channel 05/12 :    0   3   2   1
+r10i4n8:38029:38066 [0] NCCL INFO Channel 06/12 :    0   1   2   3
+r10i4n8:38029:38066 [0] NCCL INFO Channel 07/12 :    0   1   3   2
+r10i4n8:38029:38066 [0] NCCL INFO Channel 08/12 :    0   2   3   1
+r10i4n8:38030:38071 [1] NCCL INFO Trees [0] 2/-1/-1->1->0|0->1->2/-1/-1 [1] 3/-1/-1->1->-1|-1->1->3/-1/-1 [2] -1/-1/-1->1->3|3->1->-1/-1/-1 [3] 0/-1/-1->1->2|2->1->0/-1/-1 [4] 2/-1/-1->1->0|0->1->2/-1/-1 [5] 3/-1/-1->1->-1|-1->1->3/-1/-1 [6] 2/-1/-1->1->0|0->1->2/-1/-1 [7] 3/-1/-1->1->-1|-1->1->3/-1/-1 [8] -1/-1/-1->1->3|3->1->-1/-1/-1 [9] 0/-1/-1->1->2|2->1->0/-1/-1 [10] 2/-1/-1->1->0|0->1->2/-1/-1 [11] 3/-1/-1->1->-1|-1->1->3/-1/-1
+r10i4n8:38029:38066 [0] NCCL INFO Channel 09/12 :    0   2   1   3
+r10i4n8:38032:38077 [3] NCCL INFO Trees [0] -1/-1/-1->3->2|2->3->-1/-1/-1 [1] 0/-1/-1->3->1|1->3->0/-1/-1 [2] 1/-1/-1->3->0|0->3->1/-1/-1 [3] 2/-1/-1->3->-1|-1->3->2/-1/-1 [4] -1/-1/-1->3->2|2->3->-1/-1/-1 [5] 0/-1/-1->3->1|1->3->0/-1/-1 [6] -1/-1/-1->3->2|2->3->-1/-1/-1 [7] 0/-1/-1->3->1|1->3->0/-1/-1 [8] 1/-1/-1->3->0|0->3->1/-1/-1 [9] 2/-1/-1->3->-1|-1->3->2/-1/-1 [10] -1/-1/-1->3->2|2->3->-1/-1/-1 [11] 0/-1/-1->3->1|1->3->0/-1/-1
+r10i4n8:38029:38066 [0] NCCL INFO Channel 10/12 :    0   3   1   2
+r10i4n8:38029:38066 [0] NCCL INFO Channel 11/12 :    0   3   2   1
+r10i4n8:38031:38081 [2] NCCL INFO Trees [0] 3/-1/-1->2->1|1->2->3/-1/-1 [1] -1/-1/-1->2->0|0->2->-1/-1/-1 [2] 0/-1/-1->2->-1|-1->2->0/-1/-1 [3] 1/-1/-1->2->3|3->2->1/-1/-1 [4] 3/-1/-1->2->1|1->2->3/-1/-1 [5] -1/-1/-1->2->0|0->2->-1/-1/-1 [6] 3/-1/-1->2->1|1->2->3/-1/-1 [7] -1/-1/-1->2->0|0->2->-1/-1/-1 [8] 0/-1/-1->2->-1|-1->2->0/-1/-1 [9] 1/-1/-1->2->3|3->2->1/-1/-1 [10] 3/-1/-1->2->1|1->2->3/-1/-1 [11] -1/-1/-1->2->0|0->2->-1/-1/-1
+r10i4n8:38030:38071 [1] NCCL INFO Setting affinity for GPU 1 to 0fffff00,000fffff
+r10i4n8:38032:38077 [3] NCCL INFO Setting affinity for GPU 3 to ffff,f00000ff,fff00000
+r10i4n8:38031:38081 [2] NCCL INFO Setting affinity for GPU 2 to ffff,f00000ff,fff00000
+r10i4n8:38029:38066 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r10i4n8:38029:38066 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1|-1->0->1/-1/-1 [1] 2/-1/-1->0->3|3->0->2/-1/-1 [2] 3/-1/-1->0->2|2->0->3/-1/-1 [3] -1/-1/-1->0->1|1->0->-1/-1/-1 [4] 1/-1/-1->0->-1|-1->0->1/-1/-1 [5] 2/-1/-1->0->3|3->0->2/-1/-1 [6] 1/-1/-1->0->-1|-1->0->1/-1/-1 [7] 2/-1/-1->0->3|3->0->2/-1/-1 [8] 3/-1/-1->0->2|2->0->3/-1/-1 [9] -1/-1/-1->0->1|1->0->-1/-1/-1 [10] 1/-1/-1->0->-1|-1->0->1/-1/-1 [11] 2/-1/-1->0->3|3->0->2/-1/-1
+r10i4n8:38029:38066 [0] NCCL INFO Setting affinity for GPU 0 to 0fffff00,000fffff
+r10i4n8:38032:38077 [3] NCCL INFO Channel 00 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 00 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 00 : 1[1c000] -> 2[88000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 00 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 00 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 00 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 00 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 01 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 01 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 01 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 01 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 01 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 01 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 01 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 01 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 02 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 02 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 02 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 02 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 02 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 02 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 02 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 03 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 02 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 03 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 03 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 03 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 03 : 1[1c000] -> 2[88000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 03 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 03 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 03 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 03 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 04 : 1[1c000] -> 2[88000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 04 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 04 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 04 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 04 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 04 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 04 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 04 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 04 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 05 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 05 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 05 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 05 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 05 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 05 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 05 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 05 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 05 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 06 : 1[1c000] -> 2[88000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 06 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 06 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 06 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 06 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 06 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 06 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 07 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 07 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 07 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 07 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 07 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 07 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 07 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 07 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 08 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 08 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 08 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 08 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 08 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 08 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 08 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 09 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 08 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 09 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 09 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 09 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 09 : 1[1c000] -> 2[88000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 09 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 09 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 09 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 09 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 10 : 1[1c000] -> 2[88000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 10 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 10 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 10 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 10 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 10 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 10 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 10 : 0[1a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 10 : 2[88000] -> 3[8a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 11 : 1[1c000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 11 : 0[1a000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 11 : 3[8a000] -> 2[88000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 11 : 2[88000] -> 1[1c000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 11 : 3[8a000] -> 1[1c000] via P2P/IPC
+r10i4n8:38031:38081 [2] NCCL INFO Channel 11 : 2[88000] -> 0[1a000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO Channel 11 : 1[1c000] -> 3[8a000] via P2P/IPC
+r10i4n8:38032:38077 [3] NCCL INFO Channel 11 : 3[8a000] -> 0[1a000] via P2P/IPC
+r10i4n8:38029:38066 [0] NCCL INFO Channel 11 : 0[1a000] -> 2[88000] via P2P/IPC
+r10i4n8:38030:38071 [1] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r10i4n8:38030:38071 [1] NCCL INFO comm 0x14dbb0001060 rank 1 nranks 4 cudaDev 1 busId 1c000 - Init COMPLETE
+r10i4n8:38031:38081 [2] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r10i4n8:38031:38081 [2] NCCL INFO comm 0x150950001060 rank 2 nranks 4 cudaDev 2 busId 88000 - Init COMPLETE
+r10i4n8:38032:38077 [3] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r10i4n8:38032:38077 [3] NCCL INFO comm 0x14ccd8001060 rank 3 nranks 4 cudaDev 3 busId 8a000 - Init COMPLETE
+r10i4n8:38029:38066 [0] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r10i4n8:38029:38066 [0] NCCL INFO comm 0x149bac001060 rank 0 nranks 4 cudaDev 0 busId 1a000 - Init COMPLETE
+r10i4n8:38029:38029 [0] NCCL INFO Launch mode Parallel
+ignore me 1
+ignore me 1
+ignore me 1
+0:
+ duration: 0.6633 sec
+ algo throughput: 96488131490.3540 bps, 96.4881 Gbps
+ busbw: 72.3661  Gbps
+1:
+ duration: 0.4507 sec
+ algo throughput: 142007505620.8443 bps, 142.0075 Gbps
+ busbw: 106.5056  Gbps
+2:
+ duration: 0.4203 sec
+ algo throughput: 152274131784.9601 bps, 152.2741 Gbps
+ busbw: 114.2056  Gbps
+ignore me 1
+3:
+ duration: 0.4225 sec
+ algo throughput: 151490688123.0876 bps, 151.4907 Gbps
+ busbw: 113.6180  Gbps
+ignore me 7
+ignore me 7
+ignore me 7
+3:
+ duration: 0.0479 sec
+ algo throughput: 1336658447010.4644 bps, 1336.6584 Gbps
+ busbw: 1002.4938  Gbps
+ignore me 7
+1:
+ duration: 0.0483 sec
+ algo throughput: 1325019685494.1951 bps, 1325.0197 Gbps
+ busbw: 993.7648  Gbps
+0:
+ duration: 0.0483 sec
+ algo throughput: 1323924013812.1467 bps, 1323.9240 Gbps
+ busbw: 992.9430  Gbps
+2:
+ duration: 0.0483 sec
+ algo throughput: 1324507343140.4290 bps, 1324.5073 Gbps
+ busbw: 993.3805  Gbps
+ignore me 31
+ignore me 31
+ignore me 31
+ignore me 31
+3:
+ duration: 0.0479 sec
+ algo throughput: 1335850436641.9412 bps, 1335.8504 Gbps
+ busbw: 1001.8878  Gbps
+2:
+ duration: 0.0478 sec
+ algo throughput: 1338717258044.6157 bps, 1338.7173 Gbps
+ busbw: 1004.0379  Gbps
+0:
+ duration: 0.0479 sec
+ algo throughput: 1336480609710.5195 bps, 1336.4806 Gbps
+ busbw: 1002.3605  Gbps
+1:
+ duration: 0.0479 sec
+ algo throughput: 1335644997705.6060 bps, 1335.6450 Gbps
+ busbw: 1001.7337  Gbps
+ignore me 124
+ignore me 124
+ignore me 124
+2:
+ duration: 0.0479 sec
+ algo throughput: 1337297229056.0354 bps, 1337.2972 Gbps
+ busbw: 1002.9729  Gbps
+0:
+ duration: 0.0479 sec
+ algo throughput: 1337048861958.8491 bps, 1337.0489 Gbps
+ busbw: 1002.7866  Gbps
+ignore me 124
+1:
+ duration: 0.0479 sec
+ algo throughput: 1337386146372.2676 bps, 1337.3861 Gbps
+ busbw: 1003.0396  Gbps
+3:
+ duration: 0.0480 sec
+ algo throughput: 1333613993474.4404 bps, 1333.6140 Gbps
+ busbw: 1000.2105  Gbps
+ignore me 496
+ignore me 496
+ignore me 496
+ignore me 496
+2:
+ duration: 0.0481 sec
+ algo throughput: 1329998661494.7930 bps, 1329.9987 Gbps
+ busbw: 997.4990  Gbps
+3:
+ duration: 0.0480 sec
+ algo throughput: 1333082662016.4126 bps, 1333.0827 Gbps
+ busbw: 999.8120  Gbps
+1:
+ duration: 0.0481 sec
+ algo throughput: 1330394518818.0288 bps, 1330.3945 Gbps
+ busbw: 997.7959  Gbps
+0:
+ duration: 0.0481 sec
+ algo throughput: 1329424219916.1433 bps, 1329.4242 Gbps
+ busbw: 997.0682  Gbps

bigscience/experiments/bandwidth/n1_32gb_all_reduce_bench.txt

@@ -0,0 +1,264 @@
+export NCCL_DEBUG=info
+python -m torch.distributed.launch --nproc_per_node=4 all_reduce_bench.py
+
+*****************************************
+Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+*****************************************
+local_rank: 3
+local_rank: 1
+local_rank: 0
+local_rank: 2
+0 data size: 4.0 GB
+2 data size: 4.0 GB
+3 data size: 4.0 GB
+1 data size: 4.0 GB
+r7i4n1:63120:63120 [0] NCCL INFO Bootstrap : Using [0]ib0:10.148.0.76<0> [1]ib1:10.149.0.76<0>
+r7i4n1:63120:63120 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i4n1:63120:63120 [0] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.0.76<0>
+r7i4n1:63120:63120 [0] NCCL INFO Using network IB
+NCCL version 2.7.8+cuda10.2
+r7i4n1:63123:63123 [3] NCCL INFO Bootstrap : Using [0]ib0:10.148.0.76<0> [1]ib1:10.149.0.76<0>
+r7i4n1:63121:63121 [1] NCCL INFO Bootstrap : Using [0]ib0:10.148.0.76<0> [1]ib1:10.149.0.76<0>
+r7i4n1:63123:63123 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i4n1:63121:63121 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i4n1:63121:63121 [1] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.0.76<0>
+r7i4n1:63123:63123 [3] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.0.76<0>
+r7i4n1:63121:63121 [1] NCCL INFO Using network IB
+r7i4n1:63123:63123 [3] NCCL INFO Using network IB
+r7i4n1:63122:63122 [2] NCCL INFO Bootstrap : Using [0]ib0:10.148.0.76<0> [1]ib1:10.149.0.76<0>
+r7i4n1:63122:63122 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+r7i4n1:63122:63122 [2] NCCL INFO NET/IB : Using [0]hfi1_2:1/IB [1]hfi1_0:1/IB [2]hfi1_3:1/IB [3]hfi1_1:1/IB ; OOB ib0:10.148.0.76<0>
+r7i4n1:63122:63122 [2] NCCL INFO Using network IB
+r7i4n1:63120:63191 [0] NCCL INFO Channel 00/12 :    0   1   2   3
+r7i4n1:63122:63194 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r7i4n1:63121:63193 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r7i4n1:63123:63192 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r7i4n1:63120:63191 [0] NCCL INFO Channel 01/12 :    0   1   3   2
+r7i4n1:63122:63194 [2] NCCL INFO Trees [0] 3/-1/-1->2->1|1->2->3/-1/-1 [1] -1/-1/-1->2->0|0->2->-1/-1/-1 [2] 0/-1/-1->2->-1|-1->2->0/-1/-1 [3] 1/-1/-1->2->3|3->2->1/-1/-1 [4] 3/-1/-1->2->1|1->2->3/-1/-1 [5] -1/-1/-1->2->0|0->2->-1/-1/-1 [6] 3/-1/-1->2->1|1->2->3/-1/-1 [7] -1/-1/-1->2->0|0->2->-1/-1/-1 [8] 0/-1/-1->2->-1|-1->2->0/-1/-1 [9] 1/-1/-1->2->3|3->2->1/-1/-1 [10] 3/-1/-1->2->1|1->2->3/-1/-1 [11] -1/-1/-1->2->0|0->2->-1/-1/-1
+r7i4n1:63121:63193 [1] NCCL INFO Trees [0] 2/-1/-1->1->0|0->1->2/-1/-1 [1] 3/-1/-1->1->-1|-1->1->3/-1/-1 [2] -1/-1/-1->1->3|3->1->-1/-1/-1 [3] 0/-1/-1->1->2|2->1->0/-1/-1 [4] 2/-1/-1->1->0|0->1->2/-1/-1 [5] 3/-1/-1->1->-1|-1->1->3/-1/-1 [6] 2/-1/-1->1->0|0->1->2/-1/-1 [7] 3/-1/-1->1->-1|-1->1->3/-1/-1 [8] -1/-1/-1->1->3|3->1->-1/-1/-1 [9] 0/-1/-1->1->2|2->1->0/-1/-1 [10] 2/-1/-1->1->0|0->1->2/-1/-1 [11] 3/-1/-1->1->-1|-1->1->3/-1/-1
+r7i4n1:63120:63191 [0] NCCL INFO Channel 02/12 :    0   2   3   1
+r7i4n1:63123:63192 [3] NCCL INFO Trees [0] -1/-1/-1->3->2|2->3->-1/-1/-1 [1] 0/-1/-1->3->1|1->3->0/-1/-1 [2] 1/-1/-1->3->0|0->3->1/-1/-1 [3] 2/-1/-1->3->-1|-1->3->2/-1/-1 [4] -1/-1/-1->3->2|2->3->-1/-1/-1 [5] 0/-1/-1->3->1|1->3->0/-1/-1 [6] -1/-1/-1->3->2|2->3->-1/-1/-1 [7] 0/-1/-1->3->1|1->3->0/-1/-1 [8] 1/-1/-1->3->0|0->3->1/-1/-1 [9] 2/-1/-1->3->-1|-1->3->2/-1/-1 [10] -1/-1/-1->3->2|2->3->-1/-1/-1 [11] 0/-1/-1->3->1|1->3->0/-1/-1
+r7i4n1:63122:63194 [2] NCCL INFO Setting affinity for GPU 2 to ffff,f00000ff,fff00000
+r7i4n1:63120:63191 [0] NCCL INFO Channel 03/12 :    0   2   1   3
+r7i4n1:63121:63193 [1] NCCL INFO Setting affinity for GPU 1 to 0fffff00,000fffff
+r7i4n1:63123:63192 [3] NCCL INFO Setting affinity for GPU 3 to ffff,f00000ff,fff00000
+r7i4n1:63120:63191 [0] NCCL INFO Channel 04/12 :    0   3   1   2
+r7i4n1:63122:63194 [2] NCCL INFO Channel 00 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 05/12 :    0   3   2   1
+r7i4n1:63120:63191 [0] NCCL INFO Channel 06/12 :    0   1   2   3
+r7i4n1:63120:63191 [0] NCCL INFO Channel 07/12 :    0   1   3   2
+r7i4n1:63120:63191 [0] NCCL INFO Channel 08/12 :    0   2   3   1
+r7i4n1:63120:63191 [0] NCCL INFO Channel 09/12 :    0   2   1   3
+r7i4n1:63120:63191 [0] NCCL INFO Channel 10/12 :    0   3   1   2
+r7i4n1:63120:63191 [0] NCCL INFO Channel 11/12 :    0   3   2   1
+r7i4n1:63120:63191 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 8/8/64
+r7i4n1:63120:63191 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1|-1->0->1/-1/-1 [1] 2/-1/-1->0->3|3->0->2/-1/-1 [2] 3/-1/-1->0->2|2->0->3/-1/-1 [3] -1/-1/-1->0->1|1->0->-1/-1/-1 [4] 1/-1/-1->0->-1|-1->0->1/-1/-1 [5] 2/-1/-1->0->3|3->0->2/-1/-1 [6] 1/-1/-1->0->-1|-1->0->1/-1/-1 [7] 2/-1/-1->0->3|3->0->2/-1/-1 [8] 3/-1/-1->0->2|2->0->3/-1/-1 [9] -1/-1/-1->0->1|1->0->-1/-1/-1 [10] 1/-1/-1->0->-1|-1->0->1/-1/-1 [11] 2/-1/-1->0->3|3->0->2/-1/-1
+r7i4n1:63120:63191 [0] NCCL INFO Setting affinity for GPU 0 to 0fffff00,000fffff
+r7i4n1:63123:63192 [3] NCCL INFO Channel 00 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 00 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 00 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 00 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 00 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 00 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 01 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 01 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 01 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 01 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 01 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 01 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 01 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 01 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 02 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 02 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 02 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 02 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 02 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 02 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 02 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 03 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 02 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 03 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 03 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 03 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 03 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 03 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 03 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 03 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 03 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 04 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 04 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 04 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 04 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 04 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 04 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 04 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 04 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 04 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 05 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 05 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 05 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 05 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 05 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 05 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 05 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 05 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 05 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 06 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 06 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 06 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 06 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 06 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 06 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 06 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 07 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 07 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 07 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 07 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 07 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 07 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 07 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 07 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 08 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 08 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 08 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 08 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 08 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 08 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 08 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 09 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 08 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 09 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 09 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 09 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 09 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 09 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 09 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 09 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 09 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 10 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 10 : 1[1c000] -> 2[88000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 10 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 10 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 10 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 10 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 10 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 10 : 0[1a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 10 : 2[88000] -> 3[8a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 11 : 1[1c000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 11 : 0[1a000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 11 : 3[8a000] -> 2[88000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 11 : 2[88000] -> 1[1c000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 11 : 3[8a000] -> 1[1c000] via P2P/IPC
+r7i4n1:63122:63194 [2] NCCL INFO Channel 11 : 2[88000] -> 0[1a000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO Channel 11 : 1[1c000] -> 3[8a000] via P2P/IPC
+r7i4n1:63123:63192 [3] NCCL INFO Channel 11 : 3[8a000] -> 0[1a000] via P2P/IPC
+r7i4n1:63120:63191 [0] NCCL INFO Channel 11 : 0[1a000] -> 2[88000] via P2P/IPC
+r7i4n1:63121:63193 [1] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r7i4n1:63121:63193 [1] NCCL INFO comm 0x148f80001060 rank 1 nranks 4 cudaDev 1 busId 1c000 - Init COMPLETE
+r7i4n1:63122:63194 [2] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r7i4n1:63122:63194 [2] NCCL INFO comm 0x152f00001060 rank 2 nranks 4 cudaDev 2 busId 88000 - Init COMPLETE
+r7i4n1:63123:63192 [3] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r7i4n1:63120:63191 [0] NCCL INFO 12 coll channels, 16 p2p channels, 4 p2p channels per peer
+r7i4n1:63123:63192 [3] NCCL INFO comm 0x146050001060 rank 3 nranks 4 cudaDev 3 busId 8a000 - Init COMPLETE
+r7i4n1:63120:63191 [0] NCCL INFO comm 0x14f24c001060 rank 0 nranks 4 cudaDev 0 busId 1a000 - Init COMPLETE
+r7i4n1:63120:63120 [0] NCCL INFO Launch mode Parallel
+ignore me 2
+ignore me 2
+ignore me 2
+3:
+ duration: 0.6125 sec
+ algo throughput: 104487664227.6784 bps, 104.4877 Gbps
+ busbw: 78.3657  Gbps
+0:
+ duration: 0.5584 sec
+ algo throughput: 114613183387.2373 bps, 114.6132 Gbps
+ busbw: 85.9599  Gbps
+2:
+ duration: 0.5140 sec
+ algo throughput: 124513941981.7996 bps, 124.5139 Gbps
+ busbw: 93.3855  Gbps
+ignore me 2
+1:
+ duration: 0.6245 sec
+ algo throughput: 102486528362.0469 bps, 102.4865 Gbps
+ busbw: 76.8649  Gbps
+ignore me 11
+ignore me 11
+ignore me 11
+ignore me 11
+1:
+ duration: 0.0479 sec
+ algo throughput: 1337346013047.7080 bps, 1337.3460 Gbps
+ busbw: 1003.0095  Gbps
+2:
+ duration: 0.0482 sec
+ algo throughput: 1328071705904.8621 bps, 1328.0717 Gbps
+ busbw: 996.0538  Gbps
+3:
+ duration: 0.0483 sec
+ algo throughput: 1325052362787.1750 bps, 1325.0524 Gbps
+ busbw: 993.7893  Gbps
+0:
+ duration: 0.0483 sec
+ algo throughput: 1325619195876.0120 bps, 1325.6192 Gbps
+ busbw: 994.2144  Gbps
+ignore me 45
+ignore me 45
+ignore me 45
+ignore me 45
+1:
+ duration: 0.0485 sec
+ algo throughput: 1319242278750.3755 bps, 1319.2423 Gbps
+ busbw: 989.4317  Gbps
+3:
+ duration: 0.0485 sec
+ algo throughput: 1320339103321.9136 bps, 1320.3391 Gbps
+ busbw: 990.2543  Gbps
+2:
+ duration: 0.0485 sec
+ algo throughput: 1318722904549.9961 bps, 1318.7229 Gbps
+ busbw: 989.0422  Gbps
+0:
+ duration: 0.0485 sec
+ algo throughput: 1320313583319.3479 bps, 1320.3136 Gbps
+ busbw: 990.2352  Gbps
+ignore me 183
+ignore me 183
+ignore me 183
+ignore me 183
+2:
+ duration: 0.0484 sec
+ algo throughput: 1322236494553.5015 bps, 1322.2365 Gbps
+ busbw: 991.6774  Gbps
+0:
+ duration: 0.0484 sec
+ algo throughput: 1321797181142.1807 bps, 1321.7972 Gbps
+ busbw: 991.3479  Gbps
+1:
+ duration: 0.0485 sec
+ algo throughput: 1318282723325.4265 bps, 1318.2827 Gbps
+ busbw: 988.7120  Gbps
+3:
+ duration: 0.0485 sec
+ algo throughput: 1320550708735.8535 bps, 1320.5507 Gbps
+ busbw: 990.4130  Gbps
+ignore me 733
+ignore me 733
+ignore me 733
+1:
+ duration: 0.0483 sec
+ algo throughput: 1323715979433.6658 bps, 1323.7160 Gbps
+ busbw: 992.7870  Gbps
+2:
+ duration: 0.0484 sec
+ algo throughput: 1322345035832.8503 bps, 1322.3450 Gbps
+ busbw: 991.7588  Gbps
+ignore me 733
+3:
+ duration: 0.0484 sec
+ algo throughput: 1323624408929.4016 bps, 1323.6244 Gbps
+ busbw: 992.7183  Gbps
+0:
+ duration: 0.0485 sec
+ algo throughput: 1319272113636.8833 bps, 1319.2721 Gbps
+ busbw: 989.4541  Gbps

bigscience/experiments/decoder-only-lm/gpt2_deepspeed.sh

@@ -0,0 +1,25 @@
+DATASET=openwebtext
+SERIALIZATION_DIR=${ALL_CCFRWORK}/experiments/gpt2_repro
+LOGGING_DIR=${ALL_CCFRWORK}/tensorboard/gpt2_repro
+
+export CUDA_VISIBLE_DEVICES=0
+
+#python scripts/run_clm.py \
+deepspeed scripts/run_clm.py \
+    --deepspeed configs/deepspeed/ds_zero2.json \
+    --model_type gpt2 \
+    --tokenizer_name gpt2 \
+    --dataset_name openwebtext --block_size 1024 \
+    --preprocessing_num_workers 76 \
+    --group_by_length --length_column_name length \
+    --do_train --do_eval \
+    --max_steps 15000 \
+    --max_train_samples 10000000 \
+    --per_device_train_batch_size 4 --gradient_accumulation_steps 16 \
+    --per_device_eval_batch_size 8 \
+    --output_dir outputs --overwrite_output_dir \
+    --report_to tensorboard \
+    --logging_strategy steps --logging_first_step --logging_dir logs --logging_steps 20 \
+    --eval_steps 250 --evaluation_strategy steps \
+    --save_strategy steps --save_steps 500 --save_total_limit 31 \
+    --n_layer 3 --n_embd 128 --n_inner 128 --n_head 8

bigscience/experiments/decoder-only-lm/models/decoder_only_t5/__init__.py

@@ -0,0 +1,2 @@
+from .configuration_decoder_only_t5 import DecoderOnlyT5Config
+from .modeling_decoder_only_t5 import DecoderOnlyT5Model, DecoderOnlyT5LMHeadModel

bigscience/experiments/decoder-only-lm/models/decoder_only_t5/configuration_decoder_only_t5.py

@@ -0,0 +1,134 @@
+# coding=utf-8
+# Copyright 2020, The T5 Authors and HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" Decoder Only T5 model configuration """
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+# T5_PRETRAINED_CONFIG_ARCHIVE_MAP = {
+#     "t5-small": "https://huggingface.co/t5-small/resolve/main/config.json",
+#     "t5-base": "https://huggingface.co/t5-base/resolve/main/config.json",
+#     "t5-large": "https://huggingface.co/t5-large/resolve/main/config.json",
+#     "t5-3b": "https://huggingface.co/t5-3b/resolve/main/config.json",
+#     "t5-11b": "https://huggingface.co/t5-11b/resolve/main/config.json",
+# }
+
+
+class DecoderOnlyT5Config(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a :class:`~transformers.T5Model` or a
+    :class:`~transformers.TFT5Model`. It is used to instantiate a T5 model according to the specified arguments,
+    defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration
+    to that of the T5 `t5-small <https://huggingface.co/t5-small>`__ architecture.
+
+    Configuration objects inherit from :class:`~transformers.PretrainedConfig` and can be used to control the model
+    outputs. Read the documentation from :class:`~transformers.PretrainedConfig` for more information.
+
+    Arguments:
+        vocab_size (:obj:`int`, `optional`, defaults to 32128):
+            Vocabulary size of the T5 model. Defines the number of different tokens that can be represented by the
+            :obj:`inputs_ids` passed when calling :class:`~transformers.T5Model` or :class:`~transformers.TFT5Model`.
+        d_model (:obj:`int`, `optional`, defaults to 512):
+            Size of the encoder layers and the pooler layer.
+        d_kv (:obj:`int`, `optional`, defaults to 64):
+            Size of the key, query, value projections per attention head. :obj:`d_kv` has to be equal to :obj:`d_model
+            // num_heads`.
+        d_ff (:obj:`int`, `optional`, defaults to 2048):
+            Size of the intermediate feed forward layer in each :obj:`T5Block`.
+        num_layers (:obj:`int`, `optional`, defaults to 6):
+            Number of hidden layers in the Transformer encoder.
+        num_decoder_layers (:obj:`int`, `optional`):
+            Number of hidden layers in the Transformer decoder. Will use the same value as :obj:`num_layers` if not
+            set.
+        num_heads (:obj:`int`, `optional`, defaults to 8):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        relative_attention_num_buckets (:obj:`int`, `optional`, defaults to 32):
+            The number of buckets to use for each attention layer.
+        dropout_rate (:obj:`float`, `optional`, defaults to 0.1):
+            The ratio for all dropout layers.
+        layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-6):
+            The epsilon used by the layer normalization layers.
+        initializer_factor (:obj:`float`, `optional`, defaults to 1):
+            A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
+            testing).
+        feed_forward_proj (:obj:`string`, `optional`, defaults to :obj:`"relu"`):
+            Type of feed forward layer to be used. Should be one of :obj:`"relu"` or :obj:`"gated-gelu"`. T5v1.1 uses
+            the :obj:`"gated-gelu"` feed forward projection. Original T5 uses :obj:`"relu"`.
+        use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
+            Whether or not the model should return the last key/values attentions (not used by all models).
+        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
+            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
+    """
+    model_type = "decoder_only_t5"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=32128,
+        d_model=512,
+        d_kv=64,
+        d_ff=2048,
+        num_layers=6,
+        # num_decoder_layers=None,
+        num_heads=8,
+        relative_attention_num_buckets=32,
+        dropout_rate=0.1,
+        layer_norm_epsilon=1e-6,
+        initializer_factor=1.0,
+        feed_forward_proj="relu",
+        is_encoder_decoder=False,
+        use_cache=True,
+        pad_token_id=0,
+        eos_token_id=1,
+        gradient_checkpointing=False,
+        **kwargs
+    ):
+        super().__init__(
+            pad_token_id=pad_token_id,
+            eos_token_id=eos_token_id,
+            is_encoder_decoder=is_encoder_decoder,
+            **kwargs,
+        )
+        self.vocab_size = vocab_size
+        self.d_model = d_model
+        self.d_kv = d_kv
+        self.d_ff = d_ff
+        self.num_layers = num_layers
+        # self.num_decoder_layers = (
+        #     num_decoder_layers if num_decoder_layers is not None else self.num_layers
+        # )  # default = symmetry
+        self.num_heads = num_heads
+        self.relative_attention_num_buckets = relative_attention_num_buckets
+        self.dropout_rate = dropout_rate
+        self.layer_norm_epsilon = layer_norm_epsilon
+        self.initializer_factor = initializer_factor
+        self.feed_forward_proj = feed_forward_proj
+        self.use_cache = use_cache
+        self.gradient_checkpointing = gradient_checkpointing
+
+    @property
+    def hidden_size(self):
+        return self.d_model
+
+    @property
+    def num_attention_heads(self):
+        return self.num_heads
+
+    @property
+    def num_hidden_layers(self):
+        return self.num_layers

bigscience/experiments/decoder-only-lm/models/decoder_only_t5/modeling_decoder_only_t5.py

@@ -0,0 +1,1517 @@
+# coding=utf-8
+# Copyright 2018 Mesh TensorFlow authors, T5 Authors and HuggingFace Inc. team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" Decoder-only T5 model. It's quick and dirty so please don't mind the unused arguments."""
+
+
+import copy
+import math
+import os
+import warnings
+
+import torch
+import torch.nn.functional as F
+from torch import nn
+from torch.nn import CrossEntropyLoss
+from torch.utils.checkpoint import checkpoint
+
+from transformers.activations import ACT2FN
+from transformers.file_utils import (
+    DUMMY_INPUTS,
+    DUMMY_MASK,
+    add_start_docstrings,
+    add_start_docstrings_to_model_forward,
+    replace_return_docstrings,
+)
+from transformers.modeling_outputs import (
+    BaseModelOutput,
+    BaseModelOutputWithPastAndCrossAttentions,
+    CausalLMOutputWithCrossAttentions,
+)
+from transformers.modeling_utils import PreTrainedModel, find_pruneable_heads_and_indices, prune_linear_layer
+from transformers.utils import logging
+from transformers.utils.model_parallel_utils import assert_device_map, get_device_map
+from .configuration_decoder_only_t5 import DecoderOnlyT5Config
+
+logger = logging.get_logger(__name__)
+
+_CONFIG_FOR_DOC = "DecoderOnlyT5Config"
+_TOKENIZER_FOR_DOC = "T5Tokenizer"
+
+####################################################
+# This dict contains ids and associated url
+# for the pretrained weights provided with the models
+####################################################
+# T5_PRETRAINED_MODEL_ARCHIVE_LIST = [
+#     "t5-small",
+#     "t5-base",
+#     "t5-large",
+#     "t5-3b",
+#     "t5-11b",
+#     # See all T5 models at https://huggingface.co/models?filter=t5
+# ]
+
+
+####################################################
+# This is a conversion method from TF 1.0 to PyTorch
+# More details: https://medium.com/huggingface/from-tensorflow-to-pytorch-265f40ef2a28
+####################################################
+def load_tf_weights_in_t5(model, config, tf_checkpoint_path):
+    """Load tf checkpoints in a pytorch model."""
+    try:
+        import re
+
+        import numpy as np
+        import tensorflow as tf
+    except ImportError:
+        logger.error(
+            "Loading a TensorFlow model in PyTorch, requires TensorFlow to be installed. Please see "
+            "https://www.tensorflow.org/install/ for installation instructions."
+        )
+        raise
+    tf_path = os.path.abspath(tf_checkpoint_path)
+    logger.info(f"Converting TensorFlow checkpoint from {tf_path}")
+    # Load weights from TF model
+    init_vars = tf.train.list_variables(tf_path)
+    names = []
+    tf_weights = {}
+    for name, shape in init_vars:
+        logger.info(f"Loading TF weight {name} with shape {shape}")
+        array = tf.train.load_variable(tf_path, name)
+        names.append(name)
+        tf_weights[name] = array
+
+    for txt_name in names:
+        name = txt_name.split("/")
+        # adam_v and adam_m are variables used in AdamWeightDecayOptimizer to calculated m and v
+        # which are not required for using pretrained model
+        if any(
+            n in ["adam_v", "adam_m", "AdamWeightDecayOptimizer", "AdamWeightDecayOptimizer_1", "global_step"]
+            for n in name
+        ):
+            logger.info(f"Skipping {'/'.join(name)}")
+            tf_weights.pop(txt_name, None)
+            continue
+        if "_slot_" in name[-1]:
+            logger.info(f"Skipping {'/'.join(name)}")
+            tf_weights.pop(txt_name, None)
+            continue
+        pointer = model
+        array = tf_weights[txt_name]
+
+        for m_name in name:
+            if re.fullmatch(r"[A-Za-z]+_\d+", m_name):
+                scope_names = re.split(r"_(\d+)", m_name)
+            else:
+                scope_names = [m_name]
+            if scope_names[0] in ["kernel", "scale", "embedding"]:
+                pointer = getattr(pointer, "weight")
+            elif scope_names[0] == "self_attention":
+                pointer = getattr(pointer, "layer")
+                pointer = pointer[0]
+            elif scope_names[0] == "enc_dec_attention":
+                pointer = getattr(pointer, "layer")
+                pointer = pointer[1]
+            elif scope_names[0] == "dense_relu_dense":
+                pointer = getattr(pointer, "layer")
+                pointer = pointer[2]
+            elif scope_names[0] == "rms_norm":
+                if hasattr(pointer, "layer_norm"):
+                    pointer = getattr(pointer, "layer_norm")
+                elif hasattr(pointer, "final_layer_norm"):
+                    pointer = getattr(pointer, "final_layer_norm")
+            elif scope_names[0] == "scale":
+                pointer = getattr(pointer, "weight")
+            elif scope_names[0] == "output_bias" or scope_names[0] == "beta":
+                pointer = getattr(pointer, "bias")
+            elif scope_names[0] == "squad":
+                pointer = getattr(pointer, "classifier")
+            elif scope_names[0] == "decoder" and name[1] == "logits":
+                continue
+            elif scope_names[0] == "logits":
+                pointer = getattr(pointer, "lm_head")
+            elif scope_names[0] == "wi" and len(scope_names) > 1 and scope_names[1].isdigit():
+                pointer = getattr(pointer, f"wi_{scope_names[1]}")
+                continue
+            else:
+                try:
+                    pointer = getattr(pointer, scope_names[0])
+                except AttributeError:
+                    logger.info(f"Skipping {'/'.join(name)}")
+                    continue
+            if len(scope_names) >= 2:
+                num = int(scope_names[1])
+                pointer = pointer[num]
+        if scope_names[0] not in ["kernel", "scale", "embedding"]:
+            pointer = getattr(pointer, "weight")
+        if scope_names[0] != "embedding":
+            logger.info(f"Transposing numpy weight of shape {array.shape} for {name}")
+            array = np.transpose(array)
+        try:
+            assert (
+                pointer.shape == array.shape
+            ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched"
+        except AssertionError as e:
+            e.args += (pointer.shape, array.shape)
+            raise
+        logger.info(f"Initialize PyTorch weight {name}")
+        pointer.data = torch.from_numpy(array.astype(np.float32))
+        tf_weights.pop(txt_name, None)
+
+    logger.info(f"Weights not copied to PyTorch model: {', '.join(tf_weights.keys())}.")
+    return model
+
+
+####################################################
+# PyTorch Models are constructed by sub-classing
+# - torch.nn.Module for the layers and
+# - PreTrainedModel for the models (it-self a sub-class of torch.nn.Module)
+####################################################
+PARALLELIZE_DOCSTRING = r"""
+    This is an experimental feature and is a subject to change at a moment's notice.
+
+    Uses a device map to distribute attention modules of the model across several devices. If no device map is given,
+    it will evenly distribute blocks across all devices.
+
+    Args:
+        device_map (:obj:`Dict[int, list]`, optional, defaults to None):
+            A dictionary that maps attention modules to devices. Note that the embedding module and LMHead are always
+            automatically mapped to the first device (for esoteric reasons). That means that the first device should
+            have fewer attention modules mapped to it than other devices. For reference, the t5 models have the
+            following number of attention modules:
+
+                - t5-small: 6
+                - t5-base: 12
+                - t5-large: 24
+                - t5-3b: 24
+                - t5-11b: 24
+
+    Example::
+
+            # Here is an example of a device map on a machine with 4 GPUs using t5-3b, which has a total of 24 attention modules:
+            model = T5ForConditionalGeneration.from_pretrained('t5-3b')
+            device_map = {0: [0, 1, 2],
+
+                         1: [3, 4, 5, 6, 7, 8, 9],
+                         2: [10, 11, 12, 13, 14, 15, 16],
+                         3: [17, 18, 19, 20, 21, 22, 23]}
+            model.parallelize(device_map)
+"""
+DEPARALLELIZE_DOCSTRING = r"""
+    Moves the model to cpu from a model parallel state.
+
+    Example::
+
+        # On a 4 GPU machine with t5-3b:
+        model = T5ForConditionalGeneration.from_pretrained('t5-3b')
+        device_map = {0: [0, 1, 2],
+
+                     1: [3, 4, 5, 6, 7, 8, 9],
+                     2: [10, 11, 12, 13, 14, 15, 16],
+                     3: [17, 18, 19, 20, 21, 22, 23]}
+        model.parallelize(device_map) # Splits the model across several devices
+        model.deparallelize() # Put the model back on cpu and cleans memory by calling torch.cuda.empty_cache()
+"""
+
+
+class DecoderOnlyT5LayerNorm(nn.Module):
+    def __init__(self, hidden_size, eps=1e-6):
+        """
+        Construct a layernorm module in the T5 style No bias and no subtraction of mean.
+        """
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(hidden_size))
+        self.variance_epsilon = eps
+
+    def forward(self, hidden_states):
+        # layer norm should always be calculated in float32
+        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
+        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+
+        # convert into float16 if necessary
+        if self.weight.dtype == torch.float16:
+            hidden_states = hidden_states.to(torch.float16)
+        return self.weight * hidden_states
+
+
+class DecoderOnlyT5DenseReluDense(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.wi = nn.Linear(config.d_model, config.d_ff, bias=False)
+        self.wo = nn.Linear(config.d_ff, config.d_model, bias=False)
+        self.dropout = nn.Dropout(config.dropout_rate)
+
+    def forward(self, hidden_states):
+        hidden_states = self.wi(hidden_states)
+        hidden_states = F.relu(hidden_states)
+        hidden_states = self.dropout(hidden_states)
+        hidden_states = self.wo(hidden_states)
+        return hidden_states
+
+
+class DecoderOnlyT5DenseGatedGeluDense(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.wi_0 = nn.Linear(config.d_model, config.d_ff, bias=False)
+        self.wi_1 = nn.Linear(config.d_model, config.d_ff, bias=False)
+        self.wo = nn.Linear(config.d_ff, config.d_model, bias=False)
+        self.dropout = nn.Dropout(config.dropout_rate)
+        self.gelu_act = ACT2FN["gelu_new"]
+
+    def forward(self, hidden_states):
+        hidden_gelu = self.gelu_act(self.wi_0(hidden_states))
+        hidden_linear = self.wi_1(hidden_states)
+        hidden_states = hidden_gelu * hidden_linear
+        hidden_states = self.dropout(hidden_states)
+        hidden_states = self.wo(hidden_states)
+        return hidden_states
+
+
+class DecoderOnlyT5LayerFF(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        if config.feed_forward_proj == "relu":
+            self.DenseReluDense = DecoderOnlyT5DenseReluDense(config)
+        elif config.feed_forward_proj == "gated-gelu":
+            self.DenseReluDense = DecoderOnlyT5DenseGatedGeluDense(config)
+        else:
+            raise ValueError(
+                f"{self.config.feed_forward_proj} is not supported. Choose between `relu` and `gated-gelu`"
+            )
+
+        self.layer_norm = DecoderOnlyT5LayerNorm(config.d_model, eps=config.layer_norm_epsilon)
+        self.dropout = nn.Dropout(config.dropout_rate)
+
+    def forward(self, hidden_states):
+        forwarded_states = self.layer_norm(hidden_states)
+        forwarded_states = self.DenseReluDense(forwarded_states)
+        hidden_states = hidden_states + self.dropout(forwarded_states)
+        return hidden_states
+
+
+class DecoderOnlyT5Attention(nn.Module):
+    def __init__(self, config: DecoderOnlyT5Config, has_relative_attention_bias=False):
+        super().__init__()
+        # self.is_decoder = config.is_decoder
+        self.has_relative_attention_bias = has_relative_attention_bias
+
+        self.relative_attention_num_buckets = config.relative_attention_num_buckets
+        self.d_model = config.d_model
+        self.key_value_proj_dim = config.d_kv
+        self.n_heads = config.num_heads
+        self.dropout = config.dropout_rate
+        self.inner_dim = self.n_heads * self.key_value_proj_dim
+
+        # Mesh TensorFlow initialization to avoid scaling before softmax
+        self.q = nn.Linear(self.d_model, self.inner_dim, bias=False)
+        self.k = nn.Linear(self.d_model, self.inner_dim, bias=False)
+        self.v = nn.Linear(self.d_model, self.inner_dim, bias=False)
+        self.o = nn.Linear(self.inner_dim, self.d_model, bias=False)
+
+        if self.has_relative_attention_bias:
+            self.relative_attention_bias = nn.Embedding(self.relative_attention_num_buckets, self.n_heads)
+        self.pruned_heads = set()
+        self.gradient_checkpointing = getattr(config, "gradient_checkpointing", False)
+
+    def prune_heads(self, heads):
+        if len(heads) == 0:
+            return
+        heads, index = find_pruneable_heads_and_indices(
+            heads, self.n_heads, self.key_value_proj_dim, self.pruned_heads
+        )
+        # Prune linear layers
+        self.q = prune_linear_layer(self.q, index)
+        self.k = prune_linear_layer(self.k, index)
+        self.v = prune_linear_layer(self.v, index)
+        self.o = prune_linear_layer(self.o, index, dim=1)
+        # Update hyper params
+        self.n_heads = self.n_heads - len(heads)
+        self.inner_dim = self.key_value_proj_dim * self.n_heads
+        self.pruned_heads = self.pruned_heads.union(heads)
+
+    @staticmethod
+    def _relative_position_bucket(relative_position, bidirectional=True, num_buckets=32, max_distance=128):
+        """
+        Adapted from Mesh Tensorflow:
+        https://github.com/tensorflow/mesh/blob/0cb87fe07da627bf0b7e60475d59f95ed6b5be3d/mesh_tensorflow/transformer/transformer_layers.py#L593
+
+        Translate relative position to a bucket number for relative attention. The relative position is defined as
+        memory_position - query_position, i.e. the distance in tokens from the attending position to the attended-to
+        position. If bidirectional=False, then positive relative positions are invalid. We use smaller buckets for
+        small absolute relative_position and larger buckets for larger absolute relative_positions. All relative
+        positions >=max_distance map to the same bucket. All relative positions <=-max_distance map to the same bucket.
+        This should allow for more graceful generalization to longer sequences than the model has been trained on
+
+        Args:
+            relative_position: an int32 Tensor
+            bidirectional: a boolean - whether the attention is bidirectional
+            num_buckets: an integer
+            max_distance: an integer
+
+        Returns:
+            a Tensor with the same shape as relative_position, containing int32 values in the range [0, num_buckets)
+        """
+        relative_buckets = 0
+        if bidirectional:
+            num_buckets //= 2
+            relative_buckets += (relative_position > 0).to(torch.long) * num_buckets
+            relative_position = torch.abs(relative_position)
+        else:
+            relative_position = -torch.min(relative_position, torch.zeros_like(relative_position))
+        # now relative_position is in the range [0, inf)
+
+        # half of the buckets are for exact increments in positions
+        max_exact = num_buckets // 2
+        is_small = relative_position < max_exact
+
+        # The other half of the buckets are for logarithmically bigger bins in positions up to max_distance
+        relative_postion_if_large = max_exact + (
+            torch.log(relative_position.float() / max_exact)
+            / math.log(max_distance / max_exact)
+            * (num_buckets - max_exact)
+        ).to(torch.long)
+        relative_postion_if_large = torch.min(
+            relative_postion_if_large, torch.full_like(relative_postion_if_large, num_buckets - 1)
+        )
+
+        relative_buckets += torch.where(is_small, relative_position, relative_postion_if_large)
+        return relative_buckets
+
+    def compute_bias(self, query_length, key_length):
+        """Compute binned relative position bias"""
+        context_position = torch.arange(query_length, dtype=torch.long)[:, None]
+        memory_position = torch.arange(key_length, dtype=torch.long)[None, :]
+        relative_position = memory_position - context_position  # shape (query_length, key_length)
+        relative_position_bucket = self._relative_position_bucket(
+            relative_position,  # shape (query_length, key_length)
+            bidirectional=False,
+            num_buckets=self.relative_attention_num_buckets,
+        )
+        relative_position_bucket = relative_position_bucket.to(self.relative_attention_bias.weight.device)
+        values = self.relative_attention_bias(relative_position_bucket)  # shape (query_length, key_length, num_heads)
+        values = values.permute([2, 0, 1]).unsqueeze(0)  # shape (1, num_heads, query_length, key_length)
+        return values
+
+    def forward(
+        self,
+        hidden_states,
+        mask=None,
+        key_value_states=None,
+        position_bias=None,
+        past_key_value=None,
+        layer_head_mask=None,
+        query_length=None,
+        use_cache=False,
+        output_attentions=False,
+    ):
+        """
+        Self-attention (if key_value_states is None) or attention over source sentence (provided by key_value_states).
+        """
+        # Input is (batch_size, seq_length, dim)
+        # Mask is (batch_size, key_length) (non-causal) or (batch_size, key_length, key_length)
+        # past_key_value[0] is (batch_size, n_heads, q_len - 1, dim_per_head)
+        batch_size, seq_length = hidden_states.shape[:2]
+
+        int_seq_length = int(seq_length)
+
+        real_seq_length = seq_length
+
+        if past_key_value is not None:
+            assert (
+                len(past_key_value) == 2
+            ), f"past_key_value should have 2 past states: keys and values. Got { len(past_key_value)} past states"
+            real_seq_length += past_key_value[0].shape[2] if query_length is None else query_length
+
+        key_length = real_seq_length if key_value_states is None else key_value_states.shape[1]
+
+        def shape(states):
+            """projection"""
+            return states.view(batch_size, -1, self.n_heads, self.key_value_proj_dim).transpose(1, 2)
+
+        def unshape(states):
+            """reshape"""
+            return states.transpose(1, 2).contiguous().view(batch_size, -1, self.inner_dim)
+
+        def project(hidden_states, proj_layer, key_value_states, past_key_value):
+            """projects hidden states correctly to key/query states"""
+            if key_value_states is None:
+                # self-attn
+                # (batch_size, n_heads, seq_length, dim_per_head)
+                hidden_states = shape(proj_layer(hidden_states))
+            elif past_key_value is None:
+                # cross-attn
+                # (batch_size, n_heads, seq_length, dim_per_head)
+                hidden_states = shape(proj_layer(key_value_states))
+
+            if past_key_value is not None:
+                if key_value_states is None:
+                    # self-attn
+                    # (batch_size, n_heads, key_length, dim_per_head)
+                    hidden_states = torch.cat([past_key_value, hidden_states], dim=2)
+                else:
+                    # cross-attn
+                    hidden_states = past_key_value
+            return hidden_states
+
+        # get query states
+        query_states = shape(self.q(hidden_states))  # (batch_size, n_heads, seq_length, dim_per_head)
+
+        # get key/value states
+        key_states = project(
+            hidden_states, self.k, key_value_states, past_key_value[0] if past_key_value is not None else None
+        )
+        value_states = project(
+            hidden_states, self.v, key_value_states, past_key_value[1] if past_key_value is not None else None
+        )
+
+        # compute scores
+        scores = torch.matmul(
+            query_states, key_states.transpose(3, 2)
+        )  # equivalent of torch.einsum("bnqd,bnkd->bnqk", query_states, key_states), compatible with onnx op>9
+
+        if position_bias is None:
+            if not self.has_relative_attention_bias:
+                position_bias = torch.zeros(
+                    (1, self.n_heads, real_seq_length, key_length), device=scores.device, dtype=scores.dtype
+                )
+                if self.training and self.gradient_checkpointing:
+                    position_bias.requires_grad = True
+            else:
+                position_bias = self.compute_bias(real_seq_length, key_length)
+
+            # if key and values are already calculated
+            # we want only the last query position bias
+            if past_key_value is not None:
+                position_bias = position_bias[:, :, -int_seq_length:, :]
+
+            if mask is not None:
+                position_bias = position_bias + mask  # (batch_size, n_heads, seq_length, key_length)
+
+        scores += position_bias
+        attn_weights = F.softmax(scores.float(), dim=-1).type_as(
+            scores
+        )  # (batch_size, n_heads, seq_length, key_length)
+        attn_weights = F.dropout(
+            attn_weights, p=self.dropout, training=self.training
+        )  # (batch_size, n_heads, seq_length, key_length)
+
+        # Mask heads if we want to
+        if layer_head_mask is not None:
+            attn_weights = attn_weights * layer_head_mask
+
+        attn_output = unshape(torch.matmul(attn_weights, value_states))  # (batch_size, seq_length, dim)
+        attn_output = self.o(attn_output)
+
+        present_key_value_state = (key_states, value_states) if use_cache else None
+        outputs = (attn_output,) + (present_key_value_state,) + (position_bias,)
+
+        if output_attentions:
+            outputs = outputs + (attn_weights,)
+        return outputs
+
+
+class DecoderOnlyT5LayerSelfAttention(nn.Module):
+    def __init__(self, config, has_relative_attention_bias=False):
+        super().__init__()
+        self.SelfAttention = DecoderOnlyT5Attention(config, has_relative_attention_bias=has_relative_attention_bias)
+        self.layer_norm = DecoderOnlyT5LayerNorm(config.d_model, eps=config.layer_norm_epsilon)
+        self.dropout = nn.Dropout(config.dropout_rate)
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        position_bias=None,
+        layer_head_mask=None,
+        past_key_value=None,
+        use_cache=False,
+        output_attentions=False,
+    ):
+        normed_hidden_states = self.layer_norm(hidden_states)
+        attention_output = self.SelfAttention(
+            normed_hidden_states,
+            mask=attention_mask,
+            position_bias=position_bias,
+            layer_head_mask=layer_head_mask,
+            past_key_value=past_key_value,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+        )
+        hidden_states = hidden_states + self.dropout(attention_output[0])
+        outputs = (hidden_states,) + attention_output[1:]  # add attentions if we output them
+        return outputs
+
+
+# class T5LayerCrossAttention(nn.Module):
+#     def __init__(self, config):
+#         super().__init__()
+#         self.EncDecAttention = T5Attention(config, has_relative_attention_bias=False)
+#         self.layer_norm = DecoderOnlyT5LayerNorm(config.d_model, eps=config.layer_norm_epsilon)
+#         self.dropout = nn.Dropout(config.dropout_rate)
+
+#     def forward(
+#         self,
+#         hidden_states,
+#         key_value_states,
+#         attention_mask=None,
+#         position_bias=None,
+#         layer_head_mask=None,
+#         past_key_value=None,
+#         use_cache=False,
+#         query_length=None,
+#         output_attentions=False,
+#     ):
+#         normed_hidden_states = self.layer_norm(hidden_states)
+#         attention_output = self.EncDecAttention(
+#             normed_hidden_states,
+#             mask=attention_mask,
+#             key_value_states=key_value_states,
+#             position_bias=position_bias,
+#             layer_head_mask=layer_head_mask,
+#             past_key_value=past_key_value,
+#             use_cache=use_cache,
+#             query_length=query_length,
+#             output_attentions=output_attentions,
+#         )
+#         layer_output = hidden_states + self.dropout(attention_output[0])
+#         outputs = (layer_output,) + attention_output[1:]  # add attentions if we output them
+#         return outputs
+
+
+class T5Block(nn.Module):
+    def __init__(self, config, has_relative_attention_bias=False):
+        super().__init__()
+        # self.is_decoder = config.is_decoder
+        self.layer = nn.ModuleList()
+        self.layer.append(DecoderOnlyT5LayerSelfAttention(config, has_relative_attention_bias=has_relative_attention_bias))
+        # if self.is_decoder:
+        #     self.layer.append(T5LayerCrossAttention(config))
+
+        self.layer.append(DecoderOnlyT5LayerFF(config))
+
+    def forward(
+        self,
+        hidden_states,
+        attention_mask=None,
+        position_bias=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        encoder_decoder_position_bias=None,
+        layer_head_mask=None,
+        cross_attn_layer_head_mask=None,
+        past_key_value=None,
+        use_cache=False,
+        output_attentions=False,
+        return_dict=True,
+    ):
+
+        if past_key_value is not None:
+            # assert self.is_decoder, "Only decoder can use `past_key_values`"
+            expected_num_past_key_values = 2 if encoder_hidden_states is None else 4
+
+            if len(past_key_value) != expected_num_past_key_values:
+                raise ValueError(
+                    f"There should be {expected_num_past_key_values} past states. "
+                    f"{'2 (past / key) for cross attention' if expected_num_past_key_values == 4 else ''}."
+                    f"Got {len(past_key_value)} past key / value states"
+                )
+
+            self_attn_past_key_value = past_key_value[:2]
+            cross_attn_past_key_value = past_key_value[2:]
+        else:
+            self_attn_past_key_value, cross_attn_past_key_value = None, None
+
+        self_attention_outputs = self.layer[0](
+            hidden_states,
+            attention_mask=attention_mask,
+            position_bias=position_bias,
+            layer_head_mask=layer_head_mask,
+            past_key_value=self_attn_past_key_value,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+        )
+        hidden_states, present_key_value_state = self_attention_outputs[:2]
+        attention_outputs = self_attention_outputs[2:]  # Keep self-attention outputs and relative position weights
+
+        # clamp inf values to enable fp16 training
+        if hidden_states.dtype == torch.float16 and torch.isinf(hidden_states).any():
+            clamp_value = torch.finfo(hidden_states.dtype).max - 1000
+            hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value)
+
+        do_cross_attention = encoder_hidden_states is not None
+        if do_cross_attention:
+            # the actual query length is unknown for cross attention
+            # if using past key value states. Need to inject it here
+            if present_key_value_state is not None:
+                query_length = present_key_value_state[0].shape[2]
+            else:
+                query_length = None
+
+            cross_attention_outputs = self.layer[1](
+                hidden_states,
+                key_value_states=encoder_hidden_states,
+                attention_mask=encoder_attention_mask,
+                position_bias=encoder_decoder_position_bias,
+                layer_head_mask=cross_attn_layer_head_mask,
+                past_key_value=cross_attn_past_key_value,
+                query_length=query_length,
+                use_cache=use_cache,
+                output_attentions=output_attentions,
+            )
+            hidden_states = cross_attention_outputs[0]
+
+            # clamp inf values to enable fp16 training
+            if hidden_states.dtype == torch.float16 and torch.isinf(hidden_states).any():
+                clamp_value = torch.finfo(hidden_states.dtype).max - 1000
+                hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value)
+
+            # Combine self attn and cross attn key value states
+            if present_key_value_state is not None:
+                present_key_value_state = present_key_value_state + cross_attention_outputs[1]
+
+            # Keep cross-attention outputs and relative position weights
+            attention_outputs = attention_outputs + cross_attention_outputs[2:]
+
+        # Apply Feed Forward layer
+        hidden_states = self.layer[-1](hidden_states)
+
+        # clamp inf values to enable fp16 training
+        if hidden_states.dtype == torch.float16 and torch.isinf(hidden_states).any():
+            clamp_value = torch.finfo(hidden_states.dtype).max - 1000
+            hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value)
+
+        outputs = (hidden_states,)
+
+        if use_cache:
+            outputs = outputs + (present_key_value_state,) + attention_outputs
+        else:
+            outputs = outputs + attention_outputs
+
+        return outputs  # hidden-states, present_key_value_states, (self-attention weights), (self-attention position bias), (cross-attention weights), (cross-attention position bias)
+
+
+class DecoderOnlyT5PreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
+    models.
+    """
+
+    config_class = DecoderOnlyT5Config
+    load_tf_weights = load_tf_weights_in_t5
+    base_model_prefix = "transformer"
+    is_parallelizable = True
+
+    @property
+    def dummy_inputs(self):
+        input_ids = torch.tensor(DUMMY_INPUTS)
+        input_mask = torch.tensor(DUMMY_MASK)
+        dummy_inputs = {
+            # "decoder_input_ids": input_ids,
+            "input_ids": input_ids,
+            # "decoder_attention_mask": input_mask,
+        }
+        return dummy_inputs
+
+    def _init_weights(self, module):
+        """Initialize the weights"""
+        factor = self.config.initializer_factor  # Used for testing weights initialization
+        if isinstance(module, DecoderOnlyT5LayerNorm):
+            module.weight.data.fill_(factor * 1.0)
+        elif isinstance(module, (DecoderOnlyT5Model)): #, T5ForConditionalGeneration, T5EncoderModel)):
+            # Mesh TensorFlow embeddings initialization
+            # See https://github.com/tensorflow/mesh/blob/fa19d69eafc9a482aff0b59ddd96b025c0cb207d/mesh_tensorflow/layers.py#L1624
+            module.shared.weight.data.normal_(mean=0.0, std=factor * 1.0)
+        elif isinstance(module, DecoderOnlyT5DenseReluDense):
+            # Mesh TensorFlow FF initialization
+            # See https://github.com/tensorflow/mesh/blob/master/mesh_tensorflow/transformer/transformer_layers.py#L56
+            # and https://github.com/tensorflow/mesh/blob/fa19d69eafc9a482aff0b59ddd96b025c0cb207d/mesh_tensorflow/layers.py#L89
+            module.wi.weight.data.normal_(mean=0.0, std=factor * ((self.config.d_model) ** -0.5))
+            if hasattr(module.wi, "bias") and module.wi.bias is not None:
+                module.wi.bias.data.zero_()
+            module.wo.weight.data.normal_(mean=0.0, std=factor * ((self.config.d_ff) ** -0.5))
+            if hasattr(module.wo, "bias") and module.wo.bias is not None:
+                module.wo.bias.data.zero_()
+        elif isinstance(module, DecoderOnlyT5DenseGatedGeluDense):
+            module.wi_0.weight.data.normal_(mean=0.0, std=factor * ((self.config.d_model) ** -0.5))
+            if hasattr(module.wi_0, "bias") and module.wi_0.bias is not None:
+                module.wi_0.bias.data.zero_()
+            module.wi_1.weight.data.normal_(mean=0.0, std=factor * ((self.config.d_model) ** -0.5))
+            if hasattr(module.wi_1, "bias") and module.wi_1.bias is not None:
+                module.wi_1.bias.data.zero_()
+            module.wo.weight.data.normal_(mean=0.0, std=factor * ((self.config.d_ff) ** -0.5))
+            if hasattr(module.wo, "bias") and module.wo.bias is not None:
+                module.wo.bias.data.zero_()
+        elif isinstance(module, DecoderOnlyT5Attention):
+            # Mesh TensorFlow attention initialization to avoid scaling before softmax
+            # See https://github.com/tensorflow/mesh/blob/fa19d69eafc9a482aff0b59ddd96b025c0cb207d/mesh_tensorflow/transformer/attention.py#L136
+            d_model = self.config.d_model
+            key_value_proj_dim = self.config.d_kv
+            n_heads = self.config.num_heads
+            module.q.weight.data.normal_(mean=0.0, std=factor * ((d_model * key_value_proj_dim) ** -0.5))
+            module.k.weight.data.normal_(mean=0.0, std=factor * (d_model ** -0.5))
+            module.v.weight.data.normal_(mean=0.0, std=factor * (d_model ** -0.5))
+            module.o.weight.data.normal_(mean=0.0, std=factor * ((n_heads * key_value_proj_dim) ** -0.5))
+            if module.has_relative_attention_bias:
+                module.relative_attention_bias.weight.data.normal_(mean=0.0, std=factor * ((d_model) ** -0.5))
+
+    # def _shift_right(self, input_ids):
+    #     decoder_start_token_id = self.config.decoder_start_token_id
+    #     pad_token_id = self.config.pad_token_id
+
+    #     assert (
+    #         decoder_start_token_id is not None
+    #     ), "self.model.config.decoder_start_token_id has to be defined. In T5 it is usually set to the pad_token_id. See T5 docs for more information"
+
+    #     # shift inputs to the right
+    #     if is_torch_fx_proxy(input_ids):
+    #         # Item assignment is not supported natively for proxies.
+    #         shifted_input_ids = torch.full(input_ids.shape[:-1] + (1,), decoder_start_token_id)
+    #         shifted_input_ids = torch.cat([shifted_input_ids, input_ids[..., :-1]], dim=-1)
+    #     else:
+    #         shifted_input_ids = input_ids.new_zeros(input_ids.shape)
+    #         shifted_input_ids[..., 1:] = input_ids[..., :-1].clone()
+    #         shifted_input_ids[..., 0] = decoder_start_token_id
+
+    #     assert pad_token_id is not None, "self.model.config.pad_token_id has to be defined."
+    #     # replace possible -100 values in labels by `pad_token_id`
+    #     shifted_input_ids.masked_fill_(shifted_input_ids == -100, pad_token_id)
+
+    #     assert torch.all(shifted_input_ids >= 0).item(), "Verify that `shifted_input_ids` has only positive values"
+
+    #     return shifted_input_ids
+
+
+class DecoderOnlyT5Stack(DecoderOnlyT5PreTrainedModel):
+    def __init__(self, config, embed_tokens=None):
+        super().__init__(config)
+
+        self.embed_tokens = embed_tokens
+        # self.is_decoder = config.is_decoder
+
+        self.block = nn.ModuleList(
+            [T5Block(config, has_relative_attention_bias=bool(i == 0)) for i in range(config.num_layers)]
+        )
+        self.final_layer_norm = DecoderOnlyT5LayerNorm(config.d_model, eps=config.layer_norm_epsilon)
+        self.dropout = nn.Dropout(config.dropout_rate)
+
+        self.init_weights()
+        # Model parallel
+        self.model_parallel = False
+        self.device_map = None
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        # Check validity of device_map
+        self.device_map = (
+            get_device_map(len(self.block), range(torch.cuda.device_count())) if device_map is None else device_map
+        )
+        assert_device_map(self.device_map, len(self.block))
+        self.model_parallel = True
+        self.first_device = "cpu" if "cpu" in self.device_map.keys() else "cuda:" + str(min(self.device_map.keys()))
+        self.last_device = "cuda:" + str(max(self.device_map.keys()))
+        # Load onto devices
+        for k, v in self.device_map.items():
+            for layer in v:
+                cuda_device = "cuda:" + str(k)
+                self.block[layer] = self.block[layer].to(cuda_device)
+
+        # Set embed_tokens to first layer
+        self.embed_tokens = self.embed_tokens.to(self.first_device)
+        # Set final layer norm to last device
+        self.final_layer_norm = self.final_layer_norm.to(self.last_device)
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.model_parallel = False
+        self.device_map = None
+        self.first_device = "cpu"
+        self.last_device = "cpu"
+        for i in range(len(self.block)):
+            self.block[i] = self.block[i].to("cpu")
+        self.embed_tokens = self.embed_tokens.to("cpu")
+        self.final_layer_norm = self.final_layer_norm.to("cpu")
+        torch.cuda.empty_cache()
+
+    def get_input_embeddings(self):
+        return self.embed_tokens
+
+    def set_input_embeddings(self, new_embeddings):
+        self.embed_tokens = new_embeddings
+
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        encoder_hidden_states=None,
+        encoder_attention_mask=None,
+        inputs_embeds=None,
+        head_mask=None,
+        cross_attn_head_mask=None,
+        past_key_values=None,
+        use_cache=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+    ):
+        # Model parallel
+        if self.model_parallel:
+            torch.cuda.set_device(self.first_device)
+            self.embed_tokens = self.embed_tokens.to(self.first_device)
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if input_ids is not None and inputs_embeds is not None:
+            err_msg_prefix = ""
+            raise ValueError(
+                f"You cannot specify both {err_msg_prefix}inputs and {err_msg_prefix}inputs_embeds at the same time"
+            )
+        elif input_ids is not None:
+            input_shape = input_ids.size()
+            input_ids = input_ids.view(-1, input_shape[-1])
+        elif inputs_embeds is not None:
+            input_shape = inputs_embeds.size()[:-1]
+        else:
+            err_msg_prefix = ""
+            raise ValueError(f"You have to specify either {err_msg_prefix}inputs or {err_msg_prefix}inputs_embeds")
+
+        if inputs_embeds is None:
+            assert self.embed_tokens is not None, "You have to initialize the model with valid token embeddings"
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        batch_size, seq_length = input_shape
+
+        # required mask seq length can be calculated via length of past
+        mask_seq_length = past_key_values[0][0].shape[2] + seq_length if past_key_values is not None else seq_length
+
+        # if use_cache is True:
+        #     assert self.is_decoder, f":obj:`use_cache` can only be set to `True` if {self} is used as a decoder"
+
+        if attention_mask is None:
+            attention_mask = torch.ones(batch_size, mask_seq_length).to(inputs_embeds.device)
+        if encoder_attention_mask is None and encoder_hidden_states is not None:
+            encoder_seq_length = encoder_hidden_states.shape[1]
+            encoder_attention_mask = torch.ones(
+                batch_size, encoder_seq_length, device=inputs_embeds.device, dtype=torch.long
+            )
+
+        # initialize past_key_values with `None` if past does not exist
+        if past_key_values is None:
+            past_key_values = [None] * len(self.block)
+
+        # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
+        # ourselves in which case we just need to make it broadcastable to all heads.
+        extended_attention_mask = self.get_extended_attention_mask(attention_mask, input_shape, inputs_embeds.device)
+
+        # If a 2D or 3D attention mask is provided for the cross-attention
+        # we need to make broadcastable to [batch_size, num_heads, seq_length, seq_length]
+        if encoder_hidden_states is not None:
+            encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states.size()
+            encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length)
+            if encoder_attention_mask is None:
+                encoder_attention_mask = torch.ones(encoder_hidden_shape, device=inputs_embeds.device)
+            encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask)
+        else:
+            encoder_extended_attention_mask = None
+
+        # Prepare head mask if needed
+        head_mask = self.get_head_mask(head_mask, self.config.num_layers)
+        cross_attn_head_mask = self.get_head_mask(cross_attn_head_mask, self.config.num_layers)
+        present_key_value_states = () if use_cache else None
+        all_hidden_states = () if output_hidden_states else None
+        all_attentions = () if output_attentions else None
+        all_cross_attentions = () if output_attentions else None
+        position_bias = None
+        encoder_decoder_position_bias = None
+
+        hidden_states = self.dropout(inputs_embeds)
+
+        for i, (layer_module, past_key_value) in enumerate(zip(self.block, past_key_values)):
+            layer_head_mask = head_mask[i]
+            cross_attn_layer_head_mask = cross_attn_head_mask[i]
+            # Model parallel
+            if self.model_parallel:
+                torch.cuda.set_device(hidden_states.device)
+                # Ensure that attention_mask is always on the same device as hidden_states
+                if attention_mask is not None:
+                    attention_mask = attention_mask.to(hidden_states.device)
+                if position_bias is not None:
+                    position_bias = position_bias.to(hidden_states.device)
+                if encoder_hidden_states is not None:
+                    encoder_hidden_states = encoder_hidden_states.to(hidden_states.device)
+                if encoder_extended_attention_mask is not None:
+                    encoder_extended_attention_mask = encoder_extended_attention_mask.to(hidden_states.device)
+                if encoder_decoder_position_bias is not None:
+                    encoder_decoder_position_bias = encoder_decoder_position_bias.to(hidden_states.device)
+                if layer_head_mask is not None:
+                    layer_head_mask = layer_head_mask.to(hidden_states.device)
+                if cross_attn_layer_head_mask is not None:
+                    cross_attn_layer_head_mask = cross_attn_layer_head_mask.to(hidden_states.device)
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)
+
+            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+                if use_cache:
+                    logger.warn(
+                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
+                        "`use_cache=False`..."
+                    )
+                    use_cache = False
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        return tuple(module(*inputs, use_cache, output_attentions))
+
+                    return custom_forward
+
+                layer_outputs = checkpoint(
+                    create_custom_forward(layer_module),
+                    hidden_states,
+                    extended_attention_mask,
+                    position_bias,
+                    encoder_hidden_states,
+                    encoder_extended_attention_mask,
+                    encoder_decoder_position_bias,
+                    layer_head_mask,
+                    cross_attn_layer_head_mask,
+                    None,  # past_key_value is always None with gradient checkpointing
+                )
+            else:
+                layer_outputs = layer_module(
+                    hidden_states,
+                    attention_mask=extended_attention_mask,
+                    position_bias=position_bias,
+                    encoder_hidden_states=encoder_hidden_states,
+                    encoder_attention_mask=encoder_extended_attention_mask,
+                    encoder_decoder_position_bias=encoder_decoder_position_bias,
+                    layer_head_mask=layer_head_mask,
+                    cross_attn_layer_head_mask=cross_attn_layer_head_mask,
+                    past_key_value=past_key_value,
+                    use_cache=use_cache,
+                    output_attentions=output_attentions,
+                )
+
+            # layer_outputs is a tuple with:
+            # hidden-states, key-value-states, (self-attention weights), (self-attention position bias), (cross-attention weights), (cross-attention position bias)
+            if use_cache is False:
+                layer_outputs = layer_outputs[:1] + (None,) + layer_outputs[1:]
+            hidden_states, present_key_value_state = layer_outputs[:2]
+
+            # We share the position biases between the layers - the first layer store them
+            # layer_outputs = hidden-states, key-value-states (self-attention weights),
+            # (self-attention position bias), (cross-attention weights), (cross-attention position bias)
+            position_bias = layer_outputs[2]
+            if encoder_hidden_states is not None:
+                encoder_decoder_position_bias = layer_outputs[4 if output_attentions else 3]
+            # append next layer key value states
+            if use_cache:
+                present_key_value_states = present_key_value_states + (present_key_value_state,)
+
+            if output_attentions:
+                all_attentions = all_attentions + (layer_outputs[3],)
+                # if self.is_decoder:
+                #     all_cross_attentions = all_cross_attentions + (layer_outputs[5],)
+
+            # Model Parallel: If it's the last layer for that device, put things on the next device
+            if self.model_parallel:
+                for k, v in self.device_map.items():
+                    if i == v[-1] and "cuda:" + str(k) != self.last_device:
+                        hidden_states = hidden_states.to("cuda:" + str(k + 1))
+
+        hidden_states = self.final_layer_norm(hidden_states)
+        hidden_states = self.dropout(hidden_states)
+
+        # Add last layer
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)
+
+        if not return_dict:
+            return tuple(
+                v
+                for v in [
+                    hidden_states,
+                    present_key_value_states,
+                    all_hidden_states,
+                    all_attentions,
+                    all_cross_attentions,
+                ]
+                if v is not None
+            )
+        return BaseModelOutputWithPastAndCrossAttentions(
+            last_hidden_state=hidden_states,
+            past_key_values=present_key_value_states,
+            hidden_states=all_hidden_states,
+            attentions=all_attentions,
+            cross_attentions=all_cross_attentions,
+        )
+
+
+T5_START_DOCSTRING = r"""
+
+    The T5 model was proposed in `Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
+    <https://arxiv.org/abs/1910.10683>`__ by Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang,
+    Michael Matena, Yanqi Zhou, Wei Li, Peter J. Liu. It's an encoder decoder transformer pre-trained in a text-to-text
+    denoising generative setting.
+
+    This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic
+    methods the library implements for all its model (such as downloading or saving, resizing the input embeddings,
+    pruning heads etc.)
+
+    This model is also a PyTorch `torch.nn.Module <https://pytorch.org/docs/stable/nn.html#torch.nn.Module>`__
+    subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to
+    general usage and behavior.
+
+    Parameters:
+        config (:class:`~transformers.T5Config`): Model configuration class with all the parameters of the model.
+            Initializing with a config file does not load the weights associated with the model, only the
+            configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model
+            weights.
+"""
+
+T5_INPUTS_DOCSTRING = r"""
+    Args:
+        input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
+            Indices of input sequence tokens in the vocabulary. T5 is a model with relative position embeddings so you
+            should be able to pad the inputs on both the right and the left.
+
+            Indices can be obtained using :class:`~transformers.T5Tokenizer`. See
+            :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
+            detail.
+
+            `What are input IDs? <../glossary.html#input-ids>`__
+
+            To know more on how to prepare :obj:`input_ids` for pretraining take a look a `T5 Training
+            <./t5.html#training>`__.
+        attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
+            Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+
+            `What are attention masks? <../glossary.html#attention-mask>`__
+        decoder_input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
+            Indices of decoder input sequence tokens in the vocabulary.
+
+            Indices can be obtained using :class:`~transformers.T5Tokenizer`. See
+            :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
+            details.
+
+            `What are decoder input IDs? <../glossary.html#decoder-input-ids>`__
+
+            T5 uses the :obj:`pad_token_id` as the starting token for :obj:`decoder_input_ids` generation. If
+            :obj:`past_key_values` is used, optionally only the last :obj:`decoder_input_ids` have to be input (see
+            :obj:`past_key_values`).
+
+            To know more on how to prepare :obj:`decoder_input_ids` for pretraining take a look at `T5 Training
+            <./t5.html#training>`__.
+        decoder_attention_mask (:obj:`torch.BoolTensor` of shape :obj:`(batch_size, target_sequence_length)`, `optional`):
+            Default behavior: generate a tensor that ignores pad tokens in :obj:`decoder_input_ids`. Causal mask will
+            also be used by default.
+        head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
+            Mask to nullify selected heads of the self-attention modules in the encoder. Mask values selected in ``[0,
+            1]``:
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+
+        decoder_head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
+            Mask to nullify selected heads of the self-attention modules in the decoder. Mask values selected in ``[0,
+            1]``:
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+
+        cross_attn_head_mask (:obj:`torch.Tensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
+                Mask to nullify selected heads of the cross-attention modules in the decoder. Mask values selected in
+                ``[0, 1]``:
+
+                - 1 indicates the head is **not masked**,
+                - 0 indicates the head is **masked**.
+
+        encoder_outputs (:obj:`tuple(tuple(torch.FloatTensor)`, `optional`):
+            Tuple consists of (:obj:`last_hidden_state`, :obj:`optional`: `hidden_states`, :obj:`optional`:
+            `attentions`) :obj:`last_hidden_state` of shape :obj:`(batch_size, sequence_length, hidden_size)` is a
+            sequence of hidden states at the output of the last layer of the encoder. Used in the cross-attention of
+            the decoder.
+        past_key_values (:obj:`tuple(tuple(torch.FloatTensor))` of length :obj:`config.n_layers` with each tuple having 4 tensors of shape :obj:`(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
+            Contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.
+
+            If :obj:`past_key_values` are used, the user can optionally input only the last :obj:`decoder_input_ids`
+            (those that don't have their past key value states given to this model) of shape :obj:`(batch_size, 1)`
+            instead of all :obj:`decoder_input_ids` of shape :obj:`(batch_size, sequence_length)`.
+        inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
+            Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
+            This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
+            vectors than the model's internal embedding lookup matrix.
+        decoder_inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, target_sequence_length, hidden_size)`, `optional`):
+            Optionally, instead of passing :obj:`decoder_input_ids` you can choose to directly pass an embedded
+            representation. If :obj:`past_key_values` is used, optionally only the last :obj:`decoder_inputs_embeds`
+            have to be input (see :obj:`past_key_values`). This is useful if you want more control over how to convert
+            :obj:`decoder_input_ids` indices into associated vectors than the model's internal embedding lookup matrix.
+
+            If :obj:`decoder_input_ids` and :obj:`decoder_inputs_embeds` are both unset, :obj:`decoder_inputs_embeds`
+            takes the value of :obj:`inputs_embeds`.
+
+        use_cache (:obj:`bool`, `optional`):
+            If set to :obj:`True`, :obj:`past_key_values` key value states are returned and can be used to speed up
+            decoding (see :obj:`past_key_values`).
+
+        output_attentions (:obj:`bool`, `optional`):
+            Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
+            tensors for more detail.
+        output_hidden_states (:obj:`bool`, `optional`):
+            Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
+            more detail.
+        return_dict (:obj:`bool`, `optional`):
+            Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
+"""
+
+T5_ENCODER_INPUTS_DOCSTRING = r"""
+    Args:
+        input_ids (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`):
+            Indices of input sequence tokens in the vocabulary. T5 is a model with relative position embeddings so you
+            should be able to pad the inputs on both the right and the left.
+
+            Indices can be obtained using :class:`~transformers.T5Tokenizer`. See
+            :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for
+            detail.
+
+            To know more on how to prepare :obj:`input_ids` for pretraining take a look a `T5 Training
+            <./t5.html#training>`__.
+        attention_mask (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`):
+            Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+
+            `What are attention masks? <../glossary.html#attention-mask>`__
+        head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`):
+            Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``:
+
+            - 1 indicates the head is **not masked**,
+            - 0 indicates the head is **masked**.
+
+        inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length, hidden_size)`, `optional`):
+            Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation.
+            This is useful if you want more control over how to convert :obj:`input_ids` indices into associated
+            vectors than the model's internal embedding lookup matrix.
+        output_attentions (:obj:`bool`, `optional`):
+            Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned
+            tensors for more detail.
+        output_hidden_states (:obj:`bool`, `optional`):
+            Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for
+            more detail.
+        return_dict (:obj:`bool`, `optional`):
+            Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple.
+"""
+
+# Warning message for FutureWarning: head_mask was separated into two input args - head_mask, decoder_head_mask
+__HEAD_MASK_WARNING_MSG = """
+The input argument `head_mask` was split into two arguments `head_mask` and `decoder_head_mask`. Currently,
+`decoder_head_mask` is set to copy `head_mask`, but this feature is deprecated and will be removed in future versions.
+If you do not want to use any `decoder_head_mask` now, please set `decoder_head_mask = torch.ones(num_layers,
+num_heads)`.
+"""
+
+
+@add_start_docstrings(
+    "The bare T5 Model transformer outputting raw hidden-states" "without any specific head on top.",
+    T5_START_DOCSTRING,
+)
+class DecoderOnlyT5Model(DecoderOnlyT5PreTrainedModel):
+    _keys_to_ignore_on_load_missing = [
+        r"encoder\.embed_tokens\.weight",
+    ]
+
+    def __init__(self, config: DecoderOnlyT5Config):
+        super().__init__(config)
+        self.shared = nn.Embedding(config.vocab_size, config.d_model)
+
+        encoder_config = copy.deepcopy(config)
+        encoder_config.is_decoder = True
+        encoder_config.use_cache = False
+        encoder_config.is_encoder_decoder = False
+        self.encoder = DecoderOnlyT5Stack(encoder_config, self.shared)
+
+
+        self.init_weights()
+
+        # Model parallel
+        self.model_parallel = False
+        self.device_map = None
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        self.device_map = (
+            get_device_map(len(self.encoder.block), range(torch.cuda.device_count()))
+            if device_map is None
+            else device_map
+        )
+        assert_device_map(self.device_map, len(self.encoder.block))
+        self.encoder.parallelize(self.device_map)
+        self.model_parallel = True
+
+    @add_start_docstrings(DEPARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.encoder.deparallelize()
+        self.encoder = self.encoder.to("cpu")
+        self.model_parallel = False
+        self.device_map = None
+        torch.cuda.empty_cache()
+
+    def get_input_embeddings(self):
+        return self.shared
+
+    def set_input_embeddings(self, new_embeddings):
+        self.shared = new_embeddings
+        self.encoder.set_input_embeddings(new_embeddings)
+
+    def get_encoder(self):
+        return self.encoder
+
+
+    def _prune_heads(self, heads_to_prune):
+        """
+        Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base
+        class PreTrainedModel
+        """
+        for layer, heads in heads_to_prune.items():
+            self.encoder.layer[layer].attention.prune_heads(heads)
+
+    @add_start_docstrings_to_model_forward(T5_INPUTS_DOCSTRING)
+    @replace_return_docstrings(output_type=BaseModelOutput, config_class=_CONFIG_FOR_DOC)
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        head_mask=None,
+        encoder_outputs=None,
+        inputs_embeds=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+    ):
+        r"""
+        Returns:
+
+        Example::
+
+            >>> from transformers import T5Tokenizer, T5Model
+
+            >>> tokenizer = T5Tokenizer.from_pretrained('t5-small')
+            >>> model = T5Model.from_pretrained('t5-small')
+
+            >>> input_ids = tokenizer("Studies have been shown that owning a dog is good for you", return_tensors="pt").input_ids  # Batch size 1
+            >>> decoder_input_ids = tokenizer("Studies show that", return_tensors="pt").input_ids  # Batch size 1
+            >>> outputs = model(input_ids=input_ids, decoder_input_ids=decoder_input_ids)
+
+            >>> last_hidden_states = outputs.last_hidden_state
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        encoder_outputs = self.encoder(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            inputs_embeds=inputs_embeds,
+            head_mask=head_mask,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        return encoder_outputs
+
+@add_start_docstrings("""T5 Model with a `language modeling` head on top. """, T5_START_DOCSTRING)
+class DecoderOnlyT5LMHeadModel(DecoderOnlyT5PreTrainedModel):
+    _keys_to_ignore_on_load_missing = [
+        r"encoder\.embed_tokens\.weight",
+        r"lm_head\.weight",
+    ]
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.model_dim = config.d_model
+
+        self.shared = nn.Embedding(config.vocab_size, config.d_model)
+
+        encoder_config = copy.deepcopy(config)
+        encoder_config.is_decoder = True
+        encoder_config.use_cache = False
+        encoder_config.is_encoder_decoder = False
+        self.encoder = DecoderOnlyT5Stack(encoder_config, self.shared)
+
+        self.lm_head = nn.Linear(config.d_model, config.vocab_size, bias=False)
+
+        self.init_weights()
+
+        # Model parallel
+        self.model_parallel = False
+        self.device_map = None
+
+    @add_start_docstrings(PARALLELIZE_DOCSTRING)
+    def parallelize(self, device_map=None):
+        self.device_map = (
+            get_device_map(len(self.encoder.block), range(torch.cuda.device_count()))
+            if device_map is None
+            else device_map
+        )
+        assert_device_map(self.device_map, len(self.encoder.block))
+        self.encoder.parallelize(self.device_map)
+        self.lm_head = self.lm_head.to(self.encoder.first_device)
+        self.model_parallel = True
+
+    @add_start_docstrings(DEPARALLELIZE_DOCSTRING)
+    def deparallelize(self):
+        self.encoder.deparallelize()
+        self.encoder = self.encoder.to("cpu")
+        self.lm_head = self.lm_head.to("cpu")
+        self.model_parallel = False
+        self.device_map = None
+        torch.cuda.empty_cache()
+
+    def get_input_embeddings(self):
+        return self.shared
+
+    def set_input_embeddings(self, new_embeddings):
+        self.shared = new_embeddings
+        self.encoder.set_input_embeddings(new_embeddings)
+
+    def set_output_embeddings(self, new_embeddings):
+        self.lm_head = new_embeddings
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def get_encoder(self):
+        return self.encoder
+
+    @add_start_docstrings_to_model_forward(T5_INPUTS_DOCSTRING)
+    @replace_return_docstrings(output_type=CausalLMOutputWithCrossAttentions, config_class=_CONFIG_FOR_DOC)
+    def forward(
+        self,
+        input_ids=None,
+        attention_mask=None,
+        head_mask=None,
+        past_key_values=None,
+        inputs_embeds=None,
+        labels=None,
+        use_cache=None,
+        output_attentions=None,
+        output_hidden_states=None,
+        return_dict=None,
+    ):
+        r"""
+        labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`):
+            Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[-100, 0, ...,
+            config.vocab_size - 1]`. All labels set to ``-100`` are ignored (masked), the loss is only computed for
+            labels in ``[0, ..., config.vocab_size]``
+
+        Returns:
+
+        Examples::
+
+            >>> from transformers import T5Tokenizer, T5ForConditionalGeneration
+
+            >>> tokenizer = T5Tokenizer.from_pretrained('t5-small')
+            >>> model = T5ForConditionalGeneration.from_pretrained('t5-small')
+
+            >>> input_ids = tokenizer('The <extra_id_0> walks in <extra_id_1> park', return_tensors='pt').input_ids
+            >>> labels = tokenizer('<extra_id_0> cute dog <extra_id_1> the <extra_id_2> </s>', return_tensors='pt').input_ids
+            >>> outputs = model(input_ids=input_ids, labels=labels)
+            >>> loss = outputs.loss
+            >>> logits = outputs.logits
+
+            >>> input_ids = tokenizer("summarize: studies have shown that owning a dog is good for you ", return_tensors="pt").input_ids  # Batch size 1
+            >>> outputs = model.generate(input_ids)
+        """
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        encoder_outputs = self.encoder(
+            input_ids=input_ids,
+            past_key_values=past_key_values,
+            attention_mask=attention_mask,
+            inputs_embeds=inputs_embeds,
+            head_mask=head_mask,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = encoder_outputs[0]
+
+        if self.model_parallel:
+            torch.cuda.set_device(self.encoder.first_device)
+            hidden_states = hidden_states.to(self.lm_head.weight.device)
+
+        lm_logits = self.lm_head(hidden_states)
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = lm_logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss(ignore_index=-100)
+            loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))
+
+        if not return_dict:
+            output = (lm_logits,) + encoder_outputs[1:]
+            return ((loss,) + output) if loss is not None else output
+
+        return CausalLMOutputWithCrossAttentions(
+            loss=loss,
+            logits=lm_logits,
+            past_key_values=encoder_outputs.past_key_values,
+            hidden_states=encoder_outputs.hidden_states,
+            attentions=encoder_outputs.attentions,
+            cross_attentions=encoder_outputs.cross_attentions,
+        )
+
+    def prepare_inputs_for_generation(
+        self,
+        input_ids,
+        past=None,
+        attention_mask=None,
+        head_mask=None,
+        use_cache=None,
+        encoder_outputs=None,
+        **kwargs
+    ):
+
+        # cut decoder_input_ids if past is used
+        if past is not None:
+            input_ids = input_ids[:, -1:]
+
+        return {
+            "input_ids": input_ids,
+            "past_key_values": past,
+            "attention_mask": attention_mask,
+            "use_cache": use_cache,
+        }
+
+    def _reorder_cache(past, beam_idx):
+        """
+        This function is used to re-order the :obj:`past_key_values` cache if
+        :meth:`~transformers.PreTrainedModel.beam_search` or :meth:`~transformers.PreTrainedModel.beam_sample` is
+        called. This is required to match :obj:`past_key_values` with the correct beam_idx at every generation step.
+        """
+        return tuple(
+            tuple(past_state.index_select(0, beam_idx.to(past_state.device)) for past_state in layer_past)
+            for layer_past in past
+        )

bigscience/experiments/gpt2-hf-ds/hf_ds_gpt2_base_n16.slurm

@@ -0,0 +1,169 @@
+#!/bin/bash
+#SBATCH --job-name=hf_ds_gpt2_base_n16
+#SBATCH --constraint=v100-32g
+#SBATCH --nodes=16
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=40           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:4                 # number of gpus
+#SBATCH --time 00:30:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --error=%x-%j.out            # error file name (same to watch just one file)
+#SBATCH --account=six@gpu
+
+set -x -e
+export PYTHONUNBUFFERED=1
+
+source $six_ALL_CCFRWORK/start-prod
+
+nvidia-smi
+
+cd $six_ALL_CCFRWORK/code/transformers-clm-any-model-config/
+
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+DATASET="stas/openwebtext-10k"
+
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+
+# adjust depending on the number of the nodes
+
+NNODES=16
+MICRO_BATCH_SIZE=4
+
+# succeeded:
+# MSIZE=30
+# MSIZE=39
+# MSIZE=52
+# MSIZE=65 # @ MICRO_BATCH_SIZE=4
+# MSIZE=81
+
+MSIZE=97
+
+if   [[ ${MSIZE} == 7 ]];    then NHIDDEN=4096;  NLAYERS=36
+elif [[ ${MSIZE} == 14 ]];   then NHIDDEN=6144;  NLAYERS=32
+elif [[ ${MSIZE} == 18 ]];   then NHIDDEN=6144;  NLAYERS=40
+elif [[ ${MSIZE} == 25 ]];   then NHIDDEN=7168;  NLAYERS=40
+elif [[ ${MSIZE} == 30 ]];   then NHIDDEN=7168;  NLAYERS=48
+elif [[ ${MSIZE} == 39 ]];   then NHIDDEN=8192;  NLAYERS=48
+elif [[ ${MSIZE} == 52 ]];   then NHIDDEN=8192;  NLAYERS=64
+elif [[ ${MSIZE} == 65 ]];   then NHIDDEN=9216;  NLAYERS=64
+elif [[ ${MSIZE} == 81 ]];   then NHIDDEN=10240; NLAYERS=64
+elif [[ ${MSIZE} == 97 ]];   then NHIDDEN=11264; NLAYERS=64
+elif [[ ${MSIZE} == 116 ]];  then NHIDDEN=12288; NLAYERS=64
+elif [[ ${MSIZE} == 136 ]];  then NHIDDEN=13312; NLAYERS=64
+elif [[ ${MSIZE} == 158 ]];  then NHIDDEN=14336; NLAYERS=64
+elif [[ ${MSIZE} == 181 ]];  then NHIDDEN=15360; NLAYERS=64
+elif [[ ${MSIZE} == 206 ]];  then NHIDDEN=16384; NLAYERS=64
+else echo "invalid MSIZE: $MSIZE"
+fi
+
+
+GPUS_PER_NODE=4
+NHEADS=32
+SEQ_LEN=1024
+VOCAB_SIZE=50257
+
+export LAUNCHER="python -u -m torch.distributed.launch \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT \
+    "
+
+
+config_json="./ds_z3_cpu_offload.json"
+cat <<EOT > $config_json
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 8,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 3,
+        "offload_optimizer": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "offload_param": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "overlap_comm": true,
+        "contiguous_gradients": true,
+        "sub_group_size": 1e14,
+        "reduce_bucket_size": "auto",
+        "stage3_prefetch_bucket_size": "auto",
+        "stage3_param_persistence_threshold": "auto",
+        "stage3_max_live_parameters": 1e9,
+        "stage3_max_reuse_distance": 1e9,
+        "stage3_gather_fp16_weights_on_model_save": false
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}
+EOT
+
+export PYTHONPATH=src
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export USE_TF=0
+
+export CMD=" \
+    examples/pytorch/language-modeling/run_clm.py \
+    --model_type gpt2 \
+    --tokenizer_name gpt2 \
+    --config_overrides "n_embd=$NHIDDEN,n_head=$NHEADS,n_layer=$NLAYERS,n_positions=$SEQ_LEN,gradient_checkpointing=true,use_cache=False" \
+    --dataset_name $DATASET \
+    --output_dir output_dir \
+    --overwrite_output_dir \
+    --do_train \
+    --max_train_samples 1000 \
+    --per_device_train_batch_size $MICRO_BATCH_SIZE \
+    --num_train_epochs 1 \
+    --warmup_steps 8 \
+    --fp16 \
+    --report_to none \
+    --deepspeed $config_json \
+    "
+
+# clear old checkpoint as it'd mismatch while we sort things out
+rm -rf $six_ALL_CCFRWORK/checkpoints/gpt2-1-node
+
+# model size
+python -c "h=$NHIDDEN; l=$NLAYERS; s=$SEQ_LEN; v=$VOCAB_SIZE; print(f'Model size: {(l * (12*h**2 + 13*h) + (v * h) + (s * h) ) / 10**9 :.0f}B')"
+
+# to debug - add echo (it exits and prints what it would have launched)
+clear; srun --jobid $SLURM_JOBID bash -c '$LAUNCHER --node_rank $SLURM_PROCID $CMD' 2>&1 | tee -a hf_ds_gpt2_base_n16_bs4.txt

bigscience/experiments/gpt2-hf-ds/hf_ds_gpt2_base_n32.slurm

@@ -0,0 +1,168 @@
+#!/bin/bash
+#SBATCH --job-name=hf_ds_gpt2_base_n32
+#SBATCH --constraint=v100-32g
+#SBATCH --nodes=32
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=40           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:4                 # number of gpus
+#SBATCH --time 00:30:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --error=%x-%j.out            # error file name (same to watch just one file)
+#SBATCH --account=six@gpu
+
+set -x -e
+export PYTHONUNBUFFERED=1
+
+source $six_ALL_CCFRWORK/start-prod
+
+nvidia-smi
+
+cd $six_ALL_CCFRWORK/code/transformers-clm-any-model-config/
+
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+DATASET="stas/openwebtext-10k"
+
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+
+# adjust depending on the number of the nodes
+
+NNODES=32
+MICRO_BATCH_SIZE=4
+
+# succeeded:
+
+# to try
+MSIZE=158
+
+# failed
+# MSIZE=181
+
+if   [[ ${MSIZE} == 7 ]];    then NHIDDEN=4096;  NLAYERS=36
+elif [[ ${MSIZE} == 14 ]];   then NHIDDEN=6144;  NLAYERS=32
+elif [[ ${MSIZE} == 18 ]];   then NHIDDEN=6144;  NLAYERS=40
+elif [[ ${MSIZE} == 25 ]];   then NHIDDEN=7168;  NLAYERS=40
+elif [[ ${MSIZE} == 30 ]];   then NHIDDEN=7168;  NLAYERS=48
+elif [[ ${MSIZE} == 39 ]];   then NHIDDEN=8192;  NLAYERS=48
+elif [[ ${MSIZE} == 52 ]];   then NHIDDEN=8192;  NLAYERS=64
+elif [[ ${MSIZE} == 65 ]];   then NHIDDEN=9216;  NLAYERS=64
+elif [[ ${MSIZE} == 81 ]];   then NHIDDEN=10240; NLAYERS=64
+elif [[ ${MSIZE} == 97 ]];   then NHIDDEN=11264; NLAYERS=64
+elif [[ ${MSIZE} == 116 ]];  then NHIDDEN=12288; NLAYERS=64
+elif [[ ${MSIZE} == 136 ]];  then NHIDDEN=13312; NLAYERS=64
+elif [[ ${MSIZE} == 158 ]];  then NHIDDEN=14336; NLAYERS=64
+elif [[ ${MSIZE} == 181 ]];  then NHIDDEN=15360; NLAYERS=64
+elif [[ ${MSIZE} == 206 ]];  then NHIDDEN=16384; NLAYERS=64
+else echo "invalid MSIZE: $MSIZE"
+fi
+
+
+GPUS_PER_NODE=4
+NHEADS=32
+SEQ_LEN=1024
+VOCAB_SIZE=50257
+
+export LAUNCHER="python -u -m torch.distributed.launch \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT \
+    "
+
+
+config_json="./ds_z3_cpu_offload.json"
+cat <<EOT > $config_json
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 8,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 3,
+        "offload_optimizer": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "offload_param": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "overlap_comm": true,
+        "contiguous_gradients": true,
+        "sub_group_size": 1e14,
+        "reduce_bucket_size": "auto",
+        "stage3_prefetch_bucket_size": "auto",
+        "stage3_param_persistence_threshold": "auto",
+        "stage3_max_live_parameters": 1e9,
+        "stage3_max_reuse_distance": 1e9,
+        "stage3_gather_fp16_weights_on_model_save": false
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}
+EOT
+
+export PYTHONPATH=src
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export USE_TF=0
+
+export CMD=" \
+    examples/pytorch/language-modeling/run_clm.py \
+    --model_type gpt2 \
+    --tokenizer_name gpt2 \
+    --config_overrides "n_embd=$NHIDDEN,n_head=$NHEADS,n_layer=$NLAYERS,n_positions=$SEQ_LEN,gradient_checkpointing=true,use_cache=False" \
+    --dataset_name $DATASET \
+    --output_dir output_dir \
+    --overwrite_output_dir \
+    --do_train \
+    --max_train_samples 1000 \
+    --per_device_train_batch_size $MICRO_BATCH_SIZE \
+    --num_train_epochs 1 \
+    --warmup_steps 8 \
+    --fp16 \
+    --report_to none \
+    --deepspeed $config_json \
+    "
+
+# clear old checkpoint as it'd mismatch while we sort things out
+rm -rf $six_ALL_CCFRWORK/checkpoints/gpt2-1-node
+
+# model size
+python -c "h=$NHIDDEN; l=$NLAYERS; s=$SEQ_LEN; v=$VOCAB_SIZE; print(f'Model size: {(l * (12*h**2 + 13*h) + (v * h) + (s * h) ) / 10**9 :.0f}B')"
+
+# to debug - add echo (it exits and prints what it would have launched)
+srun --jobid $SLURM_JOBID bash -c '$LAUNCHER --node_rank $SLURM_PROCID $CMD' 2>&1 | tee -a hf_ds_gpt2_base_n32_bs4.txt

bigscience/experiments/gpt2-hf-ds/hf_ds_gpt2_base_n4.slurm

@@ -0,0 +1,164 @@
+#!/bin/bash
+#SBATCH --job-name=hf_ds_gpt2_base_n4
+#SBATCH --constraint=v100-32g
+#SBATCH --nodes=4
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=40           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:4                 # number of gpus
+#SBATCH --time 00:30:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --error=%x-%j.out            # error file name (same to watch just one file)
+#SBATCH --account=six@gpu
+
+set -x -e
+export PYTHONUNBUFFERED=1
+
+source $six_ALL_CCFRWORK/start-prod
+
+nvidia-smi
+
+cd $six_ALL_CCFRWORK/code/transformers-clm-any-model-config/
+
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+DATASET="stas/openwebtext-10k"
+
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+
+# adjust depending on the number of the nodes
+
+NNODES=4
+MICRO_BATCH_SIZE=4
+
+# succeeded:
+
+# to try:
+MSIZE=25
+
+if   [[ ${MSIZE} == 7 ]];    then NHIDDEN=4096;  NLAYERS=36
+elif [[ ${MSIZE} == 14 ]];   then NHIDDEN=6144;  NLAYERS=32
+elif [[ ${MSIZE} == 18 ]];   then NHIDDEN=6144;  NLAYERS=40
+elif [[ ${MSIZE} == 25 ]];   then NHIDDEN=7168;  NLAYERS=40
+elif [[ ${MSIZE} == 30 ]];   then NHIDDEN=7168;  NLAYERS=48
+elif [[ ${MSIZE} == 39 ]];   then NHIDDEN=8192;  NLAYERS=48
+elif [[ ${MSIZE} == 52 ]];   then NHIDDEN=8192;  NLAYERS=64
+elif [[ ${MSIZE} == 65 ]];   then NHIDDEN=9216;  NLAYERS=64
+elif [[ ${MSIZE} == 81 ]];   then NHIDDEN=10240; NLAYERS=64
+elif [[ ${MSIZE} == 97 ]];   then NHIDDEN=11264; NLAYERS=64
+elif [[ ${MSIZE} == 116 ]];  then NHIDDEN=12288; NLAYERS=64
+elif [[ ${MSIZE} == 136 ]];  then NHIDDEN=13312; NLAYERS=64
+elif [[ ${MSIZE} == 158 ]];  then NHIDDEN=14336; NLAYERS=64
+elif [[ ${MSIZE} == 181 ]];  then NHIDDEN=15360; NLAYERS=64
+elif [[ ${MSIZE} == 206 ]];  then NHIDDEN=16384; NLAYERS=64
+else echo "invalid MSIZE: $MSIZE"
+fi
+
+
+GPUS_PER_NODE=4
+NHEADS=32
+SEQ_LEN=1024
+VOCAB_SIZE=50257
+
+export LAUNCHER="python -u -m torch.distributed.launch \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT \
+    "
+
+
+config_json="./ds_z3_cpu_offload.json"
+cat <<EOT > $config_json
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 8,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 3,
+        "offload_optimizer": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "offload_param": {
+            "device": "none"
+        },
+        "overlap_comm": true,
+        "contiguous_gradients": true,
+        "sub_group_size": 1e14,
+        "reduce_bucket_size": "auto",
+        "stage3_prefetch_bucket_size": "auto",
+        "stage3_param_persistence_threshold": "auto",
+        "stage3_max_live_parameters": 1e9,
+        "stage3_max_reuse_distance": 1e9,
+        "stage3_gather_fp16_weights_on_model_save": false
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}
+EOT
+
+export PYTHONPATH=src
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export USE_TF=0
+
+export CMD=" \
+    examples/pytorch/language-modeling/run_clm.py \
+    --model_type gpt2 \
+    --tokenizer_name gpt2 \
+    --config_overrides "n_embd=$NHIDDEN,n_head=$NHEADS,n_layer=$NLAYERS,n_positions=$SEQ_LEN,gradient_checkpointing=true,use_cache=False" \
+    --dataset_name $DATASET \
+    --output_dir output_dir \
+    --overwrite_output_dir \
+    --do_train \
+    --max_train_samples 1000 \
+    --per_device_train_batch_size $MICRO_BATCH_SIZE \
+    --num_train_epochs 1 \
+    --warmup_steps 8 \
+    --fp16 \
+    --report_to none \
+    --deepspeed $config_json \
+    "
+
+# clear old checkpoint as it'd mismatch while we sort things out
+rm -rf $six_ALL_CCFRWORK/checkpoints/gpt2-1-node
+
+# model size
+python -c "h=$NHIDDEN; l=$NLAYERS; s=$SEQ_LEN; v=$VOCAB_SIZE; print(f'Model size: {(l * (12*h**2 + 13*h) + (v * h) + (s * h) ) / 10**9 :.0f}B')"
+
+# to debug - add echo (it exits and prints what it would have launched)
+srun --jobid $SLURM_JOBID bash -c '$LAUNCHER --node_rank $SLURM_PROCID $CMD'

bigscience/experiments/gpt2-hf-ds/hf_ds_gpt2_base_n8.slurm

@@ -0,0 +1,168 @@
+#!/bin/bash
+#SBATCH --job-name=hf_ds_gpt2_base_n8
+#SBATCH --constraint=v100-32g
+#SBATCH --nodes=8
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=40           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:4                 # number of gpus
+#SBATCH --time 00:30:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --error=%x-%j.out            # error file name (same to watch just one file)
+#SBATCH --account=six@gpu
+
+set -x -e
+export PYTHONUNBUFFERED=1
+
+source $six_ALL_CCFRWORK/start-prod
+
+nvidia-smi
+
+cd $six_ALL_CCFRWORK/code/transformers-clm-any-model-config/
+
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+DATASET="stas/openwebtext-10k"
+
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+
+# adjust depending on the number of the nodes
+
+NNODES=8
+MICRO_BATCH_SIZE=4
+
+# succeeded:
+# MSIZE=30
+#MSIZE=39
+
+# to try
+MSIZE=52
+
+
+if   [[ ${MSIZE} == 7 ]];    then NHIDDEN=4096;  NLAYERS=36
+elif [[ ${MSIZE} == 14 ]];   then NHIDDEN=6144;  NLAYERS=32
+elif [[ ${MSIZE} == 18 ]];   then NHIDDEN=6144;  NLAYERS=40
+elif [[ ${MSIZE} == 25 ]];   then NHIDDEN=7168;  NLAYERS=40
+elif [[ ${MSIZE} == 30 ]];   then NHIDDEN=7168;  NLAYERS=48
+elif [[ ${MSIZE} == 39 ]];   then NHIDDEN=8192;  NLAYERS=48
+elif [[ ${MSIZE} == 52 ]];   then NHIDDEN=8192;  NLAYERS=64
+elif [[ ${MSIZE} == 65 ]];   then NHIDDEN=9216;  NLAYERS=64
+elif [[ ${MSIZE} == 81 ]];   then NHIDDEN=10240; NLAYERS=64
+elif [[ ${MSIZE} == 97 ]];   then NHIDDEN=11264; NLAYERS=64
+elif [[ ${MSIZE} == 116 ]];  then NHIDDEN=12288; NLAYERS=64
+elif [[ ${MSIZE} == 136 ]];  then NHIDDEN=13312; NLAYERS=64
+elif [[ ${MSIZE} == 158 ]];  then NHIDDEN=14336; NLAYERS=64
+elif [[ ${MSIZE} == 181 ]];  then NHIDDEN=15360; NLAYERS=64
+elif [[ ${MSIZE} == 206 ]];  then NHIDDEN=16384; NLAYERS=64
+else echo "invalid MSIZE: $MSIZE"
+fi
+
+
+GPUS_PER_NODE=4
+NHEADS=32
+SEQ_LEN=1024
+VOCAB_SIZE=50257
+
+export LAUNCHER="python -u -m torch.distributed.launch \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT \
+    "
+
+
+config_json="./ds_z3_cpu_offload.json"
+cat <<EOT > $config_json
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 8,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 3,
+        "offload_optimizer": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "offload_param": {
+            "device": "cpu",
+            "pin_memory": true
+        },
+        "overlap_comm": true,
+        "contiguous_gradients": true,
+        "sub_group_size": 1e14,
+        "reduce_bucket_size": "auto",
+        "stage3_prefetch_bucket_size": "auto",
+        "stage3_param_persistence_threshold": "auto",
+        "stage3_max_live_parameters": 1e9,
+        "stage3_max_reuse_distance": 1e9,
+        "stage3_gather_fp16_weights_on_model_save": false
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}
+EOT
+
+export PYTHONPATH=src
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export USE_TF=0
+
+export CMD=" \
+    examples/pytorch/language-modeling/run_clm.py \
+    --model_type gpt2 \
+    --tokenizer_name gpt2 \
+    --config_overrides "n_embd=$NHIDDEN,n_head=$NHEADS,n_layer=$NLAYERS,n_positions=$SEQ_LEN,gradient_checkpointing=true,use_cache=False" \
+    --dataset_name $DATASET \
+    --output_dir output_dir \
+    --overwrite_output_dir \
+    --do_train \
+    --max_train_samples 1000 \
+    --per_device_train_batch_size $MICRO_BATCH_SIZE \
+    --num_train_epochs 1 \
+    --warmup_steps 8 \
+    --fp16 \
+    --report_to none \
+    --deepspeed $config_json \
+    "
+
+# clear old checkpoint as it'd mismatch while we sort things out
+rm -rf $six_ALL_CCFRWORK/checkpoints/gpt2-1-node
+
+# model size
+python -c "h=$NHIDDEN; l=$NLAYERS; s=$SEQ_LEN; v=$VOCAB_SIZE; print(f'Model size: {(l * (12*h**2 + 13*h) + (v * h) + (s * h) ) / 10**9 :.0f}B')"
+
+# to debug - add echo (it exits and prints what it would have launched)
+srun --jobid $SLURM_JOBID bash -c '$LAUNCHER --node_rank $SLURM_PROCID $CMD' 2>&1 | tee -a hf_ds_gpt2_base_n16_bs4.txt

bigscience/experiments/gpt2-hf-ds/hf_ds_gpt2_perf_n16.slurm

@@ -0,0 +1,169 @@
+#!/bin/bash
+#SBATCH --job-name=hf_ds_gpt2_perf_n16
+#SBATCH --constraint=v100-32g
+#SBATCH --nodes=16
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=40           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:4                 # number of gpus
+#SBATCH --time 00:30:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --error=%x-%j.out            # error file name (same to watch just one file)
+#SBATCH --account=six@gpu
+
+set -x -e
+export PYTHONUNBUFFERED=1
+
+source $six_ALL_CCFRWORK/start-prod
+
+nvidia-smi
+
+cd $six_ALL_CCFRWORK/code/transformers-clm-any-model-config/
+
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+DATASET="stas/openwebtext-10k"
+
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+
+# adjust depending on the number of the nodes
+
+NNODES=16
+MICRO_BATCH_SIZE=10 # 10 is 99% gpu
+
+# succeeded:
+MSIZE=52
+
+
+if   [[ ${MSIZE} == 7 ]];    then NHIDDEN=4096;  NLAYERS=36
+elif [[ ${MSIZE} == 14 ]];   then NHIDDEN=6144;  NLAYERS=32
+elif [[ ${MSIZE} == 18 ]];   then NHIDDEN=6144;  NLAYERS=40
+elif [[ ${MSIZE} == 25 ]];   then NHIDDEN=7168;  NLAYERS=40
+elif [[ ${MSIZE} == 30 ]];   then NHIDDEN=7168;  NLAYERS=48
+elif [[ ${MSIZE} == 39 ]];   then NHIDDEN=8192;  NLAYERS=48
+elif [[ ${MSIZE} == 52 ]];   then NHIDDEN=8192;  NLAYERS=64
+elif [[ ${MSIZE} == 65 ]];   then NHIDDEN=9216;  NLAYERS=64
+elif [[ ${MSIZE} == 81 ]];   then NHIDDEN=10240; NLAYERS=64
+elif [[ ${MSIZE} == 97 ]];   then NHIDDEN=11264; NLAYERS=64
+elif [[ ${MSIZE} == 116 ]];  then NHIDDEN=12288; NLAYERS=64
+elif [[ ${MSIZE} == 136 ]];  then NHIDDEN=13312; NLAYERS=64
+elif [[ ${MSIZE} == 158 ]];  then NHIDDEN=14336; NLAYERS=64
+elif [[ ${MSIZE} == 181 ]];  then NHIDDEN=15360; NLAYERS=64
+elif [[ ${MSIZE} == 206 ]];  then NHIDDEN=16384; NLAYERS=64
+else echo "invalid MSIZE: $MSIZE"
+fi
+
+
+GPUS_PER_NODE=4
+NHEADS=32
+SEQ_LEN=1024
+VOCAB_SIZE=50257
+
+export LAUNCHER="python -u -m torch.distributed.launch \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --master_addr $MASTER_ADDR \
+    --master_port $MASTER_PORT \
+    "
+
+
+config_json="./ds_z3_cpu_offload.json"
+cat <<EOT > $config_json
+{
+    "fp16": {
+        "enabled": "auto",
+        "loss_scale": 0,
+        "loss_scale_window": 1000,
+        "initial_scale_power": 8,
+        "hysteresis": 2,
+        "min_loss_scale": 1
+    },
+
+    "optimizer": {
+        "type": "AdamW",
+        "params": {
+            "lr": "auto",
+            "betas": "auto",
+            "eps": "auto",
+            "weight_decay": "auto"
+        }
+    },
+
+    "scheduler": {
+        "type": "WarmupLR",
+        "params": {
+            "warmup_min_lr": "auto",
+            "warmup_max_lr": "auto",
+            "warmup_num_steps": "auto"
+        }
+    },
+
+    "zero_optimization": {
+        "stage": 3,
+        "offload_optimizer": {
+            "device": "none"
+        },
+        "offload_param": {
+            "device": "none"
+        },
+        "overlap_comm": true,
+        "contiguous_gradients": true,
+        "sub_group_size": 1e14,
+        "reduce_bucket_size": "auto",
+        "stage3_prefetch_bucket_size": "auto",
+        "stage3_param_persistence_threshold": "auto",
+        "stage3_max_live_parameters": 1e9,
+        "stage3_max_reuse_distance": 1e9,
+        "stage3_gather_fp16_weights_on_model_save": false
+    },
+
+    "gradient_accumulation_steps": "auto",
+    "gradient_clipping": "auto",
+    "steps_per_print": 2000,
+    "train_batch_size": "auto",
+    "train_micro_batch_size_per_gpu": "auto",
+    "wall_clock_breakdown": false
+}
+EOT
+
+export PYTHONPATH=src
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export USE_TF=0
+
+# new arg to start using
+#    --log_on_each_node 0 \
+
+export CMD=" \
+    examples/pytorch/language-modeling/run_clm.py \
+    --model_type gpt2 \
+    --tokenizer_name gpt2 \
+    --config_overrides "n_embd=$NHIDDEN,n_head=$NHEADS,n_layer=$NLAYERS,n_positions=$SEQ_LEN,gradient_checkpointing=true,use_cache=False" \
+    --dataset_name $DATASET \
+    --output_dir output_dir \
+    --overwrite_output_dir \
+    --do_train \
+    --max_train_samples 1000 \
+    --per_device_train_batch_size $MICRO_BATCH_SIZE \
+    --num_train_epochs 1 \
+    --warmup_steps 8 \
+    --fp16 \
+    --report_to none \
+    --deepspeed $config_json \
+    "
+
+# clear old checkpoint as it'd mismatch while we sort things out
+rm -rf $six_ALL_CCFRWORK/checkpoints/gpt2-1-node
+
+# model size
+python -c "h=$NHIDDEN; l=$NLAYERS; s=$SEQ_LEN; v=$VOCAB_SIZE; print(f'Model size: {(l * (12*h**2 + 13*h) + (v * h) + (s * h) ) / 10**9 :.0f}B')"
+
+# make sure no zombies have been left behind from previous runs
+export PKILL="pkill python"
+
+echo $CMD
+# to debug - add echo (it exits and prints what it would have launched)
+clear; srun --jobid $SLURM_JOBID bash -c '$PKILL; $LAUNCHER --node_rank $SLURM_PROCID $CMD' 2>&1 | tee -a hf_ds_gpt2_perf_n16_bs4.out