Add files using upload-large-folder tool

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00010-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4807894ff54ad64db5a649d7e90cb4e3e9f8a26bfb685b7f432ea1eb4394fc8b
+size 502704664

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00040-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d50de72f983c0cd96d1d7666f0f9954c469d528b36c125cf814cdda194dac659
+size 501774384

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00059-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aac08a2949eb916e497ce22f237fb4369a0c8fd1556bceae0b2fa0b1f6b5d7bd
+size 501785176

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00070-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6020c4344c228e8755cae6ec4b1500137cc4783dcf962746166e971ca5b409f8
+size 503043536

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00072-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6e3dc92ae3b90d778e052af7ad3e04e3a7ba64590cce7cd9283833e3528e303
+size 504777888

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00074-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d25f7ef19eb2f2b03c30a64093aec0051903936d654e2b84e57ae5f3229796c
+size 503353848

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00089-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10685c132e62df801092ae15c334f2d41e6b6c4dc3a57c654aa21de0fc6d385d
+size 504698888

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00100-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cdf89b59c0c71e2a41f148011c746392a17b0bda4cee09cc32bf2f59d6abcec
+size 504413528

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00105-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac97ee67f09f4ce7d2c1d749b876d0cf12163d1e2b6259022f6992c3862c1804
+size 503532080

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00126-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06bdfb9f767e4686fbe816c6f17f7f7b30255420ea5602e5e28a5b6ccd2b3a33
+size 501036656

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00135-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72409ff2a64d68480801880f7e07f96be0fcdc6d284cc6a1cbcf603500c12ac5
+size 500714592

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00136-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3b6689190dec993e818c370580e7693afcdd8ce0ec519327e1950c3e8d20657
+size 508477840

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00140-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e13af03d22307d0acc99bbe230f09229d5a05cd02dcf4a22d85d2f6fdc347c9d
+size 505473128

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00200-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9129643c9a3b941f028355a008ed2930709edddb8ec96f9ffa536bd051a701d
+size 506929072

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00202-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44962dbf63ef79c3b59afd465b7c574e777582fa1e5a4762b72ce5e142123e1e
+size 503715992

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00204-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5360abb8c1653afae61d587e9219f360d1cb82fda2c7cd40158de1532a72a78e
+size 500887496

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00252-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51ae7970b358b5731a8f124797e981f02fd1be35c0dc76299ce936886e5da17d
+size 501961432

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00256-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5547b8a852311739cfb48ae53381c2e87f2e9673eea842e35919b9169337a301
+size 504684008

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00295-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07d7823b5bd634d72fc0814a4cd2cacdf8e12da3d3ef75e9acf8b82ee985ebbd
+size 505840960

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00320-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efeea6d8a6dca9c3ca851360cc61624a02c616ce3e53777280d807946ef1a107
+size 504561128

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00324-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1db0a503e57ccd733520c3fe775c261f06c241e68c3442a16f7d9a535fa7595e
+size 506267152

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00326-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2bb76369cc5054c121fe828e02299ee8dfa8c38758b98505468c4d9c5d6248da
+size 501325880

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00328-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd4ea31c7e8c9c4ec734346406d6155af92b2b1cceab5c0579e5470b0b852aea
+size 506335368

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00344-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db08a0f23e31ade1361180c9ec554700b6118ec4628c909f8b7812d06973c740
+size 502926768

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00360-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a1a4413cb358c425279de1492be218dd8c9584b4c70d2365a515ef0b51d2647
+size 502754056

bigscience/data/oscar/cache/oscar/unshuffled_deduplicated_zh/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2/oscar-train-00495-of-00532.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9ebc298a4c86d42ec40d85b5a5ede8739dbd2569d1dfe8cc8a5b48893780733
+size 506053128

bigscience/evaluation/results/tr11/bloom1b3/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_step340500"
+# 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_step340500_") + len("global_step340500_"):].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/bloom1b3/bslmevalfiles/tr11-1b3-ml-evalharness-results_lm-eval_global_step340500_2022-07-13-11-29-13.json

@@ -0,0 +1,172 @@
+{
+    "results": {
+        "arc_challenge": {
+            "acc": 0.23464163822525597,
+            "acc_stderr": 0.012383873560768673,
+            "acc_norm": 0.26791808873720135,
+            "acc_norm_stderr": 0.012942030195136423
+        },
+        "arc_easy": {
+            "acc": 0.5631313131313131,
+            "acc_stderr": 0.010177672928157678,
+            "acc_norm": 0.4810606060606061,
+            "acc_norm_stderr": 0.010252420496894487
+        },
+        "boolq": {
+            "acc": 0.617737003058104,
+            "acc_stderr": 0.008499149690449272
+        },
+        "copa": {
+            "acc": 0.7,
+            "acc_stderr": 0.046056618647183814
+        },
+        "headqa": {
+            "acc": 0.25419401896425964,
+            "acc_stderr": 0.008316509290190668,
+            "acc_norm": 0.29576951130561635,
+            "acc_norm_stderr": 0.008717251898361422
+        },
+        "hellaswag": {
+            "acc": 0.37621987651862177,
+            "acc_stderr": 0.004834461997944872,
+            "acc_norm": 0.46564429396534557,
+            "acc_norm_stderr": 0.004977988452502641
+        },
+        "lambada": {
+            "ppl": 12.583447597222621,
+            "ppl_stderr": 0.4021518609838198,
+            "acc": 0.46322530564719583,
+            "acc_stderr": 0.006947110835634445
+        },
+        "logiqa": {
+            "acc": 0.21658986175115208,
+            "acc_stderr": 0.016156860583178303,
+            "acc_norm": 0.28110599078341014,
+            "acc_norm_stderr": 0.017632374626460005
+        },
+        "mathqa": {
+            "acc": 0.2489112227805695,
+            "acc_stderr": 0.007915319798861361,
+            "acc_norm": 0.2422110552763819,
+            "acc_norm_stderr": 0.007842810183504986
+        },
+        "mc_taco": {
+            "em": 0.12537537537537538,
+            "f1": 0.4747075325110886
+        },
+        "mrpc": {
+            "acc": 0.6813725490196079,
+            "acc_stderr": 0.023095996571841474,
+            "f1": 0.8104956268221574,
+            "f1_stderr": 0.016329211455484924
+        },
+        "multirc": {
+            "acc": 0.011542497376705142,
+            "acc_stderr": 0.003461867320927179
+        },
+        "openbookqa": {
+            "acc": 0.214,
+            "acc_stderr": 0.01835979750238702,
+            "acc_norm": 0.298,
+            "acc_norm_stderr": 0.020475118092988978
+        },
+        "piqa": {
+            "acc": 0.6871599564744287,
+            "acc_stderr": 0.010817714425701112,
+            "acc_norm": 0.7002176278563657,
+            "acc_norm_stderr": 0.010689686967138092
+        },
+        "prost": {
+            "acc": 0.23505550811272416,
+            "acc_stderr": 0.0030979423271461875,
+            "acc_norm": 0.2670260461144321,
+            "acc_norm_stderr": 0.0032321702981822874
+        },
+        "pubmedqa": {
+            "acc": 0.56,
+            "acc_stderr": 0.015704987954361798
+        },
+        "qnli": {
+            "acc": 0.4962474830679114,
+            "acc_stderr": 0.006765220016415222
+        },
+        "qqp": {
+            "acc": 0.3681424684640119,
+            "acc_stderr": 0.0023986729832071816,
+            "f1": 0.5381138352498734,
+            "f1_stderr": 0.002555831569895799
+        },
+        "race": {
+            "acc": 0.3320574162679426,
+            "acc_stderr": 0.014575582129545914
+        },
+        "rte": {
+            "acc": 0.5342960288808665,
+            "acc_stderr": 0.030025579819366426
+        },
+        "sciq": {
+            "acc": 0.853,
+            "acc_stderr": 0.011203415395160335,
+            "acc_norm": 0.771,
+            "acc_norm_stderr": 0.013294199326613609
+        },
+        "sst": {
+            "acc": 0.6823394495412844,
+            "acc_stderr": 0.015775124845202545
+        },
+        "triviaqa": {
+            "acc": 0.0313798285158667,
+            "acc_stderr": 0.0016392014864795154
+        },
+        "webqs": {
+            "acc": 0.012795275590551181,
+            "acc_stderr": 0.0024938680596856277
+        },
+        "wic": {
+            "acc": 0.5,
+            "acc_stderr": 0.01981072129375818
+        },
+        "winogrande": {
+            "acc": 0.5730071033938438,
+            "acc_stderr": 0.013901878072575058
+        },
+        "wnli": {
+            "acc": 0.43661971830985913,
+            "acc_stderr": 0.0592793555841297
+        },
+        "wsc": {
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792
+        }
+    },
+    "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/bloom1b3/bslmevalfiles/tr11b-1b3-ml-bsevalharness-results_lm-eval_global_step340500_2022-07-14-10-03-25.json

@@ -0,0 +1,2169 @@
+{
+    "results": [
+        {
+            "task_name": "wic",
+            "prompt_name": "GPT-3-prompt",
+            "acc": 0.5,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "c3a0a5d8-cfe9-4a7f-8a3c-3c526e0ad0c6",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nQuestion: Is the word '{{word}}' used in the same sense in the two sentences above?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.01981072129375818
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "GPT-3-prompt",
+            "acc_norm": 0.5,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "c3a0a5d8-cfe9-4a7f-8a3c-3c526e0ad0c6",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nQuestion: Is the word '{{word}}' used in the same sense in the two sentences above?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "GPT-3-prompt-with-label",
+            "acc": 0.49216300940438873,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "d9e1db2a-ab0b-4621-bb41-01d5788d3873",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nQuestion: Is the word '{{word}}' used in the same sense in the two sentences above? Yes, No?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019808287657813832
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "GPT-3-prompt-with-label",
+            "acc_norm": 0.5,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "d9e1db2a-ab0b-4621-bb41-01d5788d3873",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nQuestion: Is the word '{{word}}' used in the same sense in the two sentences above? Yes, No?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "affirmation_true_or_false",
+            "acc": 0.5,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "725b5ed0-7728-4890-95a4-a74cb7ae1bb4",
+            "prompt_jinja": "Sentence A: {{sentence1}}\nSentence B: {{sentence2}}\n\n\"{{word}}\" has a similar meaning in sentences A and B. True or False?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.01981072129375818
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "affirmation_true_or_false",
+            "acc_norm": 0.5078369905956113,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "725b5ed0-7728-4890-95a4-a74cb7ae1bb4",
+            "prompt_jinja": "Sentence A: {{sentence1}}\nSentence B: {{sentence2}}\n\n\"{{word}}\" has a similar meaning in sentences A and B. True or False?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.019808287657813832
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "grammar_homework",
+            "acc": 0.5094043887147336,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "611d13dc-d414-4b9b-9204-e4f325e859e7",
+            "prompt_jinja": "Homework\n\nDecide whether the word \"{{word}}\" is used with the same meaning in the two following sentences. Answer by yes or no.\n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019807216763271497
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "grammar_homework",
+            "acc_norm": 0.49843260188087773,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "611d13dc-d414-4b9b-9204-e4f325e859e7",
+            "prompt_jinja": "Homework\n\nDecide whether the word \"{{word}}\" is used with the same meaning in the two following sentences. Answer by yes or no.\n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "polysemous",
+            "acc": 0.512539184952978,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "dd2080cf-3117-49ba-9aff-c988a21fdb69",
+            "prompt_jinja": "The word \"{{word}}\" has multiple meanings. Does it have the same meaning in sentences 1 and 2? Yes or no?\n\nSentence 1: {{sentence1}}\nSentence 2: {{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019804490588592596
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "polysemous",
+            "acc_norm": 0.49843260188087773,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "dd2080cf-3117-49ba-9aff-c988a21fdb69",
+            "prompt_jinja": "The word \"{{word}}\" has multiple meanings. Does it have the same meaning in sentences 1 and 2? Yes or no?\n\nSentence 1: {{sentence1}}\nSentence 2: {{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "question-context",
+            "acc": 0.5266457680250783,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "cfbc1637-10b8-4f20-a31c-55292f3cebd0",
+            "prompt_jinja": "Determine if the word '{{word}}' is used in the same way in the two sentences below. \n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019782570188812167
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "question-context",
+            "acc_norm": 0.5031347962382445,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "cfbc1637-10b8-4f20-a31c-55292f3cebd0",
+            "prompt_jinja": "Determine if the word '{{word}}' is used in the same way in the two sentences below. \n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.019810331932097542
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "question-context-meaning",
+            "acc": 0.5438871473354232,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "3503ead5-4fa5-4f77-95dc-f0c2ed3eecdc",
+            "prompt_jinja": "Does the word \"{{word}}\" have the same meaning in these two sentences?\n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019734259601993404
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "question-context-meaning",
+            "acc_norm": 0.5015673981191222,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "3503ead5-4fa5-4f77-95dc-f0c2ed3eecdc",
+            "prompt_jinja": "Does the word \"{{word}}\" have the same meaning in these two sentences?\n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "question-context-meaning-with-label",
+            "acc": 0.5156739811912225,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "14e73f39-a0d1-44c2-b9a4-4e48f9f1608e",
+            "prompt_jinja": "Does the word \"{{word}}\" have the same meaning in these two sentences? Yes, No?\n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019800984955347847
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "question-context-meaning-with-label",
+            "acc_norm": 0.5015673981191222,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "14e73f39-a0d1-44c2-b9a4-4e48f9f1608e",
+            "prompt_jinja": "Does the word \"{{word}}\" have the same meaning in these two sentences? Yes, No?\n{{sentence1}}\n{{sentence2}}\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "same_sense",
+            "acc": 0.5047021943573667,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "ce8b5a93-1841-4897-84db-b100f1c84f4b",
+            "prompt_jinja": "Sentence 1: {{sentence1}}\nSentence 2: {{sentence2}}\n\nDetermine whether the word \"{{word}}\" is used in the same sense in both sentences. Yes or no?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.019809845219259763
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "same_sense",
+            "acc_norm": 0.5,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "ce8b5a93-1841-4897-84db-b100f1c84f4b",
+            "prompt_jinja": "Sentence 1: {{sentence1}}\nSentence 2: {{sentence2}}\n\nDetermine whether the word \"{{word}}\" is used in the same sense in both sentences. Yes or no?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "similar-sense",
+            "acc": 0.542319749216301,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "f934a96d-fe4d-4075-aa47-5595b9a604c7",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nSimilar sense of {{word}}?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.01973963328373276
+        },
+        {
+            "task_name": "wic",
+            "prompt_name": "similar-sense",
+            "acc_norm": 0.5,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wic",
+            "subset": null,
+            "prompt_id": "f934a96d-fe4d-4075-aa47-5595b9a604c7",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nSimilar sense of {{word}}?\n||| {% if label != -1%}\n{{answer_choices[label]}}\n{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "GPT-3 Style",
+            "acc": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "7d377293-d043-4b6c-8ec1-d61eaf14ec67",
+            "prompt_jinja": "Passage: {{ text }} \n\nQuestion: In the passage above, does the pronoun \"{{ span2_text }}\" refer to {{ span1_text }}?\n\nAnswer: ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "GPT-3 Style",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "7d377293-d043-4b6c-8ec1-d61eaf14ec67",
+            "prompt_jinja": "Passage: {{ text }} \n\nQuestion: In the passage above, does the pronoun \"{{ span2_text }}\" refer to {{ span1_text }}?\n\nAnswer: ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "I think they mean",
+            "acc": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "4b3e29cc-ccb8-4e4c-a845-4935ca29cf34",
+            "prompt_jinja": "{{ text }} I think they mean \"{{ text.split(\" \")[span2_index:] | join(\" \") | replace(span2_text, span1_text) }}\" Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "I think they mean",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "4b3e29cc-ccb8-4e4c-a845-4935ca29cf34",
+            "prompt_jinja": "{{ text }} I think they mean \"{{ text.split(\" \")[span2_index:] | join(\" \") | replace(span2_text, span1_text) }}\" Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "Who or what is/are",
+            "acc": 0.40384615384615385,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "d88f3e21-42dc-49a5-924d-69b764a14816",
+            "prompt_jinja": "{{ text }} \n{% if span2_text.lower()  == \"they\" or span2_text.lower() == \"them\" %}\nQuestion: Who or what are \"{{ span2_text.lower() }}\"? {{ span1_text }}?\n{% else %}\nQuestion: Who or what is \"{{ span2_text.lower() }}\"? Is it {{ span1_text }}?\n{% endif %}\nAnswer: ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.048346889526540184
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "Who or what is/are",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "d88f3e21-42dc-49a5-924d-69b764a14816",
+            "prompt_jinja": "{{ text }} \n{% if span2_text.lower()  == \"they\" or span2_text.lower() == \"them\" %}\nQuestion: Who or what are \"{{ span2_text.lower() }}\"? {{ span1_text }}?\n{% else %}\nQuestion: Who or what is \"{{ span2_text.lower() }}\"? Is it {{ span1_text }}?\n{% endif %}\nAnswer: ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "by p they mean",
+            "acc": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "23361c5d-b67f-4c2a-9da7-16301c55d0e1",
+            "prompt_jinja": "{{ text }} Here, by \"{{ span2_text }}\" they mean \"{{ span1_text }}\". Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "by p they mean",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "23361c5d-b67f-4c2a-9da7-16301c55d0e1",
+            "prompt_jinja": "{{ text }} Here, by \"{{ span2_text }}\" they mean \"{{ span1_text }}\". Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "does p stand for",
+            "acc": 0.375,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "7482d24f-cf45-4013-b82d-369489fc958b",
+            "prompt_jinja": "{{ text }} Here, does \"{{ span2_text.lower() }}\" stand for {{ span1_text }}? Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.04770204856076104
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "does p stand for",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "7482d24f-cf45-4013-b82d-369489fc958b",
+            "prompt_jinja": "{{ text }} Here, does \"{{ span2_text.lower() }}\" stand for {{ span1_text }}? Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "does the pronoun refer to",
+            "acc": 0.5480769230769231,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "212fb8b1-8436-4f64-8f37-a9094fe029f4",
+            "prompt_jinja": "{{ text }} In the previous sentence, does the pronoun \"{{ span2_text.lower() }}\" refer to {{ span1_text }}? Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.049038186969314335
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "does the pronoun refer to",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "212fb8b1-8436-4f64-8f37-a9094fe029f4",
+            "prompt_jinja": "{{ text }} In the previous sentence, does the pronoun \"{{ span2_text.lower() }}\" refer to {{ span1_text }}? Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "in other words",
+            "acc": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "2f17f18b-6daa-44ef-a2dd-dddaf04aec0e",
+            "prompt_jinja": "{{ text }} \n\nIn other words, {{ text.split(\" \")[span2_index:] | join(\" \") | replace(span2_text, span1_text) }} True or false? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "in other words",
+            "acc_norm": 0.5288461538461539,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "2f17f18b-6daa-44ef-a2dd-dddaf04aec0e",
+            "prompt_jinja": "{{ text }} \n\nIn other words, {{ text.split(\" \")[span2_index:] | join(\" \") | replace(span2_text, span1_text) }} True or false? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.04918440626354964
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "p is/are r",
+            "acc": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "87f97aa0-1fa9-4f0b-b8e6-89d3c1f19bd6",
+            "prompt_jinja": "Context: {{ text }} \n\n{% if span2_text.lower()  == \"they\" or span2_text.lower() == \"them\" %}\nQuestion: \"{{ span2_text }}\" are {{ span1_text }}. True or false?\n{% else %}\nQuestion: \"{{ span2_text }}\" is {{ span1_text }}. True or false?\n{% endif %}\n\nAnswer: ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "p is/are r",
+            "acc_norm": 0.34615384615384615,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "87f97aa0-1fa9-4f0b-b8e6-89d3c1f19bd6",
+            "prompt_jinja": "Context: {{ text }} \n\n{% if span2_text.lower()  == \"they\" or span2_text.lower() == \"them\" %}\nQuestion: \"{{ span2_text }}\" are {{ span1_text }}. True or false?\n{% else %}\nQuestion: \"{{ span2_text }}\" is {{ span1_text }}. True or false?\n{% endif %}\n\nAnswer: ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.04687634642174987
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "replaced with",
+            "acc": 0.6153846153846154,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "809eacd0-2f6c-4e3a-b52a-57c783879d36",
+            "prompt_jinja": "{{ text }} In the previous sentence, can the pronoun \"{{ span2_text }}\" be replaced with \"{{ span1_text }}\"? Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.047936688680750406
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "replaced with",
+            "acc_norm": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "No",
+                "Yes"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "809eacd0-2f6c-4e3a-b52a-57c783879d36",
+            "prompt_jinja": "{{ text }} In the previous sentence, can the pronoun \"{{ span2_text }}\" be replaced with \"{{ span1_text }}\"? Yes or no? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "the pronoun refers to",
+            "acc": 0.36538461538461536,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "aae24b54-c3a7-4f69-8b77-f6dc115988f8",
+            "prompt_jinja": "{{ text }} \nIn the passage above, the pronoun \"{{ span2_text }}\" refers to {{ span1_text }}. True or false? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0474473339327792
+        },
+        {
+            "task_name": "wsc",
+            "prompt_name": "the pronoun refers to",
+            "acc_norm": 0.5865384615384616,
+            "fixed_answer_choice_list": [
+                "False",
+                "True"
+            ],
+            "dataset_path": "super_glue",
+            "dataset_name": "wsc.fixed",
+            "subset": null,
+            "prompt_id": "aae24b54-c3a7-4f69-8b77-f6dc115988f8",
+            "prompt_jinja": "{{ text }} \nIn the passage above, the pronoun \"{{ span2_text }}\" refers to {{ span1_text }}. True or false? ||| {% if label != -1 %}{{ answer_choices[label] }}{% endif %}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.04852294969729053
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "confident",
+            "acc": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "not confident",
+                "very confident"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "10c354ee-6f4e-4b04-91e1-29e999a8f3e7",
+            "prompt_jinja": "If it's true that\n{{sentence1}}\nhow {{\"confident\"}} should I be that\n{{sentence2}}\n{{\"very confident or not confident?\"}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "confident",
+            "acc_norm": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "not confident",
+                "very confident"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "10c354ee-6f4e-4b04-91e1-29e999a8f3e7",
+            "prompt_jinja": "If it's true that\n{{sentence1}}\nhow {{\"confident\"}} should I be that\n{{sentence2}}\n{{\"very confident or not confident?\"}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "entailment explained",
+            "acc": 0.39436619718309857,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "3a0e46cb-0b96-4972-83f6-29a6c6a09ba9",
+            "prompt_jinja": "{{\"Entailment\"}} means that the second sentence follows from the first sentence. Are the following two sentences an example of entailment?\n{{sentence1}}\n{{sentence2}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.058412510854444266
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "entailment explained",
+            "acc_norm": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "3a0e46cb-0b96-4972-83f6-29a6c6a09ba9",
+            "prompt_jinja": "{{\"Entailment\"}} means that the second sentence follows from the first sentence. Are the following two sentences an example of entailment?\n{{sentence1}}\n{{sentence2}}\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "imply",
+            "acc": 0.4225352112676056,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "a2ce492b-dfd0-4f04-bc44-70c7867ba231",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nDoes the first sentence imply the second sentence?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.05903984205682581
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "imply",
+            "acc_norm": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "a2ce492b-dfd0-4f04-bc44-70c7867ba231",
+            "prompt_jinja": "{{sentence1}}\n{{sentence2}}\nDoes the first sentence imply the second sentence?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "justified",
+            "acc": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "a244158a-a248-4e34-bef7-66e269dd0815",
+            "prompt_jinja": "Someone told me \"{{sentence1}}\" Now, I think that \"{{sentence2}}\" Am I justified in thinking this?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "justified",
+            "acc_norm": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "a244158a-a248-4e34-bef7-66e269dd0815",
+            "prompt_jinja": "Someone told me \"{{sentence1}}\" Now, I think that \"{{sentence2}}\" Am I justified in thinking this?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "mean",
+            "acc": 0.6619718309859155,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "75f89b05-5a81-401b-8a04-8239211a9a95",
+            "prompt_jinja": "Assume that the following is true:\n{{sentence1}}\nDoes this mean that \"{{sentence2}}\"?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_stderr": 0.05653887739133513
+        },
+        {
+            "task_name": "wnli",
+            "prompt_name": "mean",
+            "acc_norm": 0.43661971830985913,
+            "fixed_answer_choice_list": [
+                "no",
+                "yes"
+            ],
+            "dataset_path": "glue",
+            "dataset_name": "wnli",
+            "subset": null,
+            "prompt_id": "75f89b05-5a81-401b-8a04-8239211a9a95",
+            "prompt_jinja": "Assume that the following is true:\n{{sentence1}}\nDoes this mean that \"{{sentence2}}\"?\n|||\n{{answer_choices[label]}}",
+            "prompt_original_task": true,
+            "comment": "",
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        {
+            "task_name": "gsarti/flores_101_afr",
+            "prompt_name": null,
+            "word_perplexity": 139324.0466654445
+        },
+        {
+            "task_name": "gsarti/flores_101_afr",
+            "prompt_name": null,
+            "byte_perplexity": 7.049422805555328
+        },
+        {
+            "task_name": "gsarti/flores_101_afr",
+            "prompt_name": null,
+            "bits_per_byte": 2.8175051369933213
+        },
+        {
+            "task_name": "gsarti/flores_101_amh",
+            "prompt_name": null,
+            "word_perplexity": 105036774.30501972
+        },
+        {
+            "task_name": "gsarti/flores_101_amh",
+            "prompt_name": null,
+            "byte_perplexity": 4.172368790188039
+        },
+        {
+            "task_name": "gsarti/flores_101_amh",
+            "prompt_name": null,
+            "bits_per_byte": 2.0608666814101815
+        },
+        {
+            "task_name": "gsarti/flores_101_ara",
+            "prompt_name": null,
+            "word_perplexity": 674.8640314665696
+        },
+        {
+            "task_name": "gsarti/flores_101_ara",
+            "prompt_name": null,
+            "byte_perplexity": 1.8400375612633983
+        },
+        {
+            "task_name": "gsarti/flores_101_ara",
+            "prompt_name": null,
+            "bits_per_byte": 0.8797352167688847
+        },
+        {
+            "task_name": "gsarti/flores_101_hye",
+            "prompt_name": null,
+            "word_perplexity": 99262887.01092263
+        },
+        {
+            "task_name": "gsarti/flores_101_hye",
+            "prompt_name": null,
+            "byte_perplexity": 3.7481249397064547
+        },
+        {
+            "task_name": "gsarti/flores_101_hye",
+            "prompt_name": null,
+            "bits_per_byte": 1.906169044483402
+        },
+        {
+            "task_name": "gsarti/flores_101_asm",
+            "prompt_name": null,
+            "word_perplexity": 6763188828222.085
+        },
+        {
+            "task_name": "gsarti/flores_101_asm",
+            "prompt_name": null,
+            "byte_perplexity": 5.497254736157445
+        },
+        {
+            "task_name": "gsarti/flores_101_asm",
+            "prompt_name": null,
+            "bits_per_byte": 2.458711333673663
+        },
+        {
+            "task_name": "gsarti/flores_101_ast",
+            "prompt_name": null,
+            "word_perplexity": 10657.272913539553
+        },
+        {
+            "task_name": "gsarti/flores_101_ast",
+            "prompt_name": null,
+            "byte_perplexity": 4.260251728273795
+        },
+        {
+            "task_name": "gsarti/flores_101_ast",
+            "prompt_name": null,
+            "bits_per_byte": 2.0909386784329675
+        },
+        {
+            "task_name": "gsarti/flores_101_azj",
+            "prompt_name": null,
+            "word_perplexity": 45923924.18878753
+        },
+        {
+            "task_name": "gsarti/flores_101_azj",
+            "prompt_name": null,
+            "byte_perplexity": 7.691396328945705
+        },
+        {
+            "task_name": "gsarti/flores_101_azj",
+            "prompt_name": null,
+            "bits_per_byte": 2.9432455349850195
+        },
+        {
+            "task_name": "gsarti/flores_101_bel",
+            "prompt_name": null,
+            "word_perplexity": 23935692.781315073
+        },
+        {
+            "task_name": "gsarti/flores_101_bel",
+            "prompt_name": null,
+            "byte_perplexity": 3.7706591215465943
+        },
+        {
+            "task_name": "gsarti/flores_101_bel",
+            "prompt_name": null,
+            "bits_per_byte": 1.914816732584341
+        },
+        {
+            "task_name": "gsarti/flores_101_ben",
+            "prompt_name": null,
+            "word_perplexity": 2480418685142.412
+        },
+        {
+            "task_name": "gsarti/flores_101_ben",
+            "prompt_name": null,
+            "byte_perplexity": 5.074281765515423
+        },
+        {
+            "task_name": "gsarti/flores_101_ben",
+            "prompt_name": null,
+            "bits_per_byte": 2.3432036318231058
+        },
+        {
+            "task_name": "gsarti/flores_101_bos",
+            "prompt_name": null,
+            "word_perplexity": 229622.13691086147
+        },
+        {
+            "task_name": "gsarti/flores_101_bos",
+            "prompt_name": null,
+            "byte_perplexity": 6.343363734045183
+        },
+        {
+            "task_name": "gsarti/flores_101_bos",
+            "prompt_name": null,
+            "bits_per_byte": 2.665248069942796
+        },
+        {
+            "task_name": "gsarti/flores_101_bul",
+            "prompt_name": null,
+            "word_perplexity": 194851.13344620814
+        },
+        {
+            "task_name": "gsarti/flores_101_bul",
+            "prompt_name": null,
+            "byte_perplexity": 2.8553687444403257
+        },
+        {
+            "task_name": "gsarti/flores_101_bul",
+            "prompt_name": null,
+            "bits_per_byte": 1.5136770683283687
+        },
+        {
+            "task_name": "gsarti/flores_101_mya",
+            "prompt_name": null,
+            "word_perplexity": 5.887577237013639e+18
+        },
+        {
+            "task_name": "gsarti/flores_101_mya",
+            "prompt_name": null,
+            "byte_perplexity": 2.657561458464019
+        },
+        {
+            "task_name": "gsarti/flores_101_mya",
+            "prompt_name": null,
+            "bits_per_byte": 1.4101030557435918
+        },
+        {
+            "task_name": "gsarti/flores_101_cat",
+            "prompt_name": null,
+            "word_perplexity": 179.13123174533087
+        },
+        {
+            "task_name": "gsarti/flores_101_cat",
+            "prompt_name": null,
+            "byte_perplexity": 2.358207169698056
+        },
+        {
+            "task_name": "gsarti/flores_101_cat",
+            "prompt_name": null,
+            "bits_per_byte": 1.2376904653775254
+        },
+        {
+            "task_name": "gsarti/flores_101_ceb",
+            "prompt_name": null,
+            "word_perplexity": 113330.67154113152
+        },
+        {
+            "task_name": "gsarti/flores_101_ceb",
+            "prompt_name": null,
+            "byte_perplexity": 6.896481056329736
+        },
+        {
+            "task_name": "gsarti/flores_101_ceb",
+            "prompt_name": null,
+            "bits_per_byte": 2.7858604115174295
+        },
+        {
+            "task_name": "gsarti/flores_101_zho_simpl",
+            "prompt_name": null,
+            "word_perplexity": 1.0554528210220222e+21
+        },
+        {
+            "task_name": "gsarti/flores_101_zho_simpl",
+            "prompt_name": null,
+            "byte_perplexity": 2.322457417595381
+        },
+        {
+            "task_name": "gsarti/flores_101_zho_simpl",
+            "prompt_name": null,
+            "bits_per_byte": 1.2156521449449949
+        },
+        {
+            "task_name": "gsarti/flores_101_zho_trad",
+            "prompt_name": null,
+            "word_perplexity": 4.787781515987923e+24
+        },
+        {
+            "task_name": "gsarti/flores_101_zho_trad",
+            "prompt_name": null,
+            "byte_perplexity": 2.5709177552415134
+        },
+        {
+            "task_name": "gsarti/flores_101_zho_trad",
+            "prompt_name": null,
+            "bits_per_byte": 1.3622834584784203
+        },
+        {
+            "task_name": "gsarti/flores_101_hrv",
+            "prompt_name": null,
+            "word_perplexity": 307789.1462790266
+        },
+        {
+            "task_name": "gsarti/flores_101_hrv",
+            "prompt_name": null,
+            "byte_perplexity": 6.50559790827845
+        },
+        {
+            "task_name": "gsarti/flores_101_hrv",
+            "prompt_name": null,
+            "bits_per_byte": 2.7016816564307984
+        },
+        {
+            "task_name": "gsarti/flores_101_ces",
+            "prompt_name": null,
+            "word_perplexity": 625101.1441414964
+        },
+        {
+            "task_name": "gsarti/flores_101_ces",
+            "prompt_name": null,
+            "byte_perplexity": 6.126526835715164
+        },
+        {
+            "task_name": "gsarti/flores_101_ces",
+            "prompt_name": null,
+            "bits_per_byte": 2.6150694333085327
+        },
+        {
+            "task_name": "gsarti/flores_101_dan",
+            "prompt_name": null,
+            "word_perplexity": 71695.50336412797
+        },
+        {
+            "task_name": "gsarti/flores_101_dan",
+            "prompt_name": null,
+            "byte_perplexity": 5.778786323448377
+        },
+        {
+            "task_name": "gsarti/flores_101_dan",
+            "prompt_name": null,
+            "bits_per_byte": 2.5307665257708245
+        },
+        {
+            "task_name": "gsarti/flores_101_nld",
+            "prompt_name": null,
+            "word_perplexity": 13951.877058430618
+        },
+        {
+            "task_name": "gsarti/flores_101_nld",
+            "prompt_name": null,
+            "byte_perplexity": 4.535651709856251
+        },
+        {
+            "task_name": "gsarti/flores_101_nld",
+            "prompt_name": null,
+            "bits_per_byte": 2.1813098607926804
+        },
+        {
+            "task_name": "gsarti/flores_101_eng",
+            "prompt_name": null,
+            "word_perplexity": 75.56480997823662
+        },
+        {
+            "task_name": "gsarti/flores_101_eng",
+            "prompt_name": null,
+            "byte_perplexity": 2.061283234268159
+        },
+        {
+            "task_name": "gsarti/flores_101_eng",
+            "prompt_name": null,
+            "bits_per_byte": 1.0435427545613876
+        },
+        {
+            "task_name": "gsarti/flores_101_est",
+            "prompt_name": null,
+            "word_perplexity": 92602633.82439691
+        },
+        {
+            "task_name": "gsarti/flores_101_est",
+            "prompt_name": null,
+            "byte_perplexity": 10.131736127467489
+        },
+        {
+            "task_name": "gsarti/flores_101_est",
+            "prompt_name": null,
+            "bits_per_byte": 3.340809503762674
+        },
+        {
+            "task_name": "gsarti/flores_101_tgl",
+            "prompt_name": null,
+            "word_perplexity": 87554.31770184237
+        },
+        {
+            "task_name": "gsarti/flores_101_tgl",
+            "prompt_name": null,
+            "byte_perplexity": 6.256957969905079
+        },
+        {
+            "task_name": "gsarti/flores_101_tgl",
+            "prompt_name": null,
+            "bits_per_byte": 2.645461413001105
+        },
+        {
+            "task_name": "gsarti/flores_101_fin",
+            "prompt_name": null,
+            "word_perplexity": 91621886.60145952
+        },
+        {
+            "task_name": "gsarti/flores_101_fin",
+            "prompt_name": null,
+            "byte_perplexity": 7.5129644427067355
+        },
+        {
+            "task_name": "gsarti/flores_101_fin",
+            "prompt_name": null,
+            "bits_per_byte": 2.9093822743068216
+        },
+        {
+            "task_name": "gsarti/flores_101_fra",
+            "prompt_name": null,
+            "word_perplexity": 89.45884576931464
+        },
+        {
+            "task_name": "gsarti/flores_101_fra",
+            "prompt_name": null,
+            "byte_perplexity": 2.0177390037335385
+        },
+        {
+            "task_name": "gsarti/flores_101_fra",
+            "prompt_name": null,
+            "bits_per_byte": 1.0127395726746855
+        },
+        {
+            "task_name": "gsarti/flores_101_ful",
+            "prompt_name": null,
+            "word_perplexity": 908715.1423017589
+        },
+        {
+            "task_name": "gsarti/flores_101_ful",
+            "prompt_name": null,
+            "byte_perplexity": 11.810263420287875
+        },
+        {
+            "task_name": "gsarti/flores_101_ful",
+            "prompt_name": null,
+            "bits_per_byte": 3.561969238361191
+        },
+        {
+            "task_name": "gsarti/flores_101_glg",
+            "prompt_name": null,
+            "word_perplexity": 1537.3193913761668
+        },
+        {
+            "task_name": "gsarti/flores_101_glg",
+            "prompt_name": null,
+            "byte_perplexity": 3.2214647330840154
+        },
+        {
+            "task_name": "gsarti/flores_101_glg",
+            "prompt_name": null,
+            "bits_per_byte": 1.6877168009728167
+        },
+        {
+            "task_name": "gsarti/flores_101_lug",
+            "prompt_name": null,
+            "word_perplexity": 32046806.791237485
+        },
+        {
+            "task_name": "gsarti/flores_101_lug",
+            "prompt_name": null,
+            "byte_perplexity": 9.285708185212261
+        },
+        {
+            "task_name": "gsarti/flores_101_lug",
+            "prompt_name": null,
+            "bits_per_byte": 3.2150119431528754
+        },
+        {
+            "task_name": "gsarti/flores_101_kat",
+            "prompt_name": null,
+            "word_perplexity": 1133105340.614723
+        },
+        {
+            "task_name": "gsarti/flores_101_kat",
+            "prompt_name": null,
+            "byte_perplexity": 2.5184571084900518
+        },
+        {
+            "task_name": "gsarti/flores_101_kat",
+            "prompt_name": null,
+            "bits_per_byte": 1.3325401608568794
+        },
+        {
+            "task_name": "gsarti/flores_101_deu",
+            "prompt_name": null,
+            "word_perplexity": 5647.282599404732
+        },
+        {
+            "task_name": "gsarti/flores_101_deu",
+            "prompt_name": null,
+            "byte_perplexity": 3.361758059911202
+        },
+        {
+            "task_name": "gsarti/flores_101_deu",
+            "prompt_name": null,
+            "bits_per_byte": 1.7492158999678582
+        },
+        {
+            "task_name": "gsarti/flores_101_ell",
+            "prompt_name": null,
+            "word_perplexity": 102751.5248402687
+        },
+        {
+            "task_name": "gsarti/flores_101_ell",
+            "prompt_name": null,
+            "byte_perplexity": 2.6139607239932805
+        },
+        {
+            "task_name": "gsarti/flores_101_ell",
+            "prompt_name": null,
+            "bits_per_byte": 1.3862374641150543
+        },
+        {
+            "task_name": "gsarti/flores_101_guj",
+            "prompt_name": null,
+            "word_perplexity": 133216198508.6925
+        },
+        {
+            "task_name": "gsarti/flores_101_guj",
+            "prompt_name": null,
+            "byte_perplexity": 5.125904532570054
+        },
+        {
+            "task_name": "gsarti/flores_101_guj",
+            "prompt_name": null,
+            "bits_per_byte": 2.357806609400009
+        },
+        {
+            "task_name": "gsarti/flores_101_hau",
+            "prompt_name": null,
+            "word_perplexity": 730749.6449046461
+        },
+        {
+            "task_name": "gsarti/flores_101_hau",
+            "prompt_name": null,
+            "byte_perplexity": 11.049458818357667
+        },
+        {
+            "task_name": "gsarti/flores_101_hau",
+            "prompt_name": null,
+            "bits_per_byte": 3.4659038057537184
+        },
+        {
+            "task_name": "gsarti/flores_101_heb",
+            "prompt_name": null,
+            "word_perplexity": 880255.4148832298
+        },
+        {
+            "task_name": "gsarti/flores_101_heb",
+            "prompt_name": null,
+            "byte_perplexity": 3.7036842387723694
+        },
+        {
+            "task_name": "gsarti/flores_101_heb",
+            "prompt_name": null,
+            "bits_per_byte": 1.8889611054621571
+        },
+        {
+            "task_name": "gsarti/flores_101_hin",
+            "prompt_name": null,
+            "word_perplexity": 453226793.5348556
+        },
+        {
+            "task_name": "gsarti/flores_101_hin",
+            "prompt_name": null,
+            "byte_perplexity": 4.581311639568996
+        },
+        {
+            "task_name": "gsarti/flores_101_hin",
+            "prompt_name": null,
+            "bits_per_byte": 2.195760704215568
+        },
+        {
+            "task_name": "gsarti/flores_101_hun",
+            "prompt_name": null,
+            "word_perplexity": 8545882.19823639
+        },
+        {
+            "task_name": "gsarti/flores_101_hun",
+            "prompt_name": null,
+            "byte_perplexity": 7.19531655942431
+        },
+        {
+            "task_name": "gsarti/flores_101_hun",
+            "prompt_name": null,
+            "bits_per_byte": 2.8470581600253615
+        },
+        {
+            "task_name": "gsarti/flores_101_isl",
+            "prompt_name": null,
+            "word_perplexity": 3947458.536983725
+        },
+        {
+            "task_name": "gsarti/flores_101_isl",
+            "prompt_name": null,
+            "byte_perplexity": 8.812045732299993
+        },
+        {
+            "task_name": "gsarti/flores_101_isl",
+            "prompt_name": null,
+            "bits_per_byte": 3.1394769822824644
+        },
+        {
+            "task_name": "gsarti/flores_101_ibo",
+            "prompt_name": null,
+            "word_perplexity": 99576.38125028457
+        },
+        {
+            "task_name": "gsarti/flores_101_ibo",
+            "prompt_name": null,
+            "byte_perplexity": 6.06807351892086
+        },
+        {
+            "task_name": "gsarti/flores_101_ibo",
+            "prompt_name": null,
+            "bits_per_byte": 2.6012385649422316
+        },
+        {
+            "task_name": "gsarti/flores_101_ind",
+            "prompt_name": null,
+            "word_perplexity": 299.41864562936706
+        },
+        {
+            "task_name": "gsarti/flores_101_ind",
+            "prompt_name": null,
+            "byte_perplexity": 2.2193428661828962
+        },
+        {
+            "task_name": "gsarti/flores_101_ind",
+            "prompt_name": null,
+            "bits_per_byte": 1.1501325666473412
+        },
+        {
+            "task_name": "gsarti/flores_101_gle",
+            "prompt_name": null,
+            "word_perplexity": 1548851.5929806433
+        },
+        {
+            "task_name": "gsarti/flores_101_gle",
+            "prompt_name": null,
+            "byte_perplexity": 9.712259930753122
+        },
+        {
+            "task_name": "gsarti/flores_101_gle",
+            "prompt_name": null,
+            "bits_per_byte": 3.2798070331865063
+        },
+        {
+            "task_name": "gsarti/flores_101_ita",
+            "prompt_name": null,
+            "word_perplexity": 1951.0663459405935
+        },
+        {
+            "task_name": "gsarti/flores_101_ita",
+            "prompt_name": null,
+            "byte_perplexity": 3.238337491305615
+        },
+        {
+            "task_name": "gsarti/flores_101_ita",
+            "prompt_name": null,
+            "bits_per_byte": 1.695253347487448
+        },
+        {
+            "task_name": "gsarti/flores_101_jpn",
+            "prompt_name": null,
+            "word_perplexity": 6.0024027118732196e+69
+        },
+        {
+            "task_name": "gsarti/flores_101_jpn",
+            "prompt_name": null,
+            "byte_perplexity": 2.907038023970581
+        },
+        {
+            "task_name": "gsarti/flores_101_jpn",
+            "prompt_name": null,
+            "bits_per_byte": 1.539549942005635
+        },
+        {
+            "task_name": "gsarti/flores_101_jav",
+            "prompt_name": null,
+            "word_perplexity": 956961.3940329206
+        },
+        {
+            "task_name": "gsarti/flores_101_jav",
+            "prompt_name": null,
+            "byte_perplexity": 7.460632752007581
+        },
+        {
+            "task_name": "gsarti/flores_101_jav",
+            "prompt_name": null,
+            "bits_per_byte": 2.899297993680408
+        },
+        {
+            "task_name": "gsarti/flores_101_kea",
+            "prompt_name": null,
+            "word_perplexity": 438558.0012817139
+        },
+        {
+            "task_name": "gsarti/flores_101_kea",
+            "prompt_name": null,
+            "byte_perplexity": 9.281572608888562
+        },
+        {
+            "task_name": "gsarti/flores_101_kea",
+            "prompt_name": null,
+            "bits_per_byte": 3.2143692668645976
+        },
+        {
+            "task_name": "gsarti/flores_101_kam",
+            "prompt_name": null,
+            "word_perplexity": 4288601.196402131
+        },
+        {
+            "task_name": "gsarti/flores_101_kam",
+            "prompt_name": null,
+            "byte_perplexity": 11.436917146974627
+        },
+        {
+            "task_name": "gsarti/flores_101_kam",
+            "prompt_name": null,
+            "bits_per_byte": 3.515626316920499
+        },
+        {
+            "task_name": "gsarti/flores_101_kan",
+            "prompt_name": null,
+            "word_perplexity": 5.3861539364992216e+16
+        },
+        {
+            "task_name": "gsarti/flores_101_kan",
+            "prompt_name": null,
+            "byte_perplexity": 5.274956219477929
+        },
+        {
+            "task_name": "gsarti/flores_101_kan",
+            "prompt_name": null,
+            "bits_per_byte": 2.3991591199422513
+        },
+        {
+            "task_name": "gsarti/flores_101_kaz",
+            "prompt_name": null,
+            "word_perplexity": 89537342.10068764
+        },
+        {
+            "task_name": "gsarti/flores_101_kaz",
+            "prompt_name": null,
+            "byte_perplexity": 3.5945005448756477
+        },
+        {
+            "task_name": "gsarti/flores_101_kaz",
+            "prompt_name": null,
+            "bits_per_byte": 1.845791322405974
+        }
+    ],
+    "versions": {
+        "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,
+        "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,
+        "wnli+confident": 1,
+        "wnli+entailment explained": 1,
+        "wnli+imply": 1,
+        "wnli+justified": 1,
+        "wnli+mean": 1,
+        "gsarti/flores_101_afr+null": 0,
+        "gsarti/flores_101_amh+null": 0,
+        "gsarti/flores_101_ara+null": 0,
+        "gsarti/flores_101_hye+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_mya+null": 0,
+        "gsarti/flores_101_cat+null": 0,
+        "gsarti/flores_101_ceb+null": 0,
+        "gsarti/flores_101_zho_simpl+null": 0,
+        "gsarti/flores_101_zho_trad+null": 0,
+        "gsarti/flores_101_hrv+null": 0,
+        "gsarti/flores_101_ces+null": 0,
+        "gsarti/flores_101_dan+null": 0,
+        "gsarti/flores_101_nld+null": 0,
+        "gsarti/flores_101_eng+null": 0,
+        "gsarti/flores_101_est+null": 0,
+        "gsarti/flores_101_tgl+null": 0,
+        "gsarti/flores_101_fin+null": 0,
+        "gsarti/flores_101_fra+null": 0,
+        "gsarti/flores_101_ful+null": 0,
+        "gsarti/flores_101_glg+null": 0,
+        "gsarti/flores_101_lug+null": 0,
+        "gsarti/flores_101_kat+null": 0,
+        "gsarti/flores_101_deu+null": 0,
+        "gsarti/flores_101_ell+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_hun+null": 0,
+        "gsarti/flores_101_isl+null": 0,
+        "gsarti/flores_101_ibo+null": 0,
+        "gsarti/flores_101_ind+null": 0,
+        "gsarti/flores_101_gle+null": 0,
+        "gsarti/flores_101_ita+null": 0,
+        "gsarti/flores_101_jpn+null": 0,
+        "gsarti/flores_101_jav+null": 0,
+        "gsarti/flores_101_kea+null": 0,
+        "gsarti/flores_101_kam+null": 0,
+        "gsarti/flores_101_kan+null": 0,
+        "gsarti/flores_101_kaz+null": 0
+    },
+    "table_results": {
+        "wic+GPT-3-prompt": {
+            "task_name": "wic",
+            "prompt_name": "GPT-3-prompt",
+            "acc": 0.5,
+            "acc_stderr": 0.01981072129375818,
+            "acc_norm": 0.5,
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        "wic+GPT-3-prompt-with-label": {
+            "task_name": "wic",
+            "prompt_name": "GPT-3-prompt-with-label",
+            "acc": 0.49216300940438873,
+            "acc_stderr": 0.019808287657813832,
+            "acc_norm": 0.5,
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        "wic+affirmation_true_or_false": {
+            "task_name": "wic",
+            "prompt_name": "affirmation_true_or_false",
+            "acc": 0.5,
+            "acc_stderr": 0.01981072129375818,
+            "acc_norm": 0.5078369905956113,
+            "acc_norm_stderr": 0.019808287657813832
+        },
+        "wic+grammar_homework": {
+            "task_name": "wic",
+            "prompt_name": "grammar_homework",
+            "acc": 0.5094043887147336,
+            "acc_stderr": 0.019807216763271497,
+            "acc_norm": 0.49843260188087773,
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        "wic+polysemous": {
+            "task_name": "wic",
+            "prompt_name": "polysemous",
+            "acc": 0.512539184952978,
+            "acc_stderr": 0.019804490588592596,
+            "acc_norm": 0.49843260188087773,
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        "wic+question-context": {
+            "task_name": "wic",
+            "prompt_name": "question-context",
+            "acc": 0.5266457680250783,
+            "acc_stderr": 0.019782570188812167,
+            "acc_norm": 0.5031347962382445,
+            "acc_norm_stderr": 0.019810331932097542
+        },
+        "wic+question-context-meaning": {
+            "task_name": "wic",
+            "prompt_name": "question-context-meaning",
+            "acc": 0.5438871473354232,
+            "acc_stderr": 0.019734259601993404,
+            "acc_norm": 0.5015673981191222,
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        "wic+question-context-meaning-with-label": {
+            "task_name": "wic",
+            "prompt_name": "question-context-meaning-with-label",
+            "acc": 0.5156739811912225,
+            "acc_stderr": 0.019800984955347847,
+            "acc_norm": 0.5015673981191222,
+            "acc_norm_stderr": 0.019810623954060382
+        },
+        "wic+same_sense": {
+            "task_name": "wic",
+            "prompt_name": "same_sense",
+            "acc": 0.5047021943573667,
+            "acc_stderr": 0.019809845219259763,
+            "acc_norm": 0.5,
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        "wic+similar-sense": {
+            "task_name": "wic",
+            "prompt_name": "similar-sense",
+            "acc": 0.542319749216301,
+            "acc_stderr": 0.01973963328373276,
+            "acc_norm": 0.5,
+            "acc_norm_stderr": 0.01981072129375818
+        },
+        "wsc+GPT-3 Style": {
+            "task_name": "wsc",
+            "prompt_name": "GPT-3 Style",
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+I think they mean": {
+            "task_name": "wsc",
+            "prompt_name": "I think they mean",
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+Who or what is/are": {
+            "task_name": "wsc",
+            "prompt_name": "Who or what is/are",
+            "acc": 0.40384615384615385,
+            "acc_stderr": 0.048346889526540184,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+by p they mean": {
+            "task_name": "wsc",
+            "prompt_name": "by p they mean",
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+does p stand for": {
+            "task_name": "wsc",
+            "prompt_name": "does p stand for",
+            "acc": 0.375,
+            "acc_stderr": 0.04770204856076104,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+does the pronoun refer to": {
+            "task_name": "wsc",
+            "prompt_name": "does the pronoun refer to",
+            "acc": 0.5480769230769231,
+            "acc_stderr": 0.049038186969314335,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+in other words": {
+            "task_name": "wsc",
+            "prompt_name": "in other words",
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792,
+            "acc_norm": 0.5288461538461539,
+            "acc_norm_stderr": 0.04918440626354964
+        },
+        "wsc+p is/are r": {
+            "task_name": "wsc",
+            "prompt_name": "p is/are r",
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792,
+            "acc_norm": 0.34615384615384615,
+            "acc_norm_stderr": 0.04687634642174987
+        },
+        "wsc+replaced with": {
+            "task_name": "wsc",
+            "prompt_name": "replaced with",
+            "acc": 0.6153846153846154,
+            "acc_stderr": 0.047936688680750406,
+            "acc_norm": 0.36538461538461536,
+            "acc_norm_stderr": 0.0474473339327792
+        },
+        "wsc+the pronoun refers to": {
+            "task_name": "wsc",
+            "prompt_name": "the pronoun refers to",
+            "acc": 0.36538461538461536,
+            "acc_stderr": 0.0474473339327792,
+            "acc_norm": 0.5865384615384616,
+            "acc_norm_stderr": 0.04852294969729053
+        },
+        "wnli+confident": {
+            "task_name": "wnli",
+            "prompt_name": "confident",
+            "acc": 0.43661971830985913,
+            "acc_stderr": 0.0592793555841297,
+            "acc_norm": 0.43661971830985913,
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        "wnli+entailment explained": {
+            "task_name": "wnli",
+            "prompt_name": "entailment explained",
+            "acc": 0.39436619718309857,
+            "acc_stderr": 0.058412510854444266,
+            "acc_norm": 0.43661971830985913,
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        "wnli+imply": {
+            "task_name": "wnli",
+            "prompt_name": "imply",
+            "acc": 0.4225352112676056,
+            "acc_stderr": 0.05903984205682581,
+            "acc_norm": 0.43661971830985913,
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        "wnli+justified": {
+            "task_name": "wnli",
+            "prompt_name": "justified",
+            "acc": 0.43661971830985913,
+            "acc_stderr": 0.0592793555841297,
+            "acc_norm": 0.43661971830985913,
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        "wnli+mean": {
+            "task_name": "wnli",
+            "prompt_name": "mean",
+            "acc": 0.6619718309859155,
+            "acc_stderr": 0.05653887739133513,
+            "acc_norm": 0.43661971830985913,
+            "acc_norm_stderr": 0.0592793555841297
+        },
+        "gsarti/flores_101_afr+null": {
+            "task_name": "gsarti/flores_101_afr",
+            "prompt_name": "null",
+            "word_perplexity": 139324.0466654445,
+            "byte_perplexity": 7.049422805555328,
+            "bits_per_byte": 2.8175051369933213
+        },
+        "gsarti/flores_101_amh+null": {
+            "task_name": "gsarti/flores_101_amh",
+            "prompt_name": "null",
+            "word_perplexity": 105036774.30501972,
+            "byte_perplexity": 4.172368790188039,
+            "bits_per_byte": 2.0608666814101815
+        },
+        "gsarti/flores_101_ara+null": {
+            "task_name": "gsarti/flores_101_ara",
+            "prompt_name": "null",
+            "word_perplexity": 674.8640314665696,
+            "byte_perplexity": 1.8400375612633983,
+            "bits_per_byte": 0.8797352167688847
+        },
+        "gsarti/flores_101_hye+null": {
+            "task_name": "gsarti/flores_101_hye",
+            "prompt_name": "null",
+            "word_perplexity": 99262887.01092263,
+            "byte_perplexity": 3.7481249397064547,
+            "bits_per_byte": 1.906169044483402
+        },
+        "gsarti/flores_101_asm+null": {
+            "task_name": "gsarti/flores_101_asm",
+            "prompt_name": "null",
+            "word_perplexity": 6763188828222.085,
+            "byte_perplexity": 5.497254736157445,
+            "bits_per_byte": 2.458711333673663
+        },
+        "gsarti/flores_101_ast+null": {
+            "task_name": "gsarti/flores_101_ast",
+            "prompt_name": "null",
+            "word_perplexity": 10657.272913539553,
+            "byte_perplexity": 4.260251728273795,
+            "bits_per_byte": 2.0909386784329675
+        },
+        "gsarti/flores_101_azj+null": {
+            "task_name": "gsarti/flores_101_azj",
+            "prompt_name": "null",
+            "word_perplexity": 45923924.18878753,
+            "byte_perplexity": 7.691396328945705,
+            "bits_per_byte": 2.9432455349850195
+        },
+        "gsarti/flores_101_bel+null": {
+            "task_name": "gsarti/flores_101_bel",
+            "prompt_name": "null",
+            "word_perplexity": 23935692.781315073,
+            "byte_perplexity": 3.7706591215465943,
+            "bits_per_byte": 1.914816732584341
+        },
+        "gsarti/flores_101_ben+null": {
+            "task_name": "gsarti/flores_101_ben",
+            "prompt_name": "null",
+            "word_perplexity": 2480418685142.412,
+            "byte_perplexity": 5.074281765515423,
+            "bits_per_byte": 2.3432036318231058
+        },
+        "gsarti/flores_101_bos+null": {
+            "task_name": "gsarti/flores_101_bos",
+            "prompt_name": "null",
+            "word_perplexity": 229622.13691086147,
+            "byte_perplexity": 6.343363734045183,
+            "bits_per_byte": 2.665248069942796
+        },
+        "gsarti/flores_101_bul+null": {
+            "task_name": "gsarti/flores_101_bul",
+            "prompt_name": "null",
+            "word_perplexity": 194851.13344620814,
+            "byte_perplexity": 2.8553687444403257,
+            "bits_per_byte": 1.5136770683283687
+        },
+        "gsarti/flores_101_mya+null": {
+            "task_name": "gsarti/flores_101_mya",
+            "prompt_name": "null",
+            "word_perplexity": 5.887577237013639e+18,
+            "byte_perplexity": 2.657561458464019,
+            "bits_per_byte": 1.4101030557435918
+        },
+        "gsarti/flores_101_cat+null": {
+            "task_name": "gsarti/flores_101_cat",
+            "prompt_name": "null",
+            "word_perplexity": 179.13123174533087,
+            "byte_perplexity": 2.358207169698056,
+            "bits_per_byte": 1.2376904653775254
+        },
+        "gsarti/flores_101_ceb+null": {
+            "task_name": "gsarti/flores_101_ceb",
+            "prompt_name": "null",
+            "word_perplexity": 113330.67154113152,
+            "byte_perplexity": 6.896481056329736,
+            "bits_per_byte": 2.7858604115174295
+        },
+        "gsarti/flores_101_zho_simpl+null": {
+            "task_name": "gsarti/flores_101_zho_simpl",
+            "prompt_name": "null",
+            "word_perplexity": 1.0554528210220222e+21,
+            "byte_perplexity": 2.322457417595381,
+            "bits_per_byte": 1.2156521449449949
+        },
+        "gsarti/flores_101_zho_trad+null": {
+            "task_name": "gsarti/flores_101_zho_trad",
+            "prompt_name": "null",
+            "word_perplexity": 4.787781515987923e+24,
+            "byte_perplexity": 2.5709177552415134,
+            "bits_per_byte": 1.3622834584784203
+        },
+        "gsarti/flores_101_hrv+null": {
+            "task_name": "gsarti/flores_101_hrv",
+            "prompt_name": "null",
+            "word_perplexity": 307789.1462790266,
+            "byte_perplexity": 6.50559790827845,
+            "bits_per_byte": 2.7016816564307984
+        },
+        "gsarti/flores_101_ces+null": {
+            "task_name": "gsarti/flores_101_ces",
+            "prompt_name": "null",
+            "word_perplexity": 625101.1441414964,
+            "byte_perplexity": 6.126526835715164,
+            "bits_per_byte": 2.6150694333085327
+        },
+        "gsarti/flores_101_dan+null": {
+            "task_name": "gsarti/flores_101_dan",
+            "prompt_name": "null",
+            "word_perplexity": 71695.50336412797,
+            "byte_perplexity": 5.778786323448377,
+            "bits_per_byte": 2.5307665257708245
+        },
+        "gsarti/flores_101_nld+null": {
+            "task_name": "gsarti/flores_101_nld",
+            "prompt_name": "null",
+            "word_perplexity": 13951.877058430618,
+            "byte_perplexity": 4.535651709856251,
+            "bits_per_byte": 2.1813098607926804
+        },
+        "gsarti/flores_101_eng+null": {
+            "task_name": "gsarti/flores_101_eng",
+            "prompt_name": "null",
+            "word_perplexity": 75.56480997823662,
+            "byte_perplexity": 2.061283234268159,
+            "bits_per_byte": 1.0435427545613876
+        },
+        "gsarti/flores_101_est+null": {
+            "task_name": "gsarti/flores_101_est",
+            "prompt_name": "null",
+            "word_perplexity": 92602633.82439691,
+            "byte_perplexity": 10.131736127467489,
+            "bits_per_byte": 3.340809503762674
+        },
+        "gsarti/flores_101_tgl+null": {
+            "task_name": "gsarti/flores_101_tgl",
+            "prompt_name": "null",
+            "word_perplexity": 87554.31770184237,
+            "byte_perplexity": 6.256957969905079,
+            "bits_per_byte": 2.645461413001105
+        },
+        "gsarti/flores_101_fin+null": {
+            "task_name": "gsarti/flores_101_fin",
+            "prompt_name": "null",
+            "word_perplexity": 91621886.60145952,
+            "byte_perplexity": 7.5129644427067355,
+            "bits_per_byte": 2.9093822743068216
+        },
+        "gsarti/flores_101_fra+null": {
+            "task_name": "gsarti/flores_101_fra",
+            "prompt_name": "null",
+            "word_perplexity": 89.45884576931464,
+            "byte_perplexity": 2.0177390037335385,
+            "bits_per_byte": 1.0127395726746855
+        },
+        "gsarti/flores_101_ful+null": {
+            "task_name": "gsarti/flores_101_ful",
+            "prompt_name": "null",
+            "word_perplexity": 908715.1423017589,
+            "byte_perplexity": 11.810263420287875,
+            "bits_per_byte": 3.561969238361191
+        },
+        "gsarti/flores_101_glg+null": {
+            "task_name": "gsarti/flores_101_glg",
+            "prompt_name": "null",
+            "word_perplexity": 1537.3193913761668,
+            "byte_perplexity": 3.2214647330840154,
+            "bits_per_byte": 1.6877168009728167
+        },
+        "gsarti/flores_101_lug+null": {
+            "task_name": "gsarti/flores_101_lug",
+            "prompt_name": "null",
+            "word_perplexity": 32046806.791237485,
+            "byte_perplexity": 9.285708185212261,
+            "bits_per_byte": 3.2150119431528754
+        },
+        "gsarti/flores_101_kat+null": {
+            "task_name": "gsarti/flores_101_kat",
+            "prompt_name": "null",
+            "word_perplexity": 1133105340.614723,
+            "byte_perplexity": 2.5184571084900518,
+            "bits_per_byte": 1.3325401608568794
+        },
+        "gsarti/flores_101_deu+null": {
+            "task_name": "gsarti/flores_101_deu",
+            "prompt_name": "null",
+            "word_perplexity": 5647.282599404732,
+            "byte_perplexity": 3.361758059911202,
+            "bits_per_byte": 1.7492158999678582
+        },
+        "gsarti/flores_101_ell+null": {
+            "task_name": "gsarti/flores_101_ell",
+            "prompt_name": "null",
+            "word_perplexity": 102751.5248402687,
+            "byte_perplexity": 2.6139607239932805,
+            "bits_per_byte": 1.3862374641150543
+        },
+        "gsarti/flores_101_guj+null": {
+            "task_name": "gsarti/flores_101_guj",
+            "prompt_name": "null",
+            "word_perplexity": 133216198508.6925,
+            "byte_perplexity": 5.125904532570054,
+            "bits_per_byte": 2.357806609400009
+        },
+        "gsarti/flores_101_hau+null": {
+            "task_name": "gsarti/flores_101_hau",
+            "prompt_name": "null",
+            "word_perplexity": 730749.6449046461,
+            "byte_perplexity": 11.049458818357667,
+            "bits_per_byte": 3.4659038057537184
+        },
+        "gsarti/flores_101_heb+null": {
+            "task_name": "gsarti/flores_101_heb",
+            "prompt_name": "null",
+            "word_perplexity": 880255.4148832298,
+            "byte_perplexity": 3.7036842387723694,
+            "bits_per_byte": 1.8889611054621571
+        },
+        "gsarti/flores_101_hin+null": {
+            "task_name": "gsarti/flores_101_hin",
+            "prompt_name": "null",
+            "word_perplexity": 453226793.5348556,
+            "byte_perplexity": 4.581311639568996,
+            "bits_per_byte": 2.195760704215568
+        },
+        "gsarti/flores_101_hun+null": {
+            "task_name": "gsarti/flores_101_hun",
+            "prompt_name": "null",
+            "word_perplexity": 8545882.19823639,
+            "byte_perplexity": 7.19531655942431,
+            "bits_per_byte": 2.8470581600253615
+        },
+        "gsarti/flores_101_isl+null": {
+            "task_name": "gsarti/flores_101_isl",
+            "prompt_name": "null",
+            "word_perplexity": 3947458.536983725,
+            "byte_perplexity": 8.812045732299993,
+            "bits_per_byte": 3.1394769822824644
+        },
+        "gsarti/flores_101_ibo+null": {
+            "task_name": "gsarti/flores_101_ibo",
+            "prompt_name": "null",
+            "word_perplexity": 99576.38125028457,
+            "byte_perplexity": 6.06807351892086,
+            "bits_per_byte": 2.6012385649422316
+        },
+        "gsarti/flores_101_ind+null": {
+            "task_name": "gsarti/flores_101_ind",
+            "prompt_name": "null",
+            "word_perplexity": 299.41864562936706,
+            "byte_perplexity": 2.2193428661828962,
+            "bits_per_byte": 1.1501325666473412
+        },
+        "gsarti/flores_101_gle+null": {
+            "task_name": "gsarti/flores_101_gle",
+            "prompt_name": "null",
+            "word_perplexity": 1548851.5929806433,
+            "byte_perplexity": 9.712259930753122,
+            "bits_per_byte": 3.2798070331865063
+        },
+        "gsarti/flores_101_ita+null": {
+            "task_name": "gsarti/flores_101_ita",
+            "prompt_name": "null",
+            "word_perplexity": 1951.0663459405935,
+            "byte_perplexity": 3.238337491305615,
+            "bits_per_byte": 1.695253347487448
+        },
+        "gsarti/flores_101_jpn+null": {
+            "task_name": "gsarti/flores_101_jpn",
+            "prompt_name": "null",
+            "word_perplexity": 6.0024027118732196e+69,
+            "byte_perplexity": 2.907038023970581,
+            "bits_per_byte": 1.539549942005635
+        },
+        "gsarti/flores_101_jav+null": {
+            "task_name": "gsarti/flores_101_jav",
+            "prompt_name": "null",
+            "word_perplexity": 956961.3940329206,
+            "byte_perplexity": 7.460632752007581,
+            "bits_per_byte": 2.899297993680408
+        },
+        "gsarti/flores_101_kea+null": {
+            "task_name": "gsarti/flores_101_kea",
+            "prompt_name": "null",
+            "word_perplexity": 438558.0012817139,
+            "byte_perplexity": 9.281572608888562,
+            "bits_per_byte": 3.2143692668645976
+        },
+        "gsarti/flores_101_kam+null": {
+            "task_name": "gsarti/flores_101_kam",
+            "prompt_name": "null",
+            "word_perplexity": 4288601.196402131,
+            "byte_perplexity": 11.436917146974627,
+            "bits_per_byte": 3.515626316920499
+        },
+        "gsarti/flores_101_kan+null": {
+            "task_name": "gsarti/flores_101_kan",
+            "prompt_name": "null",
+            "word_perplexity": 5.3861539364992216e+16,
+            "byte_perplexity": 5.274956219477929,
+            "bits_per_byte": 2.3991591199422513
+        },
+        "gsarti/flores_101_kaz+null": {
+            "task_name": "gsarti/flores_101_kaz",
+            "prompt_name": "null",
+            "word_perplexity": 89537342.10068764,
+            "byte_perplexity": 3.5945005448756477,
+            "bits_per_byte": 1.845791322405974
+        }
+    },
+    "config": {
+        "adaptive_seq_len": true,
+        "num_fewshot": 0,
+        "bootstrap_iters": 100000
+    }
+}

bigscience/evaluation/results/tr11/conversion/json_to_markdown.py

@@ -0,0 +1,307 @@
+"""
+Table example:
+
+| Task | Language | Metric | BLOOM-176B | OPT-176B |
+|:--------|:-----------------|:------------------------|-------------:|------------:|
+| arc_challenge | eng | acc | 0.4112627986348123 | 0.4121160409556314 |
+
+
+Metadata example:
+
+model-index:
+- name: bart-large-cnn-samsum
+  results:
+    - task:
+      type: summarization
+      name: Summarization
+    dataset:
+      name: 'SAMSum Corpus: A Human-annotated Dialogue Dataset for Abstractive Summarization'
+      type: samsum
+    metrics:
+    - name: Validation ROGUE-1
+      type: rogue-1
+      value: 42.621
+    - name: Validation ROGUE-2
+      type: rogue-2
+      value: 21.9825
+    - name: Validation ROGUE-L
+      type: rogue-l
+      value: 33.034
+    - name: Test ROGUE-1
+      type: rogue-1
+      value: 41.3174
+    - name: Test ROGUE-2
+      type: rogue-2
+      value: 20.8716
+    - name: Test ROGUE-L
+      type: rogue-l
+      value: 32.1337
+  - task:
+      type: summarization
+      name: Summarization
+    dataset:
+      name: samsum
+      type: samsum
+      config: samsum
+      split: test
+    metrics:
+    - name: ROUGE-1
+      type: rouge
+      value: 41.3282
+      verified: true
+    - name: ROUGE-2
+      type: rouge
+      value: 20.8755
+      verified: true
+    - name: ROUGE-L
+      type: rouge
+      value: 32.1353
+      verified: true
+    - name: ROUGE-LSUM
+      type: rouge
+      value: 38.401
+      verified: true
+    - name: loss
+      type: loss
+      value: 1.4297215938568115
+      verified: true
+    - name: gen_len
+      type: gen_len
+      value: 60.0757
+      verified: true
+"""
+
+import json
+import statistics
+
+FILE_NAMES = ["bslmeval", "humaneval_temp02", "humaneval_temp06", "humaneval_temp08"]
+
+# Optionally subselect tasks
+SELECTED_LIST = [
+    "winogrande"
+]
+
+with open("bloom2b5/bslmeval.json", "r") as f:
+    bloom_bslmeval = json.load(f)
+
+with open("opt/bslmeval.json", "r") as f:
+    opt_bslmeval = json.load(f)
+
+
+
+results_formatted = {}
+for task_name in bloom_bslmeval["results"]:
+    #if task_name not in SELECTED_LIST:
+    #    continue
+    date_keys = list(bloom_bslmeval["results"][task_name].keys())
+    assert len(date_keys) == 1
+    metrics = bloom_bslmeval["results"][task_name][date_keys[0]]
+
+    lang = "eng"
+    if "gsarti/flores_101_" in task_name:
+        lang = task_name.replace("gsarti/flores_101_", "").replace("+null", "")
+    elif "lambada_mt_de" in task_name:
+        lang = "deu"
+    elif "lambada_mt_en" in task_name:
+        lang = "eng"
+    elif "lambada_mt_es" in task_name:
+        lang = "esp"
+    elif "lambada_mt_it" in task_name:
+        lang = "ita"
+    elif "lambada" == task_name:
+        continue
+    elif "crows_pairs_french" in task_name:
+        lang = "fra"
+    elif "headqa" == task_name:
+        lang = "esp"
+
+    if "acc" in metrics:
+        main_metric_name = "acc ↑"
+    elif "byte_perplexity" in metrics:
+        main_metric_name = "byte_perplexity ↓"
+    elif "pass@100" in metrics:
+        main_metric_name = "pass@100 ↑"
+    elif "em" in metrics:
+        main_metric_name = "em ↑"
+
+    date_keys_opt = list(opt_bslmeval["results"][task_name].keys())
+    score_opt = opt_bslmeval["results"][task_name][date_keys_opt[0]][main_metric_name[:-2]]
+
+    fin_task_name = metrics.get("task_name", task_name)
+    
+    results_formatted.setdefault(fin_task_name, {})
+    results_formatted[fin_task_name].setdefault("prompts", [])
+    results_formatted[fin_task_name].setdefault("all_metrics", [])
+    results_formatted[fin_task_name].setdefault("main_metrics", [])
+
+    if "prompt_name" in metrics:
+        results_formatted[fin_task_name]["prompts"].append(metrics["prompt_name"])
+    results_formatted[fin_task_name]["name"] = fin_task_name
+    results_formatted[fin_task_name]["lang"] = lang
+    results_formatted[fin_task_name]["all_metrics"].append(metrics) # [{name: score}]
+    results_formatted[fin_task_name]["main_metrics"].append((main_metric_name, metrics[main_metric_name[:-2]], score_opt))
+    results_formatted[fin_task_name]["type"] = "text-generation"
+
+# Take Median of scores
+for k, v in results_formatted.items():
+    if "prompts" in v and len(v["prompts"]) > 1:
+        assert len(v["all_metrics"]) == len(v["main_metrics"])
+        num_scores = len(v["main_metrics"])
+
+        bloom_median = statistics.median([triplet[1] for triplet in v["main_metrics"]])
+        opt_median = statistics.median([triplet[2] for triplet in v["main_metrics"]])
+
+        results_formatted[k]["main_metrics"] = [(
+            v["main_metrics"][0][0],
+            bloom_median,
+            opt_median,
+        )]
+
+        results_formatted[k]["name"] = results_formatted[k]["name"] + f" (Median of {num_scores} prompts)"
+
+
+
+def keep_best_score(new_eval, old_eval):
+    for k, v in new_eval.items():
+        old_eval[k] = max(old_eval[k], v) 
+    return old_eval
+
+for i, temp in enumerate(["02", "06", "08"]):
+    with open(f"bloom/humaneval_temp{temp}.json", "r") as f:
+        if i > 0:
+            keep_best_score(json.load(f), bloom_humaneval)
+        else:
+            bloom_humaneval = json.load(f)
+    with open(f"opt/humaneval_temp{temp}.json", "r") as f:
+        if i > 0:
+            keep_best_score(json.load(f), opt_humaneval)
+        else:
+            opt_humaneval = json.load(f)
+
+results_formatted["humaneval"] = {
+    "name": "humaneval",
+    "lang": "python",
+    "all_metrics": [bloom_humaneval], # [{name: score}]
+    "main_metrics": [(f"{name} ↑", score, opt_humaneval[name]) for name, score in bloom_humaneval.items()],
+    "type": "text-generation"
+}
+
+
+
+# Add multilingual average
+for k, v in results_formatted.items():
+    if "prompts" in v and len(v["prompts"]) > 1 and len(v["main_metrics"]) > 1:
+        assert len(v["all_metrics"]) == len(v["main_metrics"]), f"{k}, {len(v['all_metrics'])}, {len(v['main_metrics'])}"
+        num_scores = len(v["main_metrics"])
+
+        bloom_median = statistics.median([triplet[1] for triplet in v["main_metrics"]])
+        opt_median = statistics.median([triplet[2] for triplet in v["main_metrics"]])
+
+        results_formatted[k]["main_metrics"] = [(
+            v["main_metrics"][0][0],
+            bloom_median,
+            opt_median,
+        )]
+
+        results_formatted[k]["name"] = results_formatted[k]["name"] + f" (Median of {num_scores} prompts)"
+
+"""Optional aggregated statistics
+bloom_mean = statistics.mean([triplet[1] for k,v in results_formatted.items() for triplet in v["main_metrics"] if v["lang"] == "eng"])
+opt_mean = statistics.mean([triplet[2] for k,v in results_formatted.items() for triplet in v["main_metrics"] if v["lang"] == "eng"])
+
+results_formatted["mean_eng"] = {
+    "name": "mean_eng ↑",
+    "lang": "eng",
+    "all_metrics": [{"mean": bloom_mean}], # [{name: score}]
+    "main_metrics": [("mean", bloom_mean, opt_mean)],
+    "type": "text-generation"
+}
+
+bloom_mean = statistics.mean([triplet[1] for k,v in results_formatted.items() for triplet in v["main_metrics"] if "flores" in k])
+opt_mean = statistics.mean([triplet[2] for k,v in results_formatted.items() for triplet in v["main_metrics"] if "flores" in k])
+
+results_formatted["mean_multilingual"] = {
+    "name": "mean_multilingual (Flores) ↓",
+    "lang": "mul",
+    "all_metrics": [{"mean": bloom_mean}], # [{name: score}]
+    "main_metrics": [("mean", bloom_mean, opt_mean)],
+    "type": "text-generation"
+}
+
+main_metrics = ([triplet for k,v in results_formatted.items() for triplet in v["main_metrics"]])
+
+bloom_best_on, opt_best_on = 0,0
+for (name, bloom, opt) in main_metrics:
+    if name[:-2] in ["acc", "em"] or "pass" in name:
+        if bloom > opt:
+            bloom_best_on += 1
+        elif bloom < opt:
+            opt_best_on += 1
+    elif name[:-2] in ["byte_perplexity"]:
+        if bloom < opt:
+            bloom_best_on += 1
+        elif bloom > opt:
+            opt_best_on += 1
+"""
+### Markdown Table ###
+
+HEADER = "| Task | Language | Metric | BLOOM-350M | BLOOM-750M | BLOOM-1B3 | BLOOM-2B5 | BLOOM-6B3 | BLOOM-176B |"
+SEP = "|:----|:----|:----|:----:|"
+ONE_LINE = "| {} | {} | {} | {} |"
+
+TABLE_STRING = "\n".join([HEADER, SEP])
+
+for task_name, res_dict in results_formatted.items():
+    for (name, score, score_opt) in res_dict["main_metrics"]:
+        TABLE_STRING += "\n" + ONE_LINE.format(
+            res_dict["name"],
+            res_dict["lang"],
+            name,
+            round(score, 3),
+            round(score_opt, 3),
+        )
+
+with open("./mdtable.txt", "w") as f:
+    f.write(TABLE_STRING)
+
+
+
+### Metadata ###
+
+HEADER = "model-index:"
+MODEL = "- name: bloom"
+RES = "  results:"
+
+META_STRING = "\n".join([HEADER, MODEL, RES])
+
+ONE_TASK = "  - task:\n      type: {}\n      name: {}\n    dataset:\n      name: {}\n      type: {}\n    metrics:"
+ONE_METRIC = "    - name: {}\n      type: {}\n      value: {}\n      verified: false"
+
+for task_name, res_dict in results_formatted.items():
+    META_STRING += "\n" + ONE_TASK.format(
+        res_dict["type"],
+        res_dict["type"].replace("-", " "),
+        task_name,
+        task_name,
+    )
+    for (name, score, score_opt) in res_dict["main_metrics"]:
+            META_STRING += "\n" + ONE_METRIC.format(
+                name.split(" ")[0],
+                name.split(" ")[0],
+                score
+            )   
+"""
+    for metrics in res_dict["all_metrics"]:
+        for metric_name, metric in metrics.items():
+            if isinstance(metric, str):
+                continue
+            META_STRING += "\n" + ONE_METRIC.format(
+                metric_name,
+                metric_name,
+                metric
+            )   
+"""
+
+
+with open("./mdmeta.txt", "w") as f:
+    f.write(META_STRING)

bigscience/evaluation/results/tr11/opt/humaneval_temp02.json

@@ -0,0 +1 @@
+{"pass@1": 0.0, "pass@10": 0.0, "pass@100": 0.0}

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

@@ -0,0 +1 @@
+{"pass@1": 3.0487804878048808e-05, "pass@10": 0.0003048780487804881, "pass@100": 0.003048780487804878}

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

@@ -0,0 +1 @@
+{"pass@1": 0.0, "pass@10": 0.0, "pass@100": 0.0}

bigscience/evaluation/results/tr11/scripts/download_bsevalharness.py

@@ -0,0 +1,21 @@
+# Downloads the specified taks in the evaluation harness
+# This is particularly useful when running in environments where the GPU nodes 
+# do not have internet access. This way we can pre-download them and use the cached data-set during evaluation.
+
+from lm_eval import tasks
+from lm_eval.tasks import ALL_TASKS
+import argparse
+import os
+
+
+parser = argparse.ArgumentParser(description='Download evaluation harness', allow_abbrev=False)
+parser.add_argument('--task_list', type=str, default = "all", help='Either "all" or comma separated list of tasks to download.')
+args = parser.parse_args()
+
+def main():
+    task_list = ALL_TASKS if args.task_list == 'all' else args.task_list.split(',')
+    tasks.get_task_dict_promptsource(task_list)
+
+if __name__ == '__main__':
+    main()
+

bigscience/evaluation/results/tr11/scripts/multi_eurlex_tmp.slurm

@@ -0,0 +1,63 @@
+#!/bin/bash
+#SBATCH --job-name=multieurlex
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=8           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:1                 # number of gpus
+#SBATCH --constraint=a100
+#SBATCH --reservation=hug
+#SBATCH --time 20:00:00             # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-tr13f-6B3-ml-t0
+conda activate muennighofflmevalgen
+
+echo "START TIME: $(date)"
+
+# defining the right environment variables
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export TOKENIZERS_PARALLELISM=false
+
+# Converted transformer checkpoint
+#MODEL_CKPT=/gpfsscratch/rech/six/commun/experiments/muennighoff/bloomckpt/6b3/bloom-7b1
+MODEL_CKPT=/gpfsscratch/rech/six/commun/experiments/muennighoff/bloomckpt/6b3t0/tr13f-6b3-ml-t0-lmtoks341b-t0toks13b-xp3capmixv2lossseq
+
+cd /gpfsscratch/rech/six/commun/experiments/muennighoff/bslmevalgeneration/lm-evaluation-harness
+
+DATASETS_AND_CONFIGS=(
+multi_eurlex_mt,multi,"version-fr-en-source+target"
+multi_eurlex_mt,multi,"version-en-fr-source+target"
+multi_eurlex_mt,multi,"a_good_translation-fr-en-source+target"
+multi_eurlex_mt,multi,"a_good_translation-en-fr-source+target"
+multi_eurlex_mt,multi,"prev_doc-en-fr"
+multi_eurlex_mt,multi,"prev_doc-fr-en"
+)
+
+DATASET_AND_CONFIG=${DATASETS_AND_CONFIGS[$SLURM_ARRAY_TASK_ID]}
+echo $ARGUMENT
+
+IFS=',' read dataset_name lang template_name <<< "${DATASET_AND_CONFIG}"
+
+# Use this fork of lm-eval: https://github.com/bigscience-workshop/lm-evaluation-harness/pull/109
+python main.py \
+    --model_api_name 'hf-causal' \
+    --model_args pretrained=$MODEL_CKPT,use_accelerate=True,tokenizer=$MODEL_CKPT,dtype=float16 \
+    --device cuda \
+    --batch_size 16 \
+    --no_tracking \
+    --task_name $dataset_name \
+    --template_names $template_name \
+    --bootstrap_iters 10 \
+    --num_fewshot 0 \
+    --limit 500
+
+echo "END TIME: $(date)"

bigscience/evaluation/results/tr11/scripts/report-to-csv.py

@@ -0,0 +1,58 @@
+#!/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 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:
+    results = json.load(f)
+
+with io.open(csv_file, 'w', encoding='utf-8') as f:
+
+    writer = csv.writer(f)
+    writer.writerow(["task", "metric", "value", "err", "version"])
+
+    versions = results["versions"]
+
+    for k,v in sorted(results["results"].items()):
+        if k not in versions:
+            versions[k] = -1
+
+        if "acc" in v:
+            writer.writerow([k, "acc", v["acc"], v["acc_stderr"], versions[k]])
+        if "acc_norm" in v:
+            writer.writerow([k, "acc_norm", v["acc_norm"], v["acc_norm_stderr"], versions[k]])
+        if "f1" in v:
+            writer.writerow([k, "f1", v["f1"], v["f1_stderr"] if "f1_stderr" in v else "", versions[k]])
+        # if "ppl" in v:
+        #     writer.writerow([k, "ppl", v["ppl"], v["ppl_stderr"], versions[k]])
+        # if "em" in v:
+        #     writer.writerow([k, "em", v["em"], v["em_stderr"] if "em_stderr" in v else "", versions[k]])

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_generation_176b.slurm

@@ -0,0 +1,128 @@
+#!/bin/bash
+#SBATCH --job-name=genbseval
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --reservation=hug
+#SBATCH --qos=qos_gpu-gc             # up to 100h
+#SBATCH --nodes=1
+#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 100:00:00             # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-tr13f-6B3-ml-t0
+conda activate muennighofflmevalgen
+
+echo "START TIME: $(date)"
+
+# defining the right environment variables
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export TOKENIZERS_PARALLELISM=false
+
+# Converted transformer checkpoint
+MODEL_CKPT=/gpfsscratch/rech/six/commun/uan68tv-model-conversion/bloom
+
+cd /gpfsscratch/rech/six/commun/experiments/muennighoff/bslmevalgeneration/lm-evaluation-harness
+
+
+DATASETS_AND_CONFIGS=(
+GEM/wiki_lingua_ar,ar,"article_summary_ar"
+GEM/wiki_lingua_ar,ar,"write_abstract_ar"
+GEM/wiki_lingua_ar,ar,"summarize_above_ar"
+GEM/wiki_lingua_ar,ar,"rephrase_ar"
+GEM/wiki_lingua_ar,ar,"tldr_ar"
+GEM/wiki_lingua_en,en,"article_summary_en"
+GEM/wiki_lingua_en,en,"write_abstract_en"
+GEM/wiki_lingua_en,en,"summarize_above_en"
+GEM/wiki_lingua_en,en,"rephrase_en"
+GEM/wiki_lingua_en,en,"tldr_en"
+GEM/wiki_lingua_es,es,"article_summary_es"
+GEM/wiki_lingua_es,es,"write_abstract_es"
+GEM/wiki_lingua_es,es,"summarize_above_es"
+GEM/wiki_lingua_es,es,"rephrase_es"
+GEM/wiki_lingua_es,es,"tldr_es"
+GEM/wiki_lingua_fr,fr,"article_summary_fr"
+GEM/wiki_lingua_fr,fr,"write_abstract_fr"
+GEM/wiki_lingua_fr,fr,"summarize_above_fr"
+GEM/wiki_lingua_fr,fr,"rephrase_fr"
+GEM/wiki_lingua_fr,fr,"tldr_fr"
+GEM/wiki_lingua_hi,hi,"article_summary_hi"
+GEM/wiki_lingua_hi,hi,"write_abstract_hi"
+GEM/wiki_lingua_hi,hi,"summarize_above_hi"
+GEM/wiki_lingua_hi,hi,"rephrase_hi"
+GEM/wiki_lingua_hi,hi,"tldr_hi"
+GEM/wiki_lingua_id,id,"article_summary_id"
+GEM/wiki_lingua_id,id,"write_abstract_id"
+GEM/wiki_lingua_id,id,"summarize_above_id"
+GEM/wiki_lingua_id,id,"rephrase_id"
+GEM/wiki_lingua_id,id,"tldr_id"
+GEM/wiki_lingua_pt,pt,"article_summary_pt"
+GEM/wiki_lingua_pt,pt,"write_abstract_pt"
+GEM/wiki_lingua_pt,pt,"summarize_above_pt"
+GEM/wiki_lingua_pt,pt,"rephrase_pt"
+GEM/wiki_lingua_pt,pt,"tldr_pt"
+GEM/wiki_lingua_vi,vi,"article_summary_vi"
+GEM/wiki_lingua_vi,vi,"write_abstract_vi"
+GEM/wiki_lingua_vi,vi,"summarize_above_vi"
+GEM/wiki_lingua_vi,vi,"rephrase_vi"
+GEM/wiki_lingua_vi,vi,"tldr_vi"
+GEM/wiki_lingua_zh,zh,"article_summary_zh"
+GEM/wiki_lingua_zh,zh,"write_abstract_zh"
+GEM/wiki_lingua_zh,zh,"summarize_above_zh"
+GEM/wiki_lingua_zh,zh,"rephrase_zh"
+GEM/wiki_lingua_zh,zh,"tldr_zh"
+)
+
+DATASETS_AND_CONFIGS=(
+wmt14_fr_en,fr-en,"version-en-fr-target"
+wmt14_fr_en,fr-en,"a_good_translation-en-fr-target"
+wmt14_fr_en,fr-en,"a_good_translation-en-fr-source+target"
+wmt14_fr_en,fr-en,"xglm-en-fr-target"
+wmt14_fr_en,fr-en,"gpt3-en-fr"
+wmt14_fr_en,fr-en,"version-fr-en-target"
+wmt14_fr_en,fr-en,"a_good_translation-fr-en-target"
+wmt14_fr_en,fr-en,"a_good_translation-fr-en-source+target"
+wmt14_fr_en,fr-en,"xglm-fr-en-target"
+wmt14_fr_en,fr-en,"gpt3-fr-en"
+)
+
+DATASETS_AND_CONFIGS=(
+GEM/web_nlg_en,en,"PALM_prompt"
+GEM/web_nlg_en,en,"explicit-graph-description-2"
+GEM/web_nlg_en,en,"implicit-graph-description"
+GEM/web_nlg_en,en,"non-explicit-description"
+GEM/web_nlg_en,en,"use-category"
+GEM/web_nlg_ru,ru,"PALM_prompt"
+GEM/web_nlg_ru,ru,"explicit-graph-description-2-Russian"
+GEM/web_nlg_ru,ru,"implicit-graph-description-Russian"
+GEM/web_nlg_ru,ru,"non-explicit-description-Russian"
+GEM/web_nlg_ru,ru,"use-category-Russian"
+)
+
+DATASET_AND_CONFIG=${DATASETS_AND_CONFIGS[$SLURM_ARRAY_TASK_ID]}
+echo $ARGUMENT
+
+IFS=',' read dataset_name lang template_name <<< "${DATASET_AND_CONFIG}"
+
+# Use this fork of lm-eval: https://github.com/bigscience-workshop/lm-evaluation-harness/pull/109
+python main.py \
+    --model_api_name 'hf-causal' \
+    --model_args pretrained=$MODEL_CKPT,use_accelerate=True,tokenizer=$MODEL_CKPT,dtype=bfloat16 \
+    --device cuda \
+    --batch_size 8 \
+    --no_tracking \
+    --task_name $dataset_name \
+    --template_names $template_name \
+    --bootstrap_iters 10
+
+echo "END TIME: $(date)"

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_generation_350m.slurm

@@ -0,0 +1,110 @@
+#!/bin/bash
+#SBATCH --job-name=run_bsevalharness-generation-350m
+#SBATCH --constraint=v100-32g
+#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 --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@v100
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-tr13f-6B3-ml-t0
+conda activate muennighofflmevalgen
+
+echo "START TIME: $(date)"
+
+# defining the right environment variables
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export TOKENIZERS_PARALLELISM=false
+
+# Converted transformer checkpoint
+MODEL_CKPT=/gpfsscratch/rech/six/commun/commun/experiments/muennighoff/bloomckpt/350m/bloom-350m
+
+cd /gpfsscratch/rech/six/commun/experiments/muennighoff/bslmevalgeneration/lm-evaluation-harness
+
+# WMT19 ZH-EN does not work
+DATASETS_AND_CONFIGS=(
+GEM/wiki_lingua_ar,ar,"article_summary_ar"
+GEM/wiki_lingua_ar,ar,"write_abstract_ar"
+GEM/wiki_lingua_ar,ar,"summarize_above_ar"
+GEM/wiki_lingua_ar,ar,"rephrase_ar"
+GEM/wiki_lingua_ar,ar,"tldr_ar"
+GEM/wiki_lingua_en,en,"article_summary_en"
+GEM/wiki_lingua_en,en,"write_abstract_en"
+GEM/wiki_lingua_en,en,"summarize_above_en"
+GEM/wiki_lingua_en,en,"rephrase_en"
+GEM/wiki_lingua_en,en,"tldr_en"
+GEM/wiki_lingua_es,es,"article_summary_es"
+GEM/wiki_lingua_es,es,"write_abstract_es"
+GEM/wiki_lingua_es,es,"summarize_above_es"
+GEM/wiki_lingua_es,es,"rephrase_es"
+GEM/wiki_lingua_es,es,"tldr_es"
+GEM/wiki_lingua_fr,fr,"article_summary_fr"
+GEM/wiki_lingua_fr,fr,"write_abstract_fr"
+GEM/wiki_lingua_fr,fr,"summarize_above_fr"
+GEM/wiki_lingua_fr,fr,"rephrase_fr"
+GEM/wiki_lingua_fr,fr,"tldr_fr"
+GEM/wiki_lingua_hi,hi,"article_summary_hi"
+GEM/wiki_lingua_hi,hi,"write_abstract_hi"
+GEM/wiki_lingua_hi,hi,"summarize_above_hi"
+GEM/wiki_lingua_hi,hi,"rephrase_hi"
+GEM/wiki_lingua_hi,hi,"tldr_hi"
+GEM/wiki_lingua_id,id,"article_summary_id"
+GEM/wiki_lingua_id,id,"write_abstract_id"
+GEM/wiki_lingua_id,id,"summarize_above_id"
+GEM/wiki_lingua_id,id,"rephrase_id"
+GEM/wiki_lingua_id,id,"tldr_id"
+GEM/wiki_lingua_pt,pt,"article_summary_pt"
+GEM/wiki_lingua_pt,pt,"write_abstract_pt"
+GEM/wiki_lingua_pt,pt,"summarize_above_pt"
+GEM/wiki_lingua_pt,pt,"rephrase_pt"
+GEM/wiki_lingua_pt,pt,"tldr_pt"
+GEM/wiki_lingua_vi,vi,"article_summary_vi"
+GEM/wiki_lingua_vi,vi,"write_abstract_vi"
+GEM/wiki_lingua_vi,vi,"summarize_above_vi"
+GEM/wiki_lingua_vi,vi,"rephrase_vi"
+GEM/wiki_lingua_vi,vi,"tldr_vi"
+GEM/wiki_lingua_zh,zh,"article_summary_zh"
+GEM/wiki_lingua_zh,zh,"write_abstract_zh"
+GEM/wiki_lingua_zh,zh,"summarize_above_zh"
+GEM/wiki_lingua_zh,zh,"rephrase_zh"
+GEM/wiki_lingua_zh,zh,"tldr_zh"
+)
+
+#GEM/wiki_lingua_ar,ar,"article_summary_ar"
+#GEM/wiki_lingua_ar,ar,"write_abstract_ar"
+#GEM/wiki_lingua_ar,ar,"summarize_above_ar"
+#GEM/wiki_lingua_ar,ar,"rephrase_ar"
+#GEM/wiki_lingua_ar,ar,"tldr_ar"
+#GEM/wiki_lingua_zh,zh,"article_summary_zh"
+#GEM/wiki_lingua_zh,zh,"write_abstract_zh"
+#GEM/wiki_lingua_zh,zh,"summarize_above_zh"
+#GEM/wiki_lingua_zh,zh,"rephrase_zh"
+#GEM/wiki_lingua_zh,zh,"tldr_zh"
+
+DATASET_AND_CONFIG=${DATASETS_AND_CONFIGS[$SLURM_ARRAY_TASK_ID]}
+echo $ARGUMENT
+
+IFS=',' read dataset_name lang template_name <<< "${DATASET_AND_CONFIG}"
+
+# Use this fork of lm-eval: https://github.com/bigscience-workshop/lm-evaluation-harness/pull/109
+python main.py \
+    --model_api_name 'hf-causal' \
+    --model_args pretrained=$MODEL_CKPT,use_accelerate=True,tokenizer=$MODEL_CKPT,dtype=float16 \
+    --device cuda \
+    --batch_size 16 \
+    --no_tracking \
+    --task_name $dataset_name \
+    --template_names $template_name \
+    --bootstrap_iters 10
+
+echo "END TIME: $(date)"

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_generation_6b3.slurm

@@ -0,0 +1,101 @@
+#!/bin/bash
+#SBATCH --job-name=evaluate_t0
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=8           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:1                 # number of gpus
+#SBATCH --constraint=a100
+#SBATCH --reservation=hug
+#SBATCH --time 20:00:00             # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-tr13f-6B3-ml-t0
+conda activate muennighofflmevalgen
+
+echo "START TIME: $(date)"
+
+# defining the right environment variables
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export TOKENIZERS_PARALLELISM=false
+
+# Converted transformer checkpoint
+MODEL_CKPT=/gpfsscratch/rech/six/commun/experiments/muennighoff/bloomckpt/6b3/bloom-7b1
+
+cd /gpfsscratch/rech/six/commun/experiments/muennighoff/bslmevalgeneration/lm-evaluation-harness
+
+# WMT19 ZH-EN does not work
+DATASETS_AND_CONFIGS=(
+GEM/wiki_lingua_en,en,"article_summary_en"
+GEM/wiki_lingua_en,en,"write_abstract_en"
+GEM/wiki_lingua_en,en,"summarize_above_en"
+GEM/wiki_lingua_en,en,"rephrase_en"
+GEM/wiki_lingua_en,en,"tldr_en"
+GEM/wiki_lingua_es,es,"article_summary_es"
+GEM/wiki_lingua_es,es,"write_abstract_es"
+GEM/wiki_lingua_es,es,"summarize_above_es"
+GEM/wiki_lingua_es,es,"rephrase_es"
+GEM/wiki_lingua_es,es,"tldr_es"
+GEM/wiki_lingua_fr,fr,"article_summary_fr"
+GEM/wiki_lingua_fr,fr,"write_abstract_fr"
+GEM/wiki_lingua_fr,fr,"summarize_above_fr"
+GEM/wiki_lingua_fr,fr,"rephrase_fr"
+GEM/wiki_lingua_fr,fr,"tldr_fr"
+GEM/wiki_lingua_hi,hi,"article_summary_hi"
+GEM/wiki_lingua_hi,hi,"write_abstract_hi"
+GEM/wiki_lingua_hi,hi,"summarize_above_hi"
+GEM/wiki_lingua_hi,hi,"rephrase_hi"
+GEM/wiki_lingua_hi,hi,"tldr_hi"
+GEM/wiki_lingua_id,id,"article_summary_id"
+GEM/wiki_lingua_id,id,"write_abstract_id"
+GEM/wiki_lingua_id,id,"summarize_above_id"
+GEM/wiki_lingua_id,id,"rephrase_id"
+GEM/wiki_lingua_id,id,"tldr_id"
+GEM/wiki_lingua_pt,pt,"article_summary_pt"
+GEM/wiki_lingua_pt,pt,"write_abstract_pt"
+GEM/wiki_lingua_pt,pt,"summarize_above_pt"
+GEM/wiki_lingua_pt,pt,"rephrase_pt"
+GEM/wiki_lingua_pt,pt,"tldr_pt"
+GEM/wiki_lingua_vi,vi,"article_summary_vi"
+GEM/wiki_lingua_vi,vi,"write_abstract_vi"
+GEM/wiki_lingua_vi,vi,"summarize_above_vi"
+GEM/wiki_lingua_vi,vi,"rephrase_vi"
+GEM/wiki_lingua_vi,vi,"tldr_vi"
+)
+
+#GEM/wiki_lingua_ar,ar,"article_summary_ar"
+#GEM/wiki_lingua_ar,ar,"write_abstract_ar"
+#GEM/wiki_lingua_ar,ar,"summarize_above_ar"
+#GEM/wiki_lingua_ar,ar,"rephrase_ar"
+#GEM/wiki_lingua_ar,ar,"tldr_ar"
+#GEM/wiki_lingua_zh,zh,"article_summary_zh"
+#GEM/wiki_lingua_zh,zh,"write_abstract_zh"
+#GEM/wiki_lingua_zh,zh,"summarize_above_zh"
+#GEM/wiki_lingua_zh,zh,"rephrase_zh"
+#GEM/wiki_lingua_zh,zh,"tldr_zh"
+
+DATASET_AND_CONFIG=${DATASETS_AND_CONFIGS[$SLURM_ARRAY_TASK_ID]}
+echo $ARGUMENT
+
+IFS=',' read dataset_name lang template_name <<< "${DATASET_AND_CONFIG}"
+
+# Use this fork of lm-eval: https://github.com/bigscience-workshop/lm-evaluation-harness/pull/109
+python main.py \
+    --model_api_name 'hf-causal' \
+    --model_args pretrained=$MODEL_CKPT,use_accelerate=True,tokenizer=$MODEL_CKPT,dtype=float16 \
+    --device cuda \
+    --batch_size 16 \
+    --no_tracking \
+    --task_name $dataset_name \
+    --template_names $template_name \
+    --bootstrap_iters 10
+
+echo "END TIME: $(date)"

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_generation_760m.slurm

@@ -0,0 +1,110 @@
+#!/bin/bash
+#SBATCH --job-name=run_bsevalharness-generation-760m
+#SBATCH --constraint=v100-32g
+#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 --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@v100
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-tr13f-6B3-ml-t0
+conda activate muennighofflmevalgen
+
+echo "START TIME: $(date)"
+
+# defining the right environment variables
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+export TOKENIZERS_PARALLELISM=false
+
+# Converted transformer checkpoint
+MODEL_CKPT=/gpfsscratch/rech/six/commun/experiments/muennighoff/bloomckpt/760m/bloom-760m
+
+cd /gpfsscratch/rech/six/commun/experiments/muennighoff/bslmevalgeneration/lm-evaluation-harness
+
+# WMT19 ZH-EN does not work
+DATASETS_AND_CONFIGS=(
+GEM/wiki_lingua_ar,ar,"article_summary_ar"
+GEM/wiki_lingua_ar,ar,"write_abstract_ar"
+GEM/wiki_lingua_ar,ar,"summarize_above_ar"
+GEM/wiki_lingua_ar,ar,"rephrase_ar"
+GEM/wiki_lingua_ar,ar,"tldr_ar"
+GEM/wiki_lingua_en,en,"article_summary_en"
+GEM/wiki_lingua_en,en,"write_abstract_en"
+GEM/wiki_lingua_en,en,"summarize_above_en"
+GEM/wiki_lingua_en,en,"rephrase_en"
+GEM/wiki_lingua_en,en,"tldr_en"
+GEM/wiki_lingua_es,es,"article_summary_es"
+GEM/wiki_lingua_es,es,"write_abstract_es"
+GEM/wiki_lingua_es,es,"summarize_above_es"
+GEM/wiki_lingua_es,es,"rephrase_es"
+GEM/wiki_lingua_es,es,"tldr_es"
+GEM/wiki_lingua_fr,fr,"article_summary_fr"
+GEM/wiki_lingua_fr,fr,"write_abstract_fr"
+GEM/wiki_lingua_fr,fr,"summarize_above_fr"
+GEM/wiki_lingua_fr,fr,"rephrase_fr"
+GEM/wiki_lingua_fr,fr,"tldr_fr"
+GEM/wiki_lingua_hi,hi,"article_summary_hi"
+GEM/wiki_lingua_hi,hi,"write_abstract_hi"
+GEM/wiki_lingua_hi,hi,"summarize_above_hi"
+GEM/wiki_lingua_hi,hi,"rephrase_hi"
+GEM/wiki_lingua_hi,hi,"tldr_hi"
+GEM/wiki_lingua_id,id,"article_summary_id"
+GEM/wiki_lingua_id,id,"write_abstract_id"
+GEM/wiki_lingua_id,id,"summarize_above_id"
+GEM/wiki_lingua_id,id,"rephrase_id"
+GEM/wiki_lingua_id,id,"tldr_id"
+GEM/wiki_lingua_pt,pt,"article_summary_pt"
+GEM/wiki_lingua_pt,pt,"write_abstract_pt"
+GEM/wiki_lingua_pt,pt,"summarize_above_pt"
+GEM/wiki_lingua_pt,pt,"rephrase_pt"
+GEM/wiki_lingua_pt,pt,"tldr_pt"
+GEM/wiki_lingua_vi,vi,"article_summary_vi"
+GEM/wiki_lingua_vi,vi,"write_abstract_vi"
+GEM/wiki_lingua_vi,vi,"summarize_above_vi"
+GEM/wiki_lingua_vi,vi,"rephrase_vi"
+GEM/wiki_lingua_vi,vi,"tldr_vi"
+GEM/wiki_lingua_zh,zh,"article_summary_zh"
+GEM/wiki_lingua_zh,zh,"write_abstract_zh"
+GEM/wiki_lingua_zh,zh,"summarize_above_zh"
+GEM/wiki_lingua_zh,zh,"rephrase_zh"
+GEM/wiki_lingua_zh,zh,"tldr_zh"
+)
+
+#GEM/wiki_lingua_ar,ar,"article_summary_ar"
+#GEM/wiki_lingua_ar,ar,"write_abstract_ar"
+#GEM/wiki_lingua_ar,ar,"summarize_above_ar"
+#GEM/wiki_lingua_ar,ar,"rephrase_ar"
+#GEM/wiki_lingua_ar,ar,"tldr_ar"
+#GEM/wiki_lingua_zh,zh,"article_summary_zh"
+#GEM/wiki_lingua_zh,zh,"write_abstract_zh"
+#GEM/wiki_lingua_zh,zh,"summarize_above_zh"
+#GEM/wiki_lingua_zh,zh,"rephrase_zh"
+#GEM/wiki_lingua_zh,zh,"tldr_zh"
+
+DATASET_AND_CONFIG=${DATASETS_AND_CONFIGS[$SLURM_ARRAY_TASK_ID]}
+echo $ARGUMENT
+
+IFS=',' read dataset_name lang template_name <<< "${DATASET_AND_CONFIG}"
+
+# Use this fork of lm-eval: https://github.com/bigscience-workshop/lm-evaluation-harness/pull/109
+python main.py \
+    --model_api_name 'hf-causal' \
+    --model_args pretrained=$MODEL_CKPT,use_accelerate=True,tokenizer=$MODEL_CKPT,dtype=float16 \
+    --device cuda \
+    --batch_size 16 \
+    --no_tracking \
+    --task_name $dataset_name \
+    --template_names $template_name \
+    --bootstrap_iters 10
+
+echo "END TIME: $(date)"

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_tr11-176b-ml.slurm

@@ -0,0 +1,122 @@
+#!/bin/bash
+#SBATCH --job-name=run_bsevalharness-tr11-176b-ml
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --nodes=1
+#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 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+#SBATCH --reservation=hug
+
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-muennighofflmeval
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval ideally correspnding to the modelname
+VARIANT="tr11-176b-ml-bsevalharness"
+
+
+CHECKPOINT_PATH=$six_ALL_CCFRSCRATCH/checkpoints/tr11-176B-ml/checkpoints/main/global_step90000
+MEGATRON_DEEPSPEED_REPO=$six_ALL_CCFRSCRATCH/commun/experiments/muennighoff/megdsbslmeval/Megatron-DeepSpeed
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+cd $MEGATRON_DEEPSPEED_REPO
+
+TOKENIZER_NAME_OR_PATH=bigscience-catalogue-data-dev/byte-level-bpe-tokenizer-no-norm-250k-whitespace-and-eos-regex-alpha-v3-dedup-lines-articles
+
+PP_SIZE=8
+TP_SIZE=1
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+EVAL_MICRO_BATCH_SIZE=1
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1 \
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+
+# Deepspeed figures out GAS dynamically from dynamic GBS via set_train_batch_size()
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "gradient_clipping": 1.0,
+  "zero_optimization": {
+    "stage": $ZERO_STAGE
+  },
+  "bf16": {
+    "enabled": true
+  },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+
+CMD="./tasks/eval_harness/evaluate_bsevalharness.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path $TOKENIZER_NAME_OR_PATH \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --bf16 \
+    --inference \
+    --seq-length $SEQ_LEN \
+    --task_list wnli \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --intermed_results \
+    --adaptive_seq_len \
+    --micro_bs_multiplier 16 \
+    --offloadearly \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+GPUS_PER_NODE=8
+NNODES=$SLURM_NNODES
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+export LAUNCHER="python -u -m torch.distributed.run \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    --rdzv_backend c10d \
+    --max_restarts 0 \
+    --tee 3 \
+    "
+
+export CUDA_LAUNCH_BLOCKING=1
+
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_tr11b-1b3-ml.slurm

@@ -0,0 +1,122 @@
+#!/bin/bash
+#SBATCH --job-name=run_bsevalharness-tr11b-1b3-ml
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=8           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+#SBATCH --reservation=hug
+
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-muennighofflmeval
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval ideally correspnding to the modelname
+VARIANT="tr11b-1b3-ml-bsevalharness"
+
+
+CHECKPOINT_PATH=$six_ALL_CCFRSCRATCH/checkpoints/tr11b-1B3-ml/checkpoints/main/global_step340500
+MEGATRON_DEEPSPEED_REPO=$six_ALL_CCFRSCRATCH/commun/experiments/muennighoff/megdsbslmeval/Megatron-DeepSpeed
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$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 $MEGATRON_DEEPSPEED_REPO
+
+TOKENIZER_NAME_OR_PATH=bigscience-catalogue-data-dev/byte-level-bpe-tokenizer-no-norm-250k-whitespace-and-eos-regex-alpha-v3-dedup-lines-articles
+
+PP_SIZE=1
+TP_SIZE=1
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+EVAL_MICRO_BATCH_SIZE=1
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1 \
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+
+# Deepspeed figures out GAS dynamically from dynamic GBS via set_train_batch_size()
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "gradient_clipping": 1.0,
+  "zero_optimization": {
+    "stage": $ZERO_STAGE
+  },
+  "bf16": {
+    "enabled": false
+  },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+
+CMD="./tasks/eval_harness/evaluate_bsevalharness.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path $TOKENIZER_NAME_OR_PATH \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --inference \
+    --seq-length $SEQ_LEN \
+    --task_list axb,axg,boolq,cb,cola,copa,crows_pairs_english,crows_pairs_french,diabla,e2e_nlg_cleaned,mnli,mnli_mismatched,multirc,piaf,qqp,rte,sst,tydiqa_primary,tydiqa_secondary,wic,wsc,wnli,wino_bias_type1_anti,wino_bias_type1_pro,wino_bias_type2_anti,wino_bias_type2_pro,xquad_ar,xquad_en,gsarti/flores_101_afr,gsarti/flores_101_amh,gsarti/flores_101_ara,gsarti/flores_101_hye,gsarti/flores_101_asm,gsarti/flores_101_ast,gsarti/flores_101_azj,gsarti/flores_101_bel,gsarti/flores_101_ben,gsarti/flores_101_bos,gsarti/flores_101_bul,gsarti/flores_101_mya,gsarti/flores_101_cat,gsarti/flores_101_ceb,gsarti/flores_101_zho_simpl,gsarti/flores_101_zho_trad,gsarti/flores_101_hrv,gsarti/flores_101_ces,gsarti/flores_101_dan,gsarti/flores_101_nld,gsarti/flores_101_eng,gsarti/flores_101_est,gsarti/flores_101_tgl,gsarti/flores_101_fin,gsarti/flores_101_fra,gsarti/flores_101_ful,gsarti/flores_101_glg,gsarti/flores_101_lug,gsarti/flores_101_kat,gsarti/flores_101_deu,gsarti/flores_101_ell,gsarti/flores_101_guj,gsarti/flores_101_hau,gsarti/flores_101_heb,gsarti/flores_101_hin,gsarti/flores_101_hun,gsarti/flores_101_isl,gsarti/flores_101_ibo,gsarti/flores_101_ind,gsarti/flores_101_gle,gsarti/flores_101_ita,gsarti/flores_101_jpn,gsarti/flores_101_jav,gsarti/flores_101_kea,gsarti/flores_101_kam,gsarti/flores_101_kan,gsarti/flores_101_kaz,gsarti/flores_101_khm,gsarti/flores_101_kor,gsarti/flores_101_kir,gsarti/flores_101_lao,gsarti/flores_101_lav,gsarti/flores_101_lin,gsarti/flores_101_lit,gsarti/flores_101_luo,gsarti/flores_101_ltz,gsarti/flores_101_mkd,gsarti/flores_101_msa,gsarti/flores_101_mal,gsarti/flores_101_mlt,gsarti/flores_101_mri,gsarti/flores_101_mar,gsarti/flores_101_mon,gsarti/flores_101_npi,gsarti/flores_101_nso,gsarti/flores_101_nob,gsarti/flores_101_nya,gsarti/flores_101_oci,gsarti/flores_101_ory,gsarti/flores_101_orm,gsarti/flores_101_pus,gsarti/flores_101_fas,gsarti/flores_101_pol,gsarti/flores_101_por,gsarti/flores_101_pan,gsarti/flores_101_ron,gsarti/flores_101_rus,gsarti/flores_101_srp,gsarti/flores_101_sna,gsarti/flores_101_snd,gsarti/flores_101_slk,gsarti/flores_101_slv,gsarti/flores_101_som,gsarti/flores_101_ckb,gsarti/flores_101_spa,gsarti/flores_101_swh,gsarti/flores_101_swe,gsarti/flores_101_tgk,gsarti/flores_101_tam,gsarti/flores_101_tel,gsarti/flores_101_tha,gsarti/flores_101_tur,gsarti/flores_101_ukr,gsarti/flores_101_umb,gsarti/flores_101_urd,gsarti/flores_101_uzb,gsarti/flores_101_vie,gsarti/flores_101_cym,gsarti/flores_101_wol,gsarti/flores_101_xho,gsarti/flores_101_yor,gsarti/flores_101_zul \
+    --eval_fp32 \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --intermed_results \
+    --adaptive_seq_len \
+    --micro_bs_multiplier 8 \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+GPUS_PER_NODE=1
+NNODES=$SLURM_NNODES
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+export LAUNCHER="python -u -m torch.distributed.run \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    --rdzv_backend c10d \
+    --max_restarts 0 \
+    --tee 3 \
+    "
+
+export CUDA_LAUNCH_BLOCKING=1
+
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_tr11d-750m-ml.slurm

@@ -0,0 +1,120 @@
+#!/bin/bash
+#SBATCH --job-name=run_bsevalharness-tr11d-760m-ml
+#SBATCH --constraint=v100-32g
+#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 --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@v100
+
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-muennighofflmeval
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval ideally correspnding to the modelname
+VARIANT="tr11d-760m-ml-bsevalharness"
+
+
+CHECKPOINT_PATH=$six_ALL_CCFRSCRATCH/checkpoints/tr11d-760M-ml/checkpoints/main/global_step660750
+MEGATRON_DEEPSPEED_REPO=$six_ALL_CCFRSCRATCH/commun/experiments/muennighoff/bslmeval/Megatron-DeepSpeed
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+export TOKENIZERS_PARALLELISM=false
+
+cd $MEGATRON_DEEPSPEED_REPO
+
+TOKENIZER_NAME_OR_PATH=bigscience-catalogue-data-dev/byte-level-bpe-tokenizer-no-norm-250k-whitespace-and-eos-regex-alpha-v3-dedup-lines-articles
+
+PP_SIZE=1
+TP_SIZE=1
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+EVAL_MICRO_BATCH_SIZE=1
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1 \
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+
+# Deepspeed figures out GAS dynamically from dynamic GBS via set_train_batch_size()
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "gradient_clipping": 1.0,
+  "zero_optimization": {
+    "stage": $ZERO_STAGE
+  },
+  "bf16": {
+    "enabled": false
+  },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+
+CMD="./tasks/eval_harness/evaluate_bsevalharness.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path $TOKENIZER_NAME_OR_PATH \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --inference \
+    --seq-length $SEQ_LEN \
+    --task_list axb,axg,boolq,cb,cola,copa,crows_pairs_english,crows_pairs_french,diabla,e2e_nlg_cleaned,mnli,mnli_mismatched,multirc,piaf,qqp,rte,sst,tydiqa_primary,tydiqa_secondary,wic,wsc,wnli,wino_bias_type1_anti,wino_bias_type1_pro,wino_bias_type2_anti,wino_bias_type2_pro,xquad_ar,xquad_en,gsarti/flores_101_afr,gsarti/flores_101_amh,gsarti/flores_101_ara,gsarti/flores_101_hye,gsarti/flores_101_asm,gsarti/flores_101_ast,gsarti/flores_101_azj,gsarti/flores_101_bel,gsarti/flores_101_ben,gsarti/flores_101_bos,gsarti/flores_101_bul,gsarti/flores_101_mya,gsarti/flores_101_cat,gsarti/flores_101_ceb,gsarti/flores_101_zho_simpl,gsarti/flores_101_zho_trad,gsarti/flores_101_hrv,gsarti/flores_101_ces,gsarti/flores_101_dan,gsarti/flores_101_nld,gsarti/flores_101_eng,gsarti/flores_101_est,gsarti/flores_101_tgl,gsarti/flores_101_fin,gsarti/flores_101_fra,gsarti/flores_101_ful,gsarti/flores_101_glg,gsarti/flores_101_lug,gsarti/flores_101_kat,gsarti/flores_101_deu,gsarti/flores_101_ell,gsarti/flores_101_guj,gsarti/flores_101_hau,gsarti/flores_101_heb,gsarti/flores_101_hin,gsarti/flores_101_hun,gsarti/flores_101_isl,gsarti/flores_101_ibo,gsarti/flores_101_ind,gsarti/flores_101_gle,gsarti/flores_101_ita,gsarti/flores_101_jpn,gsarti/flores_101_jav,gsarti/flores_101_kea,gsarti/flores_101_kam,gsarti/flores_101_kan,gsarti/flores_101_kaz,gsarti/flores_101_khm,gsarti/flores_101_kor,gsarti/flores_101_kir,gsarti/flores_101_lao,gsarti/flores_101_lav,gsarti/flores_101_lin,gsarti/flores_101_lit,gsarti/flores_101_luo,gsarti/flores_101_ltz,gsarti/flores_101_mkd,gsarti/flores_101_msa,gsarti/flores_101_mal,gsarti/flores_101_mlt,gsarti/flores_101_mri,gsarti/flores_101_mar,gsarti/flores_101_mon,gsarti/flores_101_npi,gsarti/flores_101_nso,gsarti/flores_101_nob,gsarti/flores_101_nya,gsarti/flores_101_oci,gsarti/flores_101_ory,gsarti/flores_101_orm,gsarti/flores_101_pus,gsarti/flores_101_fas,gsarti/flores_101_pol,gsarti/flores_101_por,gsarti/flores_101_pan,gsarti/flores_101_ron,gsarti/flores_101_rus,gsarti/flores_101_srp,gsarti/flores_101_sna,gsarti/flores_101_snd,gsarti/flores_101_slk,gsarti/flores_101_slv,gsarti/flores_101_som,gsarti/flores_101_ckb,gsarti/flores_101_spa,gsarti/flores_101_swh,gsarti/flores_101_swe,gsarti/flores_101_tgk,gsarti/flores_101_tam,gsarti/flores_101_tel,gsarti/flores_101_tha,gsarti/flores_101_tur,gsarti/flores_101_ukr,gsarti/flores_101_umb,gsarti/flores_101_urd,gsarti/flores_101_uzb,gsarti/flores_101_vie,gsarti/flores_101_cym,gsarti/flores_101_wol,gsarti/flores_101_xho,gsarti/flores_101_yor,gsarti/flores_101_zul \
+    --eval_fp32 \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --intermed_results \
+    --adaptive_seq_len \
+    --micro_bs_multiplier 4 \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+GPUS_PER_NODE=1
+NNODES=$SLURM_NNODES
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6002
+export LAUNCHER="python -u -m torch.distributed.run \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    --rdzv_backend c10d \
+    --max_restarts 0 \
+    --tee 3 \
+    "
+
+export CUDA_LAUNCH_BLOCKING=1
+
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log

bigscience/evaluation/results/tr11/scripts/run_bsevalharness_tr11f-6b3-ml.slurm

@@ -0,0 +1,121 @@
+#!/bin/bash
+#SBATCH --job-name=run_bsevalharness-tr11f-6b3-ml
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=8           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+#SBATCH --reservation=hug
+
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-muennighofflmeval
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval ideally correspnding to the modelname
+VARIANT="tr11f-6b3-ml-bsevalharness"
+
+
+CHECKPOINT_PATH=$six_ALL_CCFRSCRATCH/checkpoints/tr11f-6B3-ml/checkpoints/main/global_step337500
+MEGATRON_DEEPSPEED_REPO=$six_ALL_CCFRSCRATCH/commun/experiments/muennighoff/bslmeval/Megatron-DeepSpeed
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$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 $MEGATRON_DEEPSPEED_REPO
+
+TOKENIZER_NAME_OR_PATH=bigscience-catalogue-data-dev/byte-level-bpe-tokenizer-no-norm-250k-whitespace-and-eos-regex-alpha-v3-dedup-lines-articles
+
+PP_SIZE=1
+TP_SIZE=1
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+EVAL_MICRO_BATCH_SIZE=1
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1 \
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+
+# Deepspeed figures out GAS dynamically from dynamic GBS via set_train_batch_size()
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "gradient_clipping": 1.0,
+  "zero_optimization": {
+    "stage": $ZERO_STAGE
+  },
+  "bf16": {
+    "enabled": false
+  },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+CMD="./tasks/eval_harness/evaluate_bsevalharness.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path $TOKENIZER_NAME_OR_PATH \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --inference \
+    --seq-length $SEQ_LEN \
+    --task_list axb,axg,boolq,cb,cola,copa,crows_pairs_english,crows_pairs_french,diabla,e2e_nlg_cleaned,mnli,mnli_mismatched,multirc,piaf,qqp,rte,sst,tydiqa_primary,tydiqa_secondary,wic,wsc,wnli,wino_bias_type1_anti,wino_bias_type1_pro,wino_bias_type2_anti,wino_bias_type2_pro,xquad_ar,xquad_en,gsarti/flores_101_afr,gsarti/flores_101_amh,gsarti/flores_101_ara,gsarti/flores_101_hye,gsarti/flores_101_asm,gsarti/flores_101_ast,gsarti/flores_101_azj,gsarti/flores_101_bel,gsarti/flores_101_ben,gsarti/flores_101_bos,gsarti/flores_101_bul,gsarti/flores_101_mya,gsarti/flores_101_cat,gsarti/flores_101_ceb,gsarti/flores_101_zho_simpl,gsarti/flores_101_zho_trad,gsarti/flores_101_hrv,gsarti/flores_101_ces,gsarti/flores_101_dan,gsarti/flores_101_nld,gsarti/flores_101_eng,gsarti/flores_101_est,gsarti/flores_101_tgl,gsarti/flores_101_fin,gsarti/flores_101_fra,gsarti/flores_101_ful,gsarti/flores_101_glg,gsarti/flores_101_lug,gsarti/flores_101_kat,gsarti/flores_101_deu,gsarti/flores_101_ell,gsarti/flores_101_guj,gsarti/flores_101_hau,gsarti/flores_101_heb,gsarti/flores_101_hin,gsarti/flores_101_hun,gsarti/flores_101_isl,gsarti/flores_101_ibo,gsarti/flores_101_ind,gsarti/flores_101_gle,gsarti/flores_101_ita,gsarti/flores_101_jpn,gsarti/flores_101_jav,gsarti/flores_101_kea,gsarti/flores_101_kam,gsarti/flores_101_kan,gsarti/flores_101_kaz,gsarti/flores_101_khm,gsarti/flores_101_kor,gsarti/flores_101_kir,gsarti/flores_101_lao,gsarti/flores_101_lav,gsarti/flores_101_lin,gsarti/flores_101_lit,gsarti/flores_101_luo,gsarti/flores_101_ltz,gsarti/flores_101_mkd,gsarti/flores_101_msa,gsarti/flores_101_mal,gsarti/flores_101_mlt,gsarti/flores_101_mri,gsarti/flores_101_mar,gsarti/flores_101_mon,gsarti/flores_101_npi,gsarti/flores_101_nso,gsarti/flores_101_nob,gsarti/flores_101_nya,gsarti/flores_101_oci,gsarti/flores_101_ory,gsarti/flores_101_orm,gsarti/flores_101_pus,gsarti/flores_101_fas,gsarti/flores_101_pol,gsarti/flores_101_por,gsarti/flores_101_pan,gsarti/flores_101_ron,gsarti/flores_101_rus,gsarti/flores_101_srp,gsarti/flores_101_sna,gsarti/flores_101_snd,gsarti/flores_101_slk,gsarti/flores_101_slv,gsarti/flores_101_som,gsarti/flores_101_ckb,gsarti/flores_101_spa,gsarti/flores_101_swh,gsarti/flores_101_swe,gsarti/flores_101_tgk,gsarti/flores_101_tam,gsarti/flores_101_tel,gsarti/flores_101_tha,gsarti/flores_101_tur,gsarti/flores_101_ukr,gsarti/flores_101_umb,gsarti/flores_101_urd,gsarti/flores_101_uzb,gsarti/flores_101_vie,gsarti/flores_101_cym,gsarti/flores_101_wol,gsarti/flores_101_xho,gsarti/flores_101_yor,gsarti/flores_101_zul \
+    --eval_fp32 \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --intermed_results \
+    --adaptive_seq_len \
+    --micro_bs_multiplier 8 \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+GPUS_PER_NODE=1
+NNODES=$SLURM_NNODES
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+export LAUNCHER="python -u -m torch.distributed.run \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    --rdzv_backend c10d \
+    --max_restarts 0 \
+    --tee 3 \
+    "
+
+export CUDA_LAUNCH_BLOCKING=1
+
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log

bigscience/evaluation/results/tr11/scripts/run_evalharness_deepspeed.md

@@ -0,0 +1,158 @@
+# How to run lm-eval on Megatron-DeepSpeed checkpoint using the original setup
+
+This particular setup uses the normal deepspeed checkpoint and requires no conversion to Megatron-LM.
+
+This doc assumes usage on JZ, so some peculiar requirements in places. Ignore these if you're not running this on JZ.
+
+## Prerequisites
+
+1. Install software
+
+On login console with external network
+
+Get lm-eval harness (https://github.com/EleutherAI/lm-evaluation-harness) and `best-download==0.0.7` needed to download some tasks.
+```
+start-prod
+pip install best-download==0.0.7
+pip install git+https://github.com/EleutherAI/lm-evaluation-harness
+```
+
+2. Pre-download needed datasets
+
+some symlinks due to lm-harness' issues with relative position of data
+```
+mkdir data
+ln -s `pwd`/data tasks/eval_harness/data
+```
+Also make sure `data` is not on one of the limited paritions like WORKSF.
+
+Then install datasets for the tasks:
+```
+python ./tasks/eval_harness/download.py --task_list
+arc_challenge,arc_easy,boolq,copa,hellaswag,lambada,logiqa,mathqa,mc_taco,mrpc,multirc,openbookqa,piqa,prost,pubmedqa,qnli,qqp,race,rte,sciq,sst,triviaqa,webqs,wic,winogrande,wnli,wsc
+```
+and make sure that `export HF_DATASETS_OFFLINE=1`
+
+If there are things like custom tokenizers, pre-download those too, e.g.:
+
+```
+python -c "from transformers import AutoTokenizer; AutoTokenizer.from_pretrained('bigscience/oscar_13_languages_alpha_weight')"
+```
+and make sure that `export TRANSFORMERS_OFFLINE=1` is in the script.
+You know there is a custom tokenizer if the training script had something like:
+
+```
+--tokenizer-type PretrainedFromHF \
+ --tokenizer-name-or-path bigscience/oscar_13_languages_alpha_weight \
+```
+
+3. Prepare the slurm script
+
+Prepare the run script, replace `variant` with a unique identifier for the current eval so that multiple evals could run in parallel and not all log into the same `results.json` file. so, e.g., `tr9c-1B3-swiglu`
+
+```
+cp examples/run_evalharness_deepspeed.slurm run_evalharness-variant.slurm
+```
+
+now edit `run_evalharness-variant.slurm`
+
+
+Note that the eval code knows to pull the original training args from the checkpoint, so we don't need to pass any of those. And we just need to setup the evaluation args.
+
+Note that for the bigscience lm-eval-harness fork (https://github.com/bigscience-workshop/lm-evaluation-harness), the corresponding scripts are `evaluate_bsevalharness.py` & `run_bsevalharness_tr11-176b-ml.slurm`.
+
+1. Edit:
+
+```
+PP_SIZE=1
+TP_SIZE=1
+```
+to match the eval topology. If the model fits into 1 gpu, then there is nothing to change.
+
+The eval script will automatically reshape the model if it was of a different topology.
+
+
+2. Adjust the following to fit the chosen GPU. As of last check for 1.3B model the settings are one of:
+```
+EVAL_MICRO_BATCH_SIZE=6  # 16GB GPU 1.3B model
+EVAL_MICRO_BATCH_SIZE=12 # 32GB GPU 1.3B model
+```
+
+If you get OOM lower it further.
+
+3. If not using the Deepspeed path, disable it by removing:
+
+```
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+```
+
+If you didn't disable it and the program crashed on checkpoint loading unable to find some key, disable deepspeed as explained above.
+
+4. Additional flags
+
+- To reduce the amount of iterations for stderr estimation, use e.g. `--bootstrap_iters 2`. This saves 1-2 minutes per dataset.
+- To print intermediate results when running multiple tasks use `--intermed_results`.
+- To reduce the bubble when setting PP use the flag `--micro_bs_multiplier`. Reducing `--micro-batch-size` may be needed when increasing the multiplier. 
+    - Running the 176B model with PP=8, `--micro_bs_multiplier 8` & `--micro-batch-size 4` produced the fastest results for PiQA on 1 node in 2min18s.
+
+## Eval
+
+Currently it takes 2-3 hours to run on 32GB for 1.3B model, 6-7h for 16GB GPU, so a 20h slurm job should be enough.
+
+When ready, launch:
+```
+sbatch ./run_evalharness-variant.slurm
+```
+
+To monitor progress:
+```
+tail -f tail -f $VARIANT-eval-harness.log
+```
+where the variant is what you set `$VARIANT` to in the slurm script.
+
+The template is set up for 16GB gpu since they are easier to get by. If you change to 32GB, adjust:
+```
+#SBATCH --constraint=v100-32g
+...
+EVAL_MICRO_BATCH_SIZE=12 # 32GB GPU 1.3B model
+```
+
+
+Note that the original ETA at the start of the run can be 10x too longer than the actual outcome. For example it may suggest 18 hours but will complete in 2 hours.
+
+
+## Short eval
+
+if you just want to quickly test that everything can run to the end, edit `tasks/eval_harness/evaluate.py`,  e.g. to run only 10 batches:
+```
+- results = evaluator.evaluate(adaptor, task_dict, False, 0, None)
++ results = evaluator.evaluate(adaptor, task_dict, False, 0, 10)
+```
+
+(XXX: could be a cmd line option so that code won't need to be modified)
+
+
+## Import into spreadsheet
+
+https://docs.google.com/spreadsheets/d/1CI8Q9RCblLRzUOPJ6ViqBmo284-8ojluQ-CmaEuhuv0/edit?usp=sharing
+
+Note that the spreadsheet format is quite different, so use this script:
+```
+./tasks/eval_harness/report-to-csv.py results.json
+```
+to reformat the json results into csv while changing its shape to match the spreadsheet format
+
+Since some records might be missing or extraneous here is the best way to do it:
+
+1. copy the data from first 2 columns to some place under the main spreadsheet
+
+2. put the pointer to the 3rd column next to where the 2 first columns were copied.
+
+3. import `results.csv` using file-> import -> file ->
+
+Import location: Replace data at selected cell
+
+4. Now it should be easy to align the new records with the old ones - delete irrelevant records and Insert->Cells where data is missing until the first 2 columns match
+
+5. now create 2 cols in the main table on top and now it should be safe to Copy-n-Paste the 2-col data range, without the task/metrics columns into the newly created space.

bigscience/evaluation/results/tr11/scripts/run_evalharness_deepspeed.slurm

@@ -0,0 +1,98 @@
+#!/bin/bash
+#SBATCH --job-name=eval-harness-deepspeed
+#SBATCH --constraint=v100-16g
+#SBATCH --nodes=1
+#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:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@gpu
+
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-prod
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval so that multiple evals could run in parallel and not all log into the same "results.json" file.
+VARIANT="tr9c-1B3-swiglu"
+
+CHECKPOINT_PATH=/gpfsdsstore/projects/rech/six/commun/checkpoints/tr3m-1B3-emb-norm-pile/global_step296023
+MEGATRON_DEEPSPEED_REPO=/gpfsssd/worksf/projects/rech/six/commun/code/eval/Megatron-DeepSpeed
+
+# you want these 2 on JZ, and pre-download/cache any datasets/tokenizers/models
+# but comment these out if you're running on a node with Internet access
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+cd $MEGATRON_DEEPSPEED_REPO
+
+# eval topology
+PP_SIZE=1
+TP_SIZE=1
+
+VOCAB_FILE=$MEGATRON_DEEPSPEED_REPO/data/gpt2-vocab.json
+MERGE_FILE=$MEGATRON_DEEPSPEED_REPO/data/gpt2-merges.txt
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+
+EVAL_MICRO_BATCH_SIZE=6  # 16GB GPU 1.3B model
+#EVAL_MICRO_BATCH_SIZE=12 # 32GB GPU 1.3B model
+
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "zero_optimization": { "stage": $ZERO_STAGE },
+  "fp16": { "enabled": true },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+CMD="./tasks/eval_harness/evaluate.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --vocab-file $VOCAB_FILE \
+    --merge-file $MERGE_FILE \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --inference \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --seq-length $SEQ_LEN \
+    --adaptive_seq_len \
+    --eval_fp32 \
+    --task_list arc_challenge,arc_easy,boolq,copa,hellaswag,lambada,logiqa,mathqa,mc_taco,mrpc,multirc,openbookqa,piqa,prost,pubmedqa,qnli,qqp,race,rte,sst,webqs,wic,winogrande,wnli,wsc,triviaqa,sciq \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+N_GPUS=1
+LAUNCHER="deepspeed --num_gpus $N_GPUS"
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log

bigscience/evaluation/results/tr11/scripts/run_evalharness_tr11b-1b3-ml.slurm

@@ -0,0 +1,120 @@
+#!/bin/bash
+#SBATCH --job-name=run_evalharness-tr11b-2b5-ml
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=8           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+#SBATCH --reservation=hug
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-py38-pt111
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval ideally correspnding to the modelname
+VARIANT="tr11b-1b3-ml-evalharness"
+
+
+CHECKPOINT_PATH=$six_ALL_CCFRSCRATCH/checkpoints/tr11b-1B3-ml/checkpoints/main/global_step340500
+MEGATRON_DEEPSPEED_REPO=$six_ALL_CCFRSCRATCH/commun/experiments/muennighoff/megdsbslmeval/Megatron-DeepSpeed
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+cd $MEGATRON_DEEPSPEED_REPO
+
+TOKENIZER_NAME_OR_PATH=bigscience-catalogue-data-dev/byte-level-bpe-tokenizer-no-norm-250k-whitespace-and-eos-regex-alpha-v3-dedup-lines-articles
+
+PP_SIZE=1
+TP_SIZE=1
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+EVAL_MICRO_BATCH_SIZE=1
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1 \
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+
+# Deepspeed figures out GAS dynamically from dynamic GBS via set_train_batch_size()
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "gradient_clipping": 1.0,
+  "zero_optimization": {
+    "stage": $ZERO_STAGE
+  },
+  "bf16": {
+    "enabled": false
+  },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+
+CMD="./tasks/eval_harness/evaluate.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path $TOKENIZER_NAME_OR_PATH \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --eval_fp32 \
+    --inference \
+    --seq-length $SEQ_LEN \
+    --task_list arc_challenge,arc_easy,boolq,copa,headqa,hellaswag,lambada,logiqa,mathqa,mc_taco,mrpc,multirc,openbookqa,piqa,prost,pubmedqa,qnli,qqp,race,rte,sciq,sst,triviaqa,webqs,wic,winogrande,wnli,wsc \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --intermed_results \
+    --adaptive_seq_len \
+    --micro_bs_multiplier 8 \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+GPUS_PER_NODE=1
+NNODES=$SLURM_NNODES
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+export LAUNCHER="python -u -m torch.distributed.run \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    --rdzv_backend c10d \
+    --max_restarts 0 \
+    --tee 3 \
+    "
+
+export CUDA_LAUNCH_BLOCKING=1
+
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log

bigscience/evaluation/results/tr11/scripts/run_evalharness_tr11c-2b5-ml.slurm

@@ -0,0 +1,120 @@
+#!/bin/bash
+#SBATCH --job-name=run_evalharness-tr11b-2b5-ml
+#SBATCH --partition=gpu_p5
+#SBATCH --constraint=a100
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1          # crucial - only 1 task per dist per node!
+#SBATCH --cpus-per-task=8           # number of cores per tasks
+#SBATCH --hint=nomultithread         # we get physical cores not logical
+#SBATCH --gres=gpu:1                 # number of gpus
+#SBATCH --time 20:00:00              # maximum execution time (HH:MM:SS)
+#SBATCH --output=%x-%j.out           # output file name
+#SBATCH --account=six@a100
+#SBATCH --reservation=hug
+
+set -x -e
+
+source $six_ALL_CCFRWORK/start-py38-pt111
+
+echo "START TIME: $(date)"
+
+# a unique identifier for the current eval ideally correspnding to the modelname
+VARIANT="tr11b-2b5-ml-evalharness"
+
+
+CHECKPOINT_PATH=$six_ALL_CCFRSCRATCH/checkpoints/tr11c-2B5-ml/checkpoints/main/global_step337250
+MEGATRON_DEEPSPEED_REPO=$six_ALL_CCFRSCRATCH/commun/experiments/muennighoff/megdsbslmeval/Megatron-DeepSpeed
+export HF_DATASETS_OFFLINE=1
+export TRANSFORMERS_OFFLINE=1
+
+export TRANSFORMERS_CACHE=$six_ALL_CCFRWORK/models
+export HF_DATASETS_CACHE=$six_ALL_CCFRWORK/datasets
+export HF_MODULES_CACHE=$six_ALL_CCFRWORK/modules
+export HF_METRICS_CACHE=$six_ALL_CCFRWORK/metrics
+
+cd $MEGATRON_DEEPSPEED_REPO
+
+TOKENIZER_NAME_OR_PATH=bigscience-catalogue-data-dev/byte-level-bpe-tokenizer-no-norm-250k-whitespace-and-eos-regex-alpha-v3-dedup-lines-articles
+
+PP_SIZE=1
+TP_SIZE=1
+SEQ_LEN=2048
+
+# different from the training MICRO_BATCH_SIZE - no optim memory, so can do bigger BS
+# make as big as it can fit into gpu w/o OOM, but not too close to 100%
+EVAL_MICRO_BATCH_SIZE=1
+
+#dummy arguments to make megatron happy.
+MEGATRON_REQUIRED_ARGS=" \
+    --num-layers -1 \
+    --hidden-size -1 \
+    --num-attention-heads -1 \
+    --seq-length -1  \
+    --max-position-embeddings -1 \
+"
+
+
+ZERO_STAGE=0
+
+config_json="./ds_config.json"
+
+# Deepspeed figures out GAS dynamically from dynamic GBS via set_train_batch_size()
+cat <<EOT > $config_json
+{
+  "train_micro_batch_size_per_gpu": 1,
+  "train_batch_size": 1,
+  "gradient_clipping": 1.0,
+  "zero_optimization": {
+    "stage": $ZERO_STAGE
+  },
+  "bf16": {
+    "enabled": false
+  },
+  "steps_per_print": 2000,
+  "wall_clock_breakdown": false
+}
+EOT
+
+
+CMD="./tasks/eval_harness/evaluate.py  \
+    --load $CHECKPOINT_PATH \
+    --results_path $VARIANT-results.json \
+    --tensor-model-parallel-size $TP_SIZE  \
+    --pipeline-model-parallel-size $PP_SIZE \
+    --tokenizer-type PretrainedFromHF \
+    --tokenizer-name-or-path $TOKENIZER_NAME_OR_PATH \
+    --micro-batch-size $EVAL_MICRO_BATCH_SIZE \
+    --no-load-optim \
+    --no-load-rng \
+    --eval_fp32 \
+    --inference \
+    --seq-length $SEQ_LEN \
+    --task_list arc_challenge,arc_easy,boolq,copa,headqa,hellaswag,lambada,logiqa,mathqa,mc_taco,mrpc,multirc,openbookqa,piqa,prost,pubmedqa,qnli,qqp,race,rte,sciq,sst,triviaqa,webqs,wic,winogrande,wnli,wsc \
+    --deepspeed \
+    --deepspeed_config ds_config.json \
+    --intermed_results \
+    --adaptive_seq_len \
+    --micro_bs_multiplier 8 \
+    $MEGATRON_REQUIRED_ARGS \
+    "
+
+GPUS_PER_NODE=1
+NNODES=$SLURM_NNODES
+MASTER_ADDR=$(scontrol show hostnames $SLURM_JOB_NODELIST | head -n 1)
+MASTER_PORT=6000
+export LAUNCHER="python -u -m torch.distributed.run \
+    --nproc_per_node $GPUS_PER_NODE \
+    --nnodes $NNODES \
+    --rdzv_endpoint $MASTER_ADDR:$MASTER_PORT \
+    --rdzv_backend c10d \
+    --max_restarts 0 \
+    --tee 3 \
+    "
+
+export CUDA_LAUNCH_BLOCKING=1
+
+echo $LAUNCHER $CMD
+
+export PYTHONPATH=$MEGATRON_DEEPSPEED_REPO
+
+$LAUNCHER $CMD 2>&1 | tee $VARIANT-eval-harness.log