Hugging Face's logo

LACAI
/
roberta-large-dialog-narrative

Fill-Mask
Transformers
PyTorch
TensorBoard
roberta
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
1
AbrahamSanders commited on
Commit
5f8a770
·
1 Parent(s): 5fd5118

First model version

Browse files
Files changed (16) hide show
  1. README.md +72 -0
  2. added_tokens.json +1 -0
  3. all_results.json +14 -0
  4. config.json +27 -0
  5. eval_results.json +9 -0
  6. merges.txt +0 -0
  7. pytorch_model.bin +3 -0
  8. runs/Oct30_22-32-03_panacea/1635647539.4848197/events.out.tfevents.1635647539.panacea.11977.1 +3 -0
  9. runs/Oct30_22-32-03_panacea/events.out.tfevents.1635647539.panacea.11977.0 +3 -0
  10. runs/Oct30_22-32-03_panacea/events.out.tfevents.1636346097.panacea.11977.2 +3 -0
  11. special_tokens_map.json +1 -0
  12. tokenizer_config.json +1 -0
  13. train_results.json +8 -0
  14. trainer_state.json +2941 -0
  15. training_args.bin +3 -0
  16. vocab.json +0 -0
README.md ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ tags:
4
+ - generated_from_trainer
5
+ model-index:
6
+ - name: output_mlm
7
+ results: []
8
+ ---
9
+
10
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
11
+ should probably proofread and complete it, then remove this comment. -->
12
+
13
+ # output_mlm
14
+
15
+ This model is a fine-tuned version of [roberta-large](https://huggingface.co/roberta-large) on an unknown dataset.
16
+ It achieves the following results on the evaluation set:
17
+ - Loss: 1.2024
18
+
19
+ ## Model description
20
+
21
+ More information needed
22
+
23
+ ## Intended uses & limitations
24
+
25
+ More information needed
26
+
27
+ ## Training and evaluation data
28
+
29
+ More information needed
30
+
31
+ ## Training procedure
32
+
33
+ ### Training hyperparameters
34
+
35
+ The following hyperparameters were used during training:
36
+ - learning_rate: 5e-05
37
+ - train_batch_size: 16
38
+ - eval_batch_size: 8
39
+ - seed: 42
40
+ - gradient_accumulation_steps: 2
41
+ - total_train_batch_size: 32
42
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
43
+ - lr_scheduler_type: linear
44
+ - num_epochs: 3.0
45
+
46
+ ### Training results
47
+
48
+ | Training Loss | Epoch | Step | Validation Loss |
49
+ |:-------------:|:-----:|:------:|:---------------:|
50
+ | 1.5832 | 0.19 | 15000 | 1.4992 |
51
+ | 1.5325 | 0.39 | 30000 | 1.4653 |
52
+ | 1.4979 | 0.58 | 45000 | 1.4359 |
53
+ | 1.4715 | 0.77 | 60000 | 1.4039 |
54
+ | 1.4448 | 0.97 | 75000 | 1.3877 |
55
+ | 1.4191 | 1.16 | 90000 | 1.3603 |
56
+ | 1.3988 | 1.35 | 105000 | 1.3425 |
57
+ | 1.3699 | 1.54 | 120000 | 1.3230 |
58
+ | 1.3493 | 1.74 | 135000 | 1.3012 |
59
+ | 1.3201 | 1.93 | 150000 | 1.2773 |
60
+ | 1.2993 | 2.12 | 165000 | 1.2617 |
61
+ | 1.2745 | 2.32 | 180000 | 1.2490 |
62
+ | 1.2614 | 2.51 | 195000 | 1.2283 |
63
+ | 1.2424 | 2.7 | 210000 | 1.2152 |
64
+ | 1.2296 | 2.9 | 225000 | 1.2052 |
65
+
66
+
67
+ ### Framework versions
68
+
69
+ - Transformers 4.11.2
70
+ - Pytorch 1.9.0
71
+ - Datasets 1.12.1
72
+ - Tokenizers 0.10.3
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<n>": 50265, "<d>": 50266}
all_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_loss": 1.202375054359436,
4
+ "eval_runtime": 3381.3728,
5
+ "eval_samples": 132344,
6
+ "eval_samples_per_second": 39.139,
7
+ "eval_steps_per_second": 4.892,
8
+ "perplexity": 3.3280117525215283,
9
+ "train_loss": 1.3875968830903724,
10
+ "train_runtime": 695161.7821,
11
+ "train_samples": 2486745,
12
+ "train_samples_per_second": 10.732,
13
+ "train_steps_per_second": 0.335
14
+ }
config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "roberta-large",
3
+ "architectures": [
4
+ "RobertaForMaskedLM"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 4096,
15
+ "layer_norm_eps": 1e-05,
16
+ "max_position_embeddings": 514,
17
+ "model_type": "roberta",
18
+ "num_attention_heads": 16,
19
+ "num_hidden_layers": 24,
20
+ "pad_token_id": 1,
21
+ "position_embedding_type": "absolute",
22
+ "torch_dtype": "float32",
23
+ "transformers_version": "4.11.2",
24
+ "type_vocab_size": 1,
25
+ "use_cache": true,
26
+ "vocab_size": 50267
27
+ }
eval_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_loss": 1.202375054359436,
4
+ "eval_runtime": 3381.3728,
5
+ "eval_samples": 132344,
6
+ "eval_samples_per_second": 39.139,
7
+ "eval_steps_per_second": 4.892,
8
+ "perplexity": 3.3280117525215283
9
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b78822e93ee9a2a77cb006a599cb4de40d5e194dc82251aedf91c289a7ebd770
3
+ size 1421793835
runs/Oct30_22-32-03_panacea/1635647539.4848197/events.out.tfevents.1635647539.panacea.11977.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:548b0111a02b42cfc3f2fd76714f8a8ce77103fd7ab8fc1a0e4446f5394230cf
3
+ size 4513
runs/Oct30_22-32-03_panacea/events.out.tfevents.1635647539.panacea.11977.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68c07db1ae40fa45b4b24785697f67b10fee681de961453bec16f65722617263
3
+ size 82116
runs/Oct30_22-32-03_panacea/events.out.tfevents.1636346097.panacea.11977.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:162eeae418a86d910e1369dbedf990be8e61365eed07a9471dbf37d0f5a9d1a5
3
+ size 316
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "eos_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "unk_token": {"content": "<unk>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "sep_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "pad_token": {"content": "<pad>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "cls_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true}, "additional_special_tokens": ["<n>", "<d>"]}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"errors": "replace", "unk_token": {"content": "<unk>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "bos_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "eos_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "add_prefix_space": false, "sep_token": {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "cls_token": {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "pad_token": {"content": "<pad>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "use_fast": true, "model_max_length": 512, "special_tokens_map_file": null, "tokenizer_file": "/home/deyd/.cache/huggingface/transformers/e16a2590deb9e6d73711d6e05bf27d832fa8c1162d807222e043ca650a556964.fc9576039592f026ad76a1c231b89aee8668488c671dfbe6616bab2ed298d730", "name_or_path": "roberta-large", "tokenizer_class": "RobertaTokenizer"}
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "train_loss": 1.3875968830903724,
4
+ "train_runtime": 695161.7821,
5
+ "train_samples": 2486745,
6
+ "train_samples_per_second": 10.732,
7
+ "train_steps_per_second": 0.335
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,2941 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.2051972150802612,
3
+ "best_model_checkpoint": "gator/output_mlm/checkpoint-225000",
4
+ "epoch": 3.0,
5
+ "global_step": 233133,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 4.9892765074013544e-05,
13
+ "loss": 1.6961,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 4.978553014802709e-05,
19
+ "loss": 1.6895,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.02,
24
+ "learning_rate": 4.9678295222040635e-05,
25
+ "loss": 1.6703,
26
+ "step": 1500
27
+ },
28
+ {
29
+ "epoch": 0.03,
30
+ "learning_rate": 4.9571060296054184e-05,
31
+ "loss": 1.6708,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 0.03,
36
+ "learning_rate": 4.946382537006773e-05,
37
+ "loss": 1.6626,
38
+ "step": 2500
39
+ },
40
+ {
41
+ "epoch": 0.04,
42
+ "learning_rate": 4.935659044408128e-05,
43
+ "loss": 1.659,
44
+ "step": 3000
45
+ },
46
+ {
47
+ "epoch": 0.05,
48
+ "learning_rate": 4.9249355518094824e-05,
49
+ "loss": 1.652,
50
+ "step": 3500
51
+ },
52
+ {
53
+ "epoch": 0.05,
54
+ "learning_rate": 4.914212059210837e-05,
55
+ "loss": 1.6472,
56
+ "step": 4000
57
+ },
58
+ {
59
+ "epoch": 0.06,
60
+ "learning_rate": 4.9034885666121915e-05,
61
+ "loss": 1.6317,
62
+ "step": 4500
63
+ },
64
+ {
65
+ "epoch": 0.06,
66
+ "learning_rate": 4.8927650740135464e-05,
67
+ "loss": 1.6339,
68
+ "step": 5000
69
+ },
70
+ {
71
+ "epoch": 0.07,
72
+ "learning_rate": 4.8820415814149006e-05,
73
+ "loss": 1.6211,
74
+ "step": 5500
75
+ },
76
+ {
77
+ "epoch": 0.08,
78
+ "learning_rate": 4.8713180888162555e-05,
79
+ "loss": 1.632,
80
+ "step": 6000
81
+ },
82
+ {
83
+ "epoch": 0.08,
84
+ "learning_rate": 4.86059459621761e-05,
85
+ "loss": 1.624,
86
+ "step": 6500
87
+ },
88
+ {
89
+ "epoch": 0.09,
90
+ "learning_rate": 4.8498711036189646e-05,
91
+ "loss": 1.6209,
92
+ "step": 7000
93
+ },
94
+ {
95
+ "epoch": 0.1,
96
+ "learning_rate": 4.8391476110203195e-05,
97
+ "loss": 1.6198,
98
+ "step": 7500
99
+ },
100
+ {
101
+ "epoch": 0.1,
102
+ "learning_rate": 4.828424118421674e-05,
103
+ "loss": 1.6184,
104
+ "step": 8000
105
+ },
106
+ {
107
+ "epoch": 0.11,
108
+ "learning_rate": 4.8177006258230286e-05,
109
+ "loss": 1.6141,
110
+ "step": 8500
111
+ },
112
+ {
113
+ "epoch": 0.12,
114
+ "learning_rate": 4.806977133224383e-05,
115
+ "loss": 1.6055,
116
+ "step": 9000
117
+ },
118
+ {
119
+ "epoch": 0.12,
120
+ "learning_rate": 4.7962536406257376e-05,
121
+ "loss": 1.6018,
122
+ "step": 9500
123
+ },
124
+ {
125
+ "epoch": 0.13,
126
+ "learning_rate": 4.785530148027092e-05,
127
+ "loss": 1.6092,
128
+ "step": 10000
129
+ },
130
+ {
131
+ "epoch": 0.14,
132
+ "learning_rate": 4.774806655428447e-05,
133
+ "loss": 1.5921,
134
+ "step": 10500
135
+ },
136
+ {
137
+ "epoch": 0.14,
138
+ "learning_rate": 4.764083162829801e-05,
139
+ "loss": 1.5968,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 0.15,
144
+ "learning_rate": 4.753359670231156e-05,
145
+ "loss": 1.6,
146
+ "step": 11500
147
+ },
148
+ {
149
+ "epoch": 0.15,
150
+ "learning_rate": 4.742636177632511e-05,
151
+ "loss": 1.5957,
152
+ "step": 12000
153
+ },
154
+ {
155
+ "epoch": 0.16,
156
+ "learning_rate": 4.731912685033865e-05,
157
+ "loss": 1.5944,
158
+ "step": 12500
159
+ },
160
+ {
161
+ "epoch": 0.17,
162
+ "learning_rate": 4.72118919243522e-05,
163
+ "loss": 1.5849,
164
+ "step": 13000
165
+ },
166
+ {
167
+ "epoch": 0.17,
168
+ "learning_rate": 4.710465699836574e-05,
169
+ "loss": 1.5823,
170
+ "step": 13500
171
+ },
172
+ {
173
+ "epoch": 0.18,
174
+ "learning_rate": 4.699742207237929e-05,
175
+ "loss": 1.5947,
176
+ "step": 14000
177
+ },
178
+ {
179
+ "epoch": 0.19,
180
+ "learning_rate": 4.689018714639283e-05,
181
+ "loss": 1.5874,
182
+ "step": 14500
183
+ },
184
+ {
185
+ "epoch": 0.19,
186
+ "learning_rate": 4.678295222040638e-05,
187
+ "loss": 1.5832,
188
+ "step": 15000
189
+ },
190
+ {
191
+ "epoch": 0.19,
192
+ "eval_loss": 1.4992027282714844,
193
+ "eval_runtime": 3385.2625,
194
+ "eval_samples_per_second": 39.094,
195
+ "eval_steps_per_second": 4.887,
196
+ "step": 15000
197
+ },
198
+ {
199
+ "epoch": 0.2,
200
+ "learning_rate": 4.667571729441992e-05,
201
+ "loss": 1.5678,
202
+ "step": 15500
203
+ },
204
+ {
205
+ "epoch": 0.21,
206
+ "learning_rate": 4.656848236843347e-05,
207
+ "loss": 1.5862,
208
+ "step": 16000
209
+ },
210
+ {
211
+ "epoch": 0.21,
212
+ "learning_rate": 4.646124744244702e-05,
213
+ "loss": 1.5828,
214
+ "step": 16500
215
+ },
216
+ {
217
+ "epoch": 0.22,
218
+ "learning_rate": 4.635401251646057e-05,
219
+ "loss": 1.5777,
220
+ "step": 17000
221
+ },
222
+ {
223
+ "epoch": 0.23,
224
+ "learning_rate": 4.624677759047411e-05,
225
+ "loss": 1.5722,
226
+ "step": 17500
227
+ },
228
+ {
229
+ "epoch": 0.23,
230
+ "learning_rate": 4.613954266448765e-05,
231
+ "loss": 1.5723,
232
+ "step": 18000
233
+ },
234
+ {
235
+ "epoch": 0.24,
236
+ "learning_rate": 4.60323077385012e-05,
237
+ "loss": 1.5827,
238
+ "step": 18500
239
+ },
240
+ {
241
+ "epoch": 0.24,
242
+ "learning_rate": 4.5925072812514744e-05,
243
+ "loss": 1.572,
244
+ "step": 19000
245
+ },
246
+ {
247
+ "epoch": 0.25,
248
+ "learning_rate": 4.581783788652829e-05,
249
+ "loss": 1.565,
250
+ "step": 19500
251
+ },
252
+ {
253
+ "epoch": 0.26,
254
+ "learning_rate": 4.5710602960541835e-05,
255
+ "loss": 1.5628,
256
+ "step": 20000
257
+ },
258
+ {
259
+ "epoch": 0.26,
260
+ "learning_rate": 4.5603368034555384e-05,
261
+ "loss": 1.5598,
262
+ "step": 20500
263
+ },
264
+ {
265
+ "epoch": 0.27,
266
+ "learning_rate": 4.5496133108568926e-05,
267
+ "loss": 1.5571,
268
+ "step": 21000
269
+ },
270
+ {
271
+ "epoch": 0.28,
272
+ "learning_rate": 4.538889818258248e-05,
273
+ "loss": 1.5546,
274
+ "step": 21500
275
+ },
276
+ {
277
+ "epoch": 0.28,
278
+ "learning_rate": 4.5281663256596023e-05,
279
+ "loss": 1.5578,
280
+ "step": 22000
281
+ },
282
+ {
283
+ "epoch": 0.29,
284
+ "learning_rate": 4.517442833060957e-05,
285
+ "loss": 1.554,
286
+ "step": 22500
287
+ },
288
+ {
289
+ "epoch": 0.3,
290
+ "learning_rate": 4.5067193404623114e-05,
291
+ "loss": 1.5488,
292
+ "step": 23000
293
+ },
294
+ {
295
+ "epoch": 0.3,
296
+ "learning_rate": 4.495995847863666e-05,
297
+ "loss": 1.5512,
298
+ "step": 23500
299
+ },
300
+ {
301
+ "epoch": 0.31,
302
+ "learning_rate": 4.4852723552650205e-05,
303
+ "loss": 1.5636,
304
+ "step": 24000
305
+ },
306
+ {
307
+ "epoch": 0.32,
308
+ "learning_rate": 4.4745488626663754e-05,
309
+ "loss": 1.5437,
310
+ "step": 24500
311
+ },
312
+ {
313
+ "epoch": 0.32,
314
+ "learning_rate": 4.4638253700677296e-05,
315
+ "loss": 1.5515,
316
+ "step": 25000
317
+ },
318
+ {
319
+ "epoch": 0.33,
320
+ "learning_rate": 4.453101877469084e-05,
321
+ "loss": 1.5587,
322
+ "step": 25500
323
+ },
324
+ {
325
+ "epoch": 0.33,
326
+ "learning_rate": 4.4423783848704394e-05,
327
+ "loss": 1.5413,
328
+ "step": 26000
329
+ },
330
+ {
331
+ "epoch": 0.34,
332
+ "learning_rate": 4.4316548922717936e-05,
333
+ "loss": 1.5381,
334
+ "step": 26500
335
+ },
336
+ {
337
+ "epoch": 0.35,
338
+ "learning_rate": 4.4209313996731485e-05,
339
+ "loss": 1.5483,
340
+ "step": 27000
341
+ },
342
+ {
343
+ "epoch": 0.35,
344
+ "learning_rate": 4.410207907074503e-05,
345
+ "loss": 1.5436,
346
+ "step": 27500
347
+ },
348
+ {
349
+ "epoch": 0.36,
350
+ "learning_rate": 4.3994844144758576e-05,
351
+ "loss": 1.5457,
352
+ "step": 28000
353
+ },
354
+ {
355
+ "epoch": 0.37,
356
+ "learning_rate": 4.388760921877212e-05,
357
+ "loss": 1.5367,
358
+ "step": 28500
359
+ },
360
+ {
361
+ "epoch": 0.37,
362
+ "learning_rate": 4.378037429278567e-05,
363
+ "loss": 1.5404,
364
+ "step": 29000
365
+ },
366
+ {
367
+ "epoch": 0.38,
368
+ "learning_rate": 4.367313936679921e-05,
369
+ "loss": 1.5414,
370
+ "step": 29500
371
+ },
372
+ {
373
+ "epoch": 0.39,
374
+ "learning_rate": 4.356590444081276e-05,
375
+ "loss": 1.5325,
376
+ "step": 30000
377
+ },
378
+ {
379
+ "epoch": 0.39,
380
+ "eval_loss": 1.4652618169784546,
381
+ "eval_runtime": 3382.9946,
382
+ "eval_samples_per_second": 39.12,
383
+ "eval_steps_per_second": 4.89,
384
+ "step": 30000
385
+ },
386
+ {
387
+ "epoch": 0.39,
388
+ "learning_rate": 4.3458669514826307e-05,
389
+ "loss": 1.5386,
390
+ "step": 30500
391
+ },
392
+ {
393
+ "epoch": 0.4,
394
+ "learning_rate": 4.335143458883985e-05,
395
+ "loss": 1.5513,
396
+ "step": 31000
397
+ },
398
+ {
399
+ "epoch": 0.41,
400
+ "learning_rate": 4.32441996628534e-05,
401
+ "loss": 1.5384,
402
+ "step": 31500
403
+ },
404
+ {
405
+ "epoch": 0.41,
406
+ "learning_rate": 4.313696473686694e-05,
407
+ "loss": 1.5381,
408
+ "step": 32000
409
+ },
410
+ {
411
+ "epoch": 0.42,
412
+ "learning_rate": 4.302972981088049e-05,
413
+ "loss": 1.5429,
414
+ "step": 32500
415
+ },
416
+ {
417
+ "epoch": 0.42,
418
+ "learning_rate": 4.292249488489403e-05,
419
+ "loss": 1.5325,
420
+ "step": 33000
421
+ },
422
+ {
423
+ "epoch": 0.43,
424
+ "learning_rate": 4.281525995890758e-05,
425
+ "loss": 1.5388,
426
+ "step": 33500
427
+ },
428
+ {
429
+ "epoch": 0.44,
430
+ "learning_rate": 4.270802503292112e-05,
431
+ "loss": 1.5353,
432
+ "step": 34000
433
+ },
434
+ {
435
+ "epoch": 0.44,
436
+ "learning_rate": 4.260079010693467e-05,
437
+ "loss": 1.5311,
438
+ "step": 34500
439
+ },
440
+ {
441
+ "epoch": 0.45,
442
+ "learning_rate": 4.249355518094821e-05,
443
+ "loss": 1.5356,
444
+ "step": 35000
445
+ },
446
+ {
447
+ "epoch": 0.46,
448
+ "learning_rate": 4.238632025496176e-05,
449
+ "loss": 1.5347,
450
+ "step": 35500
451
+ },
452
+ {
453
+ "epoch": 0.46,
454
+ "learning_rate": 4.227908532897531e-05,
455
+ "loss": 1.5198,
456
+ "step": 36000
457
+ },
458
+ {
459
+ "epoch": 0.47,
460
+ "learning_rate": 4.217185040298885e-05,
461
+ "loss": 1.5196,
462
+ "step": 36500
463
+ },
464
+ {
465
+ "epoch": 0.48,
466
+ "learning_rate": 4.20646154770024e-05,
467
+ "loss": 1.5173,
468
+ "step": 37000
469
+ },
470
+ {
471
+ "epoch": 0.48,
472
+ "learning_rate": 4.195738055101594e-05,
473
+ "loss": 1.52,
474
+ "step": 37500
475
+ },
476
+ {
477
+ "epoch": 0.49,
478
+ "learning_rate": 4.185014562502949e-05,
479
+ "loss": 1.5155,
480
+ "step": 38000
481
+ },
482
+ {
483
+ "epoch": 0.5,
484
+ "learning_rate": 4.1742910699043034e-05,
485
+ "loss": 1.512,
486
+ "step": 38500
487
+ },
488
+ {
489
+ "epoch": 0.5,
490
+ "learning_rate": 4.163567577305658e-05,
491
+ "loss": 1.5146,
492
+ "step": 39000
493
+ },
494
+ {
495
+ "epoch": 0.51,
496
+ "learning_rate": 4.1528440847070125e-05,
497
+ "loss": 1.5138,
498
+ "step": 39500
499
+ },
500
+ {
501
+ "epoch": 0.51,
502
+ "learning_rate": 4.142120592108368e-05,
503
+ "loss": 1.5101,
504
+ "step": 40000
505
+ },
506
+ {
507
+ "epoch": 0.52,
508
+ "learning_rate": 4.131397099509722e-05,
509
+ "loss": 1.4977,
510
+ "step": 40500
511
+ },
512
+ {
513
+ "epoch": 0.53,
514
+ "learning_rate": 4.120673606911077e-05,
515
+ "loss": 1.5153,
516
+ "step": 41000
517
+ },
518
+ {
519
+ "epoch": 0.53,
520
+ "learning_rate": 4.1099501143124314e-05,
521
+ "loss": 1.5053,
522
+ "step": 41500
523
+ },
524
+ {
525
+ "epoch": 0.54,
526
+ "learning_rate": 4.0992266217137856e-05,
527
+ "loss": 1.5088,
528
+ "step": 42000
529
+ },
530
+ {
531
+ "epoch": 0.55,
532
+ "learning_rate": 4.0885031291151405e-05,
533
+ "loss": 1.5111,
534
+ "step": 42500
535
+ },
536
+ {
537
+ "epoch": 0.55,
538
+ "learning_rate": 4.077779636516495e-05,
539
+ "loss": 1.5069,
540
+ "step": 43000
541
+ },
542
+ {
543
+ "epoch": 0.56,
544
+ "learning_rate": 4.0670561439178496e-05,
545
+ "loss": 1.5085,
546
+ "step": 43500
547
+ },
548
+ {
549
+ "epoch": 0.57,
550
+ "learning_rate": 4.056332651319204e-05,
551
+ "loss": 1.5064,
552
+ "step": 44000
553
+ },
554
+ {
555
+ "epoch": 0.57,
556
+ "learning_rate": 4.0456091587205586e-05,
557
+ "loss": 1.5004,
558
+ "step": 44500
559
+ },
560
+ {
561
+ "epoch": 0.58,
562
+ "learning_rate": 4.0348856661219135e-05,
563
+ "loss": 1.4979,
564
+ "step": 45000
565
+ },
566
+ {
567
+ "epoch": 0.58,
568
+ "eval_loss": 1.4358571767807007,
569
+ "eval_runtime": 3383.7351,
570
+ "eval_samples_per_second": 39.112,
571
+ "eval_steps_per_second": 4.889,
572
+ "step": 45000
573
+ },
574
+ {
575
+ "epoch": 0.59,
576
+ "learning_rate": 4.0241621735232684e-05,
577
+ "loss": 1.5037,
578
+ "step": 45500
579
+ },
580
+ {
581
+ "epoch": 0.59,
582
+ "learning_rate": 4.0134386809246226e-05,
583
+ "loss": 1.4942,
584
+ "step": 46000
585
+ },
586
+ {
587
+ "epoch": 0.6,
588
+ "learning_rate": 4.0027151883259775e-05,
589
+ "loss": 1.4907,
590
+ "step": 46500
591
+ },
592
+ {
593
+ "epoch": 0.6,
594
+ "learning_rate": 3.991991695727332e-05,
595
+ "loss": 1.4985,
596
+ "step": 47000
597
+ },
598
+ {
599
+ "epoch": 0.61,
600
+ "learning_rate": 3.9812682031286866e-05,
601
+ "loss": 1.4996,
602
+ "step": 47500
603
+ },
604
+ {
605
+ "epoch": 0.62,
606
+ "learning_rate": 3.970544710530041e-05,
607
+ "loss": 1.4959,
608
+ "step": 48000
609
+ },
610
+ {
611
+ "epoch": 0.62,
612
+ "learning_rate": 3.959821217931396e-05,
613
+ "loss": 1.4876,
614
+ "step": 48500
615
+ },
616
+ {
617
+ "epoch": 0.63,
618
+ "learning_rate": 3.94909772533275e-05,
619
+ "loss": 1.5026,
620
+ "step": 49000
621
+ },
622
+ {
623
+ "epoch": 0.64,
624
+ "learning_rate": 3.938374232734105e-05,
625
+ "loss": 1.4985,
626
+ "step": 49500
627
+ },
628
+ {
629
+ "epoch": 0.64,
630
+ "learning_rate": 3.92765074013546e-05,
631
+ "loss": 1.4923,
632
+ "step": 50000
633
+ },
634
+ {
635
+ "epoch": 0.65,
636
+ "learning_rate": 3.916927247536814e-05,
637
+ "loss": 1.5005,
638
+ "step": 50500
639
+ },
640
+ {
641
+ "epoch": 0.66,
642
+ "learning_rate": 3.906203754938169e-05,
643
+ "loss": 1.5018,
644
+ "step": 51000
645
+ },
646
+ {
647
+ "epoch": 0.66,
648
+ "learning_rate": 3.895480262339523e-05,
649
+ "loss": 1.4829,
650
+ "step": 51500
651
+ },
652
+ {
653
+ "epoch": 0.67,
654
+ "learning_rate": 3.884756769740878e-05,
655
+ "loss": 1.4858,
656
+ "step": 52000
657
+ },
658
+ {
659
+ "epoch": 0.68,
660
+ "learning_rate": 3.874033277142232e-05,
661
+ "loss": 1.497,
662
+ "step": 52500
663
+ },
664
+ {
665
+ "epoch": 0.68,
666
+ "learning_rate": 3.863309784543587e-05,
667
+ "loss": 1.4849,
668
+ "step": 53000
669
+ },
670
+ {
671
+ "epoch": 0.69,
672
+ "learning_rate": 3.852586291944941e-05,
673
+ "loss": 1.4851,
674
+ "step": 53500
675
+ },
676
+ {
677
+ "epoch": 0.69,
678
+ "learning_rate": 3.841862799346296e-05,
679
+ "loss": 1.4893,
680
+ "step": 54000
681
+ },
682
+ {
683
+ "epoch": 0.7,
684
+ "learning_rate": 3.831139306747651e-05,
685
+ "loss": 1.4887,
686
+ "step": 54500
687
+ },
688
+ {
689
+ "epoch": 0.71,
690
+ "learning_rate": 3.820415814149005e-05,
691
+ "loss": 1.4805,
692
+ "step": 55000
693
+ },
694
+ {
695
+ "epoch": 0.71,
696
+ "learning_rate": 3.80969232155036e-05,
697
+ "loss": 1.4815,
698
+ "step": 55500
699
+ },
700
+ {
701
+ "epoch": 0.72,
702
+ "learning_rate": 3.798968828951714e-05,
703
+ "loss": 1.477,
704
+ "step": 56000
705
+ },
706
+ {
707
+ "epoch": 0.73,
708
+ "learning_rate": 3.788245336353069e-05,
709
+ "loss": 1.4791,
710
+ "step": 56500
711
+ },
712
+ {
713
+ "epoch": 0.73,
714
+ "learning_rate": 3.777521843754423e-05,
715
+ "loss": 1.4915,
716
+ "step": 57000
717
+ },
718
+ {
719
+ "epoch": 0.74,
720
+ "learning_rate": 3.766798351155778e-05,
721
+ "loss": 1.4935,
722
+ "step": 57500
723
+ },
724
+ {
725
+ "epoch": 0.75,
726
+ "learning_rate": 3.7560748585571324e-05,
727
+ "loss": 1.4711,
728
+ "step": 58000
729
+ },
730
+ {
731
+ "epoch": 0.75,
732
+ "learning_rate": 3.745351365958487e-05,
733
+ "loss": 1.4789,
734
+ "step": 58500
735
+ },
736
+ {
737
+ "epoch": 0.76,
738
+ "learning_rate": 3.734627873359842e-05,
739
+ "loss": 1.471,
740
+ "step": 59000
741
+ },
742
+ {
743
+ "epoch": 0.77,
744
+ "learning_rate": 3.7239043807611964e-05,
745
+ "loss": 1.4733,
746
+ "step": 59500
747
+ },
748
+ {
749
+ "epoch": 0.77,
750
+ "learning_rate": 3.713180888162551e-05,
751
+ "loss": 1.4715,
752
+ "step": 60000
753
+ },
754
+ {
755
+ "epoch": 0.77,
756
+ "eval_loss": 1.4038593769073486,
757
+ "eval_runtime": 3383.4345,
758
+ "eval_samples_per_second": 39.115,
759
+ "eval_steps_per_second": 4.889,
760
+ "step": 60000
761
+ },
762
+ {
763
+ "epoch": 0.78,
764
+ "learning_rate": 3.7024573955639055e-05,
765
+ "loss": 1.4746,
766
+ "step": 60500
767
+ },
768
+ {
769
+ "epoch": 0.78,
770
+ "learning_rate": 3.6917339029652604e-05,
771
+ "loss": 1.4745,
772
+ "step": 61000
773
+ },
774
+ {
775
+ "epoch": 0.79,
776
+ "learning_rate": 3.6810104103666146e-05,
777
+ "loss": 1.4699,
778
+ "step": 61500
779
+ },
780
+ {
781
+ "epoch": 0.8,
782
+ "learning_rate": 3.6702869177679695e-05,
783
+ "loss": 1.4672,
784
+ "step": 62000
785
+ },
786
+ {
787
+ "epoch": 0.8,
788
+ "learning_rate": 3.659563425169324e-05,
789
+ "loss": 1.4654,
790
+ "step": 62500
791
+ },
792
+ {
793
+ "epoch": 0.81,
794
+ "learning_rate": 3.6488399325706786e-05,
795
+ "loss": 1.4628,
796
+ "step": 63000
797
+ },
798
+ {
799
+ "epoch": 0.82,
800
+ "learning_rate": 3.6381164399720335e-05,
801
+ "loss": 1.4698,
802
+ "step": 63500
803
+ },
804
+ {
805
+ "epoch": 0.82,
806
+ "learning_rate": 3.6273929473733884e-05,
807
+ "loss": 1.4667,
808
+ "step": 64000
809
+ },
810
+ {
811
+ "epoch": 0.83,
812
+ "learning_rate": 3.6166694547747426e-05,
813
+ "loss": 1.4591,
814
+ "step": 64500
815
+ },
816
+ {
817
+ "epoch": 0.84,
818
+ "learning_rate": 3.6059459621760974e-05,
819
+ "loss": 1.4674,
820
+ "step": 65000
821
+ },
822
+ {
823
+ "epoch": 0.84,
824
+ "learning_rate": 3.5952224695774517e-05,
825
+ "loss": 1.467,
826
+ "step": 65500
827
+ },
828
+ {
829
+ "epoch": 0.85,
830
+ "learning_rate": 3.5844989769788065e-05,
831
+ "loss": 1.4652,
832
+ "step": 66000
833
+ },
834
+ {
835
+ "epoch": 0.86,
836
+ "learning_rate": 3.573775484380161e-05,
837
+ "loss": 1.4609,
838
+ "step": 66500
839
+ },
840
+ {
841
+ "epoch": 0.86,
842
+ "learning_rate": 3.563051991781515e-05,
843
+ "loss": 1.4657,
844
+ "step": 67000
845
+ },
846
+ {
847
+ "epoch": 0.87,
848
+ "learning_rate": 3.55232849918287e-05,
849
+ "loss": 1.4614,
850
+ "step": 67500
851
+ },
852
+ {
853
+ "epoch": 0.88,
854
+ "learning_rate": 3.541605006584225e-05,
855
+ "loss": 1.4527,
856
+ "step": 68000
857
+ },
858
+ {
859
+ "epoch": 0.88,
860
+ "learning_rate": 3.5308815139855796e-05,
861
+ "loss": 1.452,
862
+ "step": 68500
863
+ },
864
+ {
865
+ "epoch": 0.89,
866
+ "learning_rate": 3.520158021386934e-05,
867
+ "loss": 1.4666,
868
+ "step": 69000
869
+ },
870
+ {
871
+ "epoch": 0.89,
872
+ "learning_rate": 3.509434528788289e-05,
873
+ "loss": 1.4544,
874
+ "step": 69500
875
+ },
876
+ {
877
+ "epoch": 0.9,
878
+ "learning_rate": 3.498711036189643e-05,
879
+ "loss": 1.448,
880
+ "step": 70000
881
+ },
882
+ {
883
+ "epoch": 0.91,
884
+ "learning_rate": 3.487987543590998e-05,
885
+ "loss": 1.4554,
886
+ "step": 70500
887
+ },
888
+ {
889
+ "epoch": 0.91,
890
+ "learning_rate": 3.477264050992352e-05,
891
+ "loss": 1.4451,
892
+ "step": 71000
893
+ },
894
+ {
895
+ "epoch": 0.92,
896
+ "learning_rate": 3.466540558393707e-05,
897
+ "loss": 1.4546,
898
+ "step": 71500
899
+ },
900
+ {
901
+ "epoch": 0.93,
902
+ "learning_rate": 3.455817065795061e-05,
903
+ "loss": 1.4573,
904
+ "step": 72000
905
+ },
906
+ {
907
+ "epoch": 0.93,
908
+ "learning_rate": 3.445093573196416e-05,
909
+ "loss": 1.4486,
910
+ "step": 72500
911
+ },
912
+ {
913
+ "epoch": 0.94,
914
+ "learning_rate": 3.434370080597771e-05,
915
+ "loss": 1.4474,
916
+ "step": 73000
917
+ },
918
+ {
919
+ "epoch": 0.95,
920
+ "learning_rate": 3.423646587999125e-05,
921
+ "loss": 1.456,
922
+ "step": 73500
923
+ },
924
+ {
925
+ "epoch": 0.95,
926
+ "learning_rate": 3.41292309540048e-05,
927
+ "loss": 1.4537,
928
+ "step": 74000
929
+ },
930
+ {
931
+ "epoch": 0.96,
932
+ "learning_rate": 3.402199602801834e-05,
933
+ "loss": 1.4532,
934
+ "step": 74500
935
+ },
936
+ {
937
+ "epoch": 0.97,
938
+ "learning_rate": 3.391476110203189e-05,
939
+ "loss": 1.4448,
940
+ "step": 75000
941
+ },
942
+ {
943
+ "epoch": 0.97,
944
+ "eval_loss": 1.3876514434814453,
945
+ "eval_runtime": 3381.619,
946
+ "eval_samples_per_second": 39.136,
947
+ "eval_steps_per_second": 4.892,
948
+ "step": 75000
949
+ },
950
+ {
951
+ "epoch": 0.97,
952
+ "learning_rate": 3.380752617604543e-05,
953
+ "loss": 1.4447,
954
+ "step": 75500
955
+ },
956
+ {
957
+ "epoch": 0.98,
958
+ "learning_rate": 3.370029125005898e-05,
959
+ "loss": 1.4423,
960
+ "step": 76000
961
+ },
962
+ {
963
+ "epoch": 0.98,
964
+ "learning_rate": 3.3593056324072524e-05,
965
+ "loss": 1.4477,
966
+ "step": 76500
967
+ },
968
+ {
969
+ "epoch": 0.99,
970
+ "learning_rate": 3.348582139808607e-05,
971
+ "loss": 1.4408,
972
+ "step": 77000
973
+ },
974
+ {
975
+ "epoch": 1.0,
976
+ "learning_rate": 3.337858647209962e-05,
977
+ "loss": 1.45,
978
+ "step": 77500
979
+ },
980
+ {
981
+ "epoch": 1.0,
982
+ "learning_rate": 3.3271351546113163e-05,
983
+ "loss": 1.4316,
984
+ "step": 78000
985
+ },
986
+ {
987
+ "epoch": 1.01,
988
+ "learning_rate": 3.316411662012671e-05,
989
+ "loss": 1.4364,
990
+ "step": 78500
991
+ },
992
+ {
993
+ "epoch": 1.02,
994
+ "learning_rate": 3.3056881694140254e-05,
995
+ "loss": 1.4296,
996
+ "step": 79000
997
+ },
998
+ {
999
+ "epoch": 1.02,
1000
+ "learning_rate": 3.29496467681538e-05,
1001
+ "loss": 1.426,
1002
+ "step": 79500
1003
+ },
1004
+ {
1005
+ "epoch": 1.03,
1006
+ "learning_rate": 3.2842411842167345e-05,
1007
+ "loss": 1.4301,
1008
+ "step": 80000
1009
+ },
1010
+ {
1011
+ "epoch": 1.04,
1012
+ "learning_rate": 3.2735176916180894e-05,
1013
+ "loss": 1.436,
1014
+ "step": 80500
1015
+ },
1016
+ {
1017
+ "epoch": 1.04,
1018
+ "learning_rate": 3.2627941990194436e-05,
1019
+ "loss": 1.4253,
1020
+ "step": 81000
1021
+ },
1022
+ {
1023
+ "epoch": 1.05,
1024
+ "learning_rate": 3.2520707064207985e-05,
1025
+ "loss": 1.4303,
1026
+ "step": 81500
1027
+ },
1028
+ {
1029
+ "epoch": 1.06,
1030
+ "learning_rate": 3.2413472138221534e-05,
1031
+ "loss": 1.4286,
1032
+ "step": 82000
1033
+ },
1034
+ {
1035
+ "epoch": 1.06,
1036
+ "learning_rate": 3.230623721223508e-05,
1037
+ "loss": 1.4331,
1038
+ "step": 82500
1039
+ },
1040
+ {
1041
+ "epoch": 1.07,
1042
+ "learning_rate": 3.2199002286248625e-05,
1043
+ "loss": 1.4307,
1044
+ "step": 83000
1045
+ },
1046
+ {
1047
+ "epoch": 1.07,
1048
+ "learning_rate": 3.2091767360262174e-05,
1049
+ "loss": 1.419,
1050
+ "step": 83500
1051
+ },
1052
+ {
1053
+ "epoch": 1.08,
1054
+ "learning_rate": 3.1984532434275716e-05,
1055
+ "loss": 1.422,
1056
+ "step": 84000
1057
+ },
1058
+ {
1059
+ "epoch": 1.09,
1060
+ "learning_rate": 3.187729750828926e-05,
1061
+ "loss": 1.4204,
1062
+ "step": 84500
1063
+ },
1064
+ {
1065
+ "epoch": 1.09,
1066
+ "learning_rate": 3.177006258230281e-05,
1067
+ "loss": 1.4301,
1068
+ "step": 85000
1069
+ },
1070
+ {
1071
+ "epoch": 1.1,
1072
+ "learning_rate": 3.166282765631635e-05,
1073
+ "loss": 1.4126,
1074
+ "step": 85500
1075
+ },
1076
+ {
1077
+ "epoch": 1.11,
1078
+ "learning_rate": 3.15555927303299e-05,
1079
+ "loss": 1.4248,
1080
+ "step": 86000
1081
+ },
1082
+ {
1083
+ "epoch": 1.11,
1084
+ "learning_rate": 3.144835780434344e-05,
1085
+ "loss": 1.423,
1086
+ "step": 86500
1087
+ },
1088
+ {
1089
+ "epoch": 1.12,
1090
+ "learning_rate": 3.1341122878356995e-05,
1091
+ "loss": 1.4192,
1092
+ "step": 87000
1093
+ },
1094
+ {
1095
+ "epoch": 1.13,
1096
+ "learning_rate": 3.123388795237054e-05,
1097
+ "loss": 1.418,
1098
+ "step": 87500
1099
+ },
1100
+ {
1101
+ "epoch": 1.13,
1102
+ "learning_rate": 3.1126653026384086e-05,
1103
+ "loss": 1.4194,
1104
+ "step": 88000
1105
+ },
1106
+ {
1107
+ "epoch": 1.14,
1108
+ "learning_rate": 3.101941810039763e-05,
1109
+ "loss": 1.4,
1110
+ "step": 88500
1111
+ },
1112
+ {
1113
+ "epoch": 1.15,
1114
+ "learning_rate": 3.091218317441118e-05,
1115
+ "loss": 1.4023,
1116
+ "step": 89000
1117
+ },
1118
+ {
1119
+ "epoch": 1.15,
1120
+ "learning_rate": 3.080494824842472e-05,
1121
+ "loss": 1.4258,
1122
+ "step": 89500
1123
+ },
1124
+ {
1125
+ "epoch": 1.16,
1126
+ "learning_rate": 3.069771332243827e-05,
1127
+ "loss": 1.4191,
1128
+ "step": 90000
1129
+ },
1130
+ {
1131
+ "epoch": 1.16,
1132
+ "eval_loss": 1.3602887392044067,
1133
+ "eval_runtime": 3380.9917,
1134
+ "eval_samples_per_second": 39.144,
1135
+ "eval_steps_per_second": 4.893,
1136
+ "step": 90000
1137
+ },
1138
+ {
1139
+ "epoch": 1.16,
1140
+ "learning_rate": 3.059047839645181e-05,
1141
+ "loss": 1.413,
1142
+ "step": 90500
1143
+ },
1144
+ {
1145
+ "epoch": 1.17,
1146
+ "learning_rate": 3.0483243470465356e-05,
1147
+ "loss": 1.4107,
1148
+ "step": 91000
1149
+ },
1150
+ {
1151
+ "epoch": 1.18,
1152
+ "learning_rate": 3.0376008544478908e-05,
1153
+ "loss": 1.4151,
1154
+ "step": 91500
1155
+ },
1156
+ {
1157
+ "epoch": 1.18,
1158
+ "learning_rate": 3.0268773618492454e-05,
1159
+ "loss": 1.4119,
1160
+ "step": 92000
1161
+ },
1162
+ {
1163
+ "epoch": 1.19,
1164
+ "learning_rate": 3.0161538692505996e-05,
1165
+ "loss": 1.4049,
1166
+ "step": 92500
1167
+ },
1168
+ {
1169
+ "epoch": 1.2,
1170
+ "learning_rate": 3.005430376651954e-05,
1171
+ "loss": 1.4124,
1172
+ "step": 93000
1173
+ },
1174
+ {
1175
+ "epoch": 1.2,
1176
+ "learning_rate": 2.9947068840533087e-05,
1177
+ "loss": 1.4025,
1178
+ "step": 93500
1179
+ },
1180
+ {
1181
+ "epoch": 1.21,
1182
+ "learning_rate": 2.9839833914546632e-05,
1183
+ "loss": 1.4116,
1184
+ "step": 94000
1185
+ },
1186
+ {
1187
+ "epoch": 1.22,
1188
+ "learning_rate": 2.9732598988560177e-05,
1189
+ "loss": 1.4148,
1190
+ "step": 94500
1191
+ },
1192
+ {
1193
+ "epoch": 1.22,
1194
+ "learning_rate": 2.9625364062573723e-05,
1195
+ "loss": 1.4042,
1196
+ "step": 95000
1197
+ },
1198
+ {
1199
+ "epoch": 1.23,
1200
+ "learning_rate": 2.951812913658727e-05,
1201
+ "loss": 1.4009,
1202
+ "step": 95500
1203
+ },
1204
+ {
1205
+ "epoch": 1.24,
1206
+ "learning_rate": 2.9410894210600814e-05,
1207
+ "loss": 1.3971,
1208
+ "step": 96000
1209
+ },
1210
+ {
1211
+ "epoch": 1.24,
1212
+ "learning_rate": 2.9303659284614366e-05,
1213
+ "loss": 1.4069,
1214
+ "step": 96500
1215
+ },
1216
+ {
1217
+ "epoch": 1.25,
1218
+ "learning_rate": 2.919642435862791e-05,
1219
+ "loss": 1.4172,
1220
+ "step": 97000
1221
+ },
1222
+ {
1223
+ "epoch": 1.25,
1224
+ "learning_rate": 2.9089189432641457e-05,
1225
+ "loss": 1.3947,
1226
+ "step": 97500
1227
+ },
1228
+ {
1229
+ "epoch": 1.26,
1230
+ "learning_rate": 2.8981954506655003e-05,
1231
+ "loss": 1.3998,
1232
+ "step": 98000
1233
+ },
1234
+ {
1235
+ "epoch": 1.27,
1236
+ "learning_rate": 2.8874719580668548e-05,
1237
+ "loss": 1.4033,
1238
+ "step": 98500
1239
+ },
1240
+ {
1241
+ "epoch": 1.27,
1242
+ "learning_rate": 2.8767484654682093e-05,
1243
+ "loss": 1.3905,
1244
+ "step": 99000
1245
+ },
1246
+ {
1247
+ "epoch": 1.28,
1248
+ "learning_rate": 2.866024972869564e-05,
1249
+ "loss": 1.3934,
1250
+ "step": 99500
1251
+ },
1252
+ {
1253
+ "epoch": 1.29,
1254
+ "learning_rate": 2.855301480270918e-05,
1255
+ "loss": 1.3938,
1256
+ "step": 100000
1257
+ },
1258
+ {
1259
+ "epoch": 1.29,
1260
+ "learning_rate": 2.8445779876722727e-05,
1261
+ "loss": 1.4021,
1262
+ "step": 100500
1263
+ },
1264
+ {
1265
+ "epoch": 1.3,
1266
+ "learning_rate": 2.833854495073628e-05,
1267
+ "loss": 1.3977,
1268
+ "step": 101000
1269
+ },
1270
+ {
1271
+ "epoch": 1.31,
1272
+ "learning_rate": 2.8231310024749824e-05,
1273
+ "loss": 1.4034,
1274
+ "step": 101500
1275
+ },
1276
+ {
1277
+ "epoch": 1.31,
1278
+ "learning_rate": 2.812407509876337e-05,
1279
+ "loss": 1.3978,
1280
+ "step": 102000
1281
+ },
1282
+ {
1283
+ "epoch": 1.32,
1284
+ "learning_rate": 2.8016840172776915e-05,
1285
+ "loss": 1.3985,
1286
+ "step": 102500
1287
+ },
1288
+ {
1289
+ "epoch": 1.33,
1290
+ "learning_rate": 2.790960524679046e-05,
1291
+ "loss": 1.3881,
1292
+ "step": 103000
1293
+ },
1294
+ {
1295
+ "epoch": 1.33,
1296
+ "learning_rate": 2.7802370320804006e-05,
1297
+ "loss": 1.3916,
1298
+ "step": 103500
1299
+ },
1300
+ {
1301
+ "epoch": 1.34,
1302
+ "learning_rate": 2.769513539481755e-05,
1303
+ "loss": 1.3981,
1304
+ "step": 104000
1305
+ },
1306
+ {
1307
+ "epoch": 1.34,
1308
+ "learning_rate": 2.7587900468831097e-05,
1309
+ "loss": 1.392,
1310
+ "step": 104500
1311
+ },
1312
+ {
1313
+ "epoch": 1.35,
1314
+ "learning_rate": 2.7480665542844643e-05,
1315
+ "loss": 1.3988,
1316
+ "step": 105000
1317
+ },
1318
+ {
1319
+ "epoch": 1.35,
1320
+ "eval_loss": 1.3425246477127075,
1321
+ "eval_runtime": 3384.4417,
1322
+ "eval_samples_per_second": 39.104,
1323
+ "eval_steps_per_second": 4.888,
1324
+ "step": 105000
1325
+ },
1326
+ {
1327
+ "epoch": 1.36,
1328
+ "learning_rate": 2.737343061685819e-05,
1329
+ "loss": 1.3916,
1330
+ "step": 105500
1331
+ },
1332
+ {
1333
+ "epoch": 1.36,
1334
+ "learning_rate": 2.7266195690871737e-05,
1335
+ "loss": 1.3971,
1336
+ "step": 106000
1337
+ },
1338
+ {
1339
+ "epoch": 1.37,
1340
+ "learning_rate": 2.7158960764885282e-05,
1341
+ "loss": 1.3993,
1342
+ "step": 106500
1343
+ },
1344
+ {
1345
+ "epoch": 1.38,
1346
+ "learning_rate": 2.7051725838898828e-05,
1347
+ "loss": 1.3899,
1348
+ "step": 107000
1349
+ },
1350
+ {
1351
+ "epoch": 1.38,
1352
+ "learning_rate": 2.6944490912912373e-05,
1353
+ "loss": 1.3771,
1354
+ "step": 107500
1355
+ },
1356
+ {
1357
+ "epoch": 1.39,
1358
+ "learning_rate": 2.683725598692592e-05,
1359
+ "loss": 1.3988,
1360
+ "step": 108000
1361
+ },
1362
+ {
1363
+ "epoch": 1.4,
1364
+ "learning_rate": 2.6730021060939464e-05,
1365
+ "loss": 1.3913,
1366
+ "step": 108500
1367
+ },
1368
+ {
1369
+ "epoch": 1.4,
1370
+ "learning_rate": 2.662278613495301e-05,
1371
+ "loss": 1.3929,
1372
+ "step": 109000
1373
+ },
1374
+ {
1375
+ "epoch": 1.41,
1376
+ "learning_rate": 2.6515551208966555e-05,
1377
+ "loss": 1.3879,
1378
+ "step": 109500
1379
+ },
1380
+ {
1381
+ "epoch": 1.42,
1382
+ "learning_rate": 2.64083162829801e-05,
1383
+ "loss": 1.385,
1384
+ "step": 110000
1385
+ },
1386
+ {
1387
+ "epoch": 1.42,
1388
+ "learning_rate": 2.630108135699365e-05,
1389
+ "loss": 1.3798,
1390
+ "step": 110500
1391
+ },
1392
+ {
1393
+ "epoch": 1.43,
1394
+ "learning_rate": 2.6193846431007195e-05,
1395
+ "loss": 1.3844,
1396
+ "step": 111000
1397
+ },
1398
+ {
1399
+ "epoch": 1.43,
1400
+ "learning_rate": 2.608661150502074e-05,
1401
+ "loss": 1.3789,
1402
+ "step": 111500
1403
+ },
1404
+ {
1405
+ "epoch": 1.44,
1406
+ "learning_rate": 2.5979376579034286e-05,
1407
+ "loss": 1.3734,
1408
+ "step": 112000
1409
+ },
1410
+ {
1411
+ "epoch": 1.45,
1412
+ "learning_rate": 2.587214165304783e-05,
1413
+ "loss": 1.3851,
1414
+ "step": 112500
1415
+ },
1416
+ {
1417
+ "epoch": 1.45,
1418
+ "learning_rate": 2.5764906727061377e-05,
1419
+ "loss": 1.3802,
1420
+ "step": 113000
1421
+ },
1422
+ {
1423
+ "epoch": 1.46,
1424
+ "learning_rate": 2.5657671801074922e-05,
1425
+ "loss": 1.3762,
1426
+ "step": 113500
1427
+ },
1428
+ {
1429
+ "epoch": 1.47,
1430
+ "learning_rate": 2.5550436875088468e-05,
1431
+ "loss": 1.3746,
1432
+ "step": 114000
1433
+ },
1434
+ {
1435
+ "epoch": 1.47,
1436
+ "learning_rate": 2.5443201949102013e-05,
1437
+ "loss": 1.38,
1438
+ "step": 114500
1439
+ },
1440
+ {
1441
+ "epoch": 1.48,
1442
+ "learning_rate": 2.5335967023115565e-05,
1443
+ "loss": 1.3842,
1444
+ "step": 115000
1445
+ },
1446
+ {
1447
+ "epoch": 1.49,
1448
+ "learning_rate": 2.522873209712911e-05,
1449
+ "loss": 1.3718,
1450
+ "step": 115500
1451
+ },
1452
+ {
1453
+ "epoch": 1.49,
1454
+ "learning_rate": 2.5121497171142656e-05,
1455
+ "loss": 1.3839,
1456
+ "step": 116000
1457
+ },
1458
+ {
1459
+ "epoch": 1.5,
1460
+ "learning_rate": 2.5014262245156202e-05,
1461
+ "loss": 1.375,
1462
+ "step": 116500
1463
+ },
1464
+ {
1465
+ "epoch": 1.51,
1466
+ "learning_rate": 2.4907027319169744e-05,
1467
+ "loss": 1.3647,
1468
+ "step": 117000
1469
+ },
1470
+ {
1471
+ "epoch": 1.51,
1472
+ "learning_rate": 2.479979239318329e-05,
1473
+ "loss": 1.3787,
1474
+ "step": 117500
1475
+ },
1476
+ {
1477
+ "epoch": 1.52,
1478
+ "learning_rate": 2.4692557467196835e-05,
1479
+ "loss": 1.3692,
1480
+ "step": 118000
1481
+ },
1482
+ {
1483
+ "epoch": 1.52,
1484
+ "learning_rate": 2.4585322541210384e-05,
1485
+ "loss": 1.371,
1486
+ "step": 118500
1487
+ },
1488
+ {
1489
+ "epoch": 1.53,
1490
+ "learning_rate": 2.447808761522393e-05,
1491
+ "loss": 1.3635,
1492
+ "step": 119000
1493
+ },
1494
+ {
1495
+ "epoch": 1.54,
1496
+ "learning_rate": 2.4370852689237475e-05,
1497
+ "loss": 1.368,
1498
+ "step": 119500
1499
+ },
1500
+ {
1501
+ "epoch": 1.54,
1502
+ "learning_rate": 2.426361776325102e-05,
1503
+ "loss": 1.3699,
1504
+ "step": 120000
1505
+ },
1506
+ {
1507
+ "epoch": 1.54,
1508
+ "eval_loss": 1.323049545288086,
1509
+ "eval_runtime": 3382.1272,
1510
+ "eval_samples_per_second": 39.13,
1511
+ "eval_steps_per_second": 4.891,
1512
+ "step": 120000
1513
+ },
1514
+ {
1515
+ "epoch": 1.55,
1516
+ "learning_rate": 2.4156382837264566e-05,
1517
+ "loss": 1.3738,
1518
+ "step": 120500
1519
+ },
1520
+ {
1521
+ "epoch": 1.56,
1522
+ "learning_rate": 2.4049147911278114e-05,
1523
+ "loss": 1.3661,
1524
+ "step": 121000
1525
+ },
1526
+ {
1527
+ "epoch": 1.56,
1528
+ "learning_rate": 2.394191298529166e-05,
1529
+ "loss": 1.3682,
1530
+ "step": 121500
1531
+ },
1532
+ {
1533
+ "epoch": 1.57,
1534
+ "learning_rate": 2.3834678059305205e-05,
1535
+ "loss": 1.3731,
1536
+ "step": 122000
1537
+ },
1538
+ {
1539
+ "epoch": 1.58,
1540
+ "learning_rate": 2.372744313331875e-05,
1541
+ "loss": 1.3668,
1542
+ "step": 122500
1543
+ },
1544
+ {
1545
+ "epoch": 1.58,
1546
+ "learning_rate": 2.3620208207332296e-05,
1547
+ "loss": 1.3665,
1548
+ "step": 123000
1549
+ },
1550
+ {
1551
+ "epoch": 1.59,
1552
+ "learning_rate": 2.3512973281345842e-05,
1553
+ "loss": 1.3606,
1554
+ "step": 123500
1555
+ },
1556
+ {
1557
+ "epoch": 1.6,
1558
+ "learning_rate": 2.3405738355359387e-05,
1559
+ "loss": 1.3559,
1560
+ "step": 124000
1561
+ },
1562
+ {
1563
+ "epoch": 1.6,
1564
+ "learning_rate": 2.3298503429372933e-05,
1565
+ "loss": 1.3663,
1566
+ "step": 124500
1567
+ },
1568
+ {
1569
+ "epoch": 1.61,
1570
+ "learning_rate": 2.3191268503386478e-05,
1571
+ "loss": 1.3656,
1572
+ "step": 125000
1573
+ },
1574
+ {
1575
+ "epoch": 1.61,
1576
+ "learning_rate": 2.3084033577400027e-05,
1577
+ "loss": 1.3765,
1578
+ "step": 125500
1579
+ },
1580
+ {
1581
+ "epoch": 1.62,
1582
+ "learning_rate": 2.2976798651413573e-05,
1583
+ "loss": 1.3688,
1584
+ "step": 126000
1585
+ },
1586
+ {
1587
+ "epoch": 1.63,
1588
+ "learning_rate": 2.2869563725427118e-05,
1589
+ "loss": 1.3631,
1590
+ "step": 126500
1591
+ },
1592
+ {
1593
+ "epoch": 1.63,
1594
+ "learning_rate": 2.2762328799440664e-05,
1595
+ "loss": 1.3493,
1596
+ "step": 127000
1597
+ },
1598
+ {
1599
+ "epoch": 1.64,
1600
+ "learning_rate": 2.265509387345421e-05,
1601
+ "loss": 1.36,
1602
+ "step": 127500
1603
+ },
1604
+ {
1605
+ "epoch": 1.65,
1606
+ "learning_rate": 2.2547858947467758e-05,
1607
+ "loss": 1.3563,
1608
+ "step": 128000
1609
+ },
1610
+ {
1611
+ "epoch": 1.65,
1612
+ "learning_rate": 2.2440624021481303e-05,
1613
+ "loss": 1.3518,
1614
+ "step": 128500
1615
+ },
1616
+ {
1617
+ "epoch": 1.66,
1618
+ "learning_rate": 2.233338909549485e-05,
1619
+ "loss": 1.3553,
1620
+ "step": 129000
1621
+ },
1622
+ {
1623
+ "epoch": 1.67,
1624
+ "learning_rate": 2.222615416950839e-05,
1625
+ "loss": 1.3541,
1626
+ "step": 129500
1627
+ },
1628
+ {
1629
+ "epoch": 1.67,
1630
+ "learning_rate": 2.211891924352194e-05,
1631
+ "loss": 1.3482,
1632
+ "step": 130000
1633
+ },
1634
+ {
1635
+ "epoch": 1.68,
1636
+ "learning_rate": 2.2011684317535485e-05,
1637
+ "loss": 1.3502,
1638
+ "step": 130500
1639
+ },
1640
+ {
1641
+ "epoch": 1.69,
1642
+ "learning_rate": 2.190444939154903e-05,
1643
+ "loss": 1.352,
1644
+ "step": 131000
1645
+ },
1646
+ {
1647
+ "epoch": 1.69,
1648
+ "learning_rate": 2.1797214465562576e-05,
1649
+ "loss": 1.3555,
1650
+ "step": 131500
1651
+ },
1652
+ {
1653
+ "epoch": 1.7,
1654
+ "learning_rate": 2.168997953957612e-05,
1655
+ "loss": 1.3587,
1656
+ "step": 132000
1657
+ },
1658
+ {
1659
+ "epoch": 1.71,
1660
+ "learning_rate": 2.158274461358967e-05,
1661
+ "loss": 1.3456,
1662
+ "step": 132500
1663
+ },
1664
+ {
1665
+ "epoch": 1.71,
1666
+ "learning_rate": 2.1475509687603216e-05,
1667
+ "loss": 1.3486,
1668
+ "step": 133000
1669
+ },
1670
+ {
1671
+ "epoch": 1.72,
1672
+ "learning_rate": 2.136827476161676e-05,
1673
+ "loss": 1.3491,
1674
+ "step": 133500
1675
+ },
1676
+ {
1677
+ "epoch": 1.72,
1678
+ "learning_rate": 2.1261039835630307e-05,
1679
+ "loss": 1.3479,
1680
+ "step": 134000
1681
+ },
1682
+ {
1683
+ "epoch": 1.73,
1684
+ "learning_rate": 2.1153804909643852e-05,
1685
+ "loss": 1.3405,
1686
+ "step": 134500
1687
+ },
1688
+ {
1689
+ "epoch": 1.74,
1690
+ "learning_rate": 2.1046569983657398e-05,
1691
+ "loss": 1.3493,
1692
+ "step": 135000
1693
+ },
1694
+ {
1695
+ "epoch": 1.74,
1696
+ "eval_loss": 1.301226258277893,
1697
+ "eval_runtime": 3380.5336,
1698
+ "eval_samples_per_second": 39.149,
1699
+ "eval_steps_per_second": 4.894,
1700
+ "step": 135000
1701
+ },
1702
+ {
1703
+ "epoch": 1.74,
1704
+ "learning_rate": 2.0939335057670943e-05,
1705
+ "loss": 1.3478,
1706
+ "step": 135500
1707
+ },
1708
+ {
1709
+ "epoch": 1.75,
1710
+ "learning_rate": 2.083210013168449e-05,
1711
+ "loss": 1.3441,
1712
+ "step": 136000
1713
+ },
1714
+ {
1715
+ "epoch": 1.76,
1716
+ "learning_rate": 2.0724865205698034e-05,
1717
+ "loss": 1.3483,
1718
+ "step": 136500
1719
+ },
1720
+ {
1721
+ "epoch": 1.76,
1722
+ "learning_rate": 2.061763027971158e-05,
1723
+ "loss": 1.3367,
1724
+ "step": 137000
1725
+ },
1726
+ {
1727
+ "epoch": 1.77,
1728
+ "learning_rate": 2.051039535372513e-05,
1729
+ "loss": 1.3468,
1730
+ "step": 137500
1731
+ },
1732
+ {
1733
+ "epoch": 1.78,
1734
+ "learning_rate": 2.0403160427738674e-05,
1735
+ "loss": 1.3434,
1736
+ "step": 138000
1737
+ },
1738
+ {
1739
+ "epoch": 1.78,
1740
+ "learning_rate": 2.029592550175222e-05,
1741
+ "loss": 1.3377,
1742
+ "step": 138500
1743
+ },
1744
+ {
1745
+ "epoch": 1.79,
1746
+ "learning_rate": 2.0188690575765765e-05,
1747
+ "loss": 1.3383,
1748
+ "step": 139000
1749
+ },
1750
+ {
1751
+ "epoch": 1.8,
1752
+ "learning_rate": 2.0081455649779314e-05,
1753
+ "loss": 1.3427,
1754
+ "step": 139500
1755
+ },
1756
+ {
1757
+ "epoch": 1.8,
1758
+ "learning_rate": 1.997422072379286e-05,
1759
+ "loss": 1.3477,
1760
+ "step": 140000
1761
+ },
1762
+ {
1763
+ "epoch": 1.81,
1764
+ "learning_rate": 1.9866985797806405e-05,
1765
+ "loss": 1.3364,
1766
+ "step": 140500
1767
+ },
1768
+ {
1769
+ "epoch": 1.81,
1770
+ "learning_rate": 1.975975087181995e-05,
1771
+ "loss": 1.3447,
1772
+ "step": 141000
1773
+ },
1774
+ {
1775
+ "epoch": 1.82,
1776
+ "learning_rate": 1.9652515945833492e-05,
1777
+ "loss": 1.3365,
1778
+ "step": 141500
1779
+ },
1780
+ {
1781
+ "epoch": 1.83,
1782
+ "learning_rate": 1.954528101984704e-05,
1783
+ "loss": 1.3231,
1784
+ "step": 142000
1785
+ },
1786
+ {
1787
+ "epoch": 1.83,
1788
+ "learning_rate": 1.9438046093860587e-05,
1789
+ "loss": 1.3362,
1790
+ "step": 142500
1791
+ },
1792
+ {
1793
+ "epoch": 1.84,
1794
+ "learning_rate": 1.9330811167874132e-05,
1795
+ "loss": 1.3415,
1796
+ "step": 143000
1797
+ },
1798
+ {
1799
+ "epoch": 1.85,
1800
+ "learning_rate": 1.9223576241887678e-05,
1801
+ "loss": 1.3323,
1802
+ "step": 143500
1803
+ },
1804
+ {
1805
+ "epoch": 1.85,
1806
+ "learning_rate": 1.9116341315901223e-05,
1807
+ "loss": 1.3316,
1808
+ "step": 144000
1809
+ },
1810
+ {
1811
+ "epoch": 1.86,
1812
+ "learning_rate": 1.9009106389914772e-05,
1813
+ "loss": 1.3304,
1814
+ "step": 144500
1815
+ },
1816
+ {
1817
+ "epoch": 1.87,
1818
+ "learning_rate": 1.8901871463928317e-05,
1819
+ "loss": 1.329,
1820
+ "step": 145000
1821
+ },
1822
+ {
1823
+ "epoch": 1.87,
1824
+ "learning_rate": 1.8794636537941863e-05,
1825
+ "loss": 1.3268,
1826
+ "step": 145500
1827
+ },
1828
+ {
1829
+ "epoch": 1.88,
1830
+ "learning_rate": 1.8687401611955408e-05,
1831
+ "loss": 1.337,
1832
+ "step": 146000
1833
+ },
1834
+ {
1835
+ "epoch": 1.89,
1836
+ "learning_rate": 1.8580166685968957e-05,
1837
+ "loss": 1.3277,
1838
+ "step": 146500
1839
+ },
1840
+ {
1841
+ "epoch": 1.89,
1842
+ "learning_rate": 1.84729317599825e-05,
1843
+ "loss": 1.323,
1844
+ "step": 147000
1845
+ },
1846
+ {
1847
+ "epoch": 1.9,
1848
+ "learning_rate": 1.8365696833996045e-05,
1849
+ "loss": 1.3318,
1850
+ "step": 147500
1851
+ },
1852
+ {
1853
+ "epoch": 1.9,
1854
+ "learning_rate": 1.825846190800959e-05,
1855
+ "loss": 1.3227,
1856
+ "step": 148000
1857
+ },
1858
+ {
1859
+ "epoch": 1.91,
1860
+ "learning_rate": 1.8151226982023136e-05,
1861
+ "loss": 1.3334,
1862
+ "step": 148500
1863
+ },
1864
+ {
1865
+ "epoch": 1.92,
1866
+ "learning_rate": 1.8043992056036685e-05,
1867
+ "loss": 1.3255,
1868
+ "step": 149000
1869
+ },
1870
+ {
1871
+ "epoch": 1.92,
1872
+ "learning_rate": 1.793675713005023e-05,
1873
+ "loss": 1.3244,
1874
+ "step": 149500
1875
+ },
1876
+ {
1877
+ "epoch": 1.93,
1878
+ "learning_rate": 1.7829522204063775e-05,
1879
+ "loss": 1.3201,
1880
+ "step": 150000
1881
+ },
1882
+ {
1883
+ "epoch": 1.93,
1884
+ "eval_loss": 1.2773252725601196,
1885
+ "eval_runtime": 3341.8511,
1886
+ "eval_samples_per_second": 39.602,
1887
+ "eval_steps_per_second": 4.95,
1888
+ "step": 150000
1889
+ },
1890
+ {
1891
+ "epoch": 1.94,
1892
+ "learning_rate": 1.772228727807732e-05,
1893
+ "loss": 1.3188,
1894
+ "step": 150500
1895
+ },
1896
+ {
1897
+ "epoch": 1.94,
1898
+ "learning_rate": 1.7615052352090866e-05,
1899
+ "loss": 1.3203,
1900
+ "step": 151000
1901
+ },
1902
+ {
1903
+ "epoch": 1.95,
1904
+ "learning_rate": 1.7507817426104415e-05,
1905
+ "loss": 1.3123,
1906
+ "step": 151500
1907
+ },
1908
+ {
1909
+ "epoch": 1.96,
1910
+ "learning_rate": 1.740058250011796e-05,
1911
+ "loss": 1.319,
1912
+ "step": 152000
1913
+ },
1914
+ {
1915
+ "epoch": 1.96,
1916
+ "learning_rate": 1.7293347574131506e-05,
1917
+ "loss": 1.3225,
1918
+ "step": 152500
1919
+ },
1920
+ {
1921
+ "epoch": 1.97,
1922
+ "learning_rate": 1.718611264814505e-05,
1923
+ "loss": 1.3231,
1924
+ "step": 153000
1925
+ },
1926
+ {
1927
+ "epoch": 1.98,
1928
+ "learning_rate": 1.7078877722158597e-05,
1929
+ "loss": 1.3213,
1930
+ "step": 153500
1931
+ },
1932
+ {
1933
+ "epoch": 1.98,
1934
+ "learning_rate": 1.6971642796172143e-05,
1935
+ "loss": 1.315,
1936
+ "step": 154000
1937
+ },
1938
+ {
1939
+ "epoch": 1.99,
1940
+ "learning_rate": 1.6864407870185688e-05,
1941
+ "loss": 1.3203,
1942
+ "step": 154500
1943
+ },
1944
+ {
1945
+ "epoch": 1.99,
1946
+ "learning_rate": 1.6757172944199234e-05,
1947
+ "loss": 1.3208,
1948
+ "step": 155000
1949
+ },
1950
+ {
1951
+ "epoch": 2.0,
1952
+ "learning_rate": 1.664993801821278e-05,
1953
+ "loss": 1.3155,
1954
+ "step": 155500
1955
+ },
1956
+ {
1957
+ "epoch": 2.01,
1958
+ "learning_rate": 1.6542703092226328e-05,
1959
+ "loss": 1.3038,
1960
+ "step": 156000
1961
+ },
1962
+ {
1963
+ "epoch": 2.01,
1964
+ "learning_rate": 1.6435468166239873e-05,
1965
+ "loss": 1.2996,
1966
+ "step": 156500
1967
+ },
1968
+ {
1969
+ "epoch": 2.02,
1970
+ "learning_rate": 1.632823324025342e-05,
1971
+ "loss": 1.3023,
1972
+ "step": 157000
1973
+ },
1974
+ {
1975
+ "epoch": 2.03,
1976
+ "learning_rate": 1.6220998314266964e-05,
1977
+ "loss": 1.3112,
1978
+ "step": 157500
1979
+ },
1980
+ {
1981
+ "epoch": 2.03,
1982
+ "learning_rate": 1.611376338828051e-05,
1983
+ "loss": 1.3002,
1984
+ "step": 158000
1985
+ },
1986
+ {
1987
+ "epoch": 2.04,
1988
+ "learning_rate": 1.600652846229406e-05,
1989
+ "loss": 1.2967,
1990
+ "step": 158500
1991
+ },
1992
+ {
1993
+ "epoch": 2.05,
1994
+ "learning_rate": 1.58992935363076e-05,
1995
+ "loss": 1.2996,
1996
+ "step": 159000
1997
+ },
1998
+ {
1999
+ "epoch": 2.05,
2000
+ "learning_rate": 1.5792058610321146e-05,
2001
+ "loss": 1.3006,
2002
+ "step": 159500
2003
+ },
2004
+ {
2005
+ "epoch": 2.06,
2006
+ "learning_rate": 1.568482368433469e-05,
2007
+ "loss": 1.2903,
2008
+ "step": 160000
2009
+ },
2010
+ {
2011
+ "epoch": 2.07,
2012
+ "learning_rate": 1.557758875834824e-05,
2013
+ "loss": 1.3077,
2014
+ "step": 160500
2015
+ },
2016
+ {
2017
+ "epoch": 2.07,
2018
+ "learning_rate": 1.5470353832361786e-05,
2019
+ "loss": 1.2952,
2020
+ "step": 161000
2021
+ },
2022
+ {
2023
+ "epoch": 2.08,
2024
+ "learning_rate": 1.536311890637533e-05,
2025
+ "loss": 1.2984,
2026
+ "step": 161500
2027
+ },
2028
+ {
2029
+ "epoch": 2.08,
2030
+ "learning_rate": 1.5255883980388877e-05,
2031
+ "loss": 1.3062,
2032
+ "step": 162000
2033
+ },
2034
+ {
2035
+ "epoch": 2.09,
2036
+ "learning_rate": 1.5148649054402422e-05,
2037
+ "loss": 1.3005,
2038
+ "step": 162500
2039
+ },
2040
+ {
2041
+ "epoch": 2.1,
2042
+ "learning_rate": 1.5041414128415971e-05,
2043
+ "loss": 1.2961,
2044
+ "step": 163000
2045
+ },
2046
+ {
2047
+ "epoch": 2.1,
2048
+ "learning_rate": 1.4934179202429515e-05,
2049
+ "loss": 1.299,
2050
+ "step": 163500
2051
+ },
2052
+ {
2053
+ "epoch": 2.11,
2054
+ "learning_rate": 1.482694427644306e-05,
2055
+ "loss": 1.2931,
2056
+ "step": 164000
2057
+ },
2058
+ {
2059
+ "epoch": 2.12,
2060
+ "learning_rate": 1.4719709350456606e-05,
2061
+ "loss": 1.2962,
2062
+ "step": 164500
2063
+ },
2064
+ {
2065
+ "epoch": 2.12,
2066
+ "learning_rate": 1.4612474424470151e-05,
2067
+ "loss": 1.2993,
2068
+ "step": 165000
2069
+ },
2070
+ {
2071
+ "epoch": 2.12,
2072
+ "eval_loss": 1.261716604232788,
2073
+ "eval_runtime": 3339.8508,
2074
+ "eval_samples_per_second": 39.626,
2075
+ "eval_steps_per_second": 4.953,
2076
+ "step": 165000
2077
+ },
2078
+ {
2079
+ "epoch": 2.13,
2080
+ "learning_rate": 1.45052394984837e-05,
2081
+ "loss": 1.2979,
2082
+ "step": 165500
2083
+ },
2084
+ {
2085
+ "epoch": 2.14,
2086
+ "learning_rate": 1.4398004572497246e-05,
2087
+ "loss": 1.2992,
2088
+ "step": 166000
2089
+ },
2090
+ {
2091
+ "epoch": 2.14,
2092
+ "learning_rate": 1.4290769646510791e-05,
2093
+ "loss": 1.3039,
2094
+ "step": 166500
2095
+ },
2096
+ {
2097
+ "epoch": 2.15,
2098
+ "learning_rate": 1.4183534720524335e-05,
2099
+ "loss": 1.2953,
2100
+ "step": 167000
2101
+ },
2102
+ {
2103
+ "epoch": 2.16,
2104
+ "learning_rate": 1.4076299794537884e-05,
2105
+ "loss": 1.2837,
2106
+ "step": 167500
2107
+ },
2108
+ {
2109
+ "epoch": 2.16,
2110
+ "learning_rate": 1.396906486855143e-05,
2111
+ "loss": 1.2971,
2112
+ "step": 168000
2113
+ },
2114
+ {
2115
+ "epoch": 2.17,
2116
+ "learning_rate": 1.3861829942564975e-05,
2117
+ "loss": 1.2963,
2118
+ "step": 168500
2119
+ },
2120
+ {
2121
+ "epoch": 2.17,
2122
+ "learning_rate": 1.375459501657852e-05,
2123
+ "loss": 1.288,
2124
+ "step": 169000
2125
+ },
2126
+ {
2127
+ "epoch": 2.18,
2128
+ "learning_rate": 1.3647360090592066e-05,
2129
+ "loss": 1.2953,
2130
+ "step": 169500
2131
+ },
2132
+ {
2133
+ "epoch": 2.19,
2134
+ "learning_rate": 1.3540125164605613e-05,
2135
+ "loss": 1.2892,
2136
+ "step": 170000
2137
+ },
2138
+ {
2139
+ "epoch": 2.19,
2140
+ "learning_rate": 1.3432890238619158e-05,
2141
+ "loss": 1.2931,
2142
+ "step": 170500
2143
+ },
2144
+ {
2145
+ "epoch": 2.2,
2146
+ "learning_rate": 1.3325655312632704e-05,
2147
+ "loss": 1.2865,
2148
+ "step": 171000
2149
+ },
2150
+ {
2151
+ "epoch": 2.21,
2152
+ "learning_rate": 1.321842038664625e-05,
2153
+ "loss": 1.2925,
2154
+ "step": 171500
2155
+ },
2156
+ {
2157
+ "epoch": 2.21,
2158
+ "learning_rate": 1.3111185460659795e-05,
2159
+ "loss": 1.286,
2160
+ "step": 172000
2161
+ },
2162
+ {
2163
+ "epoch": 2.22,
2164
+ "learning_rate": 1.3003950534673342e-05,
2165
+ "loss": 1.2949,
2166
+ "step": 172500
2167
+ },
2168
+ {
2169
+ "epoch": 2.23,
2170
+ "learning_rate": 1.2896715608686887e-05,
2171
+ "loss": 1.2841,
2172
+ "step": 173000
2173
+ },
2174
+ {
2175
+ "epoch": 2.23,
2176
+ "learning_rate": 1.2789480682700433e-05,
2177
+ "loss": 1.2869,
2178
+ "step": 173500
2179
+ },
2180
+ {
2181
+ "epoch": 2.24,
2182
+ "learning_rate": 1.2682245756713978e-05,
2183
+ "loss": 1.2791,
2184
+ "step": 174000
2185
+ },
2186
+ {
2187
+ "epoch": 2.25,
2188
+ "learning_rate": 1.2575010830727527e-05,
2189
+ "loss": 1.2822,
2190
+ "step": 174500
2191
+ },
2192
+ {
2193
+ "epoch": 2.25,
2194
+ "learning_rate": 1.2467775904741071e-05,
2195
+ "loss": 1.2857,
2196
+ "step": 175000
2197
+ },
2198
+ {
2199
+ "epoch": 2.26,
2200
+ "learning_rate": 1.2360540978754616e-05,
2201
+ "loss": 1.2913,
2202
+ "step": 175500
2203
+ },
2204
+ {
2205
+ "epoch": 2.26,
2206
+ "learning_rate": 1.2253306052768162e-05,
2207
+ "loss": 1.273,
2208
+ "step": 176000
2209
+ },
2210
+ {
2211
+ "epoch": 2.27,
2212
+ "learning_rate": 1.2146071126781709e-05,
2213
+ "loss": 1.2783,
2214
+ "step": 176500
2215
+ },
2216
+ {
2217
+ "epoch": 2.28,
2218
+ "learning_rate": 1.2038836200795255e-05,
2219
+ "loss": 1.2707,
2220
+ "step": 177000
2221
+ },
2222
+ {
2223
+ "epoch": 2.28,
2224
+ "learning_rate": 1.1931601274808802e-05,
2225
+ "loss": 1.2747,
2226
+ "step": 177500
2227
+ },
2228
+ {
2229
+ "epoch": 2.29,
2230
+ "learning_rate": 1.1824366348822347e-05,
2231
+ "loss": 1.2814,
2232
+ "step": 178000
2233
+ },
2234
+ {
2235
+ "epoch": 2.3,
2236
+ "learning_rate": 1.1717131422835893e-05,
2237
+ "loss": 1.2862,
2238
+ "step": 178500
2239
+ },
2240
+ {
2241
+ "epoch": 2.3,
2242
+ "learning_rate": 1.1609896496849438e-05,
2243
+ "loss": 1.2773,
2244
+ "step": 179000
2245
+ },
2246
+ {
2247
+ "epoch": 2.31,
2248
+ "learning_rate": 1.1502661570862984e-05,
2249
+ "loss": 1.2783,
2250
+ "step": 179500
2251
+ },
2252
+ {
2253
+ "epoch": 2.32,
2254
+ "learning_rate": 1.139542664487653e-05,
2255
+ "loss": 1.2745,
2256
+ "step": 180000
2257
+ },
2258
+ {
2259
+ "epoch": 2.32,
2260
+ "eval_loss": 1.2489670515060425,
2261
+ "eval_runtime": 3344.7969,
2262
+ "eval_samples_per_second": 39.567,
2263
+ "eval_steps_per_second": 4.946,
2264
+ "step": 180000
2265
+ },
2266
+ {
2267
+ "epoch": 2.32,
2268
+ "learning_rate": 1.1288191718890076e-05,
2269
+ "loss": 1.2783,
2270
+ "step": 180500
2271
+ },
2272
+ {
2273
+ "epoch": 2.33,
2274
+ "learning_rate": 1.1180956792903623e-05,
2275
+ "loss": 1.2782,
2276
+ "step": 181000
2277
+ },
2278
+ {
2279
+ "epoch": 2.34,
2280
+ "learning_rate": 1.1073721866917169e-05,
2281
+ "loss": 1.2727,
2282
+ "step": 181500
2283
+ },
2284
+ {
2285
+ "epoch": 2.34,
2286
+ "learning_rate": 1.0966486940930713e-05,
2287
+ "loss": 1.2769,
2288
+ "step": 182000
2289
+ },
2290
+ {
2291
+ "epoch": 2.35,
2292
+ "learning_rate": 1.085925201494426e-05,
2293
+ "loss": 1.2743,
2294
+ "step": 182500
2295
+ },
2296
+ {
2297
+ "epoch": 2.35,
2298
+ "learning_rate": 1.0752017088957805e-05,
2299
+ "loss": 1.271,
2300
+ "step": 183000
2301
+ },
2302
+ {
2303
+ "epoch": 2.36,
2304
+ "learning_rate": 1.0644782162971352e-05,
2305
+ "loss": 1.2747,
2306
+ "step": 183500
2307
+ },
2308
+ {
2309
+ "epoch": 2.37,
2310
+ "learning_rate": 1.0537547236984898e-05,
2311
+ "loss": 1.2733,
2312
+ "step": 184000
2313
+ },
2314
+ {
2315
+ "epoch": 2.37,
2316
+ "learning_rate": 1.0430312310998443e-05,
2317
+ "loss": 1.2719,
2318
+ "step": 184500
2319
+ },
2320
+ {
2321
+ "epoch": 2.38,
2322
+ "learning_rate": 1.0323077385011989e-05,
2323
+ "loss": 1.2724,
2324
+ "step": 185000
2325
+ },
2326
+ {
2327
+ "epoch": 2.39,
2328
+ "learning_rate": 1.0215842459025534e-05,
2329
+ "loss": 1.2699,
2330
+ "step": 185500
2331
+ },
2332
+ {
2333
+ "epoch": 2.39,
2334
+ "learning_rate": 1.0108607533039081e-05,
2335
+ "loss": 1.2707,
2336
+ "step": 186000
2337
+ },
2338
+ {
2339
+ "epoch": 2.4,
2340
+ "learning_rate": 1.0001372607052627e-05,
2341
+ "loss": 1.2805,
2342
+ "step": 186500
2343
+ },
2344
+ {
2345
+ "epoch": 2.41,
2346
+ "learning_rate": 9.894137681066174e-06,
2347
+ "loss": 1.2695,
2348
+ "step": 187000
2349
+ },
2350
+ {
2351
+ "epoch": 2.41,
2352
+ "learning_rate": 9.78690275507972e-06,
2353
+ "loss": 1.2618,
2354
+ "step": 187500
2355
+ },
2356
+ {
2357
+ "epoch": 2.42,
2358
+ "learning_rate": 9.679667829093265e-06,
2359
+ "loss": 1.2636,
2360
+ "step": 188000
2361
+ },
2362
+ {
2363
+ "epoch": 2.43,
2364
+ "learning_rate": 9.57243290310681e-06,
2365
+ "loss": 1.2738,
2366
+ "step": 188500
2367
+ },
2368
+ {
2369
+ "epoch": 2.43,
2370
+ "learning_rate": 9.465197977120356e-06,
2371
+ "loss": 1.266,
2372
+ "step": 189000
2373
+ },
2374
+ {
2375
+ "epoch": 2.44,
2376
+ "learning_rate": 9.357963051133903e-06,
2377
+ "loss": 1.2702,
2378
+ "step": 189500
2379
+ },
2380
+ {
2381
+ "epoch": 2.44,
2382
+ "learning_rate": 9.250728125147449e-06,
2383
+ "loss": 1.2629,
2384
+ "step": 190000
2385
+ },
2386
+ {
2387
+ "epoch": 2.45,
2388
+ "learning_rate": 9.143493199160994e-06,
2389
+ "loss": 1.2653,
2390
+ "step": 190500
2391
+ },
2392
+ {
2393
+ "epoch": 2.46,
2394
+ "learning_rate": 9.03625827317454e-06,
2395
+ "loss": 1.2544,
2396
+ "step": 191000
2397
+ },
2398
+ {
2399
+ "epoch": 2.46,
2400
+ "learning_rate": 8.929023347188087e-06,
2401
+ "loss": 1.2661,
2402
+ "step": 191500
2403
+ },
2404
+ {
2405
+ "epoch": 2.47,
2406
+ "learning_rate": 8.821788421201632e-06,
2407
+ "loss": 1.2697,
2408
+ "step": 192000
2409
+ },
2410
+ {
2411
+ "epoch": 2.48,
2412
+ "learning_rate": 8.714553495215178e-06,
2413
+ "loss": 1.262,
2414
+ "step": 192500
2415
+ },
2416
+ {
2417
+ "epoch": 2.48,
2418
+ "learning_rate": 8.607318569228725e-06,
2419
+ "loss": 1.2724,
2420
+ "step": 193000
2421
+ },
2422
+ {
2423
+ "epoch": 2.49,
2424
+ "learning_rate": 8.50008364324227e-06,
2425
+ "loss": 1.2617,
2426
+ "step": 193500
2427
+ },
2428
+ {
2429
+ "epoch": 2.5,
2430
+ "learning_rate": 8.392848717255816e-06,
2431
+ "loss": 1.263,
2432
+ "step": 194000
2433
+ },
2434
+ {
2435
+ "epoch": 2.5,
2436
+ "learning_rate": 8.285613791269361e-06,
2437
+ "loss": 1.2629,
2438
+ "step": 194500
2439
+ },
2440
+ {
2441
+ "epoch": 2.51,
2442
+ "learning_rate": 8.178378865282908e-06,
2443
+ "loss": 1.2614,
2444
+ "step": 195000
2445
+ },
2446
+ {
2447
+ "epoch": 2.51,
2448
+ "eval_loss": 1.2283380031585693,
2449
+ "eval_runtime": 3383.2987,
2450
+ "eval_samples_per_second": 39.117,
2451
+ "eval_steps_per_second": 4.89,
2452
+ "step": 195000
2453
+ },
2454
+ {
2455
+ "epoch": 2.52,
2456
+ "learning_rate": 8.071143939296454e-06,
2457
+ "loss": 1.2589,
2458
+ "step": 195500
2459
+ },
2460
+ {
2461
+ "epoch": 2.52,
2462
+ "learning_rate": 7.96390901331e-06,
2463
+ "loss": 1.2497,
2464
+ "step": 196000
2465
+ },
2466
+ {
2467
+ "epoch": 2.53,
2468
+ "learning_rate": 7.856674087323545e-06,
2469
+ "loss": 1.256,
2470
+ "step": 196500
2471
+ },
2472
+ {
2473
+ "epoch": 2.54,
2474
+ "learning_rate": 7.74943916133709e-06,
2475
+ "loss": 1.2563,
2476
+ "step": 197000
2477
+ },
2478
+ {
2479
+ "epoch": 2.54,
2480
+ "learning_rate": 7.642204235350637e-06,
2481
+ "loss": 1.2531,
2482
+ "step": 197500
2483
+ },
2484
+ {
2485
+ "epoch": 2.55,
2486
+ "learning_rate": 7.534969309364183e-06,
2487
+ "loss": 1.2586,
2488
+ "step": 198000
2489
+ },
2490
+ {
2491
+ "epoch": 2.55,
2492
+ "learning_rate": 7.427734383377728e-06,
2493
+ "loss": 1.2589,
2494
+ "step": 198500
2495
+ },
2496
+ {
2497
+ "epoch": 2.56,
2498
+ "learning_rate": 7.320499457391275e-06,
2499
+ "loss": 1.2623,
2500
+ "step": 199000
2501
+ },
2502
+ {
2503
+ "epoch": 2.57,
2504
+ "learning_rate": 7.21326453140482e-06,
2505
+ "loss": 1.2573,
2506
+ "step": 199500
2507
+ },
2508
+ {
2509
+ "epoch": 2.57,
2510
+ "learning_rate": 7.106029605418367e-06,
2511
+ "loss": 1.2487,
2512
+ "step": 200000
2513
+ },
2514
+ {
2515
+ "epoch": 2.58,
2516
+ "learning_rate": 6.998794679431912e-06,
2517
+ "loss": 1.2522,
2518
+ "step": 200500
2519
+ },
2520
+ {
2521
+ "epoch": 2.59,
2522
+ "learning_rate": 6.891559753445459e-06,
2523
+ "loss": 1.2486,
2524
+ "step": 201000
2525
+ },
2526
+ {
2527
+ "epoch": 2.59,
2528
+ "learning_rate": 6.7843248274590046e-06,
2529
+ "loss": 1.253,
2530
+ "step": 201500
2531
+ },
2532
+ {
2533
+ "epoch": 2.6,
2534
+ "learning_rate": 6.67708990147255e-06,
2535
+ "loss": 1.2453,
2536
+ "step": 202000
2537
+ },
2538
+ {
2539
+ "epoch": 2.61,
2540
+ "learning_rate": 6.569854975486096e-06,
2541
+ "loss": 1.2463,
2542
+ "step": 202500
2543
+ },
2544
+ {
2545
+ "epoch": 2.61,
2546
+ "learning_rate": 6.462620049499642e-06,
2547
+ "loss": 1.2485,
2548
+ "step": 203000
2549
+ },
2550
+ {
2551
+ "epoch": 2.62,
2552
+ "learning_rate": 6.355385123513188e-06,
2553
+ "loss": 1.2521,
2554
+ "step": 203500
2555
+ },
2556
+ {
2557
+ "epoch": 2.63,
2558
+ "learning_rate": 6.248150197526734e-06,
2559
+ "loss": 1.2441,
2560
+ "step": 204000
2561
+ },
2562
+ {
2563
+ "epoch": 2.63,
2564
+ "learning_rate": 6.14091527154028e-06,
2565
+ "loss": 1.2513,
2566
+ "step": 204500
2567
+ },
2568
+ {
2569
+ "epoch": 2.64,
2570
+ "learning_rate": 6.033680345553825e-06,
2571
+ "loss": 1.2594,
2572
+ "step": 205000
2573
+ },
2574
+ {
2575
+ "epoch": 2.64,
2576
+ "learning_rate": 5.926445419567372e-06,
2577
+ "loss": 1.2539,
2578
+ "step": 205500
2579
+ },
2580
+ {
2581
+ "epoch": 2.65,
2582
+ "learning_rate": 5.819210493580918e-06,
2583
+ "loss": 1.2503,
2584
+ "step": 206000
2585
+ },
2586
+ {
2587
+ "epoch": 2.66,
2588
+ "learning_rate": 5.7119755675944635e-06,
2589
+ "loss": 1.2463,
2590
+ "step": 206500
2591
+ },
2592
+ {
2593
+ "epoch": 2.66,
2594
+ "learning_rate": 5.60474064160801e-06,
2595
+ "loss": 1.2378,
2596
+ "step": 207000
2597
+ },
2598
+ {
2599
+ "epoch": 2.67,
2600
+ "learning_rate": 5.497505715621555e-06,
2601
+ "loss": 1.2447,
2602
+ "step": 207500
2603
+ },
2604
+ {
2605
+ "epoch": 2.68,
2606
+ "learning_rate": 5.390270789635101e-06,
2607
+ "loss": 1.2434,
2608
+ "step": 208000
2609
+ },
2610
+ {
2611
+ "epoch": 2.68,
2612
+ "learning_rate": 5.283035863648647e-06,
2613
+ "loss": 1.2469,
2614
+ "step": 208500
2615
+ },
2616
+ {
2617
+ "epoch": 2.69,
2618
+ "learning_rate": 5.175800937662193e-06,
2619
+ "loss": 1.2505,
2620
+ "step": 209000
2621
+ },
2622
+ {
2623
+ "epoch": 2.7,
2624
+ "learning_rate": 5.068566011675739e-06,
2625
+ "loss": 1.2485,
2626
+ "step": 209500
2627
+ },
2628
+ {
2629
+ "epoch": 2.7,
2630
+ "learning_rate": 4.961331085689285e-06,
2631
+ "loss": 1.2424,
2632
+ "step": 210000
2633
+ },
2634
+ {
2635
+ "epoch": 2.7,
2636
+ "eval_loss": 1.2152148485183716,
2637
+ "eval_runtime": 3385.5322,
2638
+ "eval_samples_per_second": 39.091,
2639
+ "eval_steps_per_second": 4.886,
2640
+ "step": 210000
2641
+ },
2642
+ {
2643
+ "epoch": 2.71,
2644
+ "learning_rate": 4.854096159702831e-06,
2645
+ "loss": 1.2362,
2646
+ "step": 210500
2647
+ },
2648
+ {
2649
+ "epoch": 2.72,
2650
+ "learning_rate": 4.746861233716376e-06,
2651
+ "loss": 1.2378,
2652
+ "step": 211000
2653
+ },
2654
+ {
2655
+ "epoch": 2.72,
2656
+ "learning_rate": 4.6396263077299224e-06,
2657
+ "loss": 1.2462,
2658
+ "step": 211500
2659
+ },
2660
+ {
2661
+ "epoch": 2.73,
2662
+ "learning_rate": 4.532391381743469e-06,
2663
+ "loss": 1.2415,
2664
+ "step": 212000
2665
+ },
2666
+ {
2667
+ "epoch": 2.73,
2668
+ "learning_rate": 4.425156455757014e-06,
2669
+ "loss": 1.248,
2670
+ "step": 212500
2671
+ },
2672
+ {
2673
+ "epoch": 2.74,
2674
+ "learning_rate": 4.3179215297705605e-06,
2675
+ "loss": 1.2307,
2676
+ "step": 213000
2677
+ },
2678
+ {
2679
+ "epoch": 2.75,
2680
+ "learning_rate": 4.210686603784107e-06,
2681
+ "loss": 1.2438,
2682
+ "step": 213500
2683
+ },
2684
+ {
2685
+ "epoch": 2.75,
2686
+ "learning_rate": 4.1034516777976515e-06,
2687
+ "loss": 1.2451,
2688
+ "step": 214000
2689
+ },
2690
+ {
2691
+ "epoch": 2.76,
2692
+ "learning_rate": 3.996216751811198e-06,
2693
+ "loss": 1.2388,
2694
+ "step": 214500
2695
+ },
2696
+ {
2697
+ "epoch": 2.77,
2698
+ "learning_rate": 3.888981825824744e-06,
2699
+ "loss": 1.2413,
2700
+ "step": 215000
2701
+ },
2702
+ {
2703
+ "epoch": 2.77,
2704
+ "learning_rate": 3.78174689983829e-06,
2705
+ "loss": 1.241,
2706
+ "step": 215500
2707
+ },
2708
+ {
2709
+ "epoch": 2.78,
2710
+ "learning_rate": 3.674511973851836e-06,
2711
+ "loss": 1.2288,
2712
+ "step": 216000
2713
+ },
2714
+ {
2715
+ "epoch": 2.79,
2716
+ "learning_rate": 3.5672770478653818e-06,
2717
+ "loss": 1.2341,
2718
+ "step": 216500
2719
+ },
2720
+ {
2721
+ "epoch": 2.79,
2722
+ "learning_rate": 3.460042121878928e-06,
2723
+ "loss": 1.2368,
2724
+ "step": 217000
2725
+ },
2726
+ {
2727
+ "epoch": 2.8,
2728
+ "learning_rate": 3.352807195892473e-06,
2729
+ "loss": 1.2298,
2730
+ "step": 217500
2731
+ },
2732
+ {
2733
+ "epoch": 2.81,
2734
+ "learning_rate": 3.2455722699060195e-06,
2735
+ "loss": 1.2289,
2736
+ "step": 218000
2737
+ },
2738
+ {
2739
+ "epoch": 2.81,
2740
+ "learning_rate": 3.1383373439195654e-06,
2741
+ "loss": 1.2313,
2742
+ "step": 218500
2743
+ },
2744
+ {
2745
+ "epoch": 2.82,
2746
+ "learning_rate": 3.0311024179331113e-06,
2747
+ "loss": 1.2341,
2748
+ "step": 219000
2749
+ },
2750
+ {
2751
+ "epoch": 2.82,
2752
+ "learning_rate": 2.923867491946657e-06,
2753
+ "loss": 1.2347,
2754
+ "step": 219500
2755
+ },
2756
+ {
2757
+ "epoch": 2.83,
2758
+ "learning_rate": 2.816632565960203e-06,
2759
+ "loss": 1.2415,
2760
+ "step": 220000
2761
+ },
2762
+ {
2763
+ "epoch": 2.84,
2764
+ "learning_rate": 2.709397639973749e-06,
2765
+ "loss": 1.2351,
2766
+ "step": 220500
2767
+ },
2768
+ {
2769
+ "epoch": 2.84,
2770
+ "learning_rate": 2.602162713987295e-06,
2771
+ "loss": 1.2316,
2772
+ "step": 221000
2773
+ },
2774
+ {
2775
+ "epoch": 2.85,
2776
+ "learning_rate": 2.494927788000841e-06,
2777
+ "loss": 1.229,
2778
+ "step": 221500
2779
+ },
2780
+ {
2781
+ "epoch": 2.86,
2782
+ "learning_rate": 2.3876928620143866e-06,
2783
+ "loss": 1.2256,
2784
+ "step": 222000
2785
+ },
2786
+ {
2787
+ "epoch": 2.86,
2788
+ "learning_rate": 2.2804579360279325e-06,
2789
+ "loss": 1.2342,
2790
+ "step": 222500
2791
+ },
2792
+ {
2793
+ "epoch": 2.87,
2794
+ "learning_rate": 2.173223010041479e-06,
2795
+ "loss": 1.2367,
2796
+ "step": 223000
2797
+ },
2798
+ {
2799
+ "epoch": 2.88,
2800
+ "learning_rate": 2.0659880840550243e-06,
2801
+ "loss": 1.2343,
2802
+ "step": 223500
2803
+ },
2804
+ {
2805
+ "epoch": 2.88,
2806
+ "learning_rate": 1.95875315806857e-06,
2807
+ "loss": 1.2325,
2808
+ "step": 224000
2809
+ },
2810
+ {
2811
+ "epoch": 2.89,
2812
+ "learning_rate": 1.8515182320821163e-06,
2813
+ "loss": 1.2304,
2814
+ "step": 224500
2815
+ },
2816
+ {
2817
+ "epoch": 2.9,
2818
+ "learning_rate": 1.7442833060956624e-06,
2819
+ "loss": 1.2296,
2820
+ "step": 225000
2821
+ },
2822
+ {
2823
+ "epoch": 2.9,
2824
+ "eval_loss": 1.2051972150802612,
2825
+ "eval_runtime": 3383.7385,
2826
+ "eval_samples_per_second": 39.112,
2827
+ "eval_steps_per_second": 4.889,
2828
+ "step": 225000
2829
+ },
2830
+ {
2831
+ "epoch": 2.9,
2832
+ "learning_rate": 1.637048380109208e-06,
2833
+ "loss": 1.2192,
2834
+ "step": 225500
2835
+ },
2836
+ {
2837
+ "epoch": 2.91,
2838
+ "learning_rate": 1.529813454122754e-06,
2839
+ "loss": 1.2305,
2840
+ "step": 226000
2841
+ },
2842
+ {
2843
+ "epoch": 2.91,
2844
+ "learning_rate": 1.4225785281362999e-06,
2845
+ "loss": 1.2292,
2846
+ "step": 226500
2847
+ },
2848
+ {
2849
+ "epoch": 2.92,
2850
+ "learning_rate": 1.315343602149846e-06,
2851
+ "loss": 1.2292,
2852
+ "step": 227000
2853
+ },
2854
+ {
2855
+ "epoch": 2.93,
2856
+ "learning_rate": 1.2081086761633916e-06,
2857
+ "loss": 1.2286,
2858
+ "step": 227500
2859
+ },
2860
+ {
2861
+ "epoch": 2.93,
2862
+ "learning_rate": 1.1008737501769378e-06,
2863
+ "loss": 1.2283,
2864
+ "step": 228000
2865
+ },
2866
+ {
2867
+ "epoch": 2.94,
2868
+ "learning_rate": 9.936388241904836e-07,
2869
+ "loss": 1.2272,
2870
+ "step": 228500
2871
+ },
2872
+ {
2873
+ "epoch": 2.95,
2874
+ "learning_rate": 8.864038982040294e-07,
2875
+ "loss": 1.2292,
2876
+ "step": 229000
2877
+ },
2878
+ {
2879
+ "epoch": 2.95,
2880
+ "learning_rate": 7.791689722175754e-07,
2881
+ "loss": 1.2343,
2882
+ "step": 229500
2883
+ },
2884
+ {
2885
+ "epoch": 2.96,
2886
+ "learning_rate": 6.719340462311213e-07,
2887
+ "loss": 1.2179,
2888
+ "step": 230000
2889
+ },
2890
+ {
2891
+ "epoch": 2.97,
2892
+ "learning_rate": 5.646991202446672e-07,
2893
+ "loss": 1.2239,
2894
+ "step": 230500
2895
+ },
2896
+ {
2897
+ "epoch": 2.97,
2898
+ "learning_rate": 4.5746419425821316e-07,
2899
+ "loss": 1.2263,
2900
+ "step": 231000
2901
+ },
2902
+ {
2903
+ "epoch": 2.98,
2904
+ "learning_rate": 3.5022926827175906e-07,
2905
+ "loss": 1.2263,
2906
+ "step": 231500
2907
+ },
2908
+ {
2909
+ "epoch": 2.99,
2910
+ "learning_rate": 2.4299434228530495e-07,
2911
+ "loss": 1.2282,
2912
+ "step": 232000
2913
+ },
2914
+ {
2915
+ "epoch": 2.99,
2916
+ "learning_rate": 1.3575941629885087e-07,
2917
+ "loss": 1.2334,
2918
+ "step": 232500
2919
+ },
2920
+ {
2921
+ "epoch": 3.0,
2922
+ "learning_rate": 2.8524490312396787e-08,
2923
+ "loss": 1.224,
2924
+ "step": 233000
2925
+ },
2926
+ {
2927
+ "epoch": 3.0,
2928
+ "step": 233133,
2929
+ "total_flos": 4.975552182291128e+18,
2930
+ "train_loss": 1.3875968830903724,
2931
+ "train_runtime": 695161.7821,
2932
+ "train_samples_per_second": 10.732,
2933
+ "train_steps_per_second": 0.335
2934
+ }
2935
+ ],
2936
+ "max_steps": 233133,
2937
+ "num_train_epochs": 3,
2938
+ "total_flos": 4.975552182291128e+18,
2939
+ "trial_name": null,
2940
+ "trial_params": null
2941
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04af2ef2e138f3b7dddb143bdd1531a3b7605b3abf09d8f135a4d22fefb880b5
3
+ size 2863
vocab.json ADDED
The diff for this file is too large to render. See raw diff