Hugging Face's logo

jonasknobloch
/
gpt2-morf_s0-30-x-2_cx-en_00000-00009_50k

Text Generation
Transformers
Safetensors
English
gpt2
Generated from Trainer
Eval Results
Inference Endpoints
text-generation-inference
Model card Files Community
jonasknobloch commited on
Commit
384ca66
•
1 Parent(s): a3ca8b0

Initial commit

Browse files
Files changed (12) hide show
  1. README.md +101 -0
  2. all_results.json +15 -0
  3. config.json +31 -0
  4. eval_results.json +10 -0
  5. generation_config.json +6 -0
  6. model.safetensors +3 -0
  7. special_tokens_map.json +1 -0
  8. tokenizer.json +0 -0
  9. tokenizer_config.json +15 -0
  10. train_results.json +8 -0
  11. trainer_state.json +4143 -0
  12. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,101 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - generated_from_trainer
4
+ datasets:
5
+ - uonlp/CulturaX
6
+ metrics:
7
+ - accuracy
8
+ model-index:
9
+ - name: gpt2+morf_s0-30-x-2_cx-en_00000-00009_50k
10
+ results:
11
+ - task:
12
+ name: Causal Language Modeling
13
+ type: text-generation
14
+ dataset:
15
+ name: uonlp/CulturaX en
16
+ type: uonlp/CulturaX
17
+ args: en
18
+ metrics:
19
+ - name: Accuracy
20
+ type: accuracy
21
+ value: 0.4329592727693433
22
+ ---
23
+
24
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
25
+ should probably proofread and complete it, then remove this comment. -->
26
+
27
+ # gpt2+morf_s0-30-x-2_cx-en_00000-00009_50k
28
+
29
+ This model is a fine-tuned version of [](https://huggingface.co/) on the uonlp/CulturaX en dataset.
30
+ It achieves the following results on the evaluation set:
31
+ - Loss: 2.8423
32
+ - Accuracy: 0.4330
33
+
34
+ ## Model description
35
+
36
+ More information needed
37
+
38
+ ## Intended uses & limitations
39
+
40
+ More information needed
41
+
42
+ ## Training and evaluation data
43
+
44
+ More information needed
45
+
46
+ ## Training procedure
47
+
48
+ ### Training hyperparameters
49
+
50
+ The following hyperparameters were used during training:
51
+ - learning_rate: 5e-05
52
+ - train_batch_size: 64
53
+ - eval_batch_size: 64
54
+ - seed: 42
55
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
56
+ - lr_scheduler_type: linear
57
+ - num_epochs: 1.0
58
+
59
+ ### Training results
60
+
61
+ | Training Loss | Epoch | Step | Validation Loss | Accuracy |
62
+ |:-------------:|:-----:|:------:|:---------------:|:--------:|
63
+ | 3.6569 | 0.03 | 10000 | 3.5764 | 0.3502 |
64
+ | 3.4317 | 0.06 | 20000 | 3.3581 | 0.3727 |
65
+ | 3.3161 | 0.09 | 30000 | 3.2447 | 0.3848 |
66
+ | 3.2463 | 0.13 | 40000 | 3.1761 | 0.3924 |
67
+ | 3.1897 | 0.16 | 50000 | 3.1277 | 0.3977 |
68
+ | 3.152 | 0.19 | 60000 | 3.0910 | 0.4022 |
69
+ | 3.1341 | 0.22 | 70000 | 3.0575 | 0.4060 |
70
+ | 3.1006 | 0.25 | 80000 | 3.0363 | 0.4084 |
71
+ | 3.0806 | 0.28 | 90000 | 3.0118 | 0.4115 |
72
+ | 3.0555 | 0.31 | 100000 | 2.9919 | 0.4138 |
73
+ | 3.038 | 0.34 | 110000 | 2.9786 | 0.4156 |
74
+ | 3.0291 | 0.38 | 120000 | 2.9651 | 0.4171 |
75
+ | 3.0182 | 0.41 | 130000 | 2.9499 | 0.4191 |
76
+ | 3.0145 | 0.44 | 140000 | 2.9381 | 0.4205 |
77
+ | 2.9891 | 0.47 | 150000 | 2.9272 | 0.4219 |
78
+ | 2.9836 | 0.5 | 160000 | 2.9191 | 0.4230 |
79
+ | 2.9717 | 0.53 | 170000 | 2.9103 | 0.4241 |
80
+ | 2.9651 | 0.56 | 180000 | 2.9039 | 0.4250 |
81
+ | 2.9615 | 0.59 | 190000 | 2.8971 | 0.4258 |
82
+ | 2.9556 | 0.63 | 200000 | 2.8882 | 0.4269 |
83
+ | 2.9452 | 0.66 | 210000 | 2.8825 | 0.4277 |
84
+ | 2.9412 | 0.69 | 220000 | 2.8766 | 0.4284 |
85
+ | 2.9402 | 0.72 | 230000 | 2.8722 | 0.4290 |
86
+ | 2.9299 | 0.75 | 240000 | 2.8675 | 0.4296 |
87
+ | 2.9302 | 0.78 | 250000 | 2.8623 | 0.4304 |
88
+ | 2.9165 | 0.81 | 260000 | 2.8585 | 0.4308 |
89
+ | 2.915 | 0.84 | 270000 | 2.8537 | 0.4314 |
90
+ | 2.92 | 0.88 | 280000 | 2.8506 | 0.4319 |
91
+ | 2.9186 | 0.91 | 290000 | 2.8484 | 0.4321 |
92
+ | 2.9084 | 0.94 | 300000 | 2.8458 | 0.4325 |
93
+ | 2.9142 | 0.97 | 310000 | 2.8438 | 0.4327 |
94
+
95
+
96
+ ### Framework versions
97
+
98
+ - Transformers 4.37.1
99
+ - Pytorch 2.1.2+cu121
100
+ - Datasets 2.16.1
101
+ - Tokenizers 0.15.1
all_results.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_accuracy": 0.4329592727693433,
4
+ "eval_loss": 2.842284917831421,
5
+ "eval_runtime": 5974.983,
6
+ "eval_samples": 538312,
7
+ "eval_samples_per_second": 90.094,
8
+ "eval_steps_per_second": 1.408,
9
+ "perplexity": 17.15491836848711,
10
+ "train_loss": 3.072396659716232,
11
+ "train_runtime": 348731.7472,
12
+ "train_samples": 10229668,
13
+ "train_samples_per_second": 29.334,
14
+ "train_steps_per_second": 0.917
15
+ }
config.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_function": "gelu_new",
3
+ "architectures": [
4
+ "GPT2LMHeadModel"
5
+ ],
6
+ "attn_pdrop": 0.1,
7
+ "bos_token_id": 50256,
8
+ "embd_pdrop": 0.1,
9
+ "eos_token_id": 50256,
10
+ "initializer_range": 0.02,
11
+ "layer_norm_epsilon": 1e-05,
12
+ "model_type": "gpt2",
13
+ "n_embd": 768,
14
+ "n_head": 12,
15
+ "n_inner": null,
16
+ "n_layer": 12,
17
+ "n_positions": 1024,
18
+ "reorder_and_upcast_attn": false,
19
+ "resid_pdrop": 0.1,
20
+ "scale_attn_by_inverse_layer_idx": false,
21
+ "scale_attn_weights": true,
22
+ "summary_activation": null,
23
+ "summary_first_dropout": 0.1,
24
+ "summary_proj_to_labels": true,
25
+ "summary_type": "cls_index",
26
+ "summary_use_proj": true,
27
+ "torch_dtype": "float32",
28
+ "transformers_version": "4.37.1",
29
+ "use_cache": true,
30
+ "vocab_size": 50257
31
+ }
eval_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_accuracy": 0.4329592727693433,
4
+ "eval_loss": 2.842284917831421,
5
+ "eval_runtime": 5974.983,
6
+ "eval_samples": 538312,
7
+ "eval_samples_per_second": 90.094,
8
+ "eval_steps_per_second": 1.408,
9
+ "perplexity": 17.15491836848711
10
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 50256,
4
+ "eos_token_id": 50256,
5
+ "transformers_version": "4.37.1"
6
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51d2d4bb5511e2d2716d722127d2b5391b02f33f3635239a3109768624a53e32
3
+ size 497774208
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "50256": {
4
+ "content": "<|endoftext|>",
5
+ "lstrip": false,
6
+ "normalized": true,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ }
11
+ },
12
+ "clean_up_tokenization_spaces": true,
13
+ "model_max_length": 1000000000000000019884624838656,
14
+ "tokenizer_class": "PreTrainedTokenizerFast"
15
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 3.072396659716232,
4
+ "train_runtime": 348731.7472,
5
+ "train_samples": 10229668,
6
+ "train_samples_per_second": 29.334,
7
+ "train_steps_per_second": 0.917
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,4143 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 10000,
6
+ "global_step": 319678,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 4.992179630753446e-05,
14
+ "loss": 6.4174,
15
+ "step": 500
16
+ },
17
+ {
18
+ "epoch": 0.0,
19
+ "learning_rate": 4.984359261506892e-05,
20
+ "loss": 5.3568,
21
+ "step": 1000
22
+ },
23
+ {
24
+ "epoch": 0.0,
25
+ "learning_rate": 4.9765388922603374e-05,
26
+ "loss": 5.0234,
27
+ "step": 1500
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 4.968718523013783e-05,
32
+ "loss": 4.7982,
33
+ "step": 2000
34
+ },
35
+ {
36
+ "epoch": 0.01,
37
+ "learning_rate": 4.960898153767228e-05,
38
+ "loss": 4.5929,
39
+ "step": 2500
40
+ },
41
+ {
42
+ "epoch": 0.01,
43
+ "learning_rate": 4.9530777845206745e-05,
44
+ "loss": 4.4101,
45
+ "step": 3000
46
+ },
47
+ {
48
+ "epoch": 0.01,
49
+ "learning_rate": 4.94525741527412e-05,
50
+ "loss": 4.2672,
51
+ "step": 3500
52
+ },
53
+ {
54
+ "epoch": 0.01,
55
+ "learning_rate": 4.9374370460275654e-05,
56
+ "loss": 4.1536,
57
+ "step": 4000
58
+ },
59
+ {
60
+ "epoch": 0.01,
61
+ "learning_rate": 4.929616676781011e-05,
62
+ "loss": 4.0694,
63
+ "step": 4500
64
+ },
65
+ {
66
+ "epoch": 0.02,
67
+ "learning_rate": 4.921796307534457e-05,
68
+ "loss": 3.9953,
69
+ "step": 5000
70
+ },
71
+ {
72
+ "epoch": 0.02,
73
+ "learning_rate": 4.9139759382879026e-05,
74
+ "loss": 3.9374,
75
+ "step": 5500
76
+ },
77
+ {
78
+ "epoch": 0.02,
79
+ "learning_rate": 4.906155569041348e-05,
80
+ "loss": 3.8853,
81
+ "step": 6000
82
+ },
83
+ {
84
+ "epoch": 0.02,
85
+ "learning_rate": 4.8983351997947935e-05,
86
+ "loss": 3.8492,
87
+ "step": 6500
88
+ },
89
+ {
90
+ "epoch": 0.02,
91
+ "learning_rate": 4.89051483054824e-05,
92
+ "loss": 3.8114,
93
+ "step": 7000
94
+ },
95
+ {
96
+ "epoch": 0.02,
97
+ "learning_rate": 4.882694461301685e-05,
98
+ "loss": 3.7802,
99
+ "step": 7500
100
+ },
101
+ {
102
+ "epoch": 0.03,
103
+ "learning_rate": 4.8748740920551307e-05,
104
+ "loss": 3.7541,
105
+ "step": 8000
106
+ },
107
+ {
108
+ "epoch": 0.03,
109
+ "learning_rate": 4.867053722808577e-05,
110
+ "loss": 3.7294,
111
+ "step": 8500
112
+ },
113
+ {
114
+ "epoch": 0.03,
115
+ "learning_rate": 4.859233353562022e-05,
116
+ "loss": 3.704,
117
+ "step": 9000
118
+ },
119
+ {
120
+ "epoch": 0.03,
121
+ "learning_rate": 4.851412984315468e-05,
122
+ "loss": 3.6819,
123
+ "step": 9500
124
+ },
125
+ {
126
+ "epoch": 0.03,
127
+ "learning_rate": 4.843592615068913e-05,
128
+ "loss": 3.6569,
129
+ "step": 10000
130
+ },
131
+ {
132
+ "epoch": 0.03,
133
+ "eval_accuracy": 0.35023928642253593,
134
+ "eval_loss": 3.576359748840332,
135
+ "eval_runtime": 5980.3073,
136
+ "eval_samples_per_second": 90.014,
137
+ "eval_steps_per_second": 1.407,
138
+ "step": 10000
139
+ },
140
+ {
141
+ "epoch": 0.03,
142
+ "learning_rate": 4.8357722458223594e-05,
143
+ "loss": 3.6327,
144
+ "step": 10500
145
+ },
146
+ {
147
+ "epoch": 0.03,
148
+ "learning_rate": 4.827951876575805e-05,
149
+ "loss": 3.6243,
150
+ "step": 11000
151
+ },
152
+ {
153
+ "epoch": 0.04,
154
+ "learning_rate": 4.8201315073292504e-05,
155
+ "loss": 3.6073,
156
+ "step": 11500
157
+ },
158
+ {
159
+ "epoch": 0.04,
160
+ "learning_rate": 4.812311138082696e-05,
161
+ "loss": 3.5984,
162
+ "step": 12000
163
+ },
164
+ {
165
+ "epoch": 0.04,
166
+ "learning_rate": 4.804490768836142e-05,
167
+ "loss": 3.5788,
168
+ "step": 12500
169
+ },
170
+ {
171
+ "epoch": 0.04,
172
+ "learning_rate": 4.7966703995895875e-05,
173
+ "loss": 3.565,
174
+ "step": 13000
175
+ },
176
+ {
177
+ "epoch": 0.04,
178
+ "learning_rate": 4.788850030343033e-05,
179
+ "loss": 3.5536,
180
+ "step": 13500
181
+ },
182
+ {
183
+ "epoch": 0.04,
184
+ "learning_rate": 4.7810296610964785e-05,
185
+ "loss": 3.5403,
186
+ "step": 14000
187
+ },
188
+ {
189
+ "epoch": 0.05,
190
+ "learning_rate": 4.7732092918499246e-05,
191
+ "loss": 3.5292,
192
+ "step": 14500
193
+ },
194
+ {
195
+ "epoch": 0.05,
196
+ "learning_rate": 4.76538892260337e-05,
197
+ "loss": 3.5258,
198
+ "step": 15000
199
+ },
200
+ {
201
+ "epoch": 0.05,
202
+ "learning_rate": 4.7575685533568156e-05,
203
+ "loss": 3.5033,
204
+ "step": 15500
205
+ },
206
+ {
207
+ "epoch": 0.05,
208
+ "learning_rate": 4.749748184110261e-05,
209
+ "loss": 3.4912,
210
+ "step": 16000
211
+ },
212
+ {
213
+ "epoch": 0.05,
214
+ "learning_rate": 4.741927814863707e-05,
215
+ "loss": 3.4873,
216
+ "step": 16500
217
+ },
218
+ {
219
+ "epoch": 0.05,
220
+ "learning_rate": 4.734107445617153e-05,
221
+ "loss": 3.4821,
222
+ "step": 17000
223
+ },
224
+ {
225
+ "epoch": 0.05,
226
+ "learning_rate": 4.726287076370598e-05,
227
+ "loss": 3.4656,
228
+ "step": 17500
229
+ },
230
+ {
231
+ "epoch": 0.06,
232
+ "learning_rate": 4.7184667071240443e-05,
233
+ "loss": 3.4606,
234
+ "step": 18000
235
+ },
236
+ {
237
+ "epoch": 0.06,
238
+ "learning_rate": 4.71064633787749e-05,
239
+ "loss": 3.4492,
240
+ "step": 18500
241
+ },
242
+ {
243
+ "epoch": 0.06,
244
+ "learning_rate": 4.702825968630935e-05,
245
+ "loss": 3.4429,
246
+ "step": 19000
247
+ },
248
+ {
249
+ "epoch": 0.06,
250
+ "learning_rate": 4.695005599384381e-05,
251
+ "loss": 3.4329,
252
+ "step": 19500
253
+ },
254
+ {
255
+ "epoch": 0.06,
256
+ "learning_rate": 4.687185230137827e-05,
257
+ "loss": 3.4317,
258
+ "step": 20000
259
+ },
260
+ {
261
+ "epoch": 0.06,
262
+ "eval_accuracy": 0.37270295864352604,
263
+ "eval_loss": 3.358124256134033,
264
+ "eval_runtime": 5996.148,
265
+ "eval_samples_per_second": 89.776,
266
+ "eval_steps_per_second": 1.403,
267
+ "step": 20000
268
+ },
269
+ {
270
+ "epoch": 0.06,
271
+ "learning_rate": 4.6793648608912724e-05,
272
+ "loss": 3.4168,
273
+ "step": 20500
274
+ },
275
+ {
276
+ "epoch": 0.07,
277
+ "learning_rate": 4.671544491644718e-05,
278
+ "loss": 3.4053,
279
+ "step": 21000
280
+ },
281
+ {
282
+ "epoch": 0.07,
283
+ "learning_rate": 4.6637241223981634e-05,
284
+ "loss": 3.4086,
285
+ "step": 21500
286
+ },
287
+ {
288
+ "epoch": 0.07,
289
+ "learning_rate": 4.655903753151609e-05,
290
+ "loss": 3.4018,
291
+ "step": 22000
292
+ },
293
+ {
294
+ "epoch": 0.07,
295
+ "learning_rate": 4.648083383905055e-05,
296
+ "loss": 3.3929,
297
+ "step": 22500
298
+ },
299
+ {
300
+ "epoch": 0.07,
301
+ "learning_rate": 4.6402630146585005e-05,
302
+ "loss": 3.3869,
303
+ "step": 23000
304
+ },
305
+ {
306
+ "epoch": 0.07,
307
+ "learning_rate": 4.632442645411946e-05,
308
+ "loss": 3.3841,
309
+ "step": 23500
310
+ },
311
+ {
312
+ "epoch": 0.08,
313
+ "learning_rate": 4.6246222761653915e-05,
314
+ "loss": 3.3783,
315
+ "step": 24000
316
+ },
317
+ {
318
+ "epoch": 0.08,
319
+ "learning_rate": 4.6168019069188376e-05,
320
+ "loss": 3.3731,
321
+ "step": 24500
322
+ },
323
+ {
324
+ "epoch": 0.08,
325
+ "learning_rate": 4.608981537672283e-05,
326
+ "loss": 3.3619,
327
+ "step": 25000
328
+ },
329
+ {
330
+ "epoch": 0.08,
331
+ "learning_rate": 4.6011611684257286e-05,
332
+ "loss": 3.361,
333
+ "step": 25500
334
+ },
335
+ {
336
+ "epoch": 0.08,
337
+ "learning_rate": 4.593340799179174e-05,
338
+ "loss": 3.3522,
339
+ "step": 26000
340
+ },
341
+ {
342
+ "epoch": 0.08,
343
+ "learning_rate": 4.5855204299326196e-05,
344
+ "loss": 3.3518,
345
+ "step": 26500
346
+ },
347
+ {
348
+ "epoch": 0.08,
349
+ "learning_rate": 4.577700060686066e-05,
350
+ "loss": 3.3413,
351
+ "step": 27000
352
+ },
353
+ {
354
+ "epoch": 0.09,
355
+ "learning_rate": 4.569879691439511e-05,
356
+ "loss": 3.3311,
357
+ "step": 27500
358
+ },
359
+ {
360
+ "epoch": 0.09,
361
+ "learning_rate": 4.562059322192957e-05,
362
+ "loss": 3.3358,
363
+ "step": 28000
364
+ },
365
+ {
366
+ "epoch": 0.09,
367
+ "learning_rate": 4.554238952946402e-05,
368
+ "loss": 3.3264,
369
+ "step": 28500
370
+ },
371
+ {
372
+ "epoch": 0.09,
373
+ "learning_rate": 4.5464185836998476e-05,
374
+ "loss": 3.3236,
375
+ "step": 29000
376
+ },
377
+ {
378
+ "epoch": 0.09,
379
+ "learning_rate": 4.538598214453294e-05,
380
+ "loss": 3.3148,
381
+ "step": 29500
382
+ },
383
+ {
384
+ "epoch": 0.09,
385
+ "learning_rate": 4.530777845206739e-05,
386
+ "loss": 3.3161,
387
+ "step": 30000
388
+ },
389
+ {
390
+ "epoch": 0.09,
391
+ "eval_accuracy": 0.3848112637589684,
392
+ "eval_loss": 3.2447316646575928,
393
+ "eval_runtime": 6006.5513,
394
+ "eval_samples_per_second": 89.621,
395
+ "eval_steps_per_second": 1.4,
396
+ "step": 30000
397
+ },
398
+ {
399
+ "epoch": 0.1,
400
+ "learning_rate": 4.522957475960185e-05,
401
+ "loss": 3.3048,
402
+ "step": 30500
403
+ },
404
+ {
405
+ "epoch": 0.1,
406
+ "learning_rate": 4.51513710671363e-05,
407
+ "loss": 3.3062,
408
+ "step": 31000
409
+ },
410
+ {
411
+ "epoch": 0.1,
412
+ "learning_rate": 4.5073167374670764e-05,
413
+ "loss": 3.3039,
414
+ "step": 31500
415
+ },
416
+ {
417
+ "epoch": 0.1,
418
+ "learning_rate": 4.499496368220522e-05,
419
+ "loss": 3.3031,
420
+ "step": 32000
421
+ },
422
+ {
423
+ "epoch": 0.1,
424
+ "learning_rate": 4.4916759989739674e-05,
425
+ "loss": 3.2953,
426
+ "step": 32500
427
+ },
428
+ {
429
+ "epoch": 0.1,
430
+ "learning_rate": 4.4838556297274135e-05,
431
+ "loss": 3.2892,
432
+ "step": 33000
433
+ },
434
+ {
435
+ "epoch": 0.1,
436
+ "learning_rate": 4.476035260480859e-05,
437
+ "loss": 3.2839,
438
+ "step": 33500
439
+ },
440
+ {
441
+ "epoch": 0.11,
442
+ "learning_rate": 4.4682148912343045e-05,
443
+ "loss": 3.281,
444
+ "step": 34000
445
+ },
446
+ {
447
+ "epoch": 0.11,
448
+ "learning_rate": 4.46039452198775e-05,
449
+ "loss": 3.2831,
450
+ "step": 34500
451
+ },
452
+ {
453
+ "epoch": 0.11,
454
+ "learning_rate": 4.452574152741196e-05,
455
+ "loss": 3.2739,
456
+ "step": 35000
457
+ },
458
+ {
459
+ "epoch": 0.11,
460
+ "learning_rate": 4.4447537834946416e-05,
461
+ "loss": 3.2801,
462
+ "step": 35500
463
+ },
464
+ {
465
+ "epoch": 0.11,
466
+ "learning_rate": 4.436933414248087e-05,
467
+ "loss": 3.2715,
468
+ "step": 36000
469
+ },
470
+ {
471
+ "epoch": 0.11,
472
+ "learning_rate": 4.4291130450015326e-05,
473
+ "loss": 3.2685,
474
+ "step": 36500
475
+ },
476
+ {
477
+ "epoch": 0.12,
478
+ "learning_rate": 4.421292675754979e-05,
479
+ "loss": 3.2653,
480
+ "step": 37000
481
+ },
482
+ {
483
+ "epoch": 0.12,
484
+ "learning_rate": 4.413472306508424e-05,
485
+ "loss": 3.2563,
486
+ "step": 37500
487
+ },
488
+ {
489
+ "epoch": 0.12,
490
+ "learning_rate": 4.40565193726187e-05,
491
+ "loss": 3.2579,
492
+ "step": 38000
493
+ },
494
+ {
495
+ "epoch": 0.12,
496
+ "learning_rate": 4.397831568015315e-05,
497
+ "loss": 3.2497,
498
+ "step": 38500
499
+ },
500
+ {
501
+ "epoch": 0.12,
502
+ "learning_rate": 4.390011198768761e-05,
503
+ "loss": 3.2465,
504
+ "step": 39000
505
+ },
506
+ {
507
+ "epoch": 0.12,
508
+ "learning_rate": 4.382190829522207e-05,
509
+ "loss": 3.2437,
510
+ "step": 39500
511
+ },
512
+ {
513
+ "epoch": 0.13,
514
+ "learning_rate": 4.374370460275652e-05,
515
+ "loss": 3.2463,
516
+ "step": 40000
517
+ },
518
+ {
519
+ "epoch": 0.13,
520
+ "eval_accuracy": 0.39235570262450464,
521
+ "eval_loss": 3.176072835922241,
522
+ "eval_runtime": 6008.1585,
523
+ "eval_samples_per_second": 89.597,
524
+ "eval_steps_per_second": 1.4,
525
+ "step": 40000
526
+ },
527
+ {
528
+ "epoch": 0.13,
529
+ "learning_rate": 4.3665500910290985e-05,
530
+ "loss": 3.2465,
531
+ "step": 40500
532
+ },
533
+ {
534
+ "epoch": 0.13,
535
+ "learning_rate": 4.358729721782544e-05,
536
+ "loss": 3.2423,
537
+ "step": 41000
538
+ },
539
+ {
540
+ "epoch": 0.13,
541
+ "learning_rate": 4.3509093525359894e-05,
542
+ "loss": 3.2381,
543
+ "step": 41500
544
+ },
545
+ {
546
+ "epoch": 0.13,
547
+ "learning_rate": 4.343088983289435e-05,
548
+ "loss": 3.2327,
549
+ "step": 42000
550
+ },
551
+ {
552
+ "epoch": 0.13,
553
+ "learning_rate": 4.335268614042881e-05,
554
+ "loss": 3.2307,
555
+ "step": 42500
556
+ },
557
+ {
558
+ "epoch": 0.13,
559
+ "learning_rate": 4.3274482447963265e-05,
560
+ "loss": 3.2348,
561
+ "step": 43000
562
+ },
563
+ {
564
+ "epoch": 0.14,
565
+ "learning_rate": 4.319627875549772e-05,
566
+ "loss": 3.2245,
567
+ "step": 43500
568
+ },
569
+ {
570
+ "epoch": 0.14,
571
+ "learning_rate": 4.3118075063032175e-05,
572
+ "loss": 3.2262,
573
+ "step": 44000
574
+ },
575
+ {
576
+ "epoch": 0.14,
577
+ "learning_rate": 4.3039871370566637e-05,
578
+ "loss": 3.2169,
579
+ "step": 44500
580
+ },
581
+ {
582
+ "epoch": 0.14,
583
+ "learning_rate": 4.296166767810109e-05,
584
+ "loss": 3.2157,
585
+ "step": 45000
586
+ },
587
+ {
588
+ "epoch": 0.14,
589
+ "learning_rate": 4.2883463985635546e-05,
590
+ "loss": 3.2128,
591
+ "step": 45500
592
+ },
593
+ {
594
+ "epoch": 0.14,
595
+ "learning_rate": 4.280526029317e-05,
596
+ "loss": 3.2125,
597
+ "step": 46000
598
+ },
599
+ {
600
+ "epoch": 0.15,
601
+ "learning_rate": 4.272705660070446e-05,
602
+ "loss": 3.2079,
603
+ "step": 46500
604
+ },
605
+ {
606
+ "epoch": 0.15,
607
+ "learning_rate": 4.264885290823892e-05,
608
+ "loss": 3.2102,
609
+ "step": 47000
610
+ },
611
+ {
612
+ "epoch": 0.15,
613
+ "learning_rate": 4.257064921577337e-05,
614
+ "loss": 3.2079,
615
+ "step": 47500
616
+ },
617
+ {
618
+ "epoch": 0.15,
619
+ "learning_rate": 4.249244552330783e-05,
620
+ "loss": 3.2032,
621
+ "step": 48000
622
+ },
623
+ {
624
+ "epoch": 0.15,
625
+ "learning_rate": 4.241424183084229e-05,
626
+ "loss": 3.2009,
627
+ "step": 48500
628
+ },
629
+ {
630
+ "epoch": 0.15,
631
+ "learning_rate": 4.2336038138376743e-05,
632
+ "loss": 3.2049,
633
+ "step": 49000
634
+ },
635
+ {
636
+ "epoch": 0.15,
637
+ "learning_rate": 4.22578344459112e-05,
638
+ "loss": 3.1951,
639
+ "step": 49500
640
+ },
641
+ {
642
+ "epoch": 0.16,
643
+ "learning_rate": 4.217963075344566e-05,
644
+ "loss": 3.1897,
645
+ "step": 50000
646
+ },
647
+ {
648
+ "epoch": 0.16,
649
+ "eval_accuracy": 0.39768394188345635,
650
+ "eval_loss": 3.127730369567871,
651
+ "eval_runtime": 6028.4791,
652
+ "eval_samples_per_second": 89.295,
653
+ "eval_steps_per_second": 1.395,
654
+ "step": 50000
655
+ },
656
+ {
657
+ "epoch": 0.16,
658
+ "learning_rate": 4.2101427060980115e-05,
659
+ "loss": 3.1863,
660
+ "step": 50500
661
+ },
662
+ {
663
+ "epoch": 0.16,
664
+ "learning_rate": 4.202322336851457e-05,
665
+ "loss": 3.1942,
666
+ "step": 51000
667
+ },
668
+ {
669
+ "epoch": 0.16,
670
+ "learning_rate": 4.1945019676049024e-05,
671
+ "loss": 3.192,
672
+ "step": 51500
673
+ },
674
+ {
675
+ "epoch": 0.16,
676
+ "learning_rate": 4.1866815983583486e-05,
677
+ "loss": 3.1954,
678
+ "step": 52000
679
+ },
680
+ {
681
+ "epoch": 0.16,
682
+ "learning_rate": 4.178861229111794e-05,
683
+ "loss": 3.1879,
684
+ "step": 52500
685
+ },
686
+ {
687
+ "epoch": 0.17,
688
+ "learning_rate": 4.1710408598652395e-05,
689
+ "loss": 3.1788,
690
+ "step": 53000
691
+ },
692
+ {
693
+ "epoch": 0.17,
694
+ "learning_rate": 4.163220490618685e-05,
695
+ "loss": 3.1776,
696
+ "step": 53500
697
+ },
698
+ {
699
+ "epoch": 0.17,
700
+ "learning_rate": 4.155400121372131e-05,
701
+ "loss": 3.1808,
702
+ "step": 54000
703
+ },
704
+ {
705
+ "epoch": 0.17,
706
+ "learning_rate": 4.147579752125577e-05,
707
+ "loss": 3.1724,
708
+ "step": 54500
709
+ },
710
+ {
711
+ "epoch": 0.17,
712
+ "learning_rate": 4.139759382879022e-05,
713
+ "loss": 3.1746,
714
+ "step": 55000
715
+ },
716
+ {
717
+ "epoch": 0.17,
718
+ "learning_rate": 4.1319390136324676e-05,
719
+ "loss": 3.1694,
720
+ "step": 55500
721
+ },
722
+ {
723
+ "epoch": 0.18,
724
+ "learning_rate": 4.124118644385914e-05,
725
+ "loss": 3.1685,
726
+ "step": 56000
727
+ },
728
+ {
729
+ "epoch": 0.18,
730
+ "learning_rate": 4.116298275139359e-05,
731
+ "loss": 3.1661,
732
+ "step": 56500
733
+ },
734
+ {
735
+ "epoch": 0.18,
736
+ "learning_rate": 4.108477905892805e-05,
737
+ "loss": 3.1644,
738
+ "step": 57000
739
+ },
740
+ {
741
+ "epoch": 0.18,
742
+ "learning_rate": 4.100657536646251e-05,
743
+ "loss": 3.1659,
744
+ "step": 57500
745
+ },
746
+ {
747
+ "epoch": 0.18,
748
+ "learning_rate": 4.0928371673996964e-05,
749
+ "loss": 3.1625,
750
+ "step": 58000
751
+ },
752
+ {
753
+ "epoch": 0.18,
754
+ "learning_rate": 4.085016798153142e-05,
755
+ "loss": 3.1587,
756
+ "step": 58500
757
+ },
758
+ {
759
+ "epoch": 0.18,
760
+ "learning_rate": 4.0771964289065874e-05,
761
+ "loss": 3.1563,
762
+ "step": 59000
763
+ },
764
+ {
765
+ "epoch": 0.19,
766
+ "learning_rate": 4.0693760596600335e-05,
767
+ "loss": 3.155,
768
+ "step": 59500
769
+ },
770
+ {
771
+ "epoch": 0.19,
772
+ "learning_rate": 4.061555690413479e-05,
773
+ "loss": 3.152,
774
+ "step": 60000
775
+ },
776
+ {
777
+ "epoch": 0.19,
778
+ "eval_accuracy": 0.40218440985366416,
779
+ "eval_loss": 3.091034412384033,
780
+ "eval_runtime": 5997.8965,
781
+ "eval_samples_per_second": 89.75,
782
+ "eval_steps_per_second": 1.402,
783
+ "step": 60000
784
+ },
785
+ {
786
+ "epoch": 0.19,
787
+ "learning_rate": 4.0537353211669245e-05,
788
+ "loss": 3.1516,
789
+ "step": 60500
790
+ },
791
+ {
792
+ "epoch": 0.19,
793
+ "learning_rate": 4.04591495192037e-05,
794
+ "loss": 3.1454,
795
+ "step": 61000
796
+ },
797
+ {
798
+ "epoch": 0.19,
799
+ "learning_rate": 4.038094582673816e-05,
800
+ "loss": 3.1488,
801
+ "step": 61500
802
+ },
803
+ {
804
+ "epoch": 0.19,
805
+ "learning_rate": 4.0302742134272616e-05,
806
+ "loss": 3.1497,
807
+ "step": 62000
808
+ },
809
+ {
810
+ "epoch": 0.2,
811
+ "learning_rate": 4.022453844180707e-05,
812
+ "loss": 3.1484,
813
+ "step": 62500
814
+ },
815
+ {
816
+ "epoch": 0.2,
817
+ "learning_rate": 4.0146334749341526e-05,
818
+ "loss": 3.1441,
819
+ "step": 63000
820
+ },
821
+ {
822
+ "epoch": 0.2,
823
+ "learning_rate": 4.006813105687599e-05,
824
+ "loss": 3.1412,
825
+ "step": 63500
826
+ },
827
+ {
828
+ "epoch": 0.2,
829
+ "learning_rate": 3.998992736441044e-05,
830
+ "loss": 3.1416,
831
+ "step": 64000
832
+ },
833
+ {
834
+ "epoch": 0.2,
835
+ "learning_rate": 3.99117236719449e-05,
836
+ "loss": 3.1436,
837
+ "step": 64500
838
+ },
839
+ {
840
+ "epoch": 0.2,
841
+ "learning_rate": 3.983351997947936e-05,
842
+ "loss": 3.1409,
843
+ "step": 65000
844
+ },
845
+ {
846
+ "epoch": 0.2,
847
+ "learning_rate": 3.975531628701381e-05,
848
+ "loss": 3.1392,
849
+ "step": 65500
850
+ },
851
+ {
852
+ "epoch": 0.21,
853
+ "learning_rate": 3.967711259454827e-05,
854
+ "loss": 3.1381,
855
+ "step": 66000
856
+ },
857
+ {
858
+ "epoch": 0.21,
859
+ "learning_rate": 3.959890890208272e-05,
860
+ "loss": 3.1345,
861
+ "step": 66500
862
+ },
863
+ {
864
+ "epoch": 0.21,
865
+ "learning_rate": 3.9520705209617184e-05,
866
+ "loss": 3.131,
867
+ "step": 67000
868
+ },
869
+ {
870
+ "epoch": 0.21,
871
+ "learning_rate": 3.944250151715164e-05,
872
+ "loss": 3.1339,
873
+ "step": 67500
874
+ },
875
+ {
876
+ "epoch": 0.21,
877
+ "learning_rate": 3.9364297824686094e-05,
878
+ "loss": 3.1295,
879
+ "step": 68000
880
+ },
881
+ {
882
+ "epoch": 0.21,
883
+ "learning_rate": 3.928609413222055e-05,
884
+ "loss": 3.128,
885
+ "step": 68500
886
+ },
887
+ {
888
+ "epoch": 0.22,
889
+ "learning_rate": 3.920789043975501e-05,
890
+ "loss": 3.1308,
891
+ "step": 69000
892
+ },
893
+ {
894
+ "epoch": 0.22,
895
+ "learning_rate": 3.9129686747289465e-05,
896
+ "loss": 3.1157,
897
+ "step": 69500
898
+ },
899
+ {
900
+ "epoch": 0.22,
901
+ "learning_rate": 3.905148305482392e-05,
902
+ "loss": 3.1341,
903
+ "step": 70000
904
+ },
905
+ {
906
+ "epoch": 0.22,
907
+ "eval_accuracy": 0.4060193438823364,
908
+ "eval_loss": 3.057492256164551,
909
+ "eval_runtime": 6003.3195,
910
+ "eval_samples_per_second": 89.669,
911
+ "eval_steps_per_second": 1.401,
912
+ "step": 70000
913
+ },
914
+ {
915
+ "epoch": 0.22,
916
+ "learning_rate": 3.8973279362358375e-05,
917
+ "loss": 3.1319,
918
+ "step": 70500
919
+ },
920
+ {
921
+ "epoch": 0.22,
922
+ "learning_rate": 3.8895075669892836e-05,
923
+ "loss": 3.1217,
924
+ "step": 71000
925
+ },
926
+ {
927
+ "epoch": 0.22,
928
+ "learning_rate": 3.881687197742729e-05,
929
+ "loss": 3.1223,
930
+ "step": 71500
931
+ },
932
+ {
933
+ "epoch": 0.23,
934
+ "learning_rate": 3.8738668284961746e-05,
935
+ "loss": 3.125,
936
+ "step": 72000
937
+ },
938
+ {
939
+ "epoch": 0.23,
940
+ "learning_rate": 3.86604645924962e-05,
941
+ "loss": 3.1223,
942
+ "step": 72500
943
+ },
944
+ {
945
+ "epoch": 0.23,
946
+ "learning_rate": 3.8582260900030656e-05,
947
+ "loss": 3.1109,
948
+ "step": 73000
949
+ },
950
+ {
951
+ "epoch": 0.23,
952
+ "learning_rate": 3.850405720756512e-05,
953
+ "loss": 3.1138,
954
+ "step": 73500
955
+ },
956
+ {
957
+ "epoch": 0.23,
958
+ "learning_rate": 3.842585351509957e-05,
959
+ "loss": 3.1168,
960
+ "step": 74000
961
+ },
962
+ {
963
+ "epoch": 0.23,
964
+ "learning_rate": 3.834764982263403e-05,
965
+ "loss": 3.1137,
966
+ "step": 74500
967
+ },
968
+ {
969
+ "epoch": 0.23,
970
+ "learning_rate": 3.826944613016848e-05,
971
+ "loss": 3.1085,
972
+ "step": 75000
973
+ },
974
+ {
975
+ "epoch": 0.24,
976
+ "learning_rate": 3.8191242437702937e-05,
977
+ "loss": 3.1145,
978
+ "step": 75500
979
+ },
980
+ {
981
+ "epoch": 0.24,
982
+ "learning_rate": 3.81130387452374e-05,
983
+ "loss": 3.1045,
984
+ "step": 76000
985
+ },
986
+ {
987
+ "epoch": 0.24,
988
+ "learning_rate": 3.803483505277185e-05,
989
+ "loss": 3.1078,
990
+ "step": 76500
991
+ },
992
+ {
993
+ "epoch": 0.24,
994
+ "learning_rate": 3.795663136030631e-05,
995
+ "loss": 3.1054,
996
+ "step": 77000
997
+ },
998
+ {
999
+ "epoch": 0.24,
1000
+ "learning_rate": 3.787842766784076e-05,
1001
+ "loss": 3.1047,
1002
+ "step": 77500
1003
+ },
1004
+ {
1005
+ "epoch": 0.24,
1006
+ "learning_rate": 3.780022397537522e-05,
1007
+ "loss": 3.1038,
1008
+ "step": 78000
1009
+ },
1010
+ {
1011
+ "epoch": 0.25,
1012
+ "learning_rate": 3.772202028290968e-05,
1013
+ "loss": 3.1034,
1014
+ "step": 78500
1015
+ },
1016
+ {
1017
+ "epoch": 0.25,
1018
+ "learning_rate": 3.7643816590444134e-05,
1019
+ "loss": 3.0953,
1020
+ "step": 79000
1021
+ },
1022
+ {
1023
+ "epoch": 0.25,
1024
+ "learning_rate": 3.756561289797859e-05,
1025
+ "loss": 3.1006,
1026
+ "step": 79500
1027
+ },
1028
+ {
1029
+ "epoch": 0.25,
1030
+ "learning_rate": 3.7487409205513043e-05,
1031
+ "loss": 3.1006,
1032
+ "step": 80000
1033
+ },
1034
+ {
1035
+ "epoch": 0.25,
1036
+ "eval_accuracy": 0.4084469897262718,
1037
+ "eval_loss": 3.0362584590911865,
1038
+ "eval_runtime": 5986.8174,
1039
+ "eval_samples_per_second": 89.916,
1040
+ "eval_steps_per_second": 1.405,
1041
+ "step": 80000
1042
+ },
1043
+ {
1044
+ "epoch": 0.25,
1045
+ "learning_rate": 3.7409205513047505e-05,
1046
+ "loss": 3.0934,
1047
+ "step": 80500
1048
+ },
1049
+ {
1050
+ "epoch": 0.25,
1051
+ "learning_rate": 3.733100182058196e-05,
1052
+ "loss": 3.1024,
1053
+ "step": 81000
1054
+ },
1055
+ {
1056
+ "epoch": 0.25,
1057
+ "learning_rate": 3.7252798128116415e-05,
1058
+ "loss": 3.0957,
1059
+ "step": 81500
1060
+ },
1061
+ {
1062
+ "epoch": 0.26,
1063
+ "learning_rate": 3.7174594435650876e-05,
1064
+ "loss": 3.1003,
1065
+ "step": 82000
1066
+ },
1067
+ {
1068
+ "epoch": 0.26,
1069
+ "learning_rate": 3.709639074318533e-05,
1070
+ "loss": 3.0987,
1071
+ "step": 82500
1072
+ },
1073
+ {
1074
+ "epoch": 0.26,
1075
+ "learning_rate": 3.7018187050719786e-05,
1076
+ "loss": 3.0952,
1077
+ "step": 83000
1078
+ },
1079
+ {
1080
+ "epoch": 0.26,
1081
+ "learning_rate": 3.693998335825424e-05,
1082
+ "loss": 3.0971,
1083
+ "step": 83500
1084
+ },
1085
+ {
1086
+ "epoch": 0.26,
1087
+ "learning_rate": 3.68617796657887e-05,
1088
+ "loss": 3.0974,
1089
+ "step": 84000
1090
+ },
1091
+ {
1092
+ "epoch": 0.26,
1093
+ "learning_rate": 3.678357597332316e-05,
1094
+ "loss": 3.0896,
1095
+ "step": 84500
1096
+ },
1097
+ {
1098
+ "epoch": 0.27,
1099
+ "learning_rate": 3.670537228085761e-05,
1100
+ "loss": 3.0862,
1101
+ "step": 85000
1102
+ },
1103
+ {
1104
+ "epoch": 0.27,
1105
+ "learning_rate": 3.662716858839207e-05,
1106
+ "loss": 3.0809,
1107
+ "step": 85500
1108
+ },
1109
+ {
1110
+ "epoch": 0.27,
1111
+ "learning_rate": 3.654896489592653e-05,
1112
+ "loss": 3.088,
1113
+ "step": 86000
1114
+ },
1115
+ {
1116
+ "epoch": 0.27,
1117
+ "learning_rate": 3.647076120346098e-05,
1118
+ "loss": 3.0818,
1119
+ "step": 86500
1120
+ },
1121
+ {
1122
+ "epoch": 0.27,
1123
+ "learning_rate": 3.639255751099544e-05,
1124
+ "loss": 3.0715,
1125
+ "step": 87000
1126
+ },
1127
+ {
1128
+ "epoch": 0.27,
1129
+ "learning_rate": 3.631435381852989e-05,
1130
+ "loss": 3.0869,
1131
+ "step": 87500
1132
+ },
1133
+ {
1134
+ "epoch": 0.28,
1135
+ "learning_rate": 3.6236150126064354e-05,
1136
+ "loss": 3.0856,
1137
+ "step": 88000
1138
+ },
1139
+ {
1140
+ "epoch": 0.28,
1141
+ "learning_rate": 3.615794643359881e-05,
1142
+ "loss": 3.0812,
1143
+ "step": 88500
1144
+ },
1145
+ {
1146
+ "epoch": 0.28,
1147
+ "learning_rate": 3.6079742741133264e-05,
1148
+ "loss": 3.0768,
1149
+ "step": 89000
1150
+ },
1151
+ {
1152
+ "epoch": 0.28,
1153
+ "learning_rate": 3.6001539048667726e-05,
1154
+ "loss": 3.0823,
1155
+ "step": 89500
1156
+ },
1157
+ {
1158
+ "epoch": 0.28,
1159
+ "learning_rate": 3.592333535620218e-05,
1160
+ "loss": 3.0806,
1161
+ "step": 90000
1162
+ },
1163
+ {
1164
+ "epoch": 0.28,
1165
+ "eval_accuracy": 0.41145725219591245,
1166
+ "eval_loss": 3.011796236038208,
1167
+ "eval_runtime": 5969.2265,
1168
+ "eval_samples_per_second": 90.181,
1169
+ "eval_steps_per_second": 1.409,
1170
+ "step": 90000
1171
+ },
1172
+ {
1173
+ "epoch": 0.28,
1174
+ "learning_rate": 3.5845131663736635e-05,
1175
+ "loss": 3.0726,
1176
+ "step": 90500
1177
+ },
1178
+ {
1179
+ "epoch": 0.28,
1180
+ "learning_rate": 3.576692797127109e-05,
1181
+ "loss": 3.0725,
1182
+ "step": 91000
1183
+ },
1184
+ {
1185
+ "epoch": 0.29,
1186
+ "learning_rate": 3.568872427880555e-05,
1187
+ "loss": 3.0681,
1188
+ "step": 91500
1189
+ },
1190
+ {
1191
+ "epoch": 0.29,
1192
+ "learning_rate": 3.5610520586340006e-05,
1193
+ "loss": 3.0711,
1194
+ "step": 92000
1195
+ },
1196
+ {
1197
+ "epoch": 0.29,
1198
+ "learning_rate": 3.553231689387446e-05,
1199
+ "loss": 3.0798,
1200
+ "step": 92500
1201
+ },
1202
+ {
1203
+ "epoch": 0.29,
1204
+ "learning_rate": 3.5454113201408916e-05,
1205
+ "loss": 3.07,
1206
+ "step": 93000
1207
+ },
1208
+ {
1209
+ "epoch": 0.29,
1210
+ "learning_rate": 3.537590950894338e-05,
1211
+ "loss": 3.0729,
1212
+ "step": 93500
1213
+ },
1214
+ {
1215
+ "epoch": 0.29,
1216
+ "learning_rate": 3.529770581647783e-05,
1217
+ "loss": 3.0668,
1218
+ "step": 94000
1219
+ },
1220
+ {
1221
+ "epoch": 0.3,
1222
+ "learning_rate": 3.521950212401229e-05,
1223
+ "loss": 3.0722,
1224
+ "step": 94500
1225
+ },
1226
+ {
1227
+ "epoch": 0.3,
1228
+ "learning_rate": 3.514129843154674e-05,
1229
+ "loss": 3.0681,
1230
+ "step": 95000
1231
+ },
1232
+ {
1233
+ "epoch": 0.3,
1234
+ "learning_rate": 3.5063094739081204e-05,
1235
+ "loss": 3.0712,
1236
+ "step": 95500
1237
+ },
1238
+ {
1239
+ "epoch": 0.3,
1240
+ "learning_rate": 3.498489104661566e-05,
1241
+ "loss": 3.0684,
1242
+ "step": 96000
1243
+ },
1244
+ {
1245
+ "epoch": 0.3,
1246
+ "learning_rate": 3.490668735415011e-05,
1247
+ "loss": 3.0683,
1248
+ "step": 96500
1249
+ },
1250
+ {
1251
+ "epoch": 0.3,
1252
+ "learning_rate": 3.4828483661684575e-05,
1253
+ "loss": 3.0669,
1254
+ "step": 97000
1255
+ },
1256
+ {
1257
+ "epoch": 0.3,
1258
+ "learning_rate": 3.475027996921903e-05,
1259
+ "loss": 3.0587,
1260
+ "step": 97500
1261
+ },
1262
+ {
1263
+ "epoch": 0.31,
1264
+ "learning_rate": 3.4672076276753484e-05,
1265
+ "loss": 3.0624,
1266
+ "step": 98000
1267
+ },
1268
+ {
1269
+ "epoch": 0.31,
1270
+ "learning_rate": 3.459387258428794e-05,
1271
+ "loss": 3.0589,
1272
+ "step": 98500
1273
+ },
1274
+ {
1275
+ "epoch": 0.31,
1276
+ "learning_rate": 3.45156688918224e-05,
1277
+ "loss": 3.056,
1278
+ "step": 99000
1279
+ },
1280
+ {
1281
+ "epoch": 0.31,
1282
+ "learning_rate": 3.4437465199356856e-05,
1283
+ "loss": 3.0616,
1284
+ "step": 99500
1285
+ },
1286
+ {
1287
+ "epoch": 0.31,
1288
+ "learning_rate": 3.435926150689131e-05,
1289
+ "loss": 3.0555,
1290
+ "step": 100000
1291
+ },
1292
+ {
1293
+ "epoch": 0.31,
1294
+ "eval_accuracy": 0.41383883246085473,
1295
+ "eval_loss": 2.9919075965881348,
1296
+ "eval_runtime": 5973.2974,
1297
+ "eval_samples_per_second": 90.12,
1298
+ "eval_steps_per_second": 1.408,
1299
+ "step": 100000
1300
+ },
1301
+ {
1302
+ "epoch": 0.31,
1303
+ "learning_rate": 3.4281057814425765e-05,
1304
+ "loss": 3.0601,
1305
+ "step": 100500
1306
+ },
1307
+ {
1308
+ "epoch": 0.32,
1309
+ "learning_rate": 3.420285412196023e-05,
1310
+ "loss": 3.0567,
1311
+ "step": 101000
1312
+ },
1313
+ {
1314
+ "epoch": 0.32,
1315
+ "learning_rate": 3.412465042949468e-05,
1316
+ "loss": 3.0644,
1317
+ "step": 101500
1318
+ },
1319
+ {
1320
+ "epoch": 0.32,
1321
+ "learning_rate": 3.4046446737029137e-05,
1322
+ "loss": 3.0515,
1323
+ "step": 102000
1324
+ },
1325
+ {
1326
+ "epoch": 0.32,
1327
+ "learning_rate": 3.396824304456359e-05,
1328
+ "loss": 3.0567,
1329
+ "step": 102500
1330
+ },
1331
+ {
1332
+ "epoch": 0.32,
1333
+ "learning_rate": 3.389003935209805e-05,
1334
+ "loss": 3.0588,
1335
+ "step": 103000
1336
+ },
1337
+ {
1338
+ "epoch": 0.32,
1339
+ "learning_rate": 3.381183565963251e-05,
1340
+ "loss": 3.0492,
1341
+ "step": 103500
1342
+ },
1343
+ {
1344
+ "epoch": 0.33,
1345
+ "learning_rate": 3.373363196716696e-05,
1346
+ "loss": 3.0471,
1347
+ "step": 104000
1348
+ },
1349
+ {
1350
+ "epoch": 0.33,
1351
+ "learning_rate": 3.3655428274701424e-05,
1352
+ "loss": 3.0544,
1353
+ "step": 104500
1354
+ },
1355
+ {
1356
+ "epoch": 0.33,
1357
+ "learning_rate": 3.357722458223588e-05,
1358
+ "loss": 3.0546,
1359
+ "step": 105000
1360
+ },
1361
+ {
1362
+ "epoch": 0.33,
1363
+ "learning_rate": 3.3499020889770334e-05,
1364
+ "loss": 3.0469,
1365
+ "step": 105500
1366
+ },
1367
+ {
1368
+ "epoch": 0.33,
1369
+ "learning_rate": 3.342081719730479e-05,
1370
+ "loss": 3.0514,
1371
+ "step": 106000
1372
+ },
1373
+ {
1374
+ "epoch": 0.33,
1375
+ "learning_rate": 3.334261350483925e-05,
1376
+ "loss": 3.0494,
1377
+ "step": 106500
1378
+ },
1379
+ {
1380
+ "epoch": 0.33,
1381
+ "learning_rate": 3.3264409812373705e-05,
1382
+ "loss": 3.0522,
1383
+ "step": 107000
1384
+ },
1385
+ {
1386
+ "epoch": 0.34,
1387
+ "learning_rate": 3.318620611990816e-05,
1388
+ "loss": 3.0447,
1389
+ "step": 107500
1390
+ },
1391
+ {
1392
+ "epoch": 0.34,
1393
+ "learning_rate": 3.3108002427442615e-05,
1394
+ "loss": 3.0454,
1395
+ "step": 108000
1396
+ },
1397
+ {
1398
+ "epoch": 0.34,
1399
+ "learning_rate": 3.3029798734977076e-05,
1400
+ "loss": 3.0495,
1401
+ "step": 108500
1402
+ },
1403
+ {
1404
+ "epoch": 0.34,
1405
+ "learning_rate": 3.295159504251153e-05,
1406
+ "loss": 3.0431,
1407
+ "step": 109000
1408
+ },
1409
+ {
1410
+ "epoch": 0.34,
1411
+ "learning_rate": 3.2873391350045986e-05,
1412
+ "loss": 3.038,
1413
+ "step": 109500
1414
+ },
1415
+ {
1416
+ "epoch": 0.34,
1417
+ "learning_rate": 3.279518765758044e-05,
1418
+ "loss": 3.038,
1419
+ "step": 110000
1420
+ },
1421
+ {
1422
+ "epoch": 0.34,
1423
+ "eval_accuracy": 0.41561401879437854,
1424
+ "eval_loss": 2.9786083698272705,
1425
+ "eval_runtime": 6001.372,
1426
+ "eval_samples_per_second": 89.698,
1427
+ "eval_steps_per_second": 1.402,
1428
+ "step": 110000
1429
+ },
1430
+ {
1431
+ "epoch": 0.35,
1432
+ "learning_rate": 3.27169839651149e-05,
1433
+ "loss": 3.0403,
1434
+ "step": 110500
1435
+ },
1436
+ {
1437
+ "epoch": 0.35,
1438
+ "learning_rate": 3.263878027264936e-05,
1439
+ "loss": 3.0437,
1440
+ "step": 111000
1441
+ },
1442
+ {
1443
+ "epoch": 0.35,
1444
+ "learning_rate": 3.256057658018381e-05,
1445
+ "loss": 3.0337,
1446
+ "step": 111500
1447
+ },
1448
+ {
1449
+ "epoch": 0.35,
1450
+ "learning_rate": 3.248237288771827e-05,
1451
+ "loss": 3.0308,
1452
+ "step": 112000
1453
+ },
1454
+ {
1455
+ "epoch": 0.35,
1456
+ "learning_rate": 3.240416919525273e-05,
1457
+ "loss": 3.0356,
1458
+ "step": 112500
1459
+ },
1460
+ {
1461
+ "epoch": 0.35,
1462
+ "learning_rate": 3.232596550278718e-05,
1463
+ "loss": 3.0378,
1464
+ "step": 113000
1465
+ },
1466
+ {
1467
+ "epoch": 0.36,
1468
+ "learning_rate": 3.224776181032164e-05,
1469
+ "loss": 3.0383,
1470
+ "step": 113500
1471
+ },
1472
+ {
1473
+ "epoch": 0.36,
1474
+ "learning_rate": 3.21695581178561e-05,
1475
+ "loss": 3.0391,
1476
+ "step": 114000
1477
+ },
1478
+ {
1479
+ "epoch": 0.36,
1480
+ "learning_rate": 3.2091354425390554e-05,
1481
+ "loss": 3.0335,
1482
+ "step": 114500
1483
+ },
1484
+ {
1485
+ "epoch": 0.36,
1486
+ "learning_rate": 3.201315073292501e-05,
1487
+ "loss": 3.0356,
1488
+ "step": 115000
1489
+ },
1490
+ {
1491
+ "epoch": 0.36,
1492
+ "learning_rate": 3.1934947040459464e-05,
1493
+ "loss": 3.0354,
1494
+ "step": 115500
1495
+ },
1496
+ {
1497
+ "epoch": 0.36,
1498
+ "learning_rate": 3.1856743347993925e-05,
1499
+ "loss": 3.0346,
1500
+ "step": 116000
1501
+ },
1502
+ {
1503
+ "epoch": 0.36,
1504
+ "learning_rate": 3.177853965552838e-05,
1505
+ "loss": 3.0311,
1506
+ "step": 116500
1507
+ },
1508
+ {
1509
+ "epoch": 0.37,
1510
+ "learning_rate": 3.1700335963062835e-05,
1511
+ "loss": 3.0334,
1512
+ "step": 117000
1513
+ },
1514
+ {
1515
+ "epoch": 0.37,
1516
+ "learning_rate": 3.162213227059729e-05,
1517
+ "loss": 3.0225,
1518
+ "step": 117500
1519
+ },
1520
+ {
1521
+ "epoch": 0.37,
1522
+ "learning_rate": 3.154392857813175e-05,
1523
+ "loss": 3.0252,
1524
+ "step": 118000
1525
+ },
1526
+ {
1527
+ "epoch": 0.37,
1528
+ "learning_rate": 3.1465724885666206e-05,
1529
+ "loss": 3.0335,
1530
+ "step": 118500
1531
+ },
1532
+ {
1533
+ "epoch": 0.37,
1534
+ "learning_rate": 3.138752119320066e-05,
1535
+ "loss": 3.03,
1536
+ "step": 119000
1537
+ },
1538
+ {
1539
+ "epoch": 0.37,
1540
+ "learning_rate": 3.1309317500735116e-05,
1541
+ "loss": 3.0303,
1542
+ "step": 119500
1543
+ },
1544
+ {
1545
+ "epoch": 0.38,
1546
+ "learning_rate": 3.123111380826958e-05,
1547
+ "loss": 3.0291,
1548
+ "step": 120000
1549
+ },
1550
+ {
1551
+ "epoch": 0.38,
1552
+ "eval_accuracy": 0.41712760392004566,
1553
+ "eval_loss": 2.9651412963867188,
1554
+ "eval_runtime": 5971.0497,
1555
+ "eval_samples_per_second": 90.154,
1556
+ "eval_steps_per_second": 1.409,
1557
+ "step": 120000
1558
+ },
1559
+ {
1560
+ "epoch": 0.38,
1561
+ "learning_rate": 3.115291011580403e-05,
1562
+ "loss": 3.0349,
1563
+ "step": 120500
1564
+ },
1565
+ {
1566
+ "epoch": 0.38,
1567
+ "learning_rate": 3.107470642333849e-05,
1568
+ "loss": 3.0318,
1569
+ "step": 121000
1570
+ },
1571
+ {
1572
+ "epoch": 0.38,
1573
+ "learning_rate": 3.099650273087294e-05,
1574
+ "loss": 3.0156,
1575
+ "step": 121500
1576
+ },
1577
+ {
1578
+ "epoch": 0.38,
1579
+ "learning_rate": 3.09182990384074e-05,
1580
+ "loss": 3.0226,
1581
+ "step": 122000
1582
+ },
1583
+ {
1584
+ "epoch": 0.38,
1585
+ "learning_rate": 3.084009534594186e-05,
1586
+ "loss": 3.0269,
1587
+ "step": 122500
1588
+ },
1589
+ {
1590
+ "epoch": 0.38,
1591
+ "learning_rate": 3.076189165347631e-05,
1592
+ "loss": 3.0239,
1593
+ "step": 123000
1594
+ },
1595
+ {
1596
+ "epoch": 0.39,
1597
+ "learning_rate": 3.068368796101077e-05,
1598
+ "loss": 3.0242,
1599
+ "step": 123500
1600
+ },
1601
+ {
1602
+ "epoch": 0.39,
1603
+ "learning_rate": 3.060548426854522e-05,
1604
+ "loss": 3.0267,
1605
+ "step": 124000
1606
+ },
1607
+ {
1608
+ "epoch": 0.39,
1609
+ "learning_rate": 3.052728057607968e-05,
1610
+ "loss": 3.0167,
1611
+ "step": 124500
1612
+ },
1613
+ {
1614
+ "epoch": 0.39,
1615
+ "learning_rate": 3.0449076883614136e-05,
1616
+ "loss": 3.0179,
1617
+ "step": 125000
1618
+ },
1619
+ {
1620
+ "epoch": 0.39,
1621
+ "learning_rate": 3.0370873191148597e-05,
1622
+ "loss": 3.0145,
1623
+ "step": 125500
1624
+ },
1625
+ {
1626
+ "epoch": 0.39,
1627
+ "learning_rate": 3.0292669498683052e-05,
1628
+ "loss": 3.0234,
1629
+ "step": 126000
1630
+ },
1631
+ {
1632
+ "epoch": 0.4,
1633
+ "learning_rate": 3.0214465806217507e-05,
1634
+ "loss": 3.0254,
1635
+ "step": 126500
1636
+ },
1637
+ {
1638
+ "epoch": 0.4,
1639
+ "learning_rate": 3.0136262113751962e-05,
1640
+ "loss": 3.0209,
1641
+ "step": 127000
1642
+ },
1643
+ {
1644
+ "epoch": 0.4,
1645
+ "learning_rate": 3.0058058421286423e-05,
1646
+ "loss": 3.0188,
1647
+ "step": 127500
1648
+ },
1649
+ {
1650
+ "epoch": 0.4,
1651
+ "learning_rate": 2.9979854728820878e-05,
1652
+ "loss": 3.0142,
1653
+ "step": 128000
1654
+ },
1655
+ {
1656
+ "epoch": 0.4,
1657
+ "learning_rate": 2.9901651036355333e-05,
1658
+ "loss": 3.0214,
1659
+ "step": 128500
1660
+ },
1661
+ {
1662
+ "epoch": 0.4,
1663
+ "learning_rate": 2.9823447343889795e-05,
1664
+ "loss": 3.0254,
1665
+ "step": 129000
1666
+ },
1667
+ {
1668
+ "epoch": 0.41,
1669
+ "learning_rate": 2.974524365142425e-05,
1670
+ "loss": 3.0214,
1671
+ "step": 129500
1672
+ },
1673
+ {
1674
+ "epoch": 0.41,
1675
+ "learning_rate": 2.9667039958958704e-05,
1676
+ "loss": 3.0182,
1677
+ "step": 130000
1678
+ },
1679
+ {
1680
+ "epoch": 0.41,
1681
+ "eval_accuracy": 0.4190574389830463,
1682
+ "eval_loss": 2.949920415878296,
1683
+ "eval_runtime": 5975.917,
1684
+ "eval_samples_per_second": 90.08,
1685
+ "eval_steps_per_second": 1.408,
1686
+ "step": 130000
1687
+ },
1688
+ {
1689
+ "epoch": 0.41,
1690
+ "learning_rate": 2.958883626649316e-05,
1691
+ "loss": 3.0109,
1692
+ "step": 130500
1693
+ },
1694
+ {
1695
+ "epoch": 0.41,
1696
+ "learning_rate": 2.951063257402762e-05,
1697
+ "loss": 3.0117,
1698
+ "step": 131000
1699
+ },
1700
+ {
1701
+ "epoch": 0.41,
1702
+ "learning_rate": 2.9432428881562075e-05,
1703
+ "loss": 3.0183,
1704
+ "step": 131500
1705
+ },
1706
+ {
1707
+ "epoch": 0.41,
1708
+ "learning_rate": 2.935422518909653e-05,
1709
+ "loss": 3.0173,
1710
+ "step": 132000
1711
+ },
1712
+ {
1713
+ "epoch": 0.41,
1714
+ "learning_rate": 2.9276021496630985e-05,
1715
+ "loss": 3.0113,
1716
+ "step": 132500
1717
+ },
1718
+ {
1719
+ "epoch": 0.42,
1720
+ "learning_rate": 2.9197817804165443e-05,
1721
+ "loss": 3.0125,
1722
+ "step": 133000
1723
+ },
1724
+ {
1725
+ "epoch": 0.42,
1726
+ "learning_rate": 2.91196141116999e-05,
1727
+ "loss": 3.0126,
1728
+ "step": 133500
1729
+ },
1730
+ {
1731
+ "epoch": 0.42,
1732
+ "learning_rate": 2.9041410419234356e-05,
1733
+ "loss": 3.0149,
1734
+ "step": 134000
1735
+ },
1736
+ {
1737
+ "epoch": 0.42,
1738
+ "learning_rate": 2.896320672676881e-05,
1739
+ "loss": 3.0148,
1740
+ "step": 134500
1741
+ },
1742
+ {
1743
+ "epoch": 0.42,
1744
+ "learning_rate": 2.888500303430327e-05,
1745
+ "loss": 3.0172,
1746
+ "step": 135000
1747
+ },
1748
+ {
1749
+ "epoch": 0.42,
1750
+ "learning_rate": 2.8806799341837724e-05,
1751
+ "loss": 3.0114,
1752
+ "step": 135500
1753
+ },
1754
+ {
1755
+ "epoch": 0.43,
1756
+ "learning_rate": 2.8728595649372182e-05,
1757
+ "loss": 3.0123,
1758
+ "step": 136000
1759
+ },
1760
+ {
1761
+ "epoch": 0.43,
1762
+ "learning_rate": 2.8650391956906637e-05,
1763
+ "loss": 3.0065,
1764
+ "step": 136500
1765
+ },
1766
+ {
1767
+ "epoch": 0.43,
1768
+ "learning_rate": 2.8572188264441095e-05,
1769
+ "loss": 3.0085,
1770
+ "step": 137000
1771
+ },
1772
+ {
1773
+ "epoch": 0.43,
1774
+ "learning_rate": 2.849398457197555e-05,
1775
+ "loss": 3.0126,
1776
+ "step": 137500
1777
+ },
1778
+ {
1779
+ "epoch": 0.43,
1780
+ "learning_rate": 2.8415780879510005e-05,
1781
+ "loss": 3.0092,
1782
+ "step": 138000
1783
+ },
1784
+ {
1785
+ "epoch": 0.43,
1786
+ "learning_rate": 2.8337577187044467e-05,
1787
+ "loss": 3.0072,
1788
+ "step": 138500
1789
+ },
1790
+ {
1791
+ "epoch": 0.43,
1792
+ "learning_rate": 2.825937349457892e-05,
1793
+ "loss": 3.0088,
1794
+ "step": 139000
1795
+ },
1796
+ {
1797
+ "epoch": 0.44,
1798
+ "learning_rate": 2.8181169802113376e-05,
1799
+ "loss": 3.0117,
1800
+ "step": 139500
1801
+ },
1802
+ {
1803
+ "epoch": 0.44,
1804
+ "learning_rate": 2.810296610964783e-05,
1805
+ "loss": 3.0145,
1806
+ "step": 140000
1807
+ },
1808
+ {
1809
+ "epoch": 0.44,
1810
+ "eval_accuracy": 0.42045388991709604,
1811
+ "eval_loss": 2.93807053565979,
1812
+ "eval_runtime": 6002.1748,
1813
+ "eval_samples_per_second": 89.686,
1814
+ "eval_steps_per_second": 1.401,
1815
+ "step": 140000
1816
+ },
1817
+ {
1818
+ "epoch": 0.44,
1819
+ "learning_rate": 2.8024762417182293e-05,
1820
+ "loss": 3.0022,
1821
+ "step": 140500
1822
+ },
1823
+ {
1824
+ "epoch": 0.44,
1825
+ "learning_rate": 2.7946558724716747e-05,
1826
+ "loss": 3.0051,
1827
+ "step": 141000
1828
+ },
1829
+ {
1830
+ "epoch": 0.44,
1831
+ "learning_rate": 2.7868355032251202e-05,
1832
+ "loss": 2.9975,
1833
+ "step": 141500
1834
+ },
1835
+ {
1836
+ "epoch": 0.44,
1837
+ "learning_rate": 2.7790151339785657e-05,
1838
+ "loss": 3.0062,
1839
+ "step": 142000
1840
+ },
1841
+ {
1842
+ "epoch": 0.45,
1843
+ "learning_rate": 2.771194764732012e-05,
1844
+ "loss": 3.0083,
1845
+ "step": 142500
1846
+ },
1847
+ {
1848
+ "epoch": 0.45,
1849
+ "learning_rate": 2.7633743954854573e-05,
1850
+ "loss": 3.0046,
1851
+ "step": 143000
1852
+ },
1853
+ {
1854
+ "epoch": 0.45,
1855
+ "learning_rate": 2.7555540262389028e-05,
1856
+ "loss": 3.0013,
1857
+ "step": 143500
1858
+ },
1859
+ {
1860
+ "epoch": 0.45,
1861
+ "learning_rate": 2.7477336569923483e-05,
1862
+ "loss": 2.9973,
1863
+ "step": 144000
1864
+ },
1865
+ {
1866
+ "epoch": 0.45,
1867
+ "learning_rate": 2.7399132877457945e-05,
1868
+ "loss": 2.9979,
1869
+ "step": 144500
1870
+ },
1871
+ {
1872
+ "epoch": 0.45,
1873
+ "learning_rate": 2.73209291849924e-05,
1874
+ "loss": 3.0,
1875
+ "step": 145000
1876
+ },
1877
+ {
1878
+ "epoch": 0.46,
1879
+ "learning_rate": 2.7242725492526854e-05,
1880
+ "loss": 3.0043,
1881
+ "step": 145500
1882
+ },
1883
+ {
1884
+ "epoch": 0.46,
1885
+ "learning_rate": 2.7164521800061316e-05,
1886
+ "loss": 2.9987,
1887
+ "step": 146000
1888
+ },
1889
+ {
1890
+ "epoch": 0.46,
1891
+ "learning_rate": 2.708631810759577e-05,
1892
+ "loss": 2.9951,
1893
+ "step": 146500
1894
+ },
1895
+ {
1896
+ "epoch": 0.46,
1897
+ "learning_rate": 2.7008114415130226e-05,
1898
+ "loss": 2.9972,
1899
+ "step": 147000
1900
+ },
1901
+ {
1902
+ "epoch": 0.46,
1903
+ "learning_rate": 2.692991072266468e-05,
1904
+ "loss": 2.9968,
1905
+ "step": 147500
1906
+ },
1907
+ {
1908
+ "epoch": 0.46,
1909
+ "learning_rate": 2.6851707030199142e-05,
1910
+ "loss": 2.995,
1911
+ "step": 148000
1912
+ },
1913
+ {
1914
+ "epoch": 0.46,
1915
+ "learning_rate": 2.6773503337733597e-05,
1916
+ "loss": 2.9944,
1917
+ "step": 148500
1918
+ },
1919
+ {
1920
+ "epoch": 0.47,
1921
+ "learning_rate": 2.669529964526805e-05,
1922
+ "loss": 2.9944,
1923
+ "step": 149000
1924
+ },
1925
+ {
1926
+ "epoch": 0.47,
1927
+ "learning_rate": 2.6617095952802506e-05,
1928
+ "loss": 2.9975,
1929
+ "step": 149500
1930
+ },
1931
+ {
1932
+ "epoch": 0.47,
1933
+ "learning_rate": 2.6538892260336968e-05,
1934
+ "loss": 2.9891,
1935
+ "step": 150000
1936
+ },
1937
+ {
1938
+ "epoch": 0.47,
1939
+ "eval_accuracy": 0.42188237683918567,
1940
+ "eval_loss": 2.9272468090057373,
1941
+ "eval_runtime": 5980.3151,
1942
+ "eval_samples_per_second": 90.014,
1943
+ "eval_steps_per_second": 1.407,
1944
+ "step": 150000
1945
+ },
1946
+ {
1947
+ "epoch": 0.47,
1948
+ "learning_rate": 2.6460688567871423e-05,
1949
+ "loss": 3.007,
1950
+ "step": 150500
1951
+ },
1952
+ {
1953
+ "epoch": 0.47,
1954
+ "learning_rate": 2.6382484875405878e-05,
1955
+ "loss": 2.9883,
1956
+ "step": 151000
1957
+ },
1958
+ {
1959
+ "epoch": 0.47,
1960
+ "learning_rate": 2.6304281182940332e-05,
1961
+ "loss": 2.9949,
1962
+ "step": 151500
1963
+ },
1964
+ {
1965
+ "epoch": 0.48,
1966
+ "learning_rate": 2.6226077490474794e-05,
1967
+ "loss": 2.994,
1968
+ "step": 152000
1969
+ },
1970
+ {
1971
+ "epoch": 0.48,
1972
+ "learning_rate": 2.614787379800925e-05,
1973
+ "loss": 2.9923,
1974
+ "step": 152500
1975
+ },
1976
+ {
1977
+ "epoch": 0.48,
1978
+ "learning_rate": 2.6069670105543704e-05,
1979
+ "loss": 2.9863,
1980
+ "step": 153000
1981
+ },
1982
+ {
1983
+ "epoch": 0.48,
1984
+ "learning_rate": 2.5991466413078165e-05,
1985
+ "loss": 2.9943,
1986
+ "step": 153500
1987
+ },
1988
+ {
1989
+ "epoch": 0.48,
1990
+ "learning_rate": 2.591326272061262e-05,
1991
+ "loss": 2.9937,
1992
+ "step": 154000
1993
+ },
1994
+ {
1995
+ "epoch": 0.48,
1996
+ "learning_rate": 2.5835059028147075e-05,
1997
+ "loss": 2.9929,
1998
+ "step": 154500
1999
+ },
2000
+ {
2001
+ "epoch": 0.48,
2002
+ "learning_rate": 2.575685533568153e-05,
2003
+ "loss": 2.9884,
2004
+ "step": 155000
2005
+ },
2006
+ {
2007
+ "epoch": 0.49,
2008
+ "learning_rate": 2.567865164321599e-05,
2009
+ "loss": 2.9874,
2010
+ "step": 155500
2011
+ },
2012
+ {
2013
+ "epoch": 0.49,
2014
+ "learning_rate": 2.5600447950750446e-05,
2015
+ "loss": 2.9952,
2016
+ "step": 156000
2017
+ },
2018
+ {
2019
+ "epoch": 0.49,
2020
+ "learning_rate": 2.55222442582849e-05,
2021
+ "loss": 2.9878,
2022
+ "step": 156500
2023
+ },
2024
+ {
2025
+ "epoch": 0.49,
2026
+ "learning_rate": 2.5444040565819356e-05,
2027
+ "loss": 2.9886,
2028
+ "step": 157000
2029
+ },
2030
+ {
2031
+ "epoch": 0.49,
2032
+ "learning_rate": 2.5365836873353814e-05,
2033
+ "loss": 2.9845,
2034
+ "step": 157500
2035
+ },
2036
+ {
2037
+ "epoch": 0.49,
2038
+ "learning_rate": 2.5287633180888272e-05,
2039
+ "loss": 2.993,
2040
+ "step": 158000
2041
+ },
2042
+ {
2043
+ "epoch": 0.5,
2044
+ "learning_rate": 2.5209429488422727e-05,
2045
+ "loss": 2.9941,
2046
+ "step": 158500
2047
+ },
2048
+ {
2049
+ "epoch": 0.5,
2050
+ "learning_rate": 2.513122579595718e-05,
2051
+ "loss": 2.9907,
2052
+ "step": 159000
2053
+ },
2054
+ {
2055
+ "epoch": 0.5,
2056
+ "learning_rate": 2.505302210349164e-05,
2057
+ "loss": 2.9851,
2058
+ "step": 159500
2059
+ },
2060
+ {
2061
+ "epoch": 0.5,
2062
+ "learning_rate": 2.4974818411026095e-05,
2063
+ "loss": 2.9836,
2064
+ "step": 160000
2065
+ },
2066
+ {
2067
+ "epoch": 0.5,
2068
+ "eval_accuracy": 0.4229763453615049,
2069
+ "eval_loss": 2.919142723083496,
2070
+ "eval_runtime": 5984.1342,
2071
+ "eval_samples_per_second": 89.957,
2072
+ "eval_steps_per_second": 1.406,
2073
+ "step": 160000
2074
+ },
2075
+ {
2076
+ "epoch": 0.5,
2077
+ "learning_rate": 2.4896614718560553e-05,
2078
+ "loss": 2.9821,
2079
+ "step": 160500
2080
+ },
2081
+ {
2082
+ "epoch": 0.5,
2083
+ "learning_rate": 2.4818411026095008e-05,
2084
+ "loss": 2.9832,
2085
+ "step": 161000
2086
+ },
2087
+ {
2088
+ "epoch": 0.51,
2089
+ "learning_rate": 2.4740207333629466e-05,
2090
+ "loss": 2.9875,
2091
+ "step": 161500
2092
+ },
2093
+ {
2094
+ "epoch": 0.51,
2095
+ "learning_rate": 2.466200364116392e-05,
2096
+ "loss": 2.9932,
2097
+ "step": 162000
2098
+ },
2099
+ {
2100
+ "epoch": 0.51,
2101
+ "learning_rate": 2.458379994869838e-05,
2102
+ "loss": 2.9803,
2103
+ "step": 162500
2104
+ },
2105
+ {
2106
+ "epoch": 0.51,
2107
+ "learning_rate": 2.4505596256232834e-05,
2108
+ "loss": 2.9877,
2109
+ "step": 163000
2110
+ },
2111
+ {
2112
+ "epoch": 0.51,
2113
+ "learning_rate": 2.4427392563767292e-05,
2114
+ "loss": 2.9899,
2115
+ "step": 163500
2116
+ },
2117
+ {
2118
+ "epoch": 0.51,
2119
+ "learning_rate": 2.4349188871301747e-05,
2120
+ "loss": 2.9806,
2121
+ "step": 164000
2122
+ },
2123
+ {
2124
+ "epoch": 0.51,
2125
+ "learning_rate": 2.4270985178836205e-05,
2126
+ "loss": 2.9802,
2127
+ "step": 164500
2128
+ },
2129
+ {
2130
+ "epoch": 0.52,
2131
+ "learning_rate": 2.419278148637066e-05,
2132
+ "loss": 2.9822,
2133
+ "step": 165000
2134
+ },
2135
+ {
2136
+ "epoch": 0.52,
2137
+ "learning_rate": 2.4114577793905118e-05,
2138
+ "loss": 2.9787,
2139
+ "step": 165500
2140
+ },
2141
+ {
2142
+ "epoch": 0.52,
2143
+ "learning_rate": 2.4036374101439573e-05,
2144
+ "loss": 2.9782,
2145
+ "step": 166000
2146
+ },
2147
+ {
2148
+ "epoch": 0.52,
2149
+ "learning_rate": 2.395817040897403e-05,
2150
+ "loss": 2.9797,
2151
+ "step": 166500
2152
+ },
2153
+ {
2154
+ "epoch": 0.52,
2155
+ "learning_rate": 2.3879966716508486e-05,
2156
+ "loss": 2.9812,
2157
+ "step": 167000
2158
+ },
2159
+ {
2160
+ "epoch": 0.52,
2161
+ "learning_rate": 2.3801763024042944e-05,
2162
+ "loss": 2.9824,
2163
+ "step": 167500
2164
+ },
2165
+ {
2166
+ "epoch": 0.53,
2167
+ "learning_rate": 2.37235593315774e-05,
2168
+ "loss": 2.9816,
2169
+ "step": 168000
2170
+ },
2171
+ {
2172
+ "epoch": 0.53,
2173
+ "learning_rate": 2.3645355639111857e-05,
2174
+ "loss": 2.9752,
2175
+ "step": 168500
2176
+ },
2177
+ {
2178
+ "epoch": 0.53,
2179
+ "learning_rate": 2.3567151946646315e-05,
2180
+ "loss": 2.9867,
2181
+ "step": 169000
2182
+ },
2183
+ {
2184
+ "epoch": 0.53,
2185
+ "learning_rate": 2.348894825418077e-05,
2186
+ "loss": 2.9804,
2187
+ "step": 169500
2188
+ },
2189
+ {
2190
+ "epoch": 0.53,
2191
+ "learning_rate": 2.3410744561715228e-05,
2192
+ "loss": 2.9717,
2193
+ "step": 170000
2194
+ },
2195
+ {
2196
+ "epoch": 0.53,
2197
+ "eval_accuracy": 0.4241124752924122,
2198
+ "eval_loss": 2.9102535247802734,
2199
+ "eval_runtime": 5992.1438,
2200
+ "eval_samples_per_second": 89.836,
2201
+ "eval_steps_per_second": 1.404,
2202
+ "step": 170000
2203
+ },
2204
+ {
2205
+ "epoch": 0.53,
2206
+ "learning_rate": 2.3332540869249683e-05,
2207
+ "loss": 2.9745,
2208
+ "step": 170500
2209
+ },
2210
+ {
2211
+ "epoch": 0.53,
2212
+ "learning_rate": 2.325433717678414e-05,
2213
+ "loss": 2.9788,
2214
+ "step": 171000
2215
+ },
2216
+ {
2217
+ "epoch": 0.54,
2218
+ "learning_rate": 2.3176133484318596e-05,
2219
+ "loss": 2.9801,
2220
+ "step": 171500
2221
+ },
2222
+ {
2223
+ "epoch": 0.54,
2224
+ "learning_rate": 2.3097929791853054e-05,
2225
+ "loss": 2.9712,
2226
+ "step": 172000
2227
+ },
2228
+ {
2229
+ "epoch": 0.54,
2230
+ "learning_rate": 2.301972609938751e-05,
2231
+ "loss": 2.9767,
2232
+ "step": 172500
2233
+ },
2234
+ {
2235
+ "epoch": 0.54,
2236
+ "learning_rate": 2.2941522406921967e-05,
2237
+ "loss": 2.9699,
2238
+ "step": 173000
2239
+ },
2240
+ {
2241
+ "epoch": 0.54,
2242
+ "learning_rate": 2.2863318714456422e-05,
2243
+ "loss": 2.9698,
2244
+ "step": 173500
2245
+ },
2246
+ {
2247
+ "epoch": 0.54,
2248
+ "learning_rate": 2.278511502199088e-05,
2249
+ "loss": 2.9742,
2250
+ "step": 174000
2251
+ },
2252
+ {
2253
+ "epoch": 0.55,
2254
+ "learning_rate": 2.2706911329525335e-05,
2255
+ "loss": 2.9816,
2256
+ "step": 174500
2257
+ },
2258
+ {
2259
+ "epoch": 0.55,
2260
+ "learning_rate": 2.2628707637059793e-05,
2261
+ "loss": 2.9703,
2262
+ "step": 175000
2263
+ },
2264
+ {
2265
+ "epoch": 0.55,
2266
+ "learning_rate": 2.2550503944594248e-05,
2267
+ "loss": 2.9699,
2268
+ "step": 175500
2269
+ },
2270
+ {
2271
+ "epoch": 0.55,
2272
+ "learning_rate": 2.2472300252128706e-05,
2273
+ "loss": 2.9715,
2274
+ "step": 176000
2275
+ },
2276
+ {
2277
+ "epoch": 0.55,
2278
+ "learning_rate": 2.239409655966316e-05,
2279
+ "loss": 2.9706,
2280
+ "step": 176500
2281
+ },
2282
+ {
2283
+ "epoch": 0.55,
2284
+ "learning_rate": 2.231589286719762e-05,
2285
+ "loss": 2.9689,
2286
+ "step": 177000
2287
+ },
2288
+ {
2289
+ "epoch": 0.56,
2290
+ "learning_rate": 2.2237689174732077e-05,
2291
+ "loss": 2.9728,
2292
+ "step": 177500
2293
+ },
2294
+ {
2295
+ "epoch": 0.56,
2296
+ "learning_rate": 2.2159485482266532e-05,
2297
+ "loss": 2.9702,
2298
+ "step": 178000
2299
+ },
2300
+ {
2301
+ "epoch": 0.56,
2302
+ "learning_rate": 2.208128178980099e-05,
2303
+ "loss": 2.9703,
2304
+ "step": 178500
2305
+ },
2306
+ {
2307
+ "epoch": 0.56,
2308
+ "learning_rate": 2.2003078097335445e-05,
2309
+ "loss": 2.967,
2310
+ "step": 179000
2311
+ },
2312
+ {
2313
+ "epoch": 0.56,
2314
+ "learning_rate": 2.1924874404869904e-05,
2315
+ "loss": 2.9689,
2316
+ "step": 179500
2317
+ },
2318
+ {
2319
+ "epoch": 0.56,
2320
+ "learning_rate": 2.184667071240436e-05,
2321
+ "loss": 2.9651,
2322
+ "step": 180000
2323
+ },
2324
+ {
2325
+ "epoch": 0.56,
2326
+ "eval_accuracy": 0.4249954188646057,
2327
+ "eval_loss": 2.9038655757904053,
2328
+ "eval_runtime": 5968.0597,
2329
+ "eval_samples_per_second": 90.199,
2330
+ "eval_steps_per_second": 1.41,
2331
+ "step": 180000
2332
+ },
2333
+ {
2334
+ "epoch": 0.56,
2335
+ "learning_rate": 2.1768467019938817e-05,
2336
+ "loss": 2.9695,
2337
+ "step": 180500
2338
+ },
2339
+ {
2340
+ "epoch": 0.57,
2341
+ "learning_rate": 2.169026332747327e-05,
2342
+ "loss": 2.9705,
2343
+ "step": 181000
2344
+ },
2345
+ {
2346
+ "epoch": 0.57,
2347
+ "learning_rate": 2.161205963500773e-05,
2348
+ "loss": 2.9677,
2349
+ "step": 181500
2350
+ },
2351
+ {
2352
+ "epoch": 0.57,
2353
+ "learning_rate": 2.1533855942542184e-05,
2354
+ "loss": 2.9684,
2355
+ "step": 182000
2356
+ },
2357
+ {
2358
+ "epoch": 0.57,
2359
+ "learning_rate": 2.1455652250076643e-05,
2360
+ "loss": 2.9638,
2361
+ "step": 182500
2362
+ },
2363
+ {
2364
+ "epoch": 0.57,
2365
+ "learning_rate": 2.1377448557611097e-05,
2366
+ "loss": 2.9618,
2367
+ "step": 183000
2368
+ },
2369
+ {
2370
+ "epoch": 0.57,
2371
+ "learning_rate": 2.1299244865145556e-05,
2372
+ "loss": 2.9713,
2373
+ "step": 183500
2374
+ },
2375
+ {
2376
+ "epoch": 0.58,
2377
+ "learning_rate": 2.122104117268001e-05,
2378
+ "loss": 2.9715,
2379
+ "step": 184000
2380
+ },
2381
+ {
2382
+ "epoch": 0.58,
2383
+ "learning_rate": 2.1142837480214465e-05,
2384
+ "loss": 2.963,
2385
+ "step": 184500
2386
+ },
2387
+ {
2388
+ "epoch": 0.58,
2389
+ "learning_rate": 2.1064633787748923e-05,
2390
+ "loss": 2.9695,
2391
+ "step": 185000
2392
+ },
2393
+ {
2394
+ "epoch": 0.58,
2395
+ "learning_rate": 2.0986430095283378e-05,
2396
+ "loss": 2.9593,
2397
+ "step": 185500
2398
+ },
2399
+ {
2400
+ "epoch": 0.58,
2401
+ "learning_rate": 2.0908226402817836e-05,
2402
+ "loss": 2.9695,
2403
+ "step": 186000
2404
+ },
2405
+ {
2406
+ "epoch": 0.58,
2407
+ "learning_rate": 2.083002271035229e-05,
2408
+ "loss": 2.9696,
2409
+ "step": 186500
2410
+ },
2411
+ {
2412
+ "epoch": 0.58,
2413
+ "learning_rate": 2.075181901788675e-05,
2414
+ "loss": 2.9624,
2415
+ "step": 187000
2416
+ },
2417
+ {
2418
+ "epoch": 0.59,
2419
+ "learning_rate": 2.0673615325421204e-05,
2420
+ "loss": 2.9576,
2421
+ "step": 187500
2422
+ },
2423
+ {
2424
+ "epoch": 0.59,
2425
+ "learning_rate": 2.0595411632955662e-05,
2426
+ "loss": 2.9655,
2427
+ "step": 188000
2428
+ },
2429
+ {
2430
+ "epoch": 0.59,
2431
+ "learning_rate": 2.0517207940490117e-05,
2432
+ "loss": 2.9667,
2433
+ "step": 188500
2434
+ },
2435
+ {
2436
+ "epoch": 0.59,
2437
+ "learning_rate": 2.0439004248024575e-05,
2438
+ "loss": 2.9581,
2439
+ "step": 189000
2440
+ },
2441
+ {
2442
+ "epoch": 0.59,
2443
+ "learning_rate": 2.036080055555903e-05,
2444
+ "loss": 2.9597,
2445
+ "step": 189500
2446
+ },
2447
+ {
2448
+ "epoch": 0.59,
2449
+ "learning_rate": 2.028259686309349e-05,
2450
+ "loss": 2.9615,
2451
+ "step": 190000
2452
+ },
2453
+ {
2454
+ "epoch": 0.59,
2455
+ "eval_accuracy": 0.4257997033905501,
2456
+ "eval_loss": 2.897066116333008,
2457
+ "eval_runtime": 5974.8859,
2458
+ "eval_samples_per_second": 90.096,
2459
+ "eval_steps_per_second": 1.408,
2460
+ "step": 190000
2461
+ },
2462
+ {
2463
+ "epoch": 0.6,
2464
+ "learning_rate": 2.0204393170627943e-05,
2465
+ "loss": 2.9678,
2466
+ "step": 190500
2467
+ },
2468
+ {
2469
+ "epoch": 0.6,
2470
+ "learning_rate": 2.01261894781624e-05,
2471
+ "loss": 2.9542,
2472
+ "step": 191000
2473
+ },
2474
+ {
2475
+ "epoch": 0.6,
2476
+ "learning_rate": 2.0047985785696856e-05,
2477
+ "loss": 2.9594,
2478
+ "step": 191500
2479
+ },
2480
+ {
2481
+ "epoch": 0.6,
2482
+ "learning_rate": 1.9969782093231314e-05,
2483
+ "loss": 2.9661,
2484
+ "step": 192000
2485
+ },
2486
+ {
2487
+ "epoch": 0.6,
2488
+ "learning_rate": 1.989157840076577e-05,
2489
+ "loss": 2.9648,
2490
+ "step": 192500
2491
+ },
2492
+ {
2493
+ "epoch": 0.6,
2494
+ "learning_rate": 1.9813374708300227e-05,
2495
+ "loss": 2.9638,
2496
+ "step": 193000
2497
+ },
2498
+ {
2499
+ "epoch": 0.61,
2500
+ "learning_rate": 1.9735171015834686e-05,
2501
+ "loss": 2.9664,
2502
+ "step": 193500
2503
+ },
2504
+ {
2505
+ "epoch": 0.61,
2506
+ "learning_rate": 1.965696732336914e-05,
2507
+ "loss": 2.9585,
2508
+ "step": 194000
2509
+ },
2510
+ {
2511
+ "epoch": 0.61,
2512
+ "learning_rate": 1.95787636309036e-05,
2513
+ "loss": 2.9579,
2514
+ "step": 194500
2515
+ },
2516
+ {
2517
+ "epoch": 0.61,
2518
+ "learning_rate": 1.9500559938438054e-05,
2519
+ "loss": 2.9673,
2520
+ "step": 195000
2521
+ },
2522
+ {
2523
+ "epoch": 0.61,
2524
+ "learning_rate": 1.9422356245972512e-05,
2525
+ "loss": 2.953,
2526
+ "step": 195500
2527
+ },
2528
+ {
2529
+ "epoch": 0.61,
2530
+ "learning_rate": 1.9344152553506967e-05,
2531
+ "loss": 2.9615,
2532
+ "step": 196000
2533
+ },
2534
+ {
2535
+ "epoch": 0.61,
2536
+ "learning_rate": 1.9265948861041425e-05,
2537
+ "loss": 2.9599,
2538
+ "step": 196500
2539
+ },
2540
+ {
2541
+ "epoch": 0.62,
2542
+ "learning_rate": 1.918774516857588e-05,
2543
+ "loss": 2.9584,
2544
+ "step": 197000
2545
+ },
2546
+ {
2547
+ "epoch": 0.62,
2548
+ "learning_rate": 1.9109541476110338e-05,
2549
+ "loss": 2.9578,
2550
+ "step": 197500
2551
+ },
2552
+ {
2553
+ "epoch": 0.62,
2554
+ "learning_rate": 1.9031337783644793e-05,
2555
+ "loss": 2.9648,
2556
+ "step": 198000
2557
+ },
2558
+ {
2559
+ "epoch": 0.62,
2560
+ "learning_rate": 1.895313409117925e-05,
2561
+ "loss": 2.9576,
2562
+ "step": 198500
2563
+ },
2564
+ {
2565
+ "epoch": 0.62,
2566
+ "learning_rate": 1.8874930398713706e-05,
2567
+ "loss": 2.9555,
2568
+ "step": 199000
2569
+ },
2570
+ {
2571
+ "epoch": 0.62,
2572
+ "learning_rate": 1.8796726706248164e-05,
2573
+ "loss": 2.9612,
2574
+ "step": 199500
2575
+ },
2576
+ {
2577
+ "epoch": 0.63,
2578
+ "learning_rate": 1.871852301378262e-05,
2579
+ "loss": 2.9556,
2580
+ "step": 200000
2581
+ },
2582
+ {
2583
+ "epoch": 0.63,
2584
+ "eval_accuracy": 0.42693775816826174,
2585
+ "eval_loss": 2.8882317543029785,
2586
+ "eval_runtime": 5989.3476,
2587
+ "eval_samples_per_second": 89.878,
2588
+ "eval_steps_per_second": 1.404,
2589
+ "step": 200000
2590
+ },
2591
+ {
2592
+ "epoch": 0.63,
2593
+ "learning_rate": 1.8640319321317077e-05,
2594
+ "loss": 2.957,
2595
+ "step": 200500
2596
+ },
2597
+ {
2598
+ "epoch": 0.63,
2599
+ "learning_rate": 1.8562115628851535e-05,
2600
+ "loss": 2.9636,
2601
+ "step": 201000
2602
+ },
2603
+ {
2604
+ "epoch": 0.63,
2605
+ "learning_rate": 1.848391193638599e-05,
2606
+ "loss": 2.9576,
2607
+ "step": 201500
2608
+ },
2609
+ {
2610
+ "epoch": 0.63,
2611
+ "learning_rate": 1.8405708243920448e-05,
2612
+ "loss": 2.9571,
2613
+ "step": 202000
2614
+ },
2615
+ {
2616
+ "epoch": 0.63,
2617
+ "learning_rate": 1.8327504551454903e-05,
2618
+ "loss": 2.9543,
2619
+ "step": 202500
2620
+ },
2621
+ {
2622
+ "epoch": 0.64,
2623
+ "learning_rate": 1.824930085898936e-05,
2624
+ "loss": 2.9551,
2625
+ "step": 203000
2626
+ },
2627
+ {
2628
+ "epoch": 0.64,
2629
+ "learning_rate": 1.8171097166523816e-05,
2630
+ "loss": 2.9454,
2631
+ "step": 203500
2632
+ },
2633
+ {
2634
+ "epoch": 0.64,
2635
+ "learning_rate": 1.8092893474058274e-05,
2636
+ "loss": 2.9542,
2637
+ "step": 204000
2638
+ },
2639
+ {
2640
+ "epoch": 0.64,
2641
+ "learning_rate": 1.801468978159273e-05,
2642
+ "loss": 2.9548,
2643
+ "step": 204500
2644
+ },
2645
+ {
2646
+ "epoch": 0.64,
2647
+ "learning_rate": 1.7936486089127187e-05,
2648
+ "loss": 2.9557,
2649
+ "step": 205000
2650
+ },
2651
+ {
2652
+ "epoch": 0.64,
2653
+ "learning_rate": 1.7858282396661642e-05,
2654
+ "loss": 2.9511,
2655
+ "step": 205500
2656
+ },
2657
+ {
2658
+ "epoch": 0.64,
2659
+ "learning_rate": 1.77800787041961e-05,
2660
+ "loss": 2.9551,
2661
+ "step": 206000
2662
+ },
2663
+ {
2664
+ "epoch": 0.65,
2665
+ "learning_rate": 1.7701875011730555e-05,
2666
+ "loss": 2.9415,
2667
+ "step": 206500
2668
+ },
2669
+ {
2670
+ "epoch": 0.65,
2671
+ "learning_rate": 1.7623671319265013e-05,
2672
+ "loss": 2.9552,
2673
+ "step": 207000
2674
+ },
2675
+ {
2676
+ "epoch": 0.65,
2677
+ "learning_rate": 1.7545467626799468e-05,
2678
+ "loss": 2.9497,
2679
+ "step": 207500
2680
+ },
2681
+ {
2682
+ "epoch": 0.65,
2683
+ "learning_rate": 1.7467263934333926e-05,
2684
+ "loss": 2.9488,
2685
+ "step": 208000
2686
+ },
2687
+ {
2688
+ "epoch": 0.65,
2689
+ "learning_rate": 1.738906024186838e-05,
2690
+ "loss": 2.9536,
2691
+ "step": 208500
2692
+ },
2693
+ {
2694
+ "epoch": 0.65,
2695
+ "learning_rate": 1.731085654940284e-05,
2696
+ "loss": 2.9552,
2697
+ "step": 209000
2698
+ },
2699
+ {
2700
+ "epoch": 0.66,
2701
+ "learning_rate": 1.7232652856937294e-05,
2702
+ "loss": 2.9523,
2703
+ "step": 209500
2704
+ },
2705
+ {
2706
+ "epoch": 0.66,
2707
+ "learning_rate": 1.715444916447175e-05,
2708
+ "loss": 2.9452,
2709
+ "step": 210000
2710
+ },
2711
+ {
2712
+ "epoch": 0.66,
2713
+ "eval_accuracy": 0.4276823016234361,
2714
+ "eval_loss": 2.8825109004974365,
2715
+ "eval_runtime": 5967.0534,
2716
+ "eval_samples_per_second": 90.214,
2717
+ "eval_steps_per_second": 1.41,
2718
+ "step": 210000
2719
+ },
2720
+ {
2721
+ "epoch": 0.66,
2722
+ "learning_rate": 1.7076245472006207e-05,
2723
+ "loss": 2.95,
2724
+ "step": 210500
2725
+ },
2726
+ {
2727
+ "epoch": 0.66,
2728
+ "learning_rate": 1.6998041779540662e-05,
2729
+ "loss": 2.9454,
2730
+ "step": 211000
2731
+ },
2732
+ {
2733
+ "epoch": 0.66,
2734
+ "learning_rate": 1.691983808707512e-05,
2735
+ "loss": 2.9505,
2736
+ "step": 211500
2737
+ },
2738
+ {
2739
+ "epoch": 0.66,
2740
+ "learning_rate": 1.6841634394609575e-05,
2741
+ "loss": 2.9475,
2742
+ "step": 212000
2743
+ },
2744
+ {
2745
+ "epoch": 0.66,
2746
+ "learning_rate": 1.6763430702144033e-05,
2747
+ "loss": 2.9538,
2748
+ "step": 212500
2749
+ },
2750
+ {
2751
+ "epoch": 0.67,
2752
+ "learning_rate": 1.6685227009678488e-05,
2753
+ "loss": 2.9508,
2754
+ "step": 213000
2755
+ },
2756
+ {
2757
+ "epoch": 0.67,
2758
+ "learning_rate": 1.6607023317212946e-05,
2759
+ "loss": 2.9519,
2760
+ "step": 213500
2761
+ },
2762
+ {
2763
+ "epoch": 0.67,
2764
+ "learning_rate": 1.65288196247474e-05,
2765
+ "loss": 2.9447,
2766
+ "step": 214000
2767
+ },
2768
+ {
2769
+ "epoch": 0.67,
2770
+ "learning_rate": 1.645061593228186e-05,
2771
+ "loss": 2.9466,
2772
+ "step": 214500
2773
+ },
2774
+ {
2775
+ "epoch": 0.67,
2776
+ "learning_rate": 1.6372412239816314e-05,
2777
+ "loss": 2.9437,
2778
+ "step": 215000
2779
+ },
2780
+ {
2781
+ "epoch": 0.67,
2782
+ "learning_rate": 1.6294208547350772e-05,
2783
+ "loss": 2.9471,
2784
+ "step": 215500
2785
+ },
2786
+ {
2787
+ "epoch": 0.68,
2788
+ "learning_rate": 1.6216004854885227e-05,
2789
+ "loss": 2.952,
2790
+ "step": 216000
2791
+ },
2792
+ {
2793
+ "epoch": 0.68,
2794
+ "learning_rate": 1.6137801162419685e-05,
2795
+ "loss": 2.9465,
2796
+ "step": 216500
2797
+ },
2798
+ {
2799
+ "epoch": 0.68,
2800
+ "learning_rate": 1.6059597469954143e-05,
2801
+ "loss": 2.9418,
2802
+ "step": 217000
2803
+ },
2804
+ {
2805
+ "epoch": 0.68,
2806
+ "learning_rate": 1.5981393777488598e-05,
2807
+ "loss": 2.9459,
2808
+ "step": 217500
2809
+ },
2810
+ {
2811
+ "epoch": 0.68,
2812
+ "learning_rate": 1.5903190085023056e-05,
2813
+ "loss": 2.9403,
2814
+ "step": 218000
2815
+ },
2816
+ {
2817
+ "epoch": 0.68,
2818
+ "learning_rate": 1.582498639255751e-05,
2819
+ "loss": 2.9474,
2820
+ "step": 218500
2821
+ },
2822
+ {
2823
+ "epoch": 0.69,
2824
+ "learning_rate": 1.574678270009197e-05,
2825
+ "loss": 2.9403,
2826
+ "step": 219000
2827
+ },
2828
+ {
2829
+ "epoch": 0.69,
2830
+ "learning_rate": 1.5668579007626424e-05,
2831
+ "loss": 2.9438,
2832
+ "step": 219500
2833
+ },
2834
+ {
2835
+ "epoch": 0.69,
2836
+ "learning_rate": 1.5590375315160882e-05,
2837
+ "loss": 2.9412,
2838
+ "step": 220000
2839
+ },
2840
+ {
2841
+ "epoch": 0.69,
2842
+ "eval_accuracy": 0.42842816160118896,
2843
+ "eval_loss": 2.8766274452209473,
2844
+ "eval_runtime": 5967.0165,
2845
+ "eval_samples_per_second": 90.215,
2846
+ "eval_steps_per_second": 1.41,
2847
+ "step": 220000
2848
+ },
2849
+ {
2850
+ "epoch": 0.69,
2851
+ "learning_rate": 1.5512171622695337e-05,
2852
+ "loss": 2.9422,
2853
+ "step": 220500
2854
+ },
2855
+ {
2856
+ "epoch": 0.69,
2857
+ "learning_rate": 1.5433967930229795e-05,
2858
+ "loss": 2.9483,
2859
+ "step": 221000
2860
+ },
2861
+ {
2862
+ "epoch": 0.69,
2863
+ "learning_rate": 1.535576423776425e-05,
2864
+ "loss": 2.9449,
2865
+ "step": 221500
2866
+ },
2867
+ {
2868
+ "epoch": 0.69,
2869
+ "learning_rate": 1.5277560545298708e-05,
2870
+ "loss": 2.9443,
2871
+ "step": 222000
2872
+ },
2873
+ {
2874
+ "epoch": 0.7,
2875
+ "learning_rate": 1.5199356852833163e-05,
2876
+ "loss": 2.9446,
2877
+ "step": 222500
2878
+ },
2879
+ {
2880
+ "epoch": 0.7,
2881
+ "learning_rate": 1.5121153160367621e-05,
2882
+ "loss": 2.9383,
2883
+ "step": 223000
2884
+ },
2885
+ {
2886
+ "epoch": 0.7,
2887
+ "learning_rate": 1.5042949467902076e-05,
2888
+ "loss": 2.9339,
2889
+ "step": 223500
2890
+ },
2891
+ {
2892
+ "epoch": 0.7,
2893
+ "learning_rate": 1.4964745775436534e-05,
2894
+ "loss": 2.9406,
2895
+ "step": 224000
2896
+ },
2897
+ {
2898
+ "epoch": 0.7,
2899
+ "learning_rate": 1.4886542082970989e-05,
2900
+ "loss": 2.9334,
2901
+ "step": 224500
2902
+ },
2903
+ {
2904
+ "epoch": 0.7,
2905
+ "learning_rate": 1.4808338390505447e-05,
2906
+ "loss": 2.9424,
2907
+ "step": 225000
2908
+ },
2909
+ {
2910
+ "epoch": 0.71,
2911
+ "learning_rate": 1.4730134698039904e-05,
2912
+ "loss": 2.9418,
2913
+ "step": 225500
2914
+ },
2915
+ {
2916
+ "epoch": 0.71,
2917
+ "learning_rate": 1.465193100557436e-05,
2918
+ "loss": 2.9401,
2919
+ "step": 226000
2920
+ },
2921
+ {
2922
+ "epoch": 0.71,
2923
+ "learning_rate": 1.4573727313108817e-05,
2924
+ "loss": 2.9335,
2925
+ "step": 226500
2926
+ },
2927
+ {
2928
+ "epoch": 0.71,
2929
+ "learning_rate": 1.4495523620643273e-05,
2930
+ "loss": 2.9357,
2931
+ "step": 227000
2932
+ },
2933
+ {
2934
+ "epoch": 0.71,
2935
+ "learning_rate": 1.441731992817773e-05,
2936
+ "loss": 2.9373,
2937
+ "step": 227500
2938
+ },
2939
+ {
2940
+ "epoch": 0.71,
2941
+ "learning_rate": 1.4339116235712186e-05,
2942
+ "loss": 2.9392,
2943
+ "step": 228000
2944
+ },
2945
+ {
2946
+ "epoch": 0.71,
2947
+ "learning_rate": 1.4260912543246643e-05,
2948
+ "loss": 2.937,
2949
+ "step": 228500
2950
+ },
2951
+ {
2952
+ "epoch": 0.72,
2953
+ "learning_rate": 1.4182708850781098e-05,
2954
+ "loss": 2.9427,
2955
+ "step": 229000
2956
+ },
2957
+ {
2958
+ "epoch": 0.72,
2959
+ "learning_rate": 1.4104505158315556e-05,
2960
+ "loss": 2.9348,
2961
+ "step": 229500
2962
+ },
2963
+ {
2964
+ "epoch": 0.72,
2965
+ "learning_rate": 1.402630146585001e-05,
2966
+ "loss": 2.9402,
2967
+ "step": 230000
2968
+ },
2969
+ {
2970
+ "epoch": 0.72,
2971
+ "eval_accuracy": 0.42901379624141195,
2972
+ "eval_loss": 2.872159719467163,
2973
+ "eval_runtime": 6001.9708,
2974
+ "eval_samples_per_second": 89.689,
2975
+ "eval_steps_per_second": 1.402,
2976
+ "step": 230000
2977
+ },
2978
+ {
2979
+ "epoch": 0.72,
2980
+ "learning_rate": 1.3948097773384469e-05,
2981
+ "loss": 2.9363,
2982
+ "step": 230500
2983
+ },
2984
+ {
2985
+ "epoch": 0.72,
2986
+ "learning_rate": 1.3869894080918924e-05,
2987
+ "loss": 2.9372,
2988
+ "step": 231000
2989
+ },
2990
+ {
2991
+ "epoch": 0.72,
2992
+ "learning_rate": 1.3791690388453382e-05,
2993
+ "loss": 2.9384,
2994
+ "step": 231500
2995
+ },
2996
+ {
2997
+ "epoch": 0.73,
2998
+ "learning_rate": 1.3713486695987837e-05,
2999
+ "loss": 2.9363,
3000
+ "step": 232000
3001
+ },
3002
+ {
3003
+ "epoch": 0.73,
3004
+ "learning_rate": 1.3635283003522295e-05,
3005
+ "loss": 2.9368,
3006
+ "step": 232500
3007
+ },
3008
+ {
3009
+ "epoch": 0.73,
3010
+ "learning_rate": 1.3557079311056753e-05,
3011
+ "loss": 2.9436,
3012
+ "step": 233000
3013
+ },
3014
+ {
3015
+ "epoch": 0.73,
3016
+ "learning_rate": 1.3478875618591208e-05,
3017
+ "loss": 2.9427,
3018
+ "step": 233500
3019
+ },
3020
+ {
3021
+ "epoch": 0.73,
3022
+ "learning_rate": 1.3400671926125666e-05,
3023
+ "loss": 2.9359,
3024
+ "step": 234000
3025
+ },
3026
+ {
3027
+ "epoch": 0.73,
3028
+ "learning_rate": 1.3322468233660121e-05,
3029
+ "loss": 2.9371,
3030
+ "step": 234500
3031
+ },
3032
+ {
3033
+ "epoch": 0.74,
3034
+ "learning_rate": 1.3244264541194579e-05,
3035
+ "loss": 2.9314,
3036
+ "step": 235000
3037
+ },
3038
+ {
3039
+ "epoch": 0.74,
3040
+ "learning_rate": 1.3166060848729034e-05,
3041
+ "loss": 2.9339,
3042
+ "step": 235500
3043
+ },
3044
+ {
3045
+ "epoch": 0.74,
3046
+ "learning_rate": 1.3087857156263492e-05,
3047
+ "loss": 2.9305,
3048
+ "step": 236000
3049
+ },
3050
+ {
3051
+ "epoch": 0.74,
3052
+ "learning_rate": 1.3009653463797947e-05,
3053
+ "loss": 2.9341,
3054
+ "step": 236500
3055
+ },
3056
+ {
3057
+ "epoch": 0.74,
3058
+ "learning_rate": 1.2931449771332405e-05,
3059
+ "loss": 2.9356,
3060
+ "step": 237000
3061
+ },
3062
+ {
3063
+ "epoch": 0.74,
3064
+ "learning_rate": 1.285324607886686e-05,
3065
+ "loss": 2.9354,
3066
+ "step": 237500
3067
+ },
3068
+ {
3069
+ "epoch": 0.74,
3070
+ "learning_rate": 1.2775042386401318e-05,
3071
+ "loss": 2.9408,
3072
+ "step": 238000
3073
+ },
3074
+ {
3075
+ "epoch": 0.75,
3076
+ "learning_rate": 1.2696838693935773e-05,
3077
+ "loss": 2.9357,
3078
+ "step": 238500
3079
+ },
3080
+ {
3081
+ "epoch": 0.75,
3082
+ "learning_rate": 1.2618635001470231e-05,
3083
+ "loss": 2.9361,
3084
+ "step": 239000
3085
+ },
3086
+ {
3087
+ "epoch": 0.75,
3088
+ "learning_rate": 1.2540431309004686e-05,
3089
+ "loss": 2.9313,
3090
+ "step": 239500
3091
+ },
3092
+ {
3093
+ "epoch": 0.75,
3094
+ "learning_rate": 1.2462227616539143e-05,
3095
+ "loss": 2.9299,
3096
+ "step": 240000
3097
+ },
3098
+ {
3099
+ "epoch": 0.75,
3100
+ "eval_accuracy": 0.42963944953623323,
3101
+ "eval_loss": 2.867478847503662,
3102
+ "eval_runtime": 5981.9362,
3103
+ "eval_samples_per_second": 89.99,
3104
+ "eval_steps_per_second": 1.406,
3105
+ "step": 240000
3106
+ },
3107
+ {
3108
+ "epoch": 0.75,
3109
+ "learning_rate": 1.2384023924073599e-05,
3110
+ "loss": 2.9345,
3111
+ "step": 240500
3112
+ },
3113
+ {
3114
+ "epoch": 0.75,
3115
+ "learning_rate": 1.2305820231608056e-05,
3116
+ "loss": 2.9319,
3117
+ "step": 241000
3118
+ },
3119
+ {
3120
+ "epoch": 0.76,
3121
+ "learning_rate": 1.2227616539142512e-05,
3122
+ "loss": 2.934,
3123
+ "step": 241500
3124
+ },
3125
+ {
3126
+ "epoch": 0.76,
3127
+ "learning_rate": 1.2149412846676969e-05,
3128
+ "loss": 2.9321,
3129
+ "step": 242000
3130
+ },
3131
+ {
3132
+ "epoch": 0.76,
3133
+ "learning_rate": 1.2071209154211425e-05,
3134
+ "loss": 2.9301,
3135
+ "step": 242500
3136
+ },
3137
+ {
3138
+ "epoch": 0.76,
3139
+ "learning_rate": 1.1993005461745882e-05,
3140
+ "loss": 2.934,
3141
+ "step": 243000
3142
+ },
3143
+ {
3144
+ "epoch": 0.76,
3145
+ "learning_rate": 1.1914801769280338e-05,
3146
+ "loss": 2.9358,
3147
+ "step": 243500
3148
+ },
3149
+ {
3150
+ "epoch": 0.76,
3151
+ "learning_rate": 1.1836598076814795e-05,
3152
+ "loss": 2.9328,
3153
+ "step": 244000
3154
+ },
3155
+ {
3156
+ "epoch": 0.76,
3157
+ "learning_rate": 1.1758394384349251e-05,
3158
+ "loss": 2.9332,
3159
+ "step": 244500
3160
+ },
3161
+ {
3162
+ "epoch": 0.77,
3163
+ "learning_rate": 1.1680190691883708e-05,
3164
+ "loss": 2.9288,
3165
+ "step": 245000
3166
+ },
3167
+ {
3168
+ "epoch": 0.77,
3169
+ "learning_rate": 1.1601986999418166e-05,
3170
+ "loss": 2.9357,
3171
+ "step": 245500
3172
+ },
3173
+ {
3174
+ "epoch": 0.77,
3175
+ "learning_rate": 1.1523783306952622e-05,
3176
+ "loss": 2.9231,
3177
+ "step": 246000
3178
+ },
3179
+ {
3180
+ "epoch": 0.77,
3181
+ "learning_rate": 1.1445579614487079e-05,
3182
+ "loss": 2.9373,
3183
+ "step": 246500
3184
+ },
3185
+ {
3186
+ "epoch": 0.77,
3187
+ "learning_rate": 1.1367375922021535e-05,
3188
+ "loss": 2.9313,
3189
+ "step": 247000
3190
+ },
3191
+ {
3192
+ "epoch": 0.77,
3193
+ "learning_rate": 1.1289172229555992e-05,
3194
+ "loss": 2.9398,
3195
+ "step": 247500
3196
+ },
3197
+ {
3198
+ "epoch": 0.78,
3199
+ "learning_rate": 1.1210968537090448e-05,
3200
+ "loss": 2.9304,
3201
+ "step": 248000
3202
+ },
3203
+ {
3204
+ "epoch": 0.78,
3205
+ "learning_rate": 1.1132764844624905e-05,
3206
+ "loss": 2.9308,
3207
+ "step": 248500
3208
+ },
3209
+ {
3210
+ "epoch": 0.78,
3211
+ "learning_rate": 1.1054561152159361e-05,
3212
+ "loss": 2.9292,
3213
+ "step": 249000
3214
+ },
3215
+ {
3216
+ "epoch": 0.78,
3217
+ "learning_rate": 1.0976357459693818e-05,
3218
+ "loss": 2.9204,
3219
+ "step": 249500
3220
+ },
3221
+ {
3222
+ "epoch": 0.78,
3223
+ "learning_rate": 1.0898153767228274e-05,
3224
+ "loss": 2.9302,
3225
+ "step": 250000
3226
+ },
3227
+ {
3228
+ "epoch": 0.78,
3229
+ "eval_accuracy": 0.43037667130997825,
3230
+ "eval_loss": 2.862294912338257,
3231
+ "eval_runtime": 5973.5209,
3232
+ "eval_samples_per_second": 90.116,
3233
+ "eval_steps_per_second": 1.408,
3234
+ "step": 250000
3235
+ },
3236
+ {
3237
+ "epoch": 0.78,
3238
+ "learning_rate": 1.081995007476273e-05,
3239
+ "loss": 2.929,
3240
+ "step": 250500
3241
+ },
3242
+ {
3243
+ "epoch": 0.79,
3244
+ "learning_rate": 1.0741746382297187e-05,
3245
+ "loss": 2.9253,
3246
+ "step": 251000
3247
+ },
3248
+ {
3249
+ "epoch": 0.79,
3250
+ "learning_rate": 1.0663542689831644e-05,
3251
+ "loss": 2.9267,
3252
+ "step": 251500
3253
+ },
3254
+ {
3255
+ "epoch": 0.79,
3256
+ "learning_rate": 1.05853389973661e-05,
3257
+ "loss": 2.9289,
3258
+ "step": 252000
3259
+ },
3260
+ {
3261
+ "epoch": 0.79,
3262
+ "learning_rate": 1.0507135304900557e-05,
3263
+ "loss": 2.9273,
3264
+ "step": 252500
3265
+ },
3266
+ {
3267
+ "epoch": 0.79,
3268
+ "learning_rate": 1.0428931612435013e-05,
3269
+ "loss": 2.9326,
3270
+ "step": 253000
3271
+ },
3272
+ {
3273
+ "epoch": 0.79,
3274
+ "learning_rate": 1.035072791996947e-05,
3275
+ "loss": 2.934,
3276
+ "step": 253500
3277
+ },
3278
+ {
3279
+ "epoch": 0.79,
3280
+ "learning_rate": 1.0272524227503926e-05,
3281
+ "loss": 2.9263,
3282
+ "step": 254000
3283
+ },
3284
+ {
3285
+ "epoch": 0.8,
3286
+ "learning_rate": 1.0194320535038383e-05,
3287
+ "loss": 2.9304,
3288
+ "step": 254500
3289
+ },
3290
+ {
3291
+ "epoch": 0.8,
3292
+ "learning_rate": 1.011611684257284e-05,
3293
+ "loss": 2.9256,
3294
+ "step": 255000
3295
+ },
3296
+ {
3297
+ "epoch": 0.8,
3298
+ "learning_rate": 1.0037913150107296e-05,
3299
+ "loss": 2.9208,
3300
+ "step": 255500
3301
+ },
3302
+ {
3303
+ "epoch": 0.8,
3304
+ "learning_rate": 9.959709457641752e-06,
3305
+ "loss": 2.9256,
3306
+ "step": 256000
3307
+ },
3308
+ {
3309
+ "epoch": 0.8,
3310
+ "learning_rate": 9.881505765176209e-06,
3311
+ "loss": 2.9266,
3312
+ "step": 256500
3313
+ },
3314
+ {
3315
+ "epoch": 0.8,
3316
+ "learning_rate": 9.803302072710665e-06,
3317
+ "loss": 2.9238,
3318
+ "step": 257000
3319
+ },
3320
+ {
3321
+ "epoch": 0.81,
3322
+ "learning_rate": 9.725098380245122e-06,
3323
+ "loss": 2.924,
3324
+ "step": 257500
3325
+ },
3326
+ {
3327
+ "epoch": 0.81,
3328
+ "learning_rate": 9.646894687779578e-06,
3329
+ "loss": 2.9212,
3330
+ "step": 258000
3331
+ },
3332
+ {
3333
+ "epoch": 0.81,
3334
+ "learning_rate": 9.568690995314035e-06,
3335
+ "loss": 2.9224,
3336
+ "step": 258500
3337
+ },
3338
+ {
3339
+ "epoch": 0.81,
3340
+ "learning_rate": 9.490487302848491e-06,
3341
+ "loss": 2.9273,
3342
+ "step": 259000
3343
+ },
3344
+ {
3345
+ "epoch": 0.81,
3346
+ "learning_rate": 9.412283610382948e-06,
3347
+ "loss": 2.9249,
3348
+ "step": 259500
3349
+ },
3350
+ {
3351
+ "epoch": 0.81,
3352
+ "learning_rate": 9.334079917917404e-06,
3353
+ "loss": 2.9165,
3354
+ "step": 260000
3355
+ },
3356
+ {
3357
+ "epoch": 0.81,
3358
+ "eval_accuracy": 0.4308058122732213,
3359
+ "eval_loss": 2.858468532562256,
3360
+ "eval_runtime": 5995.5219,
3361
+ "eval_samples_per_second": 89.786,
3362
+ "eval_steps_per_second": 1.403,
3363
+ "step": 260000
3364
+ },
3365
+ {
3366
+ "epoch": 0.81,
3367
+ "learning_rate": 9.255876225451861e-06,
3368
+ "loss": 2.9231,
3369
+ "step": 260500
3370
+ },
3371
+ {
3372
+ "epoch": 0.82,
3373
+ "learning_rate": 9.177672532986317e-06,
3374
+ "loss": 2.9306,
3375
+ "step": 261000
3376
+ },
3377
+ {
3378
+ "epoch": 0.82,
3379
+ "learning_rate": 9.099468840520776e-06,
3380
+ "loss": 2.9226,
3381
+ "step": 261500
3382
+ },
3383
+ {
3384
+ "epoch": 0.82,
3385
+ "learning_rate": 9.021265148055232e-06,
3386
+ "loss": 2.9248,
3387
+ "step": 262000
3388
+ },
3389
+ {
3390
+ "epoch": 0.82,
3391
+ "learning_rate": 8.943061455589689e-06,
3392
+ "loss": 2.9246,
3393
+ "step": 262500
3394
+ },
3395
+ {
3396
+ "epoch": 0.82,
3397
+ "learning_rate": 8.864857763124145e-06,
3398
+ "loss": 2.9206,
3399
+ "step": 263000
3400
+ },
3401
+ {
3402
+ "epoch": 0.82,
3403
+ "learning_rate": 8.786654070658602e-06,
3404
+ "loss": 2.9235,
3405
+ "step": 263500
3406
+ },
3407
+ {
3408
+ "epoch": 0.83,
3409
+ "learning_rate": 8.708450378193058e-06,
3410
+ "loss": 2.918,
3411
+ "step": 264000
3412
+ },
3413
+ {
3414
+ "epoch": 0.83,
3415
+ "learning_rate": 8.630246685727513e-06,
3416
+ "loss": 2.9265,
3417
+ "step": 264500
3418
+ },
3419
+ {
3420
+ "epoch": 0.83,
3421
+ "learning_rate": 8.55204299326197e-06,
3422
+ "loss": 2.924,
3423
+ "step": 265000
3424
+ },
3425
+ {
3426
+ "epoch": 0.83,
3427
+ "learning_rate": 8.473839300796426e-06,
3428
+ "loss": 2.9226,
3429
+ "step": 265500
3430
+ },
3431
+ {
3432
+ "epoch": 0.83,
3433
+ "learning_rate": 8.395635608330883e-06,
3434
+ "loss": 2.9215,
3435
+ "step": 266000
3436
+ },
3437
+ {
3438
+ "epoch": 0.83,
3439
+ "learning_rate": 8.317431915865339e-06,
3440
+ "loss": 2.9177,
3441
+ "step": 266500
3442
+ },
3443
+ {
3444
+ "epoch": 0.84,
3445
+ "learning_rate": 8.239228223399796e-06,
3446
+ "loss": 2.9188,
3447
+ "step": 267000
3448
+ },
3449
+ {
3450
+ "epoch": 0.84,
3451
+ "learning_rate": 8.161024530934252e-06,
3452
+ "loss": 2.9306,
3453
+ "step": 267500
3454
+ },
3455
+ {
3456
+ "epoch": 0.84,
3457
+ "learning_rate": 8.082820838468709e-06,
3458
+ "loss": 2.9158,
3459
+ "step": 268000
3460
+ },
3461
+ {
3462
+ "epoch": 0.84,
3463
+ "learning_rate": 8.004617146003165e-06,
3464
+ "loss": 2.9232,
3465
+ "step": 268500
3466
+ },
3467
+ {
3468
+ "epoch": 0.84,
3469
+ "learning_rate": 7.926413453537622e-06,
3470
+ "loss": 2.919,
3471
+ "step": 269000
3472
+ },
3473
+ {
3474
+ "epoch": 0.84,
3475
+ "learning_rate": 7.84820976107208e-06,
3476
+ "loss": 2.9246,
3477
+ "step": 269500
3478
+ },
3479
+ {
3480
+ "epoch": 0.84,
3481
+ "learning_rate": 7.770006068606536e-06,
3482
+ "loss": 2.915,
3483
+ "step": 270000
3484
+ },
3485
+ {
3486
+ "epoch": 0.84,
3487
+ "eval_accuracy": 0.4314090792365293,
3488
+ "eval_loss": 2.8536534309387207,
3489
+ "eval_runtime": 5987.8082,
3490
+ "eval_samples_per_second": 89.901,
3491
+ "eval_steps_per_second": 1.405,
3492
+ "step": 270000
3493
+ },
3494
+ {
3495
+ "epoch": 0.85,
3496
+ "learning_rate": 7.691802376140993e-06,
3497
+ "loss": 2.9202,
3498
+ "step": 270500
3499
+ },
3500
+ {
3501
+ "epoch": 0.85,
3502
+ "learning_rate": 7.613598683675449e-06,
3503
+ "loss": 2.9281,
3504
+ "step": 271000
3505
+ },
3506
+ {
3507
+ "epoch": 0.85,
3508
+ "learning_rate": 7.535394991209906e-06,
3509
+ "loss": 2.921,
3510
+ "step": 271500
3511
+ },
3512
+ {
3513
+ "epoch": 0.85,
3514
+ "learning_rate": 7.457191298744362e-06,
3515
+ "loss": 2.9163,
3516
+ "step": 272000
3517
+ },
3518
+ {
3519
+ "epoch": 0.85,
3520
+ "learning_rate": 7.378987606278819e-06,
3521
+ "loss": 2.9255,
3522
+ "step": 272500
3523
+ },
3524
+ {
3525
+ "epoch": 0.85,
3526
+ "learning_rate": 7.300783913813275e-06,
3527
+ "loss": 2.9185,
3528
+ "step": 273000
3529
+ },
3530
+ {
3531
+ "epoch": 0.86,
3532
+ "learning_rate": 7.222580221347732e-06,
3533
+ "loss": 2.9219,
3534
+ "step": 273500
3535
+ },
3536
+ {
3537
+ "epoch": 0.86,
3538
+ "learning_rate": 7.1443765288821875e-06,
3539
+ "loss": 2.9281,
3540
+ "step": 274000
3541
+ },
3542
+ {
3543
+ "epoch": 0.86,
3544
+ "learning_rate": 7.066172836416644e-06,
3545
+ "loss": 2.9228,
3546
+ "step": 274500
3547
+ },
3548
+ {
3549
+ "epoch": 0.86,
3550
+ "learning_rate": 6.9879691439511005e-06,
3551
+ "loss": 2.9313,
3552
+ "step": 275000
3553
+ },
3554
+ {
3555
+ "epoch": 0.86,
3556
+ "learning_rate": 6.909765451485557e-06,
3557
+ "loss": 2.9184,
3558
+ "step": 275500
3559
+ },
3560
+ {
3561
+ "epoch": 0.86,
3562
+ "learning_rate": 6.8315617590200135e-06,
3563
+ "loss": 2.9111,
3564
+ "step": 276000
3565
+ },
3566
+ {
3567
+ "epoch": 0.86,
3568
+ "learning_rate": 6.75335806655447e-06,
3569
+ "loss": 2.9214,
3570
+ "step": 276500
3571
+ },
3572
+ {
3573
+ "epoch": 0.87,
3574
+ "learning_rate": 6.6751543740889265e-06,
3575
+ "loss": 2.9167,
3576
+ "step": 277000
3577
+ },
3578
+ {
3579
+ "epoch": 0.87,
3580
+ "learning_rate": 6.596950681623385e-06,
3581
+ "loss": 2.9143,
3582
+ "step": 277500
3583
+ },
3584
+ {
3585
+ "epoch": 0.87,
3586
+ "learning_rate": 6.518746989157841e-06,
3587
+ "loss": 2.9185,
3588
+ "step": 278000
3589
+ },
3590
+ {
3591
+ "epoch": 0.87,
3592
+ "learning_rate": 6.440543296692298e-06,
3593
+ "loss": 2.9263,
3594
+ "step": 278500
3595
+ },
3596
+ {
3597
+ "epoch": 0.87,
3598
+ "learning_rate": 6.362339604226754e-06,
3599
+ "loss": 2.9141,
3600
+ "step": 279000
3601
+ },
3602
+ {
3603
+ "epoch": 0.87,
3604
+ "learning_rate": 6.28413591176121e-06,
3605
+ "loss": 2.9197,
3606
+ "step": 279500
3607
+ },
3608
+ {
3609
+ "epoch": 0.88,
3610
+ "learning_rate": 6.205932219295666e-06,
3611
+ "loss": 2.92,
3612
+ "step": 280000
3613
+ },
3614
+ {
3615
+ "epoch": 0.88,
3616
+ "eval_accuracy": 0.43186914122938036,
3617
+ "eval_loss": 2.8506009578704834,
3618
+ "eval_runtime": 5983.5515,
3619
+ "eval_samples_per_second": 89.965,
3620
+ "eval_steps_per_second": 1.406,
3621
+ "step": 280000
3622
+ },
3623
+ {
3624
+ "epoch": 0.88,
3625
+ "learning_rate": 6.127728526830123e-06,
3626
+ "loss": 2.9182,
3627
+ "step": 280500
3628
+ },
3629
+ {
3630
+ "epoch": 0.88,
3631
+ "learning_rate": 6.0495248343645794e-06,
3632
+ "loss": 2.92,
3633
+ "step": 281000
3634
+ },
3635
+ {
3636
+ "epoch": 0.88,
3637
+ "learning_rate": 5.971321141899036e-06,
3638
+ "loss": 2.9203,
3639
+ "step": 281500
3640
+ },
3641
+ {
3642
+ "epoch": 0.88,
3643
+ "learning_rate": 5.8931174494334924e-06,
3644
+ "loss": 2.9157,
3645
+ "step": 282000
3646
+ },
3647
+ {
3648
+ "epoch": 0.88,
3649
+ "learning_rate": 5.814913756967949e-06,
3650
+ "loss": 2.9169,
3651
+ "step": 282500
3652
+ },
3653
+ {
3654
+ "epoch": 0.89,
3655
+ "learning_rate": 5.7367100645024055e-06,
3656
+ "loss": 2.9154,
3657
+ "step": 283000
3658
+ },
3659
+ {
3660
+ "epoch": 0.89,
3661
+ "learning_rate": 5.658506372036863e-06,
3662
+ "loss": 2.9159,
3663
+ "step": 283500
3664
+ },
3665
+ {
3666
+ "epoch": 0.89,
3667
+ "learning_rate": 5.580302679571319e-06,
3668
+ "loss": 2.9122,
3669
+ "step": 284000
3670
+ },
3671
+ {
3672
+ "epoch": 0.89,
3673
+ "learning_rate": 5.502098987105776e-06,
3674
+ "loss": 2.9134,
3675
+ "step": 284500
3676
+ },
3677
+ {
3678
+ "epoch": 0.89,
3679
+ "learning_rate": 5.423895294640232e-06,
3680
+ "loss": 2.9108,
3681
+ "step": 285000
3682
+ },
3683
+ {
3684
+ "epoch": 0.89,
3685
+ "learning_rate": 5.345691602174689e-06,
3686
+ "loss": 2.915,
3687
+ "step": 285500
3688
+ },
3689
+ {
3690
+ "epoch": 0.89,
3691
+ "learning_rate": 5.267487909709145e-06,
3692
+ "loss": 2.9153,
3693
+ "step": 286000
3694
+ },
3695
+ {
3696
+ "epoch": 0.9,
3697
+ "learning_rate": 5.189284217243601e-06,
3698
+ "loss": 2.9169,
3699
+ "step": 286500
3700
+ },
3701
+ {
3702
+ "epoch": 0.9,
3703
+ "learning_rate": 5.1110805247780575e-06,
3704
+ "loss": 2.9172,
3705
+ "step": 287000
3706
+ },
3707
+ {
3708
+ "epoch": 0.9,
3709
+ "learning_rate": 5.032876832312515e-06,
3710
+ "loss": 2.9206,
3711
+ "step": 287500
3712
+ },
3713
+ {
3714
+ "epoch": 0.9,
3715
+ "learning_rate": 4.954673139846971e-06,
3716
+ "loss": 2.9141,
3717
+ "step": 288000
3718
+ },
3719
+ {
3720
+ "epoch": 0.9,
3721
+ "learning_rate": 4.876469447381428e-06,
3722
+ "loss": 2.9178,
3723
+ "step": 288500
3724
+ },
3725
+ {
3726
+ "epoch": 0.9,
3727
+ "learning_rate": 4.798265754915884e-06,
3728
+ "loss": 2.9089,
3729
+ "step": 289000
3730
+ },
3731
+ {
3732
+ "epoch": 0.91,
3733
+ "learning_rate": 4.720062062450341e-06,
3734
+ "loss": 2.9175,
3735
+ "step": 289500
3736
+ },
3737
+ {
3738
+ "epoch": 0.91,
3739
+ "learning_rate": 4.641858369984797e-06,
3740
+ "loss": 2.9186,
3741
+ "step": 290000
3742
+ },
3743
+ {
3744
+ "epoch": 0.91,
3745
+ "eval_accuracy": 0.4320649544420721,
3746
+ "eval_loss": 2.8484013080596924,
3747
+ "eval_runtime": 5983.5065,
3748
+ "eval_samples_per_second": 89.966,
3749
+ "eval_steps_per_second": 1.406,
3750
+ "step": 290000
3751
+ },
3752
+ {
3753
+ "epoch": 0.91,
3754
+ "learning_rate": 4.563654677519254e-06,
3755
+ "loss": 2.9101,
3756
+ "step": 290500
3757
+ },
3758
+ {
3759
+ "epoch": 0.91,
3760
+ "learning_rate": 4.48545098505371e-06,
3761
+ "loss": 2.9148,
3762
+ "step": 291000
3763
+ },
3764
+ {
3765
+ "epoch": 0.91,
3766
+ "learning_rate": 4.407247292588168e-06,
3767
+ "loss": 2.9101,
3768
+ "step": 291500
3769
+ },
3770
+ {
3771
+ "epoch": 0.91,
3772
+ "learning_rate": 4.3290436001226234e-06,
3773
+ "loss": 2.9164,
3774
+ "step": 292000
3775
+ },
3776
+ {
3777
+ "epoch": 0.91,
3778
+ "learning_rate": 4.25083990765708e-06,
3779
+ "loss": 2.9132,
3780
+ "step": 292500
3781
+ },
3782
+ {
3783
+ "epoch": 0.92,
3784
+ "learning_rate": 4.1726362151915364e-06,
3785
+ "loss": 2.9142,
3786
+ "step": 293000
3787
+ },
3788
+ {
3789
+ "epoch": 0.92,
3790
+ "learning_rate": 4.094432522725993e-06,
3791
+ "loss": 2.9153,
3792
+ "step": 293500
3793
+ },
3794
+ {
3795
+ "epoch": 0.92,
3796
+ "learning_rate": 4.0162288302604495e-06,
3797
+ "loss": 2.9044,
3798
+ "step": 294000
3799
+ },
3800
+ {
3801
+ "epoch": 0.92,
3802
+ "learning_rate": 3.938025137794906e-06,
3803
+ "loss": 2.9113,
3804
+ "step": 294500
3805
+ },
3806
+ {
3807
+ "epoch": 0.92,
3808
+ "learning_rate": 3.8598214453293625e-06,
3809
+ "loss": 2.917,
3810
+ "step": 295000
3811
+ },
3812
+ {
3813
+ "epoch": 0.92,
3814
+ "learning_rate": 3.78161775286382e-06,
3815
+ "loss": 2.9144,
3816
+ "step": 295500
3817
+ },
3818
+ {
3819
+ "epoch": 0.93,
3820
+ "learning_rate": 3.7034140603982763e-06,
3821
+ "loss": 2.9171,
3822
+ "step": 296000
3823
+ },
3824
+ {
3825
+ "epoch": 0.93,
3826
+ "learning_rate": 3.6252103679327324e-06,
3827
+ "loss": 2.9087,
3828
+ "step": 296500
3829
+ },
3830
+ {
3831
+ "epoch": 0.93,
3832
+ "learning_rate": 3.547006675467189e-06,
3833
+ "loss": 2.9131,
3834
+ "step": 297000
3835
+ },
3836
+ {
3837
+ "epoch": 0.93,
3838
+ "learning_rate": 3.4688029830016454e-06,
3839
+ "loss": 2.9139,
3840
+ "step": 297500
3841
+ },
3842
+ {
3843
+ "epoch": 0.93,
3844
+ "learning_rate": 3.390599290536102e-06,
3845
+ "loss": 2.9134,
3846
+ "step": 298000
3847
+ },
3848
+ {
3849
+ "epoch": 0.93,
3850
+ "learning_rate": 3.3123955980705584e-06,
3851
+ "loss": 2.9051,
3852
+ "step": 298500
3853
+ },
3854
+ {
3855
+ "epoch": 0.94,
3856
+ "learning_rate": 3.234191905605015e-06,
3857
+ "loss": 2.9131,
3858
+ "step": 299000
3859
+ },
3860
+ {
3861
+ "epoch": 0.94,
3862
+ "learning_rate": 3.155988213139472e-06,
3863
+ "loss": 2.9101,
3864
+ "step": 299500
3865
+ },
3866
+ {
3867
+ "epoch": 0.94,
3868
+ "learning_rate": 3.077784520673928e-06,
3869
+ "loss": 2.9084,
3870
+ "step": 300000
3871
+ },
3872
+ {
3873
+ "epoch": 0.94,
3874
+ "eval_accuracy": 0.43248307111762724,
3875
+ "eval_loss": 2.8458340167999268,
3876
+ "eval_runtime": 5966.4445,
3877
+ "eval_samples_per_second": 90.223,
3878
+ "eval_steps_per_second": 1.41,
3879
+ "step": 300000
3880
+ },
3881
+ {
3882
+ "epoch": 0.94,
3883
+ "learning_rate": 2.999580828208385e-06,
3884
+ "loss": 2.9091,
3885
+ "step": 300500
3886
+ },
3887
+ {
3888
+ "epoch": 0.94,
3889
+ "learning_rate": 2.9213771357428414e-06,
3890
+ "loss": 2.9053,
3891
+ "step": 301000
3892
+ },
3893
+ {
3894
+ "epoch": 0.94,
3895
+ "learning_rate": 2.843173443277298e-06,
3896
+ "loss": 2.9162,
3897
+ "step": 301500
3898
+ },
3899
+ {
3900
+ "epoch": 0.94,
3901
+ "learning_rate": 2.7649697508117544e-06,
3902
+ "loss": 2.9106,
3903
+ "step": 302000
3904
+ },
3905
+ {
3906
+ "epoch": 0.95,
3907
+ "learning_rate": 2.6867660583462113e-06,
3908
+ "loss": 2.9144,
3909
+ "step": 302500
3910
+ },
3911
+ {
3912
+ "epoch": 0.95,
3913
+ "learning_rate": 2.6085623658806674e-06,
3914
+ "loss": 2.9153,
3915
+ "step": 303000
3916
+ },
3917
+ {
3918
+ "epoch": 0.95,
3919
+ "learning_rate": 2.530358673415124e-06,
3920
+ "loss": 2.9175,
3921
+ "step": 303500
3922
+ },
3923
+ {
3924
+ "epoch": 0.95,
3925
+ "learning_rate": 2.4521549809495804e-06,
3926
+ "loss": 2.9068,
3927
+ "step": 304000
3928
+ },
3929
+ {
3930
+ "epoch": 0.95,
3931
+ "learning_rate": 2.3739512884840374e-06,
3932
+ "loss": 2.9149,
3933
+ "step": 304500
3934
+ },
3935
+ {
3936
+ "epoch": 0.95,
3937
+ "learning_rate": 2.295747596018494e-06,
3938
+ "loss": 2.911,
3939
+ "step": 305000
3940
+ },
3941
+ {
3942
+ "epoch": 0.96,
3943
+ "learning_rate": 2.2175439035529504e-06,
3944
+ "loss": 2.9049,
3945
+ "step": 305500
3946
+ },
3947
+ {
3948
+ "epoch": 0.96,
3949
+ "learning_rate": 2.139340211087407e-06,
3950
+ "loss": 2.9071,
3951
+ "step": 306000
3952
+ },
3953
+ {
3954
+ "epoch": 0.96,
3955
+ "learning_rate": 2.0611365186218634e-06,
3956
+ "loss": 2.9026,
3957
+ "step": 306500
3958
+ },
3959
+ {
3960
+ "epoch": 0.96,
3961
+ "learning_rate": 1.98293282615632e-06,
3962
+ "loss": 2.9093,
3963
+ "step": 307000
3964
+ },
3965
+ {
3966
+ "epoch": 0.96,
3967
+ "learning_rate": 1.9047291336907764e-06,
3968
+ "loss": 2.9199,
3969
+ "step": 307500
3970
+ },
3971
+ {
3972
+ "epoch": 0.96,
3973
+ "learning_rate": 1.826525441225233e-06,
3974
+ "loss": 2.9103,
3975
+ "step": 308000
3976
+ },
3977
+ {
3978
+ "epoch": 0.97,
3979
+ "learning_rate": 1.7483217487596896e-06,
3980
+ "loss": 2.9138,
3981
+ "step": 308500
3982
+ },
3983
+ {
3984
+ "epoch": 0.97,
3985
+ "learning_rate": 1.6701180562941461e-06,
3986
+ "loss": 2.9132,
3987
+ "step": 309000
3988
+ },
3989
+ {
3990
+ "epoch": 0.97,
3991
+ "learning_rate": 1.5919143638286026e-06,
3992
+ "loss": 2.9156,
3993
+ "step": 309500
3994
+ },
3995
+ {
3996
+ "epoch": 0.97,
3997
+ "learning_rate": 1.5137106713630591e-06,
3998
+ "loss": 2.9142,
3999
+ "step": 310000
4000
+ },
4001
+ {
4002
+ "epoch": 0.97,
4003
+ "eval_accuracy": 0.43274495378893163,
4004
+ "eval_loss": 2.843777894973755,
4005
+ "eval_runtime": 5973.9023,
4006
+ "eval_samples_per_second": 90.111,
4007
+ "eval_steps_per_second": 1.408,
4008
+ "step": 310000
4009
+ },
4010
+ {
4011
+ "epoch": 0.97,
4012
+ "learning_rate": 1.4355069788975157e-06,
4013
+ "loss": 2.9082,
4014
+ "step": 310500
4015
+ },
4016
+ {
4017
+ "epoch": 0.97,
4018
+ "learning_rate": 1.3573032864319724e-06,
4019
+ "loss": 2.9102,
4020
+ "step": 311000
4021
+ },
4022
+ {
4023
+ "epoch": 0.97,
4024
+ "learning_rate": 1.2790995939664287e-06,
4025
+ "loss": 2.9079,
4026
+ "step": 311500
4027
+ },
4028
+ {
4029
+ "epoch": 0.98,
4030
+ "learning_rate": 1.2008959015008854e-06,
4031
+ "loss": 2.9063,
4032
+ "step": 312000
4033
+ },
4034
+ {
4035
+ "epoch": 0.98,
4036
+ "learning_rate": 1.1226922090353419e-06,
4037
+ "loss": 2.9061,
4038
+ "step": 312500
4039
+ },
4040
+ {
4041
+ "epoch": 0.98,
4042
+ "learning_rate": 1.0444885165697984e-06,
4043
+ "loss": 2.9097,
4044
+ "step": 313000
4045
+ },
4046
+ {
4047
+ "epoch": 0.98,
4048
+ "learning_rate": 9.66284824104255e-07,
4049
+ "loss": 2.91,
4050
+ "step": 313500
4051
+ },
4052
+ {
4053
+ "epoch": 0.98,
4054
+ "learning_rate": 8.880811316387115e-07,
4055
+ "loss": 2.8976,
4056
+ "step": 314000
4057
+ },
4058
+ {
4059
+ "epoch": 0.98,
4060
+ "learning_rate": 8.09877439173168e-07,
4061
+ "loss": 2.9136,
4062
+ "step": 314500
4063
+ },
4064
+ {
4065
+ "epoch": 0.99,
4066
+ "learning_rate": 7.316737467076245e-07,
4067
+ "loss": 2.9078,
4068
+ "step": 315000
4069
+ },
4070
+ {
4071
+ "epoch": 0.99,
4072
+ "learning_rate": 6.534700542420811e-07,
4073
+ "loss": 2.9078,
4074
+ "step": 315500
4075
+ },
4076
+ {
4077
+ "epoch": 0.99,
4078
+ "learning_rate": 5.752663617765376e-07,
4079
+ "loss": 2.9046,
4080
+ "step": 316000
4081
+ },
4082
+ {
4083
+ "epoch": 0.99,
4084
+ "learning_rate": 4.970626693109942e-07,
4085
+ "loss": 2.909,
4086
+ "step": 316500
4087
+ },
4088
+ {
4089
+ "epoch": 0.99,
4090
+ "learning_rate": 4.188589768454507e-07,
4091
+ "loss": 2.9017,
4092
+ "step": 317000
4093
+ },
4094
+ {
4095
+ "epoch": 0.99,
4096
+ "learning_rate": 3.406552843799073e-07,
4097
+ "loss": 2.9089,
4098
+ "step": 317500
4099
+ },
4100
+ {
4101
+ "epoch": 0.99,
4102
+ "learning_rate": 2.6245159191436383e-07,
4103
+ "loss": 2.9107,
4104
+ "step": 318000
4105
+ },
4106
+ {
4107
+ "epoch": 1.0,
4108
+ "learning_rate": 1.842478994488204e-07,
4109
+ "loss": 2.9075,
4110
+ "step": 318500
4111
+ },
4112
+ {
4113
+ "epoch": 1.0,
4114
+ "learning_rate": 1.0604420698327694e-07,
4115
+ "loss": 2.9116,
4116
+ "step": 319000
4117
+ },
4118
+ {
4119
+ "epoch": 1.0,
4120
+ "learning_rate": 2.784051451773347e-08,
4121
+ "loss": 2.9071,
4122
+ "step": 319500
4123
+ },
4124
+ {
4125
+ "epoch": 1.0,
4126
+ "step": 319678,
4127
+ "total_flos": 5.345861476810752e+18,
4128
+ "train_loss": 3.072396659716232,
4129
+ "train_runtime": 348731.7472,
4130
+ "train_samples_per_second": 29.334,
4131
+ "train_steps_per_second": 0.917
4132
+ }
4133
+ ],
4134
+ "logging_steps": 500,
4135
+ "max_steps": 319678,
4136
+ "num_input_tokens_seen": 0,
4137
+ "num_train_epochs": 1,
4138
+ "save_steps": 10000,
4139
+ "total_flos": 5.345861476810752e+18,
4140
+ "train_batch_size": 32,
4141
+ "trial_name": null,
4142
+ "trial_params": null
4143
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e49b269ec3cd1e3ec6414392afbdb707c9531ae94e0b4baea1fb8de79731a8e
3
+ size 4792