Hugging Face's logo

toloka
/
gpt2-large-supervised-prompt-writing

Text Generation
Transformers
PyTorch
Safetensors
gpt2
Generated from Trainer
text-generation-inference
Inference Endpoints
Model card Files Community
2
Nikita Pavlichenko commited on
Commit
29392c3
·
1 Parent(s): 757bc89

First model version

Browse files
Files changed (14) hide show
  1. README.md +80 -0
  2. all_results.json +15 -0
  3. config.json +39 -0
  4. eval_results.json +10 -0
  5. generation_config.json +6 -0
  6. merges.txt +0 -0
  7. pytorch_model.bin +3 -0
  8. special_tokens_map.json +5 -0
  9. tokenizer.json +0 -0
  10. tokenizer_config.json +10 -0
  11. train_results.json +8 -0
  12. trainer_state.json +3352 -0
  13. training_args.bin +3 -0
  14. vocab.json +0 -0
README.md CHANGED
@@ -1,3 +1,83 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
+ tags:
4
+ - generated_from_trainer
5
+ metrics:
6
+ - accuracy
7
+ model-index:
8
+ - name: gpt2-large-finetuned
9
+ results: []
10
  ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ # gpt2-large-finetuned
16
+
17
+ This model is a fine-tuned version of [gpt2-large](https://huggingface.co/gpt2-large) on an unknown dataset.
18
+ It achieves the following results on the evaluation set:
19
+ - Loss: 2.2374
20
+ - Accuracy: 0.5978
21
+
22
+ ## Model description
23
+
24
+ More information needed
25
+
26
+ ## Intended uses & limitations
27
+
28
+ More information needed
29
+
30
+ ## Training and evaluation data
31
+
32
+ More information needed
33
+
34
+ ## Training procedure
35
+
36
+ ### Training hyperparameters
37
+
38
+ The following hyperparameters were used during training:
39
+ - learning_rate: 2.294e-05
40
+ - train_batch_size: 4
41
+ - eval_batch_size: 4
42
+ - seed: 42
43
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
44
+ - lr_scheduler_type: linear
45
+ - num_epochs: 6.0
46
+
47
+ ### Training results
48
+
49
+ | Training Loss | Epoch | Step | Validation Loss | Accuracy |
50
+ |:-------------:|:-----:|:----:|:---------------:|:--------:|
51
+ | 2.5507 | 0.23 | 100 | 2.5061 | 0.5568 |
52
+ | 2.465 | 0.46 | 200 | 2.4254 | 0.5672 |
53
+ | 2.3919 | 0.7 | 300 | 2.3827 | 0.5726 |
54
+ | 2.4222 | 0.93 | 400 | 2.3489 | 0.5760 |
55
+ | 2.1958 | 1.16 | 500 | 2.3302 | 0.5793 |
56
+ | 2.2087 | 1.39 | 600 | 2.3123 | 0.5818 |
57
+ | 2.2436 | 1.62 | 700 | 2.2960 | 0.5841 |
58
+ | 2.1737 | 1.86 | 800 | 2.2810 | 0.5866 |
59
+ | 2.0763 | 2.09 | 900 | 2.2779 | 0.5876 |
60
+ | 2.0852 | 2.32 | 1000 | 2.2678 | 0.5894 |
61
+ | 2.0946 | 2.55 | 1100 | 2.2594 | 0.5906 |
62
+ | 2.0497 | 2.78 | 1200 | 2.2516 | 0.5920 |
63
+ | 2.0141 | 3.02 | 1300 | 2.2513 | 0.5928 |
64
+ | 2.0316 | 3.25 | 1400 | 2.2505 | 0.5932 |
65
+ | 1.9783 | 3.48 | 1500 | 2.2430 | 0.5938 |
66
+ | 1.9917 | 3.71 | 1600 | 2.2386 | 0.5948 |
67
+ | 2.0152 | 3.94 | 1700 | 2.2315 | 0.5960 |
68
+ | 1.886 | 4.18 | 1800 | 2.2420 | 0.5957 |
69
+ | 1.9151 | 4.41 | 1900 | 2.2409 | 0.5967 |
70
+ | 1.9538 | 4.64 | 2000 | 2.2379 | 0.5971 |
71
+ | 1.8886 | 4.87 | 2100 | 2.2349 | 0.5976 |
72
+ | 1.9408 | 5.1 | 2200 | 2.2410 | 0.5975 |
73
+ | 1.9168 | 5.34 | 2300 | 2.2394 | 0.5976 |
74
+ | 1.8002 | 5.57 | 2400 | 2.2381 | 0.5977 |
75
+ | 1.8888 | 5.8 | 2500 | 2.2367 | 0.5978 |
76
+
77
+
78
+ ### Framework versions
79
+
80
+ - Transformers 4.26.0
81
+ - Pytorch 1.13.1
82
+ - Datasets 2.9.0
83
+ - Tokenizers 0.13.2
all_results.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 6.0,
3
+ "eval_accuracy": 0.5978312760008184,
4
+ "eval_loss": 2.2373838424682617,
5
+ "eval_runtime": 58.1232,
6
+ "eval_samples": 430,
7
+ "eval_samples_per_second": 7.398,
8
+ "eval_steps_per_second": 1.858,
9
+ "perplexity": 9.368788970328781,
10
+ "train_loss": 2.103886123784088,
11
+ "train_runtime": 6222.0683,
12
+ "train_samples": 1723,
13
+ "train_samples_per_second": 1.662,
14
+ "train_steps_per_second": 0.416
15
+ }
config.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "gpt2-large",
3
+ "activation_function": "gelu_new",
4
+ "architectures": [
5
+ "GPT2LMHeadModel"
6
+ ],
7
+ "attn_pdrop": 0.1,
8
+ "bos_token_id": 50256,
9
+ "embd_pdrop": 0.1,
10
+ "eos_token_id": 50256,
11
+ "initializer_range": 0.02,
12
+ "layer_norm_epsilon": 1e-05,
13
+ "model_type": "gpt2",
14
+ "n_ctx": 1024,
15
+ "n_embd": 1280,
16
+ "n_head": 20,
17
+ "n_inner": null,
18
+ "n_layer": 36,
19
+ "n_positions": 1024,
20
+ "reorder_and_upcast_attn": false,
21
+ "resid_pdrop": 0.1,
22
+ "scale_attn_by_inverse_layer_idx": false,
23
+ "scale_attn_weights": true,
24
+ "summary_activation": null,
25
+ "summary_first_dropout": 0.1,
26
+ "summary_proj_to_labels": true,
27
+ "summary_type": "cls_index",
28
+ "summary_use_proj": true,
29
+ "task_specific_params": {
30
+ "text-generation": {
31
+ "do_sample": true,
32
+ "max_length": 50
33
+ }
34
+ },
35
+ "torch_dtype": "float32",
36
+ "transformers_version": "4.26.0",
37
+ "use_cache": true,
38
+ "vocab_size": 50257
39
+ }
eval_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 6.0,
3
+ "eval_accuracy": 0.5978312760008184,
4
+ "eval_loss": 2.2373838424682617,
5
+ "eval_runtime": 58.1232,
6
+ "eval_samples": 430,
7
+ "eval_samples_per_second": 7.398,
8
+ "eval_steps_per_second": 1.858,
9
+ "perplexity": 9.368788970328781
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": "4.26.0"
6
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ec777bb140f4ef1ab2fd98a9c3619d22b5b322b3ca101fbb85dbb1169259bdc
3
+ size 3134035005
special_tokens_map.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<|endoftext|>",
3
+ "eos_token": "<|endoftext|>",
4
+ "unk_token": "<|endoftext|>"
5
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "bos_token": "<|endoftext|>",
4
+ "eos_token": "<|endoftext|>",
5
+ "model_max_length": 1024,
6
+ "name_or_path": "gpt2-large",
7
+ "special_tokens_map_file": null,
8
+ "tokenizer_class": "GPT2Tokenizer",
9
+ "unk_token": "<|endoftext|>"
10
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 6.0,
3
+ "train_loss": 2.103886123784088,
4
+ "train_runtime": 6222.0683,
5
+ "train_samples": 1723,
6
+ "train_samples_per_second": 1.662,
7
+ "train_steps_per_second": 0.416
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,3352 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 6.0,
5
+ "global_step": 2586,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 2.2895645784996133e-05,
13
+ "loss": 3.3205,
14
+ "step": 5
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 2.2851291569992266e-05,
19
+ "loss": 2.9999,
20
+ "step": 10
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 2.28069373549884e-05,
25
+ "loss": 2.9241,
26
+ "step": 15
27
+ },
28
+ {
29
+ "epoch": 0.05,
30
+ "learning_rate": 2.2762583139984532e-05,
31
+ "loss": 2.7706,
32
+ "step": 20
33
+ },
34
+ {
35
+ "epoch": 0.06,
36
+ "learning_rate": 2.2718228924980665e-05,
37
+ "loss": 2.8128,
38
+ "step": 25
39
+ },
40
+ {
41
+ "epoch": 0.07,
42
+ "learning_rate": 2.2673874709976798e-05,
43
+ "loss": 2.7359,
44
+ "step": 30
45
+ },
46
+ {
47
+ "epoch": 0.08,
48
+ "learning_rate": 2.262952049497293e-05,
49
+ "loss": 2.7496,
50
+ "step": 35
51
+ },
52
+ {
53
+ "epoch": 0.09,
54
+ "learning_rate": 2.2585166279969064e-05,
55
+ "loss": 2.6543,
56
+ "step": 40
57
+ },
58
+ {
59
+ "epoch": 0.1,
60
+ "learning_rate": 2.2540812064965197e-05,
61
+ "loss": 2.6493,
62
+ "step": 45
63
+ },
64
+ {
65
+ "epoch": 0.12,
66
+ "learning_rate": 2.2496457849961333e-05,
67
+ "loss": 2.6312,
68
+ "step": 50
69
+ },
70
+ {
71
+ "epoch": 0.13,
72
+ "learning_rate": 2.2452103634957466e-05,
73
+ "loss": 2.6222,
74
+ "step": 55
75
+ },
76
+ {
77
+ "epoch": 0.14,
78
+ "learning_rate": 2.2407749419953595e-05,
79
+ "loss": 2.6328,
80
+ "step": 60
81
+ },
82
+ {
83
+ "epoch": 0.15,
84
+ "learning_rate": 2.236339520494973e-05,
85
+ "loss": 2.6085,
86
+ "step": 65
87
+ },
88
+ {
89
+ "epoch": 0.16,
90
+ "learning_rate": 2.2319040989945864e-05,
91
+ "loss": 2.5869,
92
+ "step": 70
93
+ },
94
+ {
95
+ "epoch": 0.17,
96
+ "learning_rate": 2.2274686774941994e-05,
97
+ "loss": 2.6353,
98
+ "step": 75
99
+ },
100
+ {
101
+ "epoch": 0.19,
102
+ "learning_rate": 2.223033255993813e-05,
103
+ "loss": 2.6266,
104
+ "step": 80
105
+ },
106
+ {
107
+ "epoch": 0.2,
108
+ "learning_rate": 2.2185978344934263e-05,
109
+ "loss": 2.6341,
110
+ "step": 85
111
+ },
112
+ {
113
+ "epoch": 0.21,
114
+ "learning_rate": 2.2141624129930393e-05,
115
+ "loss": 2.601,
116
+ "step": 90
117
+ },
118
+ {
119
+ "epoch": 0.22,
120
+ "learning_rate": 2.209726991492653e-05,
121
+ "loss": 2.5009,
122
+ "step": 95
123
+ },
124
+ {
125
+ "epoch": 0.23,
126
+ "learning_rate": 2.205291569992266e-05,
127
+ "loss": 2.5507,
128
+ "step": 100
129
+ },
130
+ {
131
+ "epoch": 0.23,
132
+ "eval_accuracy": 0.5567914705949215,
133
+ "eval_loss": 2.5061137676239014,
134
+ "eval_runtime": 58.2489,
135
+ "eval_samples_per_second": 7.382,
136
+ "eval_steps_per_second": 1.854,
137
+ "step": 100
138
+ },
139
+ {
140
+ "epoch": 0.24,
141
+ "learning_rate": 2.2008561484918795e-05,
142
+ "loss": 2.5848,
143
+ "step": 105
144
+ },
145
+ {
146
+ "epoch": 0.26,
147
+ "learning_rate": 2.1964207269914927e-05,
148
+ "loss": 2.5978,
149
+ "step": 110
150
+ },
151
+ {
152
+ "epoch": 0.27,
153
+ "learning_rate": 2.191985305491106e-05,
154
+ "loss": 2.4714,
155
+ "step": 115
156
+ },
157
+ {
158
+ "epoch": 0.28,
159
+ "learning_rate": 2.1875498839907193e-05,
160
+ "loss": 2.5151,
161
+ "step": 120
162
+ },
163
+ {
164
+ "epoch": 0.29,
165
+ "learning_rate": 2.1831144624903326e-05,
166
+ "loss": 2.5547,
167
+ "step": 125
168
+ },
169
+ {
170
+ "epoch": 0.3,
171
+ "learning_rate": 2.178679040989946e-05,
172
+ "loss": 2.5302,
173
+ "step": 130
174
+ },
175
+ {
176
+ "epoch": 0.31,
177
+ "learning_rate": 2.1742436194895592e-05,
178
+ "loss": 2.5061,
179
+ "step": 135
180
+ },
181
+ {
182
+ "epoch": 0.32,
183
+ "learning_rate": 2.1698081979891725e-05,
184
+ "loss": 2.4813,
185
+ "step": 140
186
+ },
187
+ {
188
+ "epoch": 0.34,
189
+ "learning_rate": 2.1653727764887858e-05,
190
+ "loss": 2.44,
191
+ "step": 145
192
+ },
193
+ {
194
+ "epoch": 0.35,
195
+ "learning_rate": 2.1609373549883994e-05,
196
+ "loss": 2.4794,
197
+ "step": 150
198
+ },
199
+ {
200
+ "epoch": 0.36,
201
+ "learning_rate": 2.1565019334880123e-05,
202
+ "loss": 2.5501,
203
+ "step": 155
204
+ },
205
+ {
206
+ "epoch": 0.37,
207
+ "learning_rate": 2.1520665119876256e-05,
208
+ "loss": 2.4405,
209
+ "step": 160
210
+ },
211
+ {
212
+ "epoch": 0.38,
213
+ "learning_rate": 2.1476310904872393e-05,
214
+ "loss": 2.4833,
215
+ "step": 165
216
+ },
217
+ {
218
+ "epoch": 0.39,
219
+ "learning_rate": 2.1431956689868522e-05,
220
+ "loss": 2.5244,
221
+ "step": 170
222
+ },
223
+ {
224
+ "epoch": 0.41,
225
+ "learning_rate": 2.1387602474864655e-05,
226
+ "loss": 2.5057,
227
+ "step": 175
228
+ },
229
+ {
230
+ "epoch": 0.42,
231
+ "learning_rate": 2.134324825986079e-05,
232
+ "loss": 2.4521,
233
+ "step": 180
234
+ },
235
+ {
236
+ "epoch": 0.43,
237
+ "learning_rate": 2.129889404485692e-05,
238
+ "loss": 2.4377,
239
+ "step": 185
240
+ },
241
+ {
242
+ "epoch": 0.44,
243
+ "learning_rate": 2.1254539829853054e-05,
244
+ "loss": 2.4831,
245
+ "step": 190
246
+ },
247
+ {
248
+ "epoch": 0.45,
249
+ "learning_rate": 2.121018561484919e-05,
250
+ "loss": 2.4314,
251
+ "step": 195
252
+ },
253
+ {
254
+ "epoch": 0.46,
255
+ "learning_rate": 2.1165831399845323e-05,
256
+ "loss": 2.465,
257
+ "step": 200
258
+ },
259
+ {
260
+ "epoch": 0.46,
261
+ "eval_accuracy": 0.5671849780626975,
262
+ "eval_loss": 2.42543625831604,
263
+ "eval_runtime": 58.1969,
264
+ "eval_samples_per_second": 7.389,
265
+ "eval_steps_per_second": 1.856,
266
+ "step": 200
267
+ },
268
+ {
269
+ "epoch": 0.48,
270
+ "learning_rate": 2.1121477184841452e-05,
271
+ "loss": 2.475,
272
+ "step": 205
273
+ },
274
+ {
275
+ "epoch": 0.49,
276
+ "learning_rate": 2.107712296983759e-05,
277
+ "loss": 2.4854,
278
+ "step": 210
279
+ },
280
+ {
281
+ "epoch": 0.5,
282
+ "learning_rate": 2.103276875483372e-05,
283
+ "loss": 2.4716,
284
+ "step": 215
285
+ },
286
+ {
287
+ "epoch": 0.51,
288
+ "learning_rate": 2.0988414539829854e-05,
289
+ "loss": 2.4928,
290
+ "step": 220
291
+ },
292
+ {
293
+ "epoch": 0.52,
294
+ "learning_rate": 2.0944060324825987e-05,
295
+ "loss": 2.4601,
296
+ "step": 225
297
+ },
298
+ {
299
+ "epoch": 0.53,
300
+ "learning_rate": 2.089970610982212e-05,
301
+ "loss": 2.4256,
302
+ "step": 230
303
+ },
304
+ {
305
+ "epoch": 0.55,
306
+ "learning_rate": 2.0855351894818253e-05,
307
+ "loss": 2.4221,
308
+ "step": 235
309
+ },
310
+ {
311
+ "epoch": 0.56,
312
+ "learning_rate": 2.0810997679814386e-05,
313
+ "loss": 2.43,
314
+ "step": 240
315
+ },
316
+ {
317
+ "epoch": 0.57,
318
+ "learning_rate": 2.076664346481052e-05,
319
+ "loss": 2.4551,
320
+ "step": 245
321
+ },
322
+ {
323
+ "epoch": 0.58,
324
+ "learning_rate": 2.072228924980665e-05,
325
+ "loss": 2.4228,
326
+ "step": 250
327
+ },
328
+ {
329
+ "epoch": 0.59,
330
+ "learning_rate": 2.0677935034802784e-05,
331
+ "loss": 2.4224,
332
+ "step": 255
333
+ },
334
+ {
335
+ "epoch": 0.6,
336
+ "learning_rate": 2.0633580819798917e-05,
337
+ "loss": 2.5017,
338
+ "step": 260
339
+ },
340
+ {
341
+ "epoch": 0.61,
342
+ "learning_rate": 2.058922660479505e-05,
343
+ "loss": 2.4203,
344
+ "step": 265
345
+ },
346
+ {
347
+ "epoch": 0.63,
348
+ "learning_rate": 2.0544872389791183e-05,
349
+ "loss": 2.4648,
350
+ "step": 270
351
+ },
352
+ {
353
+ "epoch": 0.64,
354
+ "learning_rate": 2.0500518174787316e-05,
355
+ "loss": 2.4495,
356
+ "step": 275
357
+ },
358
+ {
359
+ "epoch": 0.65,
360
+ "learning_rate": 2.045616395978345e-05,
361
+ "loss": 2.4038,
362
+ "step": 280
363
+ },
364
+ {
365
+ "epoch": 0.66,
366
+ "learning_rate": 2.0411809744779582e-05,
367
+ "loss": 2.4055,
368
+ "step": 285
369
+ },
370
+ {
371
+ "epoch": 0.67,
372
+ "learning_rate": 2.0367455529775715e-05,
373
+ "loss": 2.4111,
374
+ "step": 290
375
+ },
376
+ {
377
+ "epoch": 0.68,
378
+ "learning_rate": 2.032310131477185e-05,
379
+ "loss": 2.4462,
380
+ "step": 295
381
+ },
382
+ {
383
+ "epoch": 0.7,
384
+ "learning_rate": 2.027874709976798e-05,
385
+ "loss": 2.3919,
386
+ "step": 300
387
+ },
388
+ {
389
+ "epoch": 0.7,
390
+ "eval_accuracy": 0.5726067880606516,
391
+ "eval_loss": 2.382695436477661,
392
+ "eval_runtime": 58.2041,
393
+ "eval_samples_per_second": 7.388,
394
+ "eval_steps_per_second": 1.856,
395
+ "step": 300
396
+ },
397
+ {
398
+ "epoch": 0.71,
399
+ "learning_rate": 2.0234392884764117e-05,
400
+ "loss": 2.3573,
401
+ "step": 305
402
+ },
403
+ {
404
+ "epoch": 0.72,
405
+ "learning_rate": 2.019003866976025e-05,
406
+ "loss": 2.4338,
407
+ "step": 310
408
+ },
409
+ {
410
+ "epoch": 0.73,
411
+ "learning_rate": 2.014568445475638e-05,
412
+ "loss": 2.398,
413
+ "step": 315
414
+ },
415
+ {
416
+ "epoch": 0.74,
417
+ "learning_rate": 2.0101330239752515e-05,
418
+ "loss": 2.3701,
419
+ "step": 320
420
+ },
421
+ {
422
+ "epoch": 0.75,
423
+ "learning_rate": 2.0056976024748648e-05,
424
+ "loss": 2.4837,
425
+ "step": 325
426
+ },
427
+ {
428
+ "epoch": 0.77,
429
+ "learning_rate": 2.0012621809744778e-05,
430
+ "loss": 2.3821,
431
+ "step": 330
432
+ },
433
+ {
434
+ "epoch": 0.78,
435
+ "learning_rate": 1.9968267594740914e-05,
436
+ "loss": 2.408,
437
+ "step": 335
438
+ },
439
+ {
440
+ "epoch": 0.79,
441
+ "learning_rate": 1.9923913379737047e-05,
442
+ "loss": 2.386,
443
+ "step": 340
444
+ },
445
+ {
446
+ "epoch": 0.8,
447
+ "learning_rate": 1.987955916473318e-05,
448
+ "loss": 2.4007,
449
+ "step": 345
450
+ },
451
+ {
452
+ "epoch": 0.81,
453
+ "learning_rate": 1.9835204949729313e-05,
454
+ "loss": 2.44,
455
+ "step": 350
456
+ },
457
+ {
458
+ "epoch": 0.82,
459
+ "learning_rate": 1.9790850734725446e-05,
460
+ "loss": 2.3848,
461
+ "step": 355
462
+ },
463
+ {
464
+ "epoch": 0.84,
465
+ "learning_rate": 1.974649651972158e-05,
466
+ "loss": 2.3972,
467
+ "step": 360
468
+ },
469
+ {
470
+ "epoch": 0.85,
471
+ "learning_rate": 1.970214230471771e-05,
472
+ "loss": 2.3759,
473
+ "step": 365
474
+ },
475
+ {
476
+ "epoch": 0.86,
477
+ "learning_rate": 1.9657788089713844e-05,
478
+ "loss": 2.3842,
479
+ "step": 370
480
+ },
481
+ {
482
+ "epoch": 0.87,
483
+ "learning_rate": 1.9613433874709977e-05,
484
+ "loss": 2.3788,
485
+ "step": 375
486
+ },
487
+ {
488
+ "epoch": 0.88,
489
+ "learning_rate": 1.956907965970611e-05,
490
+ "loss": 2.4706,
491
+ "step": 380
492
+ },
493
+ {
494
+ "epoch": 0.89,
495
+ "learning_rate": 1.9524725444702243e-05,
496
+ "loss": 2.3689,
497
+ "step": 385
498
+ },
499
+ {
500
+ "epoch": 0.9,
501
+ "learning_rate": 1.948037122969838e-05,
502
+ "loss": 2.3853,
503
+ "step": 390
504
+ },
505
+ {
506
+ "epoch": 0.92,
507
+ "learning_rate": 1.943601701469451e-05,
508
+ "loss": 2.4519,
509
+ "step": 395
510
+ },
511
+ {
512
+ "epoch": 0.93,
513
+ "learning_rate": 1.939166279969064e-05,
514
+ "loss": 2.4222,
515
+ "step": 400
516
+ },
517
+ {
518
+ "epoch": 0.93,
519
+ "eval_accuracy": 0.5760144581599945,
520
+ "eval_loss": 2.3488640785217285,
521
+ "eval_runtime": 58.2409,
522
+ "eval_samples_per_second": 7.383,
523
+ "eval_steps_per_second": 1.854,
524
+ "step": 400
525
+ },
526
+ {
527
+ "epoch": 0.94,
528
+ "learning_rate": 1.9347308584686778e-05,
529
+ "loss": 2.3769,
530
+ "step": 405
531
+ },
532
+ {
533
+ "epoch": 0.95,
534
+ "learning_rate": 1.9302954369682907e-05,
535
+ "loss": 2.3834,
536
+ "step": 410
537
+ },
538
+ {
539
+ "epoch": 0.96,
540
+ "learning_rate": 1.925860015467904e-05,
541
+ "loss": 2.4063,
542
+ "step": 415
543
+ },
544
+ {
545
+ "epoch": 0.97,
546
+ "learning_rate": 1.9214245939675176e-05,
547
+ "loss": 2.3603,
548
+ "step": 420
549
+ },
550
+ {
551
+ "epoch": 0.99,
552
+ "learning_rate": 1.9169891724671306e-05,
553
+ "loss": 2.3989,
554
+ "step": 425
555
+ },
556
+ {
557
+ "epoch": 1.0,
558
+ "learning_rate": 1.912553750966744e-05,
559
+ "loss": 2.3888,
560
+ "step": 430
561
+ },
562
+ {
563
+ "epoch": 1.01,
564
+ "learning_rate": 1.9081183294663575e-05,
565
+ "loss": 2.2754,
566
+ "step": 435
567
+ },
568
+ {
569
+ "epoch": 1.02,
570
+ "learning_rate": 1.9036829079659708e-05,
571
+ "loss": 2.2926,
572
+ "step": 440
573
+ },
574
+ {
575
+ "epoch": 1.03,
576
+ "learning_rate": 1.8992474864655837e-05,
577
+ "loss": 2.2617,
578
+ "step": 445
579
+ },
580
+ {
581
+ "epoch": 1.04,
582
+ "learning_rate": 1.8948120649651974e-05,
583
+ "loss": 2.2712,
584
+ "step": 450
585
+ },
586
+ {
587
+ "epoch": 1.06,
588
+ "learning_rate": 1.8903766434648107e-05,
589
+ "loss": 2.2109,
590
+ "step": 455
591
+ },
592
+ {
593
+ "epoch": 1.07,
594
+ "learning_rate": 1.885941221964424e-05,
595
+ "loss": 2.258,
596
+ "step": 460
597
+ },
598
+ {
599
+ "epoch": 1.08,
600
+ "learning_rate": 1.8815058004640372e-05,
601
+ "loss": 2.2625,
602
+ "step": 465
603
+ },
604
+ {
605
+ "epoch": 1.09,
606
+ "learning_rate": 1.8770703789636505e-05,
607
+ "loss": 2.3066,
608
+ "step": 470
609
+ },
610
+ {
611
+ "epoch": 1.1,
612
+ "learning_rate": 1.8726349574632638e-05,
613
+ "loss": 2.2384,
614
+ "step": 475
615
+ },
616
+ {
617
+ "epoch": 1.11,
618
+ "learning_rate": 1.868199535962877e-05,
619
+ "loss": 2.2411,
620
+ "step": 480
621
+ },
622
+ {
623
+ "epoch": 1.13,
624
+ "learning_rate": 1.8637641144624904e-05,
625
+ "loss": 2.2545,
626
+ "step": 485
627
+ },
628
+ {
629
+ "epoch": 1.14,
630
+ "learning_rate": 1.8593286929621037e-05,
631
+ "loss": 2.2706,
632
+ "step": 490
633
+ },
634
+ {
635
+ "epoch": 1.15,
636
+ "learning_rate": 1.854893271461717e-05,
637
+ "loss": 2.2407,
638
+ "step": 495
639
+ },
640
+ {
641
+ "epoch": 1.16,
642
+ "learning_rate": 1.8504578499613303e-05,
643
+ "loss": 2.1958,
644
+ "step": 500
645
+ },
646
+ {
647
+ "epoch": 1.16,
648
+ "eval_accuracy": 0.5792607242719771,
649
+ "eval_loss": 2.3301777839660645,
650
+ "eval_runtime": 58.2054,
651
+ "eval_samples_per_second": 7.388,
652
+ "eval_steps_per_second": 1.855,
653
+ "step": 500
654
+ },
655
+ {
656
+ "epoch": 1.17,
657
+ "learning_rate": 1.8460224284609435e-05,
658
+ "loss": 2.2448,
659
+ "step": 505
660
+ },
661
+ {
662
+ "epoch": 1.18,
663
+ "learning_rate": 1.841587006960557e-05,
664
+ "loss": 2.3031,
665
+ "step": 510
666
+ },
667
+ {
668
+ "epoch": 1.19,
669
+ "learning_rate": 1.83715158546017e-05,
670
+ "loss": 2.196,
671
+ "step": 515
672
+ },
673
+ {
674
+ "epoch": 1.21,
675
+ "learning_rate": 1.8327161639597834e-05,
676
+ "loss": 2.3195,
677
+ "step": 520
678
+ },
679
+ {
680
+ "epoch": 1.22,
681
+ "learning_rate": 1.8282807424593967e-05,
682
+ "loss": 2.2407,
683
+ "step": 525
684
+ },
685
+ {
686
+ "epoch": 1.23,
687
+ "learning_rate": 1.82384532095901e-05,
688
+ "loss": 2.2314,
689
+ "step": 530
690
+ },
691
+ {
692
+ "epoch": 1.24,
693
+ "learning_rate": 1.8194098994586236e-05,
694
+ "loss": 2.2374,
695
+ "step": 535
696
+ },
697
+ {
698
+ "epoch": 1.25,
699
+ "learning_rate": 1.8149744779582366e-05,
700
+ "loss": 2.2567,
701
+ "step": 540
702
+ },
703
+ {
704
+ "epoch": 1.26,
705
+ "learning_rate": 1.8105390564578502e-05,
706
+ "loss": 2.2234,
707
+ "step": 545
708
+ },
709
+ {
710
+ "epoch": 1.28,
711
+ "learning_rate": 1.8061036349574635e-05,
712
+ "loss": 2.2688,
713
+ "step": 550
714
+ },
715
+ {
716
+ "epoch": 1.29,
717
+ "learning_rate": 1.8016682134570764e-05,
718
+ "loss": 2.2644,
719
+ "step": 555
720
+ },
721
+ {
722
+ "epoch": 1.3,
723
+ "learning_rate": 1.79723279195669e-05,
724
+ "loss": 2.255,
725
+ "step": 560
726
+ },
727
+ {
728
+ "epoch": 1.31,
729
+ "learning_rate": 1.7927973704563033e-05,
730
+ "loss": 2.2402,
731
+ "step": 565
732
+ },
733
+ {
734
+ "epoch": 1.32,
735
+ "learning_rate": 1.7883619489559163e-05,
736
+ "loss": 2.1857,
737
+ "step": 570
738
+ },
739
+ {
740
+ "epoch": 1.33,
741
+ "learning_rate": 1.78392652745553e-05,
742
+ "loss": 2.278,
743
+ "step": 575
744
+ },
745
+ {
746
+ "epoch": 1.35,
747
+ "learning_rate": 1.7794911059551432e-05,
748
+ "loss": 2.2623,
749
+ "step": 580
750
+ },
751
+ {
752
+ "epoch": 1.36,
753
+ "learning_rate": 1.7750556844547565e-05,
754
+ "loss": 2.2192,
755
+ "step": 585
756
+ },
757
+ {
758
+ "epoch": 1.37,
759
+ "learning_rate": 1.7706202629543698e-05,
760
+ "loss": 2.2716,
761
+ "step": 590
762
+ },
763
+ {
764
+ "epoch": 1.38,
765
+ "learning_rate": 1.766184841453983e-05,
766
+ "loss": 2.2689,
767
+ "step": 595
768
+ },
769
+ {
770
+ "epoch": 1.39,
771
+ "learning_rate": 1.7617494199535964e-05,
772
+ "loss": 2.2087,
773
+ "step": 600
774
+ },
775
+ {
776
+ "epoch": 1.39,
777
+ "eval_accuracy": 0.5817590761326695,
778
+ "eval_loss": 2.3123257160186768,
779
+ "eval_runtime": 58.2597,
780
+ "eval_samples_per_second": 7.381,
781
+ "eval_steps_per_second": 1.854,
782
+ "step": 600
783
+ },
784
+ {
785
+ "epoch": 1.4,
786
+ "learning_rate": 1.7573139984532096e-05,
787
+ "loss": 2.2795,
788
+ "step": 605
789
+ },
790
+ {
791
+ "epoch": 1.42,
792
+ "learning_rate": 1.752878576952823e-05,
793
+ "loss": 2.2978,
794
+ "step": 610
795
+ },
796
+ {
797
+ "epoch": 1.43,
798
+ "learning_rate": 1.7484431554524362e-05,
799
+ "loss": 2.202,
800
+ "step": 615
801
+ },
802
+ {
803
+ "epoch": 1.44,
804
+ "learning_rate": 1.7440077339520495e-05,
805
+ "loss": 2.2099,
806
+ "step": 620
807
+ },
808
+ {
809
+ "epoch": 1.45,
810
+ "learning_rate": 1.7395723124516628e-05,
811
+ "loss": 2.2414,
812
+ "step": 625
813
+ },
814
+ {
815
+ "epoch": 1.46,
816
+ "learning_rate": 1.7351368909512764e-05,
817
+ "loss": 2.2039,
818
+ "step": 630
819
+ },
820
+ {
821
+ "epoch": 1.47,
822
+ "learning_rate": 1.7307014694508894e-05,
823
+ "loss": 2.2568,
824
+ "step": 635
825
+ },
826
+ {
827
+ "epoch": 1.48,
828
+ "learning_rate": 1.7262660479505027e-05,
829
+ "loss": 2.2013,
830
+ "step": 640
831
+ },
832
+ {
833
+ "epoch": 1.5,
834
+ "learning_rate": 1.7218306264501163e-05,
835
+ "loss": 2.2123,
836
+ "step": 645
837
+ },
838
+ {
839
+ "epoch": 1.51,
840
+ "learning_rate": 1.7173952049497292e-05,
841
+ "loss": 2.2459,
842
+ "step": 650
843
+ },
844
+ {
845
+ "epoch": 1.52,
846
+ "learning_rate": 1.7129597834493425e-05,
847
+ "loss": 2.3154,
848
+ "step": 655
849
+ },
850
+ {
851
+ "epoch": 1.53,
852
+ "learning_rate": 1.708524361948956e-05,
853
+ "loss": 2.1588,
854
+ "step": 660
855
+ },
856
+ {
857
+ "epoch": 1.54,
858
+ "learning_rate": 1.704088940448569e-05,
859
+ "loss": 2.2633,
860
+ "step": 665
861
+ },
862
+ {
863
+ "epoch": 1.55,
864
+ "learning_rate": 1.6996535189481824e-05,
865
+ "loss": 2.2424,
866
+ "step": 670
867
+ },
868
+ {
869
+ "epoch": 1.57,
870
+ "learning_rate": 1.695218097447796e-05,
871
+ "loss": 2.2245,
872
+ "step": 675
873
+ },
874
+ {
875
+ "epoch": 1.58,
876
+ "learning_rate": 1.6907826759474093e-05,
877
+ "loss": 2.2103,
878
+ "step": 680
879
+ },
880
+ {
881
+ "epoch": 1.59,
882
+ "learning_rate": 1.6863472544470223e-05,
883
+ "loss": 2.1734,
884
+ "step": 685
885
+ },
886
+ {
887
+ "epoch": 1.6,
888
+ "learning_rate": 1.681911832946636e-05,
889
+ "loss": 2.1947,
890
+ "step": 690
891
+ },
892
+ {
893
+ "epoch": 1.61,
894
+ "learning_rate": 1.6774764114462492e-05,
895
+ "loss": 2.2474,
896
+ "step": 695
897
+ },
898
+ {
899
+ "epoch": 1.62,
900
+ "learning_rate": 1.673040989945862e-05,
901
+ "loss": 2.2436,
902
+ "step": 700
903
+ },
904
+ {
905
+ "epoch": 1.62,
906
+ "eval_accuracy": 0.5841005705971948,
907
+ "eval_loss": 2.2960166931152344,
908
+ "eval_runtime": 58.2447,
909
+ "eval_samples_per_second": 7.383,
910
+ "eval_steps_per_second": 1.854,
911
+ "step": 700
912
+ },
913
+ {
914
+ "epoch": 1.64,
915
+ "learning_rate": 1.6686055684454758e-05,
916
+ "loss": 2.2242,
917
+ "step": 705
918
+ },
919
+ {
920
+ "epoch": 1.65,
921
+ "learning_rate": 1.664170146945089e-05,
922
+ "loss": 2.1763,
923
+ "step": 710
924
+ },
925
+ {
926
+ "epoch": 1.66,
927
+ "learning_rate": 1.6597347254447023e-05,
928
+ "loss": 2.2591,
929
+ "step": 715
930
+ },
931
+ {
932
+ "epoch": 1.67,
933
+ "learning_rate": 1.6552993039443156e-05,
934
+ "loss": 2.2595,
935
+ "step": 720
936
+ },
937
+ {
938
+ "epoch": 1.68,
939
+ "learning_rate": 1.650863882443929e-05,
940
+ "loss": 2.26,
941
+ "step": 725
942
+ },
943
+ {
944
+ "epoch": 1.69,
945
+ "learning_rate": 1.6464284609435422e-05,
946
+ "loss": 2.2275,
947
+ "step": 730
948
+ },
949
+ {
950
+ "epoch": 1.71,
951
+ "learning_rate": 1.6419930394431555e-05,
952
+ "loss": 2.1895,
953
+ "step": 735
954
+ },
955
+ {
956
+ "epoch": 1.72,
957
+ "learning_rate": 1.6375576179427688e-05,
958
+ "loss": 2.1943,
959
+ "step": 740
960
+ },
961
+ {
962
+ "epoch": 1.73,
963
+ "learning_rate": 1.633122196442382e-05,
964
+ "loss": 2.2282,
965
+ "step": 745
966
+ },
967
+ {
968
+ "epoch": 1.74,
969
+ "learning_rate": 1.6286867749419954e-05,
970
+ "loss": 2.1926,
971
+ "step": 750
972
+ },
973
+ {
974
+ "epoch": 1.75,
975
+ "learning_rate": 1.6242513534416086e-05,
976
+ "loss": 2.2124,
977
+ "step": 755
978
+ },
979
+ {
980
+ "epoch": 1.76,
981
+ "learning_rate": 1.619815931941222e-05,
982
+ "loss": 2.2104,
983
+ "step": 760
984
+ },
985
+ {
986
+ "epoch": 1.77,
987
+ "learning_rate": 1.6153805104408352e-05,
988
+ "loss": 2.1821,
989
+ "step": 765
990
+ },
991
+ {
992
+ "epoch": 1.79,
993
+ "learning_rate": 1.6109450889404485e-05,
994
+ "loss": 2.2004,
995
+ "step": 770
996
+ },
997
+ {
998
+ "epoch": 1.8,
999
+ "learning_rate": 1.606509667440062e-05,
1000
+ "loss": 2.213,
1001
+ "step": 775
1002
+ },
1003
+ {
1004
+ "epoch": 1.81,
1005
+ "learning_rate": 1.602074245939675e-05,
1006
+ "loss": 2.2022,
1007
+ "step": 780
1008
+ },
1009
+ {
1010
+ "epoch": 1.82,
1011
+ "learning_rate": 1.5976388244392884e-05,
1012
+ "loss": 2.2303,
1013
+ "step": 785
1014
+ },
1015
+ {
1016
+ "epoch": 1.83,
1017
+ "learning_rate": 1.593203402938902e-05,
1018
+ "loss": 2.173,
1019
+ "step": 790
1020
+ },
1021
+ {
1022
+ "epoch": 1.84,
1023
+ "learning_rate": 1.588767981438515e-05,
1024
+ "loss": 2.1799,
1025
+ "step": 795
1026
+ },
1027
+ {
1028
+ "epoch": 1.86,
1029
+ "learning_rate": 1.5843325599381286e-05,
1030
+ "loss": 2.1737,
1031
+ "step": 800
1032
+ },
1033
+ {
1034
+ "epoch": 1.86,
1035
+ "eval_accuracy": 0.5866239287094501,
1036
+ "eval_loss": 2.2810451984405518,
1037
+ "eval_runtime": 58.2336,
1038
+ "eval_samples_per_second": 7.384,
1039
+ "eval_steps_per_second": 1.855,
1040
+ "step": 800
1041
+ },
1042
+ {
1043
+ "epoch": 1.87,
1044
+ "learning_rate": 1.579897138437742e-05,
1045
+ "loss": 2.1741,
1046
+ "step": 805
1047
+ },
1048
+ {
1049
+ "epoch": 1.88,
1050
+ "learning_rate": 1.5754617169373548e-05,
1051
+ "loss": 2.2198,
1052
+ "step": 810
1053
+ },
1054
+ {
1055
+ "epoch": 1.89,
1056
+ "learning_rate": 1.5710262954369684e-05,
1057
+ "loss": 2.2216,
1058
+ "step": 815
1059
+ },
1060
+ {
1061
+ "epoch": 1.9,
1062
+ "learning_rate": 1.5665908739365817e-05,
1063
+ "loss": 2.2283,
1064
+ "step": 820
1065
+ },
1066
+ {
1067
+ "epoch": 1.91,
1068
+ "learning_rate": 1.562155452436195e-05,
1069
+ "loss": 2.1685,
1070
+ "step": 825
1071
+ },
1072
+ {
1073
+ "epoch": 1.93,
1074
+ "learning_rate": 1.5577200309358083e-05,
1075
+ "loss": 2.2341,
1076
+ "step": 830
1077
+ },
1078
+ {
1079
+ "epoch": 1.94,
1080
+ "learning_rate": 1.5532846094354216e-05,
1081
+ "loss": 2.2037,
1082
+ "step": 835
1083
+ },
1084
+ {
1085
+ "epoch": 1.95,
1086
+ "learning_rate": 1.548849187935035e-05,
1087
+ "loss": 2.1839,
1088
+ "step": 840
1089
+ },
1090
+ {
1091
+ "epoch": 1.96,
1092
+ "learning_rate": 1.5444137664346482e-05,
1093
+ "loss": 2.1978,
1094
+ "step": 845
1095
+ },
1096
+ {
1097
+ "epoch": 1.97,
1098
+ "learning_rate": 1.5399783449342615e-05,
1099
+ "loss": 2.2348,
1100
+ "step": 850
1101
+ },
1102
+ {
1103
+ "epoch": 1.98,
1104
+ "learning_rate": 1.5355429234338747e-05,
1105
+ "loss": 2.1423,
1106
+ "step": 855
1107
+ },
1108
+ {
1109
+ "epoch": 2.0,
1110
+ "learning_rate": 1.531107501933488e-05,
1111
+ "loss": 2.2643,
1112
+ "step": 860
1113
+ },
1114
+ {
1115
+ "epoch": 2.01,
1116
+ "learning_rate": 1.5266720804331013e-05,
1117
+ "loss": 2.2016,
1118
+ "step": 865
1119
+ },
1120
+ {
1121
+ "epoch": 2.02,
1122
+ "learning_rate": 1.5222366589327146e-05,
1123
+ "loss": 2.1568,
1124
+ "step": 870
1125
+ },
1126
+ {
1127
+ "epoch": 2.03,
1128
+ "learning_rate": 1.517801237432328e-05,
1129
+ "loss": 2.1041,
1130
+ "step": 875
1131
+ },
1132
+ {
1133
+ "epoch": 2.04,
1134
+ "learning_rate": 1.5133658159319412e-05,
1135
+ "loss": 2.0597,
1136
+ "step": 880
1137
+ },
1138
+ {
1139
+ "epoch": 2.05,
1140
+ "learning_rate": 1.5089303944315546e-05,
1141
+ "loss": 2.0842,
1142
+ "step": 885
1143
+ },
1144
+ {
1145
+ "epoch": 2.06,
1146
+ "learning_rate": 1.504494972931168e-05,
1147
+ "loss": 2.09,
1148
+ "step": 890
1149
+ },
1150
+ {
1151
+ "epoch": 2.08,
1152
+ "learning_rate": 1.500059551430781e-05,
1153
+ "loss": 2.1268,
1154
+ "step": 895
1155
+ },
1156
+ {
1157
+ "epoch": 2.09,
1158
+ "learning_rate": 1.4956241299303945e-05,
1159
+ "loss": 2.0763,
1160
+ "step": 900
1161
+ },
1162
+ {
1163
+ "epoch": 2.09,
1164
+ "eval_accuracy": 0.5876264520675624,
1165
+ "eval_loss": 2.2779171466827393,
1166
+ "eval_runtime": 58.2423,
1167
+ "eval_samples_per_second": 7.383,
1168
+ "eval_steps_per_second": 1.854,
1169
+ "step": 900
1170
+ },
1171
+ {
1172
+ "epoch": 2.1,
1173
+ "learning_rate": 1.4911887084300078e-05,
1174
+ "loss": 2.0782,
1175
+ "step": 905
1176
+ },
1177
+ {
1178
+ "epoch": 2.11,
1179
+ "learning_rate": 1.486753286929621e-05,
1180
+ "loss": 2.0983,
1181
+ "step": 910
1182
+ },
1183
+ {
1184
+ "epoch": 2.12,
1185
+ "learning_rate": 1.4823178654292344e-05,
1186
+ "loss": 2.0846,
1187
+ "step": 915
1188
+ },
1189
+ {
1190
+ "epoch": 2.13,
1191
+ "learning_rate": 1.4778824439288477e-05,
1192
+ "loss": 2.0959,
1193
+ "step": 920
1194
+ },
1195
+ {
1196
+ "epoch": 2.15,
1197
+ "learning_rate": 1.473447022428461e-05,
1198
+ "loss": 2.091,
1199
+ "step": 925
1200
+ },
1201
+ {
1202
+ "epoch": 2.16,
1203
+ "learning_rate": 1.4690116009280744e-05,
1204
+ "loss": 2.1233,
1205
+ "step": 930
1206
+ },
1207
+ {
1208
+ "epoch": 2.17,
1209
+ "learning_rate": 1.4645761794276875e-05,
1210
+ "loss": 2.0943,
1211
+ "step": 935
1212
+ },
1213
+ {
1214
+ "epoch": 2.18,
1215
+ "learning_rate": 1.4601407579273008e-05,
1216
+ "loss": 2.103,
1217
+ "step": 940
1218
+ },
1219
+ {
1220
+ "epoch": 2.19,
1221
+ "learning_rate": 1.4557053364269143e-05,
1222
+ "loss": 2.1249,
1223
+ "step": 945
1224
+ },
1225
+ {
1226
+ "epoch": 2.2,
1227
+ "learning_rate": 1.4512699149265274e-05,
1228
+ "loss": 2.0495,
1229
+ "step": 950
1230
+ },
1231
+ {
1232
+ "epoch": 2.22,
1233
+ "learning_rate": 1.4468344934261409e-05,
1234
+ "loss": 2.0974,
1235
+ "step": 955
1236
+ },
1237
+ {
1238
+ "epoch": 2.23,
1239
+ "learning_rate": 1.4423990719257541e-05,
1240
+ "loss": 2.1103,
1241
+ "step": 960
1242
+ },
1243
+ {
1244
+ "epoch": 2.24,
1245
+ "learning_rate": 1.4379636504253674e-05,
1246
+ "loss": 2.1507,
1247
+ "step": 965
1248
+ },
1249
+ {
1250
+ "epoch": 2.25,
1251
+ "learning_rate": 1.4335282289249809e-05,
1252
+ "loss": 2.1357,
1253
+ "step": 970
1254
+ },
1255
+ {
1256
+ "epoch": 2.26,
1257
+ "learning_rate": 1.429092807424594e-05,
1258
+ "loss": 2.0539,
1259
+ "step": 975
1260
+ },
1261
+ {
1262
+ "epoch": 2.27,
1263
+ "learning_rate": 1.4246573859242073e-05,
1264
+ "loss": 2.0619,
1265
+ "step": 980
1266
+ },
1267
+ {
1268
+ "epoch": 2.29,
1269
+ "learning_rate": 1.4202219644238208e-05,
1270
+ "loss": 2.091,
1271
+ "step": 985
1272
+ },
1273
+ {
1274
+ "epoch": 2.3,
1275
+ "learning_rate": 1.4157865429234339e-05,
1276
+ "loss": 2.1239,
1277
+ "step": 990
1278
+ },
1279
+ {
1280
+ "epoch": 2.31,
1281
+ "learning_rate": 1.4113511214230472e-05,
1282
+ "loss": 2.0991,
1283
+ "step": 995
1284
+ },
1285
+ {
1286
+ "epoch": 2.32,
1287
+ "learning_rate": 1.4069156999226606e-05,
1288
+ "loss": 2.0852,
1289
+ "step": 1000
1290
+ },
1291
+ {
1292
+ "epoch": 2.32,
1293
+ "eval_accuracy": 0.5893882561549478,
1294
+ "eval_loss": 2.2678375244140625,
1295
+ "eval_runtime": 58.2516,
1296
+ "eval_samples_per_second": 7.382,
1297
+ "eval_steps_per_second": 1.854,
1298
+ "step": 1000
1299
+ },
1300
+ {
1301
+ "epoch": 2.33,
1302
+ "learning_rate": 1.4024802784222737e-05,
1303
+ "loss": 2.0668,
1304
+ "step": 1005
1305
+ },
1306
+ {
1307
+ "epoch": 2.34,
1308
+ "learning_rate": 1.398044856921887e-05,
1309
+ "loss": 2.1403,
1310
+ "step": 1010
1311
+ },
1312
+ {
1313
+ "epoch": 2.35,
1314
+ "learning_rate": 1.3936094354215005e-05,
1315
+ "loss": 2.1447,
1316
+ "step": 1015
1317
+ },
1318
+ {
1319
+ "epoch": 2.37,
1320
+ "learning_rate": 1.3891740139211138e-05,
1321
+ "loss": 2.0871,
1322
+ "step": 1020
1323
+ },
1324
+ {
1325
+ "epoch": 2.38,
1326
+ "learning_rate": 1.3847385924207269e-05,
1327
+ "loss": 2.1181,
1328
+ "step": 1025
1329
+ },
1330
+ {
1331
+ "epoch": 2.39,
1332
+ "learning_rate": 1.3803031709203403e-05,
1333
+ "loss": 2.0435,
1334
+ "step": 1030
1335
+ },
1336
+ {
1337
+ "epoch": 2.4,
1338
+ "learning_rate": 1.3758677494199536e-05,
1339
+ "loss": 2.0696,
1340
+ "step": 1035
1341
+ },
1342
+ {
1343
+ "epoch": 2.41,
1344
+ "learning_rate": 1.3714323279195671e-05,
1345
+ "loss": 2.1185,
1346
+ "step": 1040
1347
+ },
1348
+ {
1349
+ "epoch": 2.42,
1350
+ "learning_rate": 1.3669969064191802e-05,
1351
+ "loss": 2.0651,
1352
+ "step": 1045
1353
+ },
1354
+ {
1355
+ "epoch": 2.44,
1356
+ "learning_rate": 1.3625614849187935e-05,
1357
+ "loss": 2.0914,
1358
+ "step": 1050
1359
+ },
1360
+ {
1361
+ "epoch": 2.45,
1362
+ "learning_rate": 1.358126063418407e-05,
1363
+ "loss": 2.0895,
1364
+ "step": 1055
1365
+ },
1366
+ {
1367
+ "epoch": 2.46,
1368
+ "learning_rate": 1.35369064191802e-05,
1369
+ "loss": 2.1037,
1370
+ "step": 1060
1371
+ },
1372
+ {
1373
+ "epoch": 2.47,
1374
+ "learning_rate": 1.3492552204176334e-05,
1375
+ "loss": 2.1207,
1376
+ "step": 1065
1377
+ },
1378
+ {
1379
+ "epoch": 2.48,
1380
+ "learning_rate": 1.3448197989172468e-05,
1381
+ "loss": 2.0958,
1382
+ "step": 1070
1383
+ },
1384
+ {
1385
+ "epoch": 2.49,
1386
+ "learning_rate": 1.3403843774168601e-05,
1387
+ "loss": 2.1155,
1388
+ "step": 1075
1389
+ },
1390
+ {
1391
+ "epoch": 2.51,
1392
+ "learning_rate": 1.3359489559164732e-05,
1393
+ "loss": 2.0974,
1394
+ "step": 1080
1395
+ },
1396
+ {
1397
+ "epoch": 2.52,
1398
+ "learning_rate": 1.3315135344160867e-05,
1399
+ "loss": 2.0724,
1400
+ "step": 1085
1401
+ },
1402
+ {
1403
+ "epoch": 2.53,
1404
+ "learning_rate": 1.3270781129157e-05,
1405
+ "loss": 2.052,
1406
+ "step": 1090
1407
+ },
1408
+ {
1409
+ "epoch": 2.54,
1410
+ "learning_rate": 1.3226426914153131e-05,
1411
+ "loss": 2.0653,
1412
+ "step": 1095
1413
+ },
1414
+ {
1415
+ "epoch": 2.55,
1416
+ "learning_rate": 1.3182072699149266e-05,
1417
+ "loss": 2.0946,
1418
+ "step": 1100
1419
+ },
1420
+ {
1421
+ "epoch": 2.55,
1422
+ "eval_accuracy": 0.5905817363431767,
1423
+ "eval_loss": 2.25939679145813,
1424
+ "eval_runtime": 58.2432,
1425
+ "eval_samples_per_second": 7.383,
1426
+ "eval_steps_per_second": 1.854,
1427
+ "step": 1100
1428
+ },
1429
+ {
1430
+ "epoch": 2.56,
1431
+ "learning_rate": 1.3137718484145398e-05,
1432
+ "loss": 2.1256,
1433
+ "step": 1105
1434
+ },
1435
+ {
1436
+ "epoch": 2.58,
1437
+ "learning_rate": 1.3093364269141531e-05,
1438
+ "loss": 2.0605,
1439
+ "step": 1110
1440
+ },
1441
+ {
1442
+ "epoch": 2.59,
1443
+ "learning_rate": 1.3049010054137666e-05,
1444
+ "loss": 2.0764,
1445
+ "step": 1115
1446
+ },
1447
+ {
1448
+ "epoch": 2.6,
1449
+ "learning_rate": 1.3004655839133797e-05,
1450
+ "loss": 2.1162,
1451
+ "step": 1120
1452
+ },
1453
+ {
1454
+ "epoch": 2.61,
1455
+ "learning_rate": 1.2960301624129932e-05,
1456
+ "loss": 2.1343,
1457
+ "step": 1125
1458
+ },
1459
+ {
1460
+ "epoch": 2.62,
1461
+ "learning_rate": 1.2915947409126065e-05,
1462
+ "loss": 2.1427,
1463
+ "step": 1130
1464
+ },
1465
+ {
1466
+ "epoch": 2.63,
1467
+ "learning_rate": 1.2871593194122196e-05,
1468
+ "loss": 2.0705,
1469
+ "step": 1135
1470
+ },
1471
+ {
1472
+ "epoch": 2.65,
1473
+ "learning_rate": 1.282723897911833e-05,
1474
+ "loss": 2.0804,
1475
+ "step": 1140
1476
+ },
1477
+ {
1478
+ "epoch": 2.66,
1479
+ "learning_rate": 1.2782884764114463e-05,
1480
+ "loss": 2.0478,
1481
+ "step": 1145
1482
+ },
1483
+ {
1484
+ "epoch": 2.67,
1485
+ "learning_rate": 1.2738530549110594e-05,
1486
+ "loss": 2.0797,
1487
+ "step": 1150
1488
+ },
1489
+ {
1490
+ "epoch": 2.68,
1491
+ "learning_rate": 1.2694176334106729e-05,
1492
+ "loss": 2.0719,
1493
+ "step": 1155
1494
+ },
1495
+ {
1496
+ "epoch": 2.69,
1497
+ "learning_rate": 1.2649822119102862e-05,
1498
+ "loss": 2.0958,
1499
+ "step": 1160
1500
+ },
1501
+ {
1502
+ "epoch": 2.7,
1503
+ "learning_rate": 1.2605467904098995e-05,
1504
+ "loss": 2.0709,
1505
+ "step": 1165
1506
+ },
1507
+ {
1508
+ "epoch": 2.71,
1509
+ "learning_rate": 1.256111368909513e-05,
1510
+ "loss": 2.0724,
1511
+ "step": 1170
1512
+ },
1513
+ {
1514
+ "epoch": 2.73,
1515
+ "learning_rate": 1.251675947409126e-05,
1516
+ "loss": 2.0813,
1517
+ "step": 1175
1518
+ },
1519
+ {
1520
+ "epoch": 2.74,
1521
+ "learning_rate": 1.2472405259087393e-05,
1522
+ "loss": 2.1098,
1523
+ "step": 1180
1524
+ },
1525
+ {
1526
+ "epoch": 2.75,
1527
+ "learning_rate": 1.2428051044083528e-05,
1528
+ "loss": 2.0715,
1529
+ "step": 1185
1530
+ },
1531
+ {
1532
+ "epoch": 2.76,
1533
+ "learning_rate": 1.238369682907966e-05,
1534
+ "loss": 2.0848,
1535
+ "step": 1190
1536
+ },
1537
+ {
1538
+ "epoch": 2.77,
1539
+ "learning_rate": 1.2339342614075792e-05,
1540
+ "loss": 2.1323,
1541
+ "step": 1195
1542
+ },
1543
+ {
1544
+ "epoch": 2.78,
1545
+ "learning_rate": 1.2294988399071927e-05,
1546
+ "loss": 2.0497,
1547
+ "step": 1200
1548
+ },
1549
+ {
1550
+ "epoch": 2.78,
1551
+ "eval_accuracy": 0.5920070926822615,
1552
+ "eval_loss": 2.2515528202056885,
1553
+ "eval_runtime": 58.2138,
1554
+ "eval_samples_per_second": 7.387,
1555
+ "eval_steps_per_second": 1.855,
1556
+ "step": 1200
1557
+ },
1558
+ {
1559
+ "epoch": 2.8,
1560
+ "learning_rate": 1.2250634184068058e-05,
1561
+ "loss": 2.0428,
1562
+ "step": 1205
1563
+ },
1564
+ {
1565
+ "epoch": 2.81,
1566
+ "learning_rate": 1.2206279969064194e-05,
1567
+ "loss": 2.0776,
1568
+ "step": 1210
1569
+ },
1570
+ {
1571
+ "epoch": 2.82,
1572
+ "learning_rate": 1.2161925754060325e-05,
1573
+ "loss": 2.1259,
1574
+ "step": 1215
1575
+ },
1576
+ {
1577
+ "epoch": 2.83,
1578
+ "learning_rate": 1.2117571539056458e-05,
1579
+ "loss": 2.131,
1580
+ "step": 1220
1581
+ },
1582
+ {
1583
+ "epoch": 2.84,
1584
+ "learning_rate": 1.2073217324052593e-05,
1585
+ "loss": 2.0149,
1586
+ "step": 1225
1587
+ },
1588
+ {
1589
+ "epoch": 2.85,
1590
+ "learning_rate": 1.2028863109048724e-05,
1591
+ "loss": 2.1355,
1592
+ "step": 1230
1593
+ },
1594
+ {
1595
+ "epoch": 2.87,
1596
+ "learning_rate": 1.1984508894044857e-05,
1597
+ "loss": 2.0664,
1598
+ "step": 1235
1599
+ },
1600
+ {
1601
+ "epoch": 2.88,
1602
+ "learning_rate": 1.1940154679040991e-05,
1603
+ "loss": 2.0407,
1604
+ "step": 1240
1605
+ },
1606
+ {
1607
+ "epoch": 2.89,
1608
+ "learning_rate": 1.1895800464037123e-05,
1609
+ "loss": 2.0636,
1610
+ "step": 1245
1611
+ },
1612
+ {
1613
+ "epoch": 2.9,
1614
+ "learning_rate": 1.1851446249033255e-05,
1615
+ "loss": 2.0867,
1616
+ "step": 1250
1617
+ },
1618
+ {
1619
+ "epoch": 2.91,
1620
+ "learning_rate": 1.180709203402939e-05,
1621
+ "loss": 2.0693,
1622
+ "step": 1255
1623
+ },
1624
+ {
1625
+ "epoch": 2.92,
1626
+ "learning_rate": 1.1762737819025523e-05,
1627
+ "loss": 2.0812,
1628
+ "step": 1260
1629
+ },
1630
+ {
1631
+ "epoch": 2.94,
1632
+ "learning_rate": 1.1718383604021654e-05,
1633
+ "loss": 2.1114,
1634
+ "step": 1265
1635
+ },
1636
+ {
1637
+ "epoch": 2.95,
1638
+ "learning_rate": 1.1674029389017789e-05,
1639
+ "loss": 2.158,
1640
+ "step": 1270
1641
+ },
1642
+ {
1643
+ "epoch": 2.96,
1644
+ "learning_rate": 1.1629675174013922e-05,
1645
+ "loss": 2.0781,
1646
+ "step": 1275
1647
+ },
1648
+ {
1649
+ "epoch": 2.97,
1650
+ "learning_rate": 1.1585320959010053e-05,
1651
+ "loss": 2.1295,
1652
+ "step": 1280
1653
+ },
1654
+ {
1655
+ "epoch": 2.98,
1656
+ "learning_rate": 1.1540966744006187e-05,
1657
+ "loss": 2.0517,
1658
+ "step": 1285
1659
+ },
1660
+ {
1661
+ "epoch": 2.99,
1662
+ "learning_rate": 1.149661252900232e-05,
1663
+ "loss": 2.0954,
1664
+ "step": 1290
1665
+ },
1666
+ {
1667
+ "epoch": 3.0,
1668
+ "learning_rate": 1.1452258313998453e-05,
1669
+ "loss": 2.0296,
1670
+ "step": 1295
1671
+ },
1672
+ {
1673
+ "epoch": 3.02,
1674
+ "learning_rate": 1.1407904098994586e-05,
1675
+ "loss": 2.0141,
1676
+ "step": 1300
1677
+ },
1678
+ {
1679
+ "epoch": 3.02,
1680
+ "eval_accuracy": 0.5928277523926436,
1681
+ "eval_loss": 2.251300573348999,
1682
+ "eval_runtime": 58.199,
1683
+ "eval_samples_per_second": 7.388,
1684
+ "eval_steps_per_second": 1.856,
1685
+ "step": 1300
1686
+ },
1687
+ {
1688
+ "epoch": 3.03,
1689
+ "learning_rate": 1.1363549883990719e-05,
1690
+ "loss": 1.9846,
1691
+ "step": 1305
1692
+ },
1693
+ {
1694
+ "epoch": 3.04,
1695
+ "learning_rate": 1.1319195668986852e-05,
1696
+ "loss": 2.0017,
1697
+ "step": 1310
1698
+ },
1699
+ {
1700
+ "epoch": 3.05,
1701
+ "learning_rate": 1.1274841453982986e-05,
1702
+ "loss": 2.0075,
1703
+ "step": 1315
1704
+ },
1705
+ {
1706
+ "epoch": 3.06,
1707
+ "learning_rate": 1.123048723897912e-05,
1708
+ "loss": 1.9859,
1709
+ "step": 1320
1710
+ },
1711
+ {
1712
+ "epoch": 3.07,
1713
+ "learning_rate": 1.118613302397525e-05,
1714
+ "loss": 1.959,
1715
+ "step": 1325
1716
+ },
1717
+ {
1718
+ "epoch": 3.09,
1719
+ "learning_rate": 1.1141778808971385e-05,
1720
+ "loss": 2.0016,
1721
+ "step": 1330
1722
+ },
1723
+ {
1724
+ "epoch": 3.1,
1725
+ "learning_rate": 1.1097424593967518e-05,
1726
+ "loss": 2.0378,
1727
+ "step": 1335
1728
+ },
1729
+ {
1730
+ "epoch": 3.11,
1731
+ "learning_rate": 1.105307037896365e-05,
1732
+ "loss": 1.9978,
1733
+ "step": 1340
1734
+ },
1735
+ {
1736
+ "epoch": 3.12,
1737
+ "learning_rate": 1.1008716163959784e-05,
1738
+ "loss": 2.0025,
1739
+ "step": 1345
1740
+ },
1741
+ {
1742
+ "epoch": 3.13,
1743
+ "learning_rate": 1.0964361948955917e-05,
1744
+ "loss": 1.9221,
1745
+ "step": 1350
1746
+ },
1747
+ {
1748
+ "epoch": 3.14,
1749
+ "learning_rate": 1.0920007733952051e-05,
1750
+ "loss": 1.9976,
1751
+ "step": 1355
1752
+ },
1753
+ {
1754
+ "epoch": 3.16,
1755
+ "learning_rate": 1.0875653518948182e-05,
1756
+ "loss": 1.9718,
1757
+ "step": 1360
1758
+ },
1759
+ {
1760
+ "epoch": 3.17,
1761
+ "learning_rate": 1.0831299303944315e-05,
1762
+ "loss": 2.0334,
1763
+ "step": 1365
1764
+ },
1765
+ {
1766
+ "epoch": 3.18,
1767
+ "learning_rate": 1.078694508894045e-05,
1768
+ "loss": 2.0087,
1769
+ "step": 1370
1770
+ },
1771
+ {
1772
+ "epoch": 3.19,
1773
+ "learning_rate": 1.0742590873936581e-05,
1774
+ "loss": 2.0111,
1775
+ "step": 1375
1776
+ },
1777
+ {
1778
+ "epoch": 3.2,
1779
+ "learning_rate": 1.0698236658932716e-05,
1780
+ "loss": 2.0332,
1781
+ "step": 1380
1782
+ },
1783
+ {
1784
+ "epoch": 3.21,
1785
+ "learning_rate": 1.0653882443928848e-05,
1786
+ "loss": 1.9301,
1787
+ "step": 1385
1788
+ },
1789
+ {
1790
+ "epoch": 3.23,
1791
+ "learning_rate": 1.0609528228924981e-05,
1792
+ "loss": 1.9898,
1793
+ "step": 1390
1794
+ },
1795
+ {
1796
+ "epoch": 3.24,
1797
+ "learning_rate": 1.0565174013921114e-05,
1798
+ "loss": 1.9965,
1799
+ "step": 1395
1800
+ },
1801
+ {
1802
+ "epoch": 3.25,
1803
+ "learning_rate": 1.0520819798917247e-05,
1804
+ "loss": 2.0316,
1805
+ "step": 1400
1806
+ },
1807
+ {
1808
+ "epoch": 3.25,
1809
+ "eval_accuracy": 0.5931710200277343,
1810
+ "eval_loss": 2.2504708766937256,
1811
+ "eval_runtime": 58.2338,
1812
+ "eval_samples_per_second": 7.384,
1813
+ "eval_steps_per_second": 1.855,
1814
+ "step": 1400
1815
+ },
1816
+ {
1817
+ "epoch": 3.26,
1818
+ "learning_rate": 1.047646558391338e-05,
1819
+ "loss": 2.0107,
1820
+ "step": 1405
1821
+ },
1822
+ {
1823
+ "epoch": 3.27,
1824
+ "learning_rate": 1.0432111368909513e-05,
1825
+ "loss": 1.9955,
1826
+ "step": 1410
1827
+ },
1828
+ {
1829
+ "epoch": 3.28,
1830
+ "learning_rate": 1.0387757153905646e-05,
1831
+ "loss": 1.9849,
1832
+ "step": 1415
1833
+ },
1834
+ {
1835
+ "epoch": 3.29,
1836
+ "learning_rate": 1.0343402938901779e-05,
1837
+ "loss": 1.9644,
1838
+ "step": 1420
1839
+ },
1840
+ {
1841
+ "epoch": 3.31,
1842
+ "learning_rate": 1.0299048723897911e-05,
1843
+ "loss": 1.9953,
1844
+ "step": 1425
1845
+ },
1846
+ {
1847
+ "epoch": 3.32,
1848
+ "learning_rate": 1.0254694508894044e-05,
1849
+ "loss": 1.9783,
1850
+ "step": 1430
1851
+ },
1852
+ {
1853
+ "epoch": 3.33,
1854
+ "learning_rate": 1.0210340293890179e-05,
1855
+ "loss": 2.0231,
1856
+ "step": 1435
1857
+ },
1858
+ {
1859
+ "epoch": 3.34,
1860
+ "learning_rate": 1.0165986078886312e-05,
1861
+ "loss": 1.9832,
1862
+ "step": 1440
1863
+ },
1864
+ {
1865
+ "epoch": 3.35,
1866
+ "learning_rate": 1.0121631863882443e-05,
1867
+ "loss": 2.0026,
1868
+ "step": 1445
1869
+ },
1870
+ {
1871
+ "epoch": 3.36,
1872
+ "learning_rate": 1.0077277648878578e-05,
1873
+ "loss": 2.0112,
1874
+ "step": 1450
1875
+ },
1876
+ {
1877
+ "epoch": 3.38,
1878
+ "learning_rate": 1.003292343387471e-05,
1879
+ "loss": 2.0493,
1880
+ "step": 1455
1881
+ },
1882
+ {
1883
+ "epoch": 3.39,
1884
+ "learning_rate": 9.988569218870843e-06,
1885
+ "loss": 2.0369,
1886
+ "step": 1460
1887
+ },
1888
+ {
1889
+ "epoch": 3.4,
1890
+ "learning_rate": 9.944215003866976e-06,
1891
+ "loss": 1.9922,
1892
+ "step": 1465
1893
+ },
1894
+ {
1895
+ "epoch": 3.41,
1896
+ "learning_rate": 9.899860788863109e-06,
1897
+ "loss": 2.0055,
1898
+ "step": 1470
1899
+ },
1900
+ {
1901
+ "epoch": 3.42,
1902
+ "learning_rate": 9.855506573859244e-06,
1903
+ "loss": 1.9911,
1904
+ "step": 1475
1905
+ },
1906
+ {
1907
+ "epoch": 3.43,
1908
+ "learning_rate": 9.811152358855375e-06,
1909
+ "loss": 2.0245,
1910
+ "step": 1480
1911
+ },
1912
+ {
1913
+ "epoch": 3.45,
1914
+ "learning_rate": 9.766798143851508e-06,
1915
+ "loss": 1.9636,
1916
+ "step": 1485
1917
+ },
1918
+ {
1919
+ "epoch": 3.46,
1920
+ "learning_rate": 9.722443928847642e-06,
1921
+ "loss": 1.9372,
1922
+ "step": 1490
1923
+ },
1924
+ {
1925
+ "epoch": 3.47,
1926
+ "learning_rate": 9.678089713843774e-06,
1927
+ "loss": 1.9605,
1928
+ "step": 1495
1929
+ },
1930
+ {
1931
+ "epoch": 3.48,
1932
+ "learning_rate": 9.633735498839908e-06,
1933
+ "loss": 1.9783,
1934
+ "step": 1500
1935
+ },
1936
+ {
1937
+ "epoch": 3.48,
1938
+ "eval_accuracy": 0.593768896769647,
1939
+ "eval_loss": 2.2429721355438232,
1940
+ "eval_runtime": 58.2214,
1941
+ "eval_samples_per_second": 7.386,
1942
+ "eval_steps_per_second": 1.855,
1943
+ "step": 1500
1944
+ },
1945
+ {
1946
+ "epoch": 3.49,
1947
+ "learning_rate": 9.589381283836041e-06,
1948
+ "loss": 1.9702,
1949
+ "step": 1505
1950
+ },
1951
+ {
1952
+ "epoch": 3.5,
1953
+ "learning_rate": 9.545027068832172e-06,
1954
+ "loss": 1.9922,
1955
+ "step": 1510
1956
+ },
1957
+ {
1958
+ "epoch": 3.52,
1959
+ "learning_rate": 9.500672853828307e-06,
1960
+ "loss": 1.9904,
1961
+ "step": 1515
1962
+ },
1963
+ {
1964
+ "epoch": 3.53,
1965
+ "learning_rate": 9.45631863882444e-06,
1966
+ "loss": 2.0219,
1967
+ "step": 1520
1968
+ },
1969
+ {
1970
+ "epoch": 3.54,
1971
+ "learning_rate": 9.411964423820573e-06,
1972
+ "loss": 2.0095,
1973
+ "step": 1525
1974
+ },
1975
+ {
1976
+ "epoch": 3.55,
1977
+ "learning_rate": 9.367610208816705e-06,
1978
+ "loss": 1.9953,
1979
+ "step": 1530
1980
+ },
1981
+ {
1982
+ "epoch": 3.56,
1983
+ "learning_rate": 9.323255993812838e-06,
1984
+ "loss": 1.9974,
1985
+ "step": 1535
1986
+ },
1987
+ {
1988
+ "epoch": 3.57,
1989
+ "learning_rate": 9.278901778808971e-06,
1990
+ "loss": 1.9874,
1991
+ "step": 1540
1992
+ },
1993
+ {
1994
+ "epoch": 3.58,
1995
+ "learning_rate": 9.234547563805104e-06,
1996
+ "loss": 1.9871,
1997
+ "step": 1545
1998
+ },
1999
+ {
2000
+ "epoch": 3.6,
2001
+ "learning_rate": 9.190193348801237e-06,
2002
+ "loss": 1.997,
2003
+ "step": 1550
2004
+ },
2005
+ {
2006
+ "epoch": 3.61,
2007
+ "learning_rate": 9.145839133797372e-06,
2008
+ "loss": 2.0049,
2009
+ "step": 1555
2010
+ },
2011
+ {
2012
+ "epoch": 3.62,
2013
+ "learning_rate": 9.101484918793504e-06,
2014
+ "loss": 1.9514,
2015
+ "step": 1560
2016
+ },
2017
+ {
2018
+ "epoch": 3.63,
2019
+ "learning_rate": 9.057130703789636e-06,
2020
+ "loss": 1.9647,
2021
+ "step": 1565
2022
+ },
2023
+ {
2024
+ "epoch": 3.64,
2025
+ "learning_rate": 9.01277648878577e-06,
2026
+ "loss": 1.9638,
2027
+ "step": 1570
2028
+ },
2029
+ {
2030
+ "epoch": 3.65,
2031
+ "learning_rate": 8.968422273781903e-06,
2032
+ "loss": 2.0392,
2033
+ "step": 1575
2034
+ },
2035
+ {
2036
+ "epoch": 3.67,
2037
+ "learning_rate": 8.924068058778036e-06,
2038
+ "loss": 2.0167,
2039
+ "step": 1580
2040
+ },
2041
+ {
2042
+ "epoch": 3.68,
2043
+ "learning_rate": 8.879713843774169e-06,
2044
+ "loss": 2.0097,
2045
+ "step": 1585
2046
+ },
2047
+ {
2048
+ "epoch": 3.69,
2049
+ "learning_rate": 8.835359628770302e-06,
2050
+ "loss": 1.9624,
2051
+ "step": 1590
2052
+ },
2053
+ {
2054
+ "epoch": 3.7,
2055
+ "learning_rate": 8.791005413766435e-06,
2056
+ "loss": 2.0064,
2057
+ "step": 1595
2058
+ },
2059
+ {
2060
+ "epoch": 3.71,
2061
+ "learning_rate": 8.746651198762568e-06,
2062
+ "loss": 1.9917,
2063
+ "step": 1600
2064
+ },
2065
+ {
2066
+ "epoch": 3.71,
2067
+ "eval_accuracy": 0.5947941530837255,
2068
+ "eval_loss": 2.238585948944092,
2069
+ "eval_runtime": 58.2344,
2070
+ "eval_samples_per_second": 7.384,
2071
+ "eval_steps_per_second": 1.855,
2072
+ "step": 1600
2073
+ },
2074
+ {
2075
+ "epoch": 3.72,
2076
+ "learning_rate": 8.7022969837587e-06,
2077
+ "loss": 2.0178,
2078
+ "step": 1605
2079
+ },
2080
+ {
2081
+ "epoch": 3.74,
2082
+ "learning_rate": 8.657942768754835e-06,
2083
+ "loss": 1.995,
2084
+ "step": 1610
2085
+ },
2086
+ {
2087
+ "epoch": 3.75,
2088
+ "learning_rate": 8.613588553750966e-06,
2089
+ "loss": 2.0199,
2090
+ "step": 1615
2091
+ },
2092
+ {
2093
+ "epoch": 3.76,
2094
+ "learning_rate": 8.5692343387471e-06,
2095
+ "loss": 2.0151,
2096
+ "step": 1620
2097
+ },
2098
+ {
2099
+ "epoch": 3.77,
2100
+ "learning_rate": 8.524880123743234e-06,
2101
+ "loss": 1.9532,
2102
+ "step": 1625
2103
+ },
2104
+ {
2105
+ "epoch": 3.78,
2106
+ "learning_rate": 8.480525908739365e-06,
2107
+ "loss": 1.9242,
2108
+ "step": 1630
2109
+ },
2110
+ {
2111
+ "epoch": 3.79,
2112
+ "learning_rate": 8.4361716937355e-06,
2113
+ "loss": 1.9664,
2114
+ "step": 1635
2115
+ },
2116
+ {
2117
+ "epoch": 3.81,
2118
+ "learning_rate": 8.391817478731632e-06,
2119
+ "loss": 1.9545,
2120
+ "step": 1640
2121
+ },
2122
+ {
2123
+ "epoch": 3.82,
2124
+ "learning_rate": 8.347463263727765e-06,
2125
+ "loss": 1.9732,
2126
+ "step": 1645
2127
+ },
2128
+ {
2129
+ "epoch": 3.83,
2130
+ "learning_rate": 8.303109048723898e-06,
2131
+ "loss": 1.9884,
2132
+ "step": 1650
2133
+ },
2134
+ {
2135
+ "epoch": 3.84,
2136
+ "learning_rate": 8.258754833720031e-06,
2137
+ "loss": 2.0224,
2138
+ "step": 1655
2139
+ },
2140
+ {
2141
+ "epoch": 3.85,
2142
+ "learning_rate": 8.214400618716164e-06,
2143
+ "loss": 1.9914,
2144
+ "step": 1660
2145
+ },
2146
+ {
2147
+ "epoch": 3.86,
2148
+ "learning_rate": 8.170046403712297e-06,
2149
+ "loss": 1.9627,
2150
+ "step": 1665
2151
+ },
2152
+ {
2153
+ "epoch": 3.87,
2154
+ "learning_rate": 8.12569218870843e-06,
2155
+ "loss": 1.977,
2156
+ "step": 1670
2157
+ },
2158
+ {
2159
+ "epoch": 3.89,
2160
+ "learning_rate": 8.081337973704564e-06,
2161
+ "loss": 2.0249,
2162
+ "step": 1675
2163
+ },
2164
+ {
2165
+ "epoch": 3.9,
2166
+ "learning_rate": 8.036983758700697e-06,
2167
+ "loss": 2.067,
2168
+ "step": 1680
2169
+ },
2170
+ {
2171
+ "epoch": 3.91,
2172
+ "learning_rate": 7.992629543696828e-06,
2173
+ "loss": 1.9708,
2174
+ "step": 1685
2175
+ },
2176
+ {
2177
+ "epoch": 3.92,
2178
+ "learning_rate": 7.948275328692963e-06,
2179
+ "loss": 1.9988,
2180
+ "step": 1690
2181
+ },
2182
+ {
2183
+ "epoch": 3.93,
2184
+ "learning_rate": 7.903921113689096e-06,
2185
+ "loss": 2.0803,
2186
+ "step": 1695
2187
+ },
2188
+ {
2189
+ "epoch": 3.94,
2190
+ "learning_rate": 7.859566898685229e-06,
2191
+ "loss": 2.0152,
2192
+ "step": 1700
2193
+ },
2194
+ {
2195
+ "epoch": 3.94,
2196
+ "eval_accuracy": 0.596026279297097,
2197
+ "eval_loss": 2.2315049171447754,
2198
+ "eval_runtime": 58.2112,
2199
+ "eval_samples_per_second": 7.387,
2200
+ "eval_steps_per_second": 1.855,
2201
+ "step": 1700
2202
+ },
2203
+ {
2204
+ "epoch": 3.96,
2205
+ "learning_rate": 7.815212683681361e-06,
2206
+ "loss": 1.9705,
2207
+ "step": 1705
2208
+ },
2209
+ {
2210
+ "epoch": 3.97,
2211
+ "learning_rate": 7.770858468677494e-06,
2212
+ "loss": 1.9686,
2213
+ "step": 1710
2214
+ },
2215
+ {
2216
+ "epoch": 3.98,
2217
+ "learning_rate": 7.726504253673627e-06,
2218
+ "loss": 1.9508,
2219
+ "step": 1715
2220
+ },
2221
+ {
2222
+ "epoch": 3.99,
2223
+ "learning_rate": 7.68215003866976e-06,
2224
+ "loss": 1.9775,
2225
+ "step": 1720
2226
+ },
2227
+ {
2228
+ "epoch": 4.0,
2229
+ "learning_rate": 7.637795823665893e-06,
2230
+ "loss": 1.9657,
2231
+ "step": 1725
2232
+ },
2233
+ {
2234
+ "epoch": 4.01,
2235
+ "learning_rate": 7.593441608662027e-06,
2236
+ "loss": 1.8765,
2237
+ "step": 1730
2238
+ },
2239
+ {
2240
+ "epoch": 4.03,
2241
+ "learning_rate": 7.549087393658159e-06,
2242
+ "loss": 1.9449,
2243
+ "step": 1735
2244
+ },
2245
+ {
2246
+ "epoch": 4.04,
2247
+ "learning_rate": 7.5047331786542925e-06,
2248
+ "loss": 1.9377,
2249
+ "step": 1740
2250
+ },
2251
+ {
2252
+ "epoch": 4.05,
2253
+ "learning_rate": 7.460378963650426e-06,
2254
+ "loss": 1.8993,
2255
+ "step": 1745
2256
+ },
2257
+ {
2258
+ "epoch": 4.06,
2259
+ "learning_rate": 7.416024748646558e-06,
2260
+ "loss": 1.9302,
2261
+ "step": 1750
2262
+ },
2263
+ {
2264
+ "epoch": 4.07,
2265
+ "learning_rate": 7.371670533642691e-06,
2266
+ "loss": 1.9457,
2267
+ "step": 1755
2268
+ },
2269
+ {
2270
+ "epoch": 4.08,
2271
+ "learning_rate": 7.327316318638825e-06,
2272
+ "loss": 1.9081,
2273
+ "step": 1760
2274
+ },
2275
+ {
2276
+ "epoch": 4.1,
2277
+ "learning_rate": 7.282962103634959e-06,
2278
+ "loss": 1.906,
2279
+ "step": 1765
2280
+ },
2281
+ {
2282
+ "epoch": 4.11,
2283
+ "learning_rate": 7.238607888631091e-06,
2284
+ "loss": 1.9133,
2285
+ "step": 1770
2286
+ },
2287
+ {
2288
+ "epoch": 4.12,
2289
+ "learning_rate": 7.1942536736272235e-06,
2290
+ "loss": 1.9495,
2291
+ "step": 1775
2292
+ },
2293
+ {
2294
+ "epoch": 4.13,
2295
+ "learning_rate": 7.149899458623357e-06,
2296
+ "loss": 1.8968,
2297
+ "step": 1780
2298
+ },
2299
+ {
2300
+ "epoch": 4.14,
2301
+ "learning_rate": 7.105545243619489e-06,
2302
+ "loss": 1.8823,
2303
+ "step": 1785
2304
+ },
2305
+ {
2306
+ "epoch": 4.15,
2307
+ "learning_rate": 7.061191028615623e-06,
2308
+ "loss": 1.9294,
2309
+ "step": 1790
2310
+ },
2311
+ {
2312
+ "epoch": 4.16,
2313
+ "learning_rate": 7.016836813611756e-06,
2314
+ "loss": 1.9581,
2315
+ "step": 1795
2316
+ },
2317
+ {
2318
+ "epoch": 4.18,
2319
+ "learning_rate": 6.972482598607888e-06,
2320
+ "loss": 1.886,
2321
+ "step": 1800
2322
+ },
2323
+ {
2324
+ "epoch": 4.18,
2325
+ "eval_accuracy": 0.5957148378003592,
2326
+ "eval_loss": 2.241971969604492,
2327
+ "eval_runtime": 58.1995,
2328
+ "eval_samples_per_second": 7.388,
2329
+ "eval_steps_per_second": 1.856,
2330
+ "step": 1800
2331
+ },
2332
+ {
2333
+ "epoch": 4.19,
2334
+ "learning_rate": 6.928128383604022e-06,
2335
+ "loss": 1.9006,
2336
+ "step": 1805
2337
+ },
2338
+ {
2339
+ "epoch": 4.2,
2340
+ "learning_rate": 6.883774168600155e-06,
2341
+ "loss": 1.9951,
2342
+ "step": 1810
2343
+ },
2344
+ {
2345
+ "epoch": 4.21,
2346
+ "learning_rate": 6.839419953596288e-06,
2347
+ "loss": 1.8455,
2348
+ "step": 1815
2349
+ },
2350
+ {
2351
+ "epoch": 4.22,
2352
+ "learning_rate": 6.79506573859242e-06,
2353
+ "loss": 1.9753,
2354
+ "step": 1820
2355
+ },
2356
+ {
2357
+ "epoch": 4.23,
2358
+ "learning_rate": 6.750711523588554e-06,
2359
+ "loss": 1.9538,
2360
+ "step": 1825
2361
+ },
2362
+ {
2363
+ "epoch": 4.25,
2364
+ "learning_rate": 6.706357308584687e-06,
2365
+ "loss": 1.9943,
2366
+ "step": 1830
2367
+ },
2368
+ {
2369
+ "epoch": 4.26,
2370
+ "learning_rate": 6.66200309358082e-06,
2371
+ "loss": 1.873,
2372
+ "step": 1835
2373
+ },
2374
+ {
2375
+ "epoch": 4.27,
2376
+ "learning_rate": 6.617648878576953e-06,
2377
+ "loss": 1.9086,
2378
+ "step": 1840
2379
+ },
2380
+ {
2381
+ "epoch": 4.28,
2382
+ "learning_rate": 6.5732946635730864e-06,
2383
+ "loss": 1.9429,
2384
+ "step": 1845
2385
+ },
2386
+ {
2387
+ "epoch": 4.29,
2388
+ "learning_rate": 6.528940448569219e-06,
2389
+ "loss": 1.9145,
2390
+ "step": 1850
2391
+ },
2392
+ {
2393
+ "epoch": 4.3,
2394
+ "learning_rate": 6.484586233565351e-06,
2395
+ "loss": 1.9733,
2396
+ "step": 1855
2397
+ },
2398
+ {
2399
+ "epoch": 4.32,
2400
+ "learning_rate": 6.440232018561485e-06,
2401
+ "loss": 1.9154,
2402
+ "step": 1860
2403
+ },
2404
+ {
2405
+ "epoch": 4.33,
2406
+ "learning_rate": 6.395877803557619e-06,
2407
+ "loss": 1.9516,
2408
+ "step": 1865
2409
+ },
2410
+ {
2411
+ "epoch": 4.34,
2412
+ "learning_rate": 6.351523588553751e-06,
2413
+ "loss": 1.841,
2414
+ "step": 1870
2415
+ },
2416
+ {
2417
+ "epoch": 4.35,
2418
+ "learning_rate": 6.307169373549884e-06,
2419
+ "loss": 1.964,
2420
+ "step": 1875
2421
+ },
2422
+ {
2423
+ "epoch": 4.36,
2424
+ "learning_rate": 6.2628151585460175e-06,
2425
+ "loss": 1.897,
2426
+ "step": 1880
2427
+ },
2428
+ {
2429
+ "epoch": 4.37,
2430
+ "learning_rate": 6.2184609435421495e-06,
2431
+ "loss": 1.9151,
2432
+ "step": 1885
2433
+ },
2434
+ {
2435
+ "epoch": 4.39,
2436
+ "learning_rate": 6.174106728538283e-06,
2437
+ "loss": 1.901,
2438
+ "step": 1890
2439
+ },
2440
+ {
2441
+ "epoch": 4.4,
2442
+ "learning_rate": 6.129752513534416e-06,
2443
+ "loss": 1.9238,
2444
+ "step": 1895
2445
+ },
2446
+ {
2447
+ "epoch": 4.41,
2448
+ "learning_rate": 6.08539829853055e-06,
2449
+ "loss": 1.9151,
2450
+ "step": 1900
2451
+ },
2452
+ {
2453
+ "epoch": 4.41,
2454
+ "eval_accuracy": 0.5966559821773625,
2455
+ "eval_loss": 2.2408835887908936,
2456
+ "eval_runtime": 58.2356,
2457
+ "eval_samples_per_second": 7.384,
2458
+ "eval_steps_per_second": 1.855,
2459
+ "step": 1900
2460
+ },
2461
+ {
2462
+ "epoch": 4.42,
2463
+ "learning_rate": 6.041044083526682e-06,
2464
+ "loss": 1.8906,
2465
+ "step": 1905
2466
+ },
2467
+ {
2468
+ "epoch": 4.43,
2469
+ "learning_rate": 5.996689868522816e-06,
2470
+ "loss": 1.8945,
2471
+ "step": 1910
2472
+ },
2473
+ {
2474
+ "epoch": 4.44,
2475
+ "learning_rate": 5.9523356535189485e-06,
2476
+ "loss": 1.8905,
2477
+ "step": 1915
2478
+ },
2479
+ {
2480
+ "epoch": 4.45,
2481
+ "learning_rate": 5.9079814385150805e-06,
2482
+ "loss": 1.9209,
2483
+ "step": 1920
2484
+ },
2485
+ {
2486
+ "epoch": 4.47,
2487
+ "learning_rate": 5.863627223511214e-06,
2488
+ "loss": 1.8885,
2489
+ "step": 1925
2490
+ },
2491
+ {
2492
+ "epoch": 4.48,
2493
+ "learning_rate": 5.819273008507348e-06,
2494
+ "loss": 1.8773,
2495
+ "step": 1930
2496
+ },
2497
+ {
2498
+ "epoch": 4.49,
2499
+ "learning_rate": 5.774918793503481e-06,
2500
+ "loss": 1.9252,
2501
+ "step": 1935
2502
+ },
2503
+ {
2504
+ "epoch": 4.5,
2505
+ "learning_rate": 5.730564578499614e-06,
2506
+ "loss": 1.9655,
2507
+ "step": 1940
2508
+ },
2509
+ {
2510
+ "epoch": 4.51,
2511
+ "learning_rate": 5.686210363495747e-06,
2512
+ "loss": 1.9207,
2513
+ "step": 1945
2514
+ },
2515
+ {
2516
+ "epoch": 4.52,
2517
+ "learning_rate": 5.6418561484918795e-06,
2518
+ "loss": 1.9147,
2519
+ "step": 1950
2520
+ },
2521
+ {
2522
+ "epoch": 4.54,
2523
+ "learning_rate": 5.5975019334880124e-06,
2524
+ "loss": 1.8857,
2525
+ "step": 1955
2526
+ },
2527
+ {
2528
+ "epoch": 4.55,
2529
+ "learning_rate": 5.553147718484145e-06,
2530
+ "loss": 1.959,
2531
+ "step": 1960
2532
+ },
2533
+ {
2534
+ "epoch": 4.56,
2535
+ "learning_rate": 5.508793503480279e-06,
2536
+ "loss": 1.8938,
2537
+ "step": 1965
2538
+ },
2539
+ {
2540
+ "epoch": 4.57,
2541
+ "learning_rate": 5.464439288476412e-06,
2542
+ "loss": 1.9368,
2543
+ "step": 1970
2544
+ },
2545
+ {
2546
+ "epoch": 4.58,
2547
+ "learning_rate": 5.420085073472544e-06,
2548
+ "loss": 1.9247,
2549
+ "step": 1975
2550
+ },
2551
+ {
2552
+ "epoch": 4.59,
2553
+ "learning_rate": 5.375730858468678e-06,
2554
+ "loss": 1.9154,
2555
+ "step": 1980
2556
+ },
2557
+ {
2558
+ "epoch": 4.61,
2559
+ "learning_rate": 5.331376643464811e-06,
2560
+ "loss": 1.8864,
2561
+ "step": 1985
2562
+ },
2563
+ {
2564
+ "epoch": 4.62,
2565
+ "learning_rate": 5.2870224284609435e-06,
2566
+ "loss": 1.9635,
2567
+ "step": 1990
2568
+ },
2569
+ {
2570
+ "epoch": 4.63,
2571
+ "learning_rate": 5.242668213457076e-06,
2572
+ "loss": 1.891,
2573
+ "step": 1995
2574
+ },
2575
+ {
2576
+ "epoch": 4.64,
2577
+ "learning_rate": 5.19831399845321e-06,
2578
+ "loss": 1.9538,
2579
+ "step": 2000
2580
+ },
2581
+ {
2582
+ "epoch": 4.64,
2583
+ "eval_accuracy": 0.5971379208438473,
2584
+ "eval_loss": 2.2378673553466797,
2585
+ "eval_runtime": 58.2169,
2586
+ "eval_samples_per_second": 7.386,
2587
+ "eval_steps_per_second": 1.855,
2588
+ "step": 2000
2589
+ },
2590
+ {
2591
+ "epoch": 4.65,
2592
+ "learning_rate": 5.153959783449343e-06,
2593
+ "loss": 1.9781,
2594
+ "step": 2005
2595
+ },
2596
+ {
2597
+ "epoch": 4.66,
2598
+ "learning_rate": 5.109605568445476e-06,
2599
+ "loss": 1.9347,
2600
+ "step": 2010
2601
+ },
2602
+ {
2603
+ "epoch": 4.68,
2604
+ "learning_rate": 5.065251353441609e-06,
2605
+ "loss": 1.9551,
2606
+ "step": 2015
2607
+ },
2608
+ {
2609
+ "epoch": 4.69,
2610
+ "learning_rate": 5.020897138437742e-06,
2611
+ "loss": 1.8674,
2612
+ "step": 2020
2613
+ },
2614
+ {
2615
+ "epoch": 4.7,
2616
+ "learning_rate": 4.976542923433875e-06,
2617
+ "loss": 1.9526,
2618
+ "step": 2025
2619
+ },
2620
+ {
2621
+ "epoch": 4.71,
2622
+ "learning_rate": 4.932188708430008e-06,
2623
+ "loss": 1.9281,
2624
+ "step": 2030
2625
+ },
2626
+ {
2627
+ "epoch": 4.72,
2628
+ "learning_rate": 4.88783449342614e-06,
2629
+ "loss": 1.909,
2630
+ "step": 2035
2631
+ },
2632
+ {
2633
+ "epoch": 4.73,
2634
+ "learning_rate": 4.843480278422274e-06,
2635
+ "loss": 1.9404,
2636
+ "step": 2040
2637
+ },
2638
+ {
2639
+ "epoch": 4.74,
2640
+ "learning_rate": 4.799126063418407e-06,
2641
+ "loss": 1.9413,
2642
+ "step": 2045
2643
+ },
2644
+ {
2645
+ "epoch": 4.76,
2646
+ "learning_rate": 4.75477184841454e-06,
2647
+ "loss": 1.9653,
2648
+ "step": 2050
2649
+ },
2650
+ {
2651
+ "epoch": 4.77,
2652
+ "learning_rate": 4.710417633410673e-06,
2653
+ "loss": 1.931,
2654
+ "step": 2055
2655
+ },
2656
+ {
2657
+ "epoch": 4.78,
2658
+ "learning_rate": 4.6660634184068055e-06,
2659
+ "loss": 1.9733,
2660
+ "step": 2060
2661
+ },
2662
+ {
2663
+ "epoch": 4.79,
2664
+ "learning_rate": 4.621709203402939e-06,
2665
+ "loss": 1.8921,
2666
+ "step": 2065
2667
+ },
2668
+ {
2669
+ "epoch": 4.8,
2670
+ "learning_rate": 4.577354988399072e-06,
2671
+ "loss": 1.9374,
2672
+ "step": 2070
2673
+ },
2674
+ {
2675
+ "epoch": 4.81,
2676
+ "learning_rate": 4.533000773395205e-06,
2677
+ "loss": 1.9668,
2678
+ "step": 2075
2679
+ },
2680
+ {
2681
+ "epoch": 4.83,
2682
+ "learning_rate": 4.488646558391338e-06,
2683
+ "loss": 1.8869,
2684
+ "step": 2080
2685
+ },
2686
+ {
2687
+ "epoch": 4.84,
2688
+ "learning_rate": 4.444292343387472e-06,
2689
+ "loss": 1.9688,
2690
+ "step": 2085
2691
+ },
2692
+ {
2693
+ "epoch": 4.85,
2694
+ "learning_rate": 4.3999381283836045e-06,
2695
+ "loss": 1.9266,
2696
+ "step": 2090
2697
+ },
2698
+ {
2699
+ "epoch": 4.86,
2700
+ "learning_rate": 4.3555839133797366e-06,
2701
+ "loss": 1.907,
2702
+ "step": 2095
2703
+ },
2704
+ {
2705
+ "epoch": 4.87,
2706
+ "learning_rate": 4.31122969837587e-06,
2707
+ "loss": 1.8886,
2708
+ "step": 2100
2709
+ },
2710
+ {
2711
+ "epoch": 4.87,
2712
+ "eval_accuracy": 0.5975880333719794,
2713
+ "eval_loss": 2.2348711490631104,
2714
+ "eval_runtime": 58.2184,
2715
+ "eval_samples_per_second": 7.386,
2716
+ "eval_steps_per_second": 1.855,
2717
+ "step": 2100
2718
+ },
2719
+ {
2720
+ "epoch": 4.88,
2721
+ "learning_rate": 4.266875483372003e-06,
2722
+ "loss": 1.9057,
2723
+ "step": 2105
2724
+ },
2725
+ {
2726
+ "epoch": 4.9,
2727
+ "learning_rate": 4.222521268368136e-06,
2728
+ "loss": 1.9208,
2729
+ "step": 2110
2730
+ },
2731
+ {
2732
+ "epoch": 4.91,
2733
+ "learning_rate": 4.178167053364269e-06,
2734
+ "loss": 1.9621,
2735
+ "step": 2115
2736
+ },
2737
+ {
2738
+ "epoch": 4.92,
2739
+ "learning_rate": 4.133812838360402e-06,
2740
+ "loss": 1.9827,
2741
+ "step": 2120
2742
+ },
2743
+ {
2744
+ "epoch": 4.93,
2745
+ "learning_rate": 4.0894586233565356e-06,
2746
+ "loss": 1.925,
2747
+ "step": 2125
2748
+ },
2749
+ {
2750
+ "epoch": 4.94,
2751
+ "learning_rate": 4.0451044083526684e-06,
2752
+ "loss": 1.958,
2753
+ "step": 2130
2754
+ },
2755
+ {
2756
+ "epoch": 4.95,
2757
+ "learning_rate": 4.000750193348801e-06,
2758
+ "loss": 1.9232,
2759
+ "step": 2135
2760
+ },
2761
+ {
2762
+ "epoch": 4.97,
2763
+ "learning_rate": 3.956395978344934e-06,
2764
+ "loss": 1.8868,
2765
+ "step": 2140
2766
+ },
2767
+ {
2768
+ "epoch": 4.98,
2769
+ "learning_rate": 3.912041763341068e-06,
2770
+ "loss": 1.8728,
2771
+ "step": 2145
2772
+ },
2773
+ {
2774
+ "epoch": 4.99,
2775
+ "learning_rate": 3.867687548337201e-06,
2776
+ "loss": 1.9234,
2777
+ "step": 2150
2778
+ },
2779
+ {
2780
+ "epoch": 5.0,
2781
+ "learning_rate": 3.823333333333333e-06,
2782
+ "loss": 1.9097,
2783
+ "step": 2155
2784
+ },
2785
+ {
2786
+ "epoch": 5.01,
2787
+ "learning_rate": 3.7789791183294666e-06,
2788
+ "loss": 1.9308,
2789
+ "step": 2160
2790
+ },
2791
+ {
2792
+ "epoch": 5.02,
2793
+ "learning_rate": 3.7346249033255995e-06,
2794
+ "loss": 1.9104,
2795
+ "step": 2165
2796
+ },
2797
+ {
2798
+ "epoch": 5.03,
2799
+ "learning_rate": 3.690270688321733e-06,
2800
+ "loss": 1.8072,
2801
+ "step": 2170
2802
+ },
2803
+ {
2804
+ "epoch": 5.05,
2805
+ "learning_rate": 3.6459164733178657e-06,
2806
+ "loss": 1.9154,
2807
+ "step": 2175
2808
+ },
2809
+ {
2810
+ "epoch": 5.06,
2811
+ "learning_rate": 3.601562258313998e-06,
2812
+ "loss": 1.8595,
2813
+ "step": 2180
2814
+ },
2815
+ {
2816
+ "epoch": 5.07,
2817
+ "learning_rate": 3.557208043310132e-06,
2818
+ "loss": 1.8349,
2819
+ "step": 2185
2820
+ },
2821
+ {
2822
+ "epoch": 5.08,
2823
+ "learning_rate": 3.5128538283062643e-06,
2824
+ "loss": 1.8689,
2825
+ "step": 2190
2826
+ },
2827
+ {
2828
+ "epoch": 5.09,
2829
+ "learning_rate": 3.4684996133023976e-06,
2830
+ "loss": 1.8304,
2831
+ "step": 2195
2832
+ },
2833
+ {
2834
+ "epoch": 5.1,
2835
+ "learning_rate": 3.4241453982985305e-06,
2836
+ "loss": 1.9408,
2837
+ "step": 2200
2838
+ },
2839
+ {
2840
+ "epoch": 5.1,
2841
+ "eval_accuracy": 0.5974516356361818,
2842
+ "eval_loss": 2.2409677505493164,
2843
+ "eval_runtime": 58.2193,
2844
+ "eval_samples_per_second": 7.386,
2845
+ "eval_steps_per_second": 1.855,
2846
+ "step": 2200
2847
+ },
2848
+ {
2849
+ "epoch": 5.12,
2850
+ "learning_rate": 3.3797911832946634e-06,
2851
+ "loss": 1.8922,
2852
+ "step": 2205
2853
+ },
2854
+ {
2855
+ "epoch": 5.13,
2856
+ "learning_rate": 3.3354369682907967e-06,
2857
+ "loss": 1.8482,
2858
+ "step": 2210
2859
+ },
2860
+ {
2861
+ "epoch": 5.14,
2862
+ "learning_rate": 3.2910827532869296e-06,
2863
+ "loss": 1.9108,
2864
+ "step": 2215
2865
+ },
2866
+ {
2867
+ "epoch": 5.15,
2868
+ "learning_rate": 3.246728538283063e-06,
2869
+ "loss": 1.8998,
2870
+ "step": 2220
2871
+ },
2872
+ {
2873
+ "epoch": 5.16,
2874
+ "learning_rate": 3.2023743232791958e-06,
2875
+ "loss": 1.8573,
2876
+ "step": 2225
2877
+ },
2878
+ {
2879
+ "epoch": 5.17,
2880
+ "learning_rate": 3.158020108275329e-06,
2881
+ "loss": 1.8576,
2882
+ "step": 2230
2883
+ },
2884
+ {
2885
+ "epoch": 5.19,
2886
+ "learning_rate": 3.113665893271462e-06,
2887
+ "loss": 1.9237,
2888
+ "step": 2235
2889
+ },
2890
+ {
2891
+ "epoch": 5.2,
2892
+ "learning_rate": 3.0693116782675944e-06,
2893
+ "loss": 1.8795,
2894
+ "step": 2240
2895
+ },
2896
+ {
2897
+ "epoch": 5.21,
2898
+ "learning_rate": 3.024957463263728e-06,
2899
+ "loss": 1.8451,
2900
+ "step": 2245
2901
+ },
2902
+ {
2903
+ "epoch": 5.22,
2904
+ "learning_rate": 2.9806032482598606e-06,
2905
+ "loss": 1.8141,
2906
+ "step": 2250
2907
+ },
2908
+ {
2909
+ "epoch": 5.23,
2910
+ "learning_rate": 2.936249033255994e-06,
2911
+ "loss": 1.8714,
2912
+ "step": 2255
2913
+ },
2914
+ {
2915
+ "epoch": 5.24,
2916
+ "learning_rate": 2.891894818252127e-06,
2917
+ "loss": 1.8939,
2918
+ "step": 2260
2919
+ },
2920
+ {
2921
+ "epoch": 5.26,
2922
+ "learning_rate": 2.84754060324826e-06,
2923
+ "loss": 1.8502,
2924
+ "step": 2265
2925
+ },
2926
+ {
2927
+ "epoch": 5.27,
2928
+ "learning_rate": 2.803186388244393e-06,
2929
+ "loss": 1.8796,
2930
+ "step": 2270
2931
+ },
2932
+ {
2933
+ "epoch": 5.28,
2934
+ "learning_rate": 2.7588321732405263e-06,
2935
+ "loss": 1.884,
2936
+ "step": 2275
2937
+ },
2938
+ {
2939
+ "epoch": 5.29,
2940
+ "learning_rate": 2.7144779582366588e-06,
2941
+ "loss": 1.9446,
2942
+ "step": 2280
2943
+ },
2944
+ {
2945
+ "epoch": 5.3,
2946
+ "learning_rate": 2.670123743232792e-06,
2947
+ "loss": 1.912,
2948
+ "step": 2285
2949
+ },
2950
+ {
2951
+ "epoch": 5.31,
2952
+ "learning_rate": 2.625769528228925e-06,
2953
+ "loss": 1.9372,
2954
+ "step": 2290
2955
+ },
2956
+ {
2957
+ "epoch": 5.32,
2958
+ "learning_rate": 2.5814153132250583e-06,
2959
+ "loss": 1.873,
2960
+ "step": 2295
2961
+ },
2962
+ {
2963
+ "epoch": 5.34,
2964
+ "learning_rate": 2.537061098221191e-06,
2965
+ "loss": 1.9168,
2966
+ "step": 2300
2967
+ },
2968
+ {
2969
+ "epoch": 5.34,
2970
+ "eval_accuracy": 0.5975857600763828,
2971
+ "eval_loss": 2.239448070526123,
2972
+ "eval_runtime": 58.2108,
2973
+ "eval_samples_per_second": 7.387,
2974
+ "eval_steps_per_second": 1.855,
2975
+ "step": 2300
2976
+ },
2977
+ {
2978
+ "epoch": 5.35,
2979
+ "learning_rate": 2.492706883217324e-06,
2980
+ "loss": 1.8444,
2981
+ "step": 2305
2982
+ },
2983
+ {
2984
+ "epoch": 5.36,
2985
+ "learning_rate": 2.448352668213457e-06,
2986
+ "loss": 1.9179,
2987
+ "step": 2310
2988
+ },
2989
+ {
2990
+ "epoch": 5.37,
2991
+ "learning_rate": 2.4039984532095902e-06,
2992
+ "loss": 1.8802,
2993
+ "step": 2315
2994
+ },
2995
+ {
2996
+ "epoch": 5.38,
2997
+ "learning_rate": 2.359644238205723e-06,
2998
+ "loss": 1.8798,
2999
+ "step": 2320
3000
+ },
3001
+ {
3002
+ "epoch": 5.39,
3003
+ "learning_rate": 2.3152900232018564e-06,
3004
+ "loss": 1.922,
3005
+ "step": 2325
3006
+ },
3007
+ {
3008
+ "epoch": 5.41,
3009
+ "learning_rate": 2.2709358081979893e-06,
3010
+ "loss": 1.8982,
3011
+ "step": 2330
3012
+ },
3013
+ {
3014
+ "epoch": 5.42,
3015
+ "learning_rate": 2.226581593194122e-06,
3016
+ "loss": 1.853,
3017
+ "step": 2335
3018
+ },
3019
+ {
3020
+ "epoch": 5.43,
3021
+ "learning_rate": 2.182227378190255e-06,
3022
+ "loss": 1.8815,
3023
+ "step": 2340
3024
+ },
3025
+ {
3026
+ "epoch": 5.44,
3027
+ "learning_rate": 2.1378731631863884e-06,
3028
+ "loss": 1.8678,
3029
+ "step": 2345
3030
+ },
3031
+ {
3032
+ "epoch": 5.45,
3033
+ "learning_rate": 2.0935189481825213e-06,
3034
+ "loss": 1.7827,
3035
+ "step": 2350
3036
+ },
3037
+ {
3038
+ "epoch": 5.46,
3039
+ "learning_rate": 2.0491647331786546e-06,
3040
+ "loss": 1.797,
3041
+ "step": 2355
3042
+ },
3043
+ {
3044
+ "epoch": 5.48,
3045
+ "learning_rate": 2.0048105181747875e-06,
3046
+ "loss": 1.8769,
3047
+ "step": 2360
3048
+ },
3049
+ {
3050
+ "epoch": 5.49,
3051
+ "learning_rate": 1.9604563031709203e-06,
3052
+ "loss": 1.8924,
3053
+ "step": 2365
3054
+ },
3055
+ {
3056
+ "epoch": 5.5,
3057
+ "learning_rate": 1.9161020881670532e-06,
3058
+ "loss": 1.9016,
3059
+ "step": 2370
3060
+ },
3061
+ {
3062
+ "epoch": 5.51,
3063
+ "learning_rate": 1.8717478731631863e-06,
3064
+ "loss": 1.838,
3065
+ "step": 2375
3066
+ },
3067
+ {
3068
+ "epoch": 5.52,
3069
+ "learning_rate": 1.8273936581593194e-06,
3070
+ "loss": 1.8716,
3071
+ "step": 2380
3072
+ },
3073
+ {
3074
+ "epoch": 5.53,
3075
+ "learning_rate": 1.7830394431554525e-06,
3076
+ "loss": 1.8949,
3077
+ "step": 2385
3078
+ },
3079
+ {
3080
+ "epoch": 5.55,
3081
+ "learning_rate": 1.7386852281515856e-06,
3082
+ "loss": 1.9195,
3083
+ "step": 2390
3084
+ },
3085
+ {
3086
+ "epoch": 5.56,
3087
+ "learning_rate": 1.6943310131477185e-06,
3088
+ "loss": 1.917,
3089
+ "step": 2395
3090
+ },
3091
+ {
3092
+ "epoch": 5.57,
3093
+ "learning_rate": 1.6499767981438514e-06,
3094
+ "loss": 1.8002,
3095
+ "step": 2400
3096
+ },
3097
+ {
3098
+ "epoch": 5.57,
3099
+ "eval_accuracy": 0.5977403441769533,
3100
+ "eval_loss": 2.238140106201172,
3101
+ "eval_runtime": 58.2064,
3102
+ "eval_samples_per_second": 7.388,
3103
+ "eval_steps_per_second": 1.855,
3104
+ "step": 2400
3105
+ },
3106
+ {
3107
+ "epoch": 5.58,
3108
+ "learning_rate": 1.6056225831399845e-06,
3109
+ "loss": 1.8851,
3110
+ "step": 2405
3111
+ },
3112
+ {
3113
+ "epoch": 5.59,
3114
+ "learning_rate": 1.5612683681361176e-06,
3115
+ "loss": 1.9006,
3116
+ "step": 2410
3117
+ },
3118
+ {
3119
+ "epoch": 5.6,
3120
+ "learning_rate": 1.5169141531322507e-06,
3121
+ "loss": 1.863,
3122
+ "step": 2415
3123
+ },
3124
+ {
3125
+ "epoch": 5.61,
3126
+ "learning_rate": 1.4725599381283838e-06,
3127
+ "loss": 1.9131,
3128
+ "step": 2420
3129
+ },
3130
+ {
3131
+ "epoch": 5.63,
3132
+ "learning_rate": 1.4282057231245166e-06,
3133
+ "loss": 1.8538,
3134
+ "step": 2425
3135
+ },
3136
+ {
3137
+ "epoch": 5.64,
3138
+ "learning_rate": 1.3838515081206495e-06,
3139
+ "loss": 1.8868,
3140
+ "step": 2430
3141
+ },
3142
+ {
3143
+ "epoch": 5.65,
3144
+ "learning_rate": 1.3394972931167826e-06,
3145
+ "loss": 1.8671,
3146
+ "step": 2435
3147
+ },
3148
+ {
3149
+ "epoch": 5.66,
3150
+ "learning_rate": 1.2951430781129157e-06,
3151
+ "loss": 1.922,
3152
+ "step": 2440
3153
+ },
3154
+ {
3155
+ "epoch": 5.67,
3156
+ "learning_rate": 1.2507888631090486e-06,
3157
+ "loss": 1.816,
3158
+ "step": 2445
3159
+ },
3160
+ {
3161
+ "epoch": 5.68,
3162
+ "learning_rate": 1.2064346481051817e-06,
3163
+ "loss": 1.8375,
3164
+ "step": 2450
3165
+ },
3166
+ {
3167
+ "epoch": 5.7,
3168
+ "learning_rate": 1.1620804331013148e-06,
3169
+ "loss": 1.8747,
3170
+ "step": 2455
3171
+ },
3172
+ {
3173
+ "epoch": 5.71,
3174
+ "learning_rate": 1.1177262180974477e-06,
3175
+ "loss": 1.8866,
3176
+ "step": 2460
3177
+ },
3178
+ {
3179
+ "epoch": 5.72,
3180
+ "learning_rate": 1.0733720030935808e-06,
3181
+ "loss": 1.8335,
3182
+ "step": 2465
3183
+ },
3184
+ {
3185
+ "epoch": 5.73,
3186
+ "learning_rate": 1.0290177880897139e-06,
3187
+ "loss": 1.8518,
3188
+ "step": 2470
3189
+ },
3190
+ {
3191
+ "epoch": 5.74,
3192
+ "learning_rate": 9.846635730858468e-07,
3193
+ "loss": 1.957,
3194
+ "step": 2475
3195
+ },
3196
+ {
3197
+ "epoch": 5.75,
3198
+ "learning_rate": 9.4030935808198e-07,
3199
+ "loss": 1.9421,
3200
+ "step": 2480
3201
+ },
3202
+ {
3203
+ "epoch": 5.77,
3204
+ "learning_rate": 8.959551430781131e-07,
3205
+ "loss": 1.8889,
3206
+ "step": 2485
3207
+ },
3208
+ {
3209
+ "epoch": 5.78,
3210
+ "learning_rate": 8.516009280742459e-07,
3211
+ "loss": 1.8684,
3212
+ "step": 2490
3213
+ },
3214
+ {
3215
+ "epoch": 5.79,
3216
+ "learning_rate": 8.07246713070379e-07,
3217
+ "loss": 1.8269,
3218
+ "step": 2495
3219
+ },
3220
+ {
3221
+ "epoch": 5.8,
3222
+ "learning_rate": 7.628924980665121e-07,
3223
+ "loss": 1.8888,
3224
+ "step": 2500
3225
+ },
3226
+ {
3227
+ "epoch": 5.8,
3228
+ "eval_accuracy": 0.5977812634976926,
3229
+ "eval_loss": 2.2366693019866943,
3230
+ "eval_runtime": 58.2106,
3231
+ "eval_samples_per_second": 7.387,
3232
+ "eval_steps_per_second": 1.855,
3233
+ "step": 2500
3234
+ },
3235
+ {
3236
+ "epoch": 5.81,
3237
+ "learning_rate": 7.18538283062645e-07,
3238
+ "loss": 1.8787,
3239
+ "step": 2505
3240
+ },
3241
+ {
3242
+ "epoch": 5.82,
3243
+ "learning_rate": 6.741840680587781e-07,
3244
+ "loss": 1.8871,
3245
+ "step": 2510
3246
+ },
3247
+ {
3248
+ "epoch": 5.84,
3249
+ "learning_rate": 6.298298530549111e-07,
3250
+ "loss": 1.9249,
3251
+ "step": 2515
3252
+ },
3253
+ {
3254
+ "epoch": 5.85,
3255
+ "learning_rate": 5.854756380510441e-07,
3256
+ "loss": 1.8683,
3257
+ "step": 2520
3258
+ },
3259
+ {
3260
+ "epoch": 5.86,
3261
+ "learning_rate": 5.411214230471772e-07,
3262
+ "loss": 1.8578,
3263
+ "step": 2525
3264
+ },
3265
+ {
3266
+ "epoch": 5.87,
3267
+ "learning_rate": 4.967672080433102e-07,
3268
+ "loss": 1.9025,
3269
+ "step": 2530
3270
+ },
3271
+ {
3272
+ "epoch": 5.88,
3273
+ "learning_rate": 4.5241299303944316e-07,
3274
+ "loss": 1.9059,
3275
+ "step": 2535
3276
+ },
3277
+ {
3278
+ "epoch": 5.89,
3279
+ "learning_rate": 4.080587780355762e-07,
3280
+ "loss": 1.8879,
3281
+ "step": 2540
3282
+ },
3283
+ {
3284
+ "epoch": 5.9,
3285
+ "learning_rate": 3.6370456303170925e-07,
3286
+ "loss": 1.8826,
3287
+ "step": 2545
3288
+ },
3289
+ {
3290
+ "epoch": 5.92,
3291
+ "learning_rate": 3.1935034802784224e-07,
3292
+ "loss": 1.8687,
3293
+ "step": 2550
3294
+ },
3295
+ {
3296
+ "epoch": 5.93,
3297
+ "learning_rate": 2.749961330239753e-07,
3298
+ "loss": 1.8749,
3299
+ "step": 2555
3300
+ },
3301
+ {
3302
+ "epoch": 5.94,
3303
+ "learning_rate": 2.306419180201083e-07,
3304
+ "loss": 1.8967,
3305
+ "step": 2560
3306
+ },
3307
+ {
3308
+ "epoch": 5.95,
3309
+ "learning_rate": 1.862877030162413e-07,
3310
+ "loss": 1.8327,
3311
+ "step": 2565
3312
+ },
3313
+ {
3314
+ "epoch": 5.96,
3315
+ "learning_rate": 1.4193348801237433e-07,
3316
+ "loss": 1.8679,
3317
+ "step": 2570
3318
+ },
3319
+ {
3320
+ "epoch": 5.97,
3321
+ "learning_rate": 9.757927300850735e-08,
3322
+ "loss": 1.9444,
3323
+ "step": 2575
3324
+ },
3325
+ {
3326
+ "epoch": 5.99,
3327
+ "learning_rate": 5.322505800464037e-08,
3328
+ "loss": 1.821,
3329
+ "step": 2580
3330
+ },
3331
+ {
3332
+ "epoch": 6.0,
3333
+ "learning_rate": 8.870843000773396e-09,
3334
+ "loss": 1.9073,
3335
+ "step": 2585
3336
+ },
3337
+ {
3338
+ "epoch": 6.0,
3339
+ "step": 2586,
3340
+ "total_flos": 4.49946006847488e+16,
3341
+ "train_loss": 2.103886123784088,
3342
+ "train_runtime": 6222.0683,
3343
+ "train_samples_per_second": 1.662,
3344
+ "train_steps_per_second": 0.416
3345
+ }
3346
+ ],
3347
+ "max_steps": 2586,
3348
+ "num_train_epochs": 6,
3349
+ "total_flos": 4.49946006847488e+16,
3350
+ "trial_name": null,
3351
+ "trial_params": null
3352
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6b5dab74e476d5b6df53fd4c78148028a02e2501b93ecd2e51ad20482829c3e
3
+ size 3451
vocab.json ADDED
The diff for this file is too large to render. See raw diff