Hugging Face's logo

shayekh
/
aya8b-distillkit-hidden

Safetensors
Model card Files Community
shayekh commited on
Commit
7c91766
•
1 Parent(s): 2a3ab4e

Upload folder using huggingface_hub

Browse files
Files changed (14) hide show
  1. checkpoint-1000/added_tokens.json +5 -0
  2. checkpoint-1000/config.json +28 -0
  3. checkpoint-1000/generation_config.json +6 -0
  4. checkpoint-1000/merges.txt +0 -0
  5. checkpoint-1000/model.safetensors +3 -0
  6. checkpoint-1000/optimizer.pt +3 -0
  7. checkpoint-1000/rng_state.pth +3 -0
  8. checkpoint-1000/scheduler.pt +3 -0
  9. checkpoint-1000/special_tokens_map.json +20 -0
  10. checkpoint-1000/tokenizer.json +0 -0
  11. checkpoint-1000/tokenizer_config.json +43 -0
  12. checkpoint-1000/trainer_state.json +3533 -0
  13. checkpoint-1000/training_args.bin +3 -0
  14. checkpoint-1000/vocab.json +0 -0
checkpoint-1000/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "<|endoftext|>": 151643,
3
+ "<|im_end|>": 151645,
4
+ "<|im_start|>": 151644
5
+ }
checkpoint-1000/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "Qwen/Qwen2-1.5B",
3
+ "architectures": [
4
+ "Qwen2ForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 151643,
8
+ "eos_token_id": 151643,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 1536,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 8960,
13
+ "max_position_embeddings": 131072,
14
+ "max_window_layers": 28,
15
+ "model_type": "qwen2",
16
+ "num_attention_heads": 12,
17
+ "num_hidden_layers": 28,
18
+ "num_key_value_heads": 2,
19
+ "rms_norm_eps": 1e-06,
20
+ "rope_theta": 1000000.0,
21
+ "sliding_window": null,
22
+ "tie_word_embeddings": true,
23
+ "torch_dtype": "bfloat16",
24
+ "transformers_version": "4.44.0",
25
+ "use_cache": true,
26
+ "use_sliding_window": false,
27
+ "vocab_size": 151936
28
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "eos_token_id": 151643,
4
+ "max_new_tokens": 2048,
5
+ "transformers_version": "4.44.0"
6
+ }
checkpoint-1000/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ab02b74c75259871958ea3fcc8309f061c6671cb42caabae70411cc9d1054d4
3
+ size 3087467144
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e8405e79dee1c3994dfb917f10fb0ceb9724e286a1b9b5e896f6d570af320dc
3
+ size 6175147196
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e6bdf71b784d7fc380e872c2786bd835681df5b8b45ffd8a17dc718ecedc1d0
3
+ size 14244
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1bb085c75ddb081ce4c8a41d6181d17e522362913007350c20e7296cce021592
3
+ size 1064
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>"
5
+ ],
6
+ "eos_token": {
7
+ "content": "<|endoftext|>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false
12
+ },
13
+ "pad_token": {
14
+ "content": "<|endoftext|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false
19
+ }
20
+ }
checkpoint-1000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1000/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "151643": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "151644": {
13
+ "content": "<|im_start|>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "151645": {
21
+ "content": "<|im_end|>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ }
28
+ },
29
+ "additional_special_tokens": [
30
+ "<|im_start|>",
31
+ "<|im_end|>"
32
+ ],
33
+ "bos_token": null,
34
+ "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
35
+ "clean_up_tokenization_spaces": false,
36
+ "eos_token": "<|endoftext|>",
37
+ "errors": "replace",
38
+ "model_max_length": 32768,
39
+ "pad_token": "<|endoftext|>",
40
+ "split_special_tokens": false,
41
+ "tokenizer_class": "Qwen2Tokenizer",
42
+ "unk_token": null
43
+ }
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,3533 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.5845389449072045,
5
+ "eval_steps": 500,
6
+ "global_step": 1000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.001169077889814409,
13
+ "grad_norm": 6.78125,
14
+ "learning_rate": 3.898635477582846e-07,
15
+ "loss": 5.0851,
16
+ "step": 2
17
+ },
18
+ {
19
+ "epoch": 0.002338155779628818,
20
+ "grad_norm": 7.0625,
21
+ "learning_rate": 7.797270955165692e-07,
22
+ "loss": 5.1921,
23
+ "step": 4
24
+ },
25
+ {
26
+ "epoch": 0.0035072336694432268,
27
+ "grad_norm": 7.34375,
28
+ "learning_rate": 1.1695906432748538e-06,
29
+ "loss": 5.0756,
30
+ "step": 6
31
+ },
32
+ {
33
+ "epoch": 0.004676311559257636,
34
+ "grad_norm": 7.03125,
35
+ "learning_rate": 1.5594541910331385e-06,
36
+ "loss": 5.1097,
37
+ "step": 8
38
+ },
39
+ {
40
+ "epoch": 0.005845389449072045,
41
+ "grad_norm": 6.40625,
42
+ "learning_rate": 1.949317738791423e-06,
43
+ "loss": 5.0591,
44
+ "step": 10
45
+ },
46
+ {
47
+ "epoch": 0.0070144673388864535,
48
+ "grad_norm": 5.8125,
49
+ "learning_rate": 2.3391812865497075e-06,
50
+ "loss": 5.096,
51
+ "step": 12
52
+ },
53
+ {
54
+ "epoch": 0.008183545228700862,
55
+ "grad_norm": 5.1875,
56
+ "learning_rate": 2.729044834307992e-06,
57
+ "loss": 5.0234,
58
+ "step": 14
59
+ },
60
+ {
61
+ "epoch": 0.009352623118515271,
62
+ "grad_norm": 5.03125,
63
+ "learning_rate": 3.118908382066277e-06,
64
+ "loss": 5.0266,
65
+ "step": 16
66
+ },
67
+ {
68
+ "epoch": 0.01052170100832968,
69
+ "grad_norm": 4.90625,
70
+ "learning_rate": 3.5087719298245615e-06,
71
+ "loss": 5.031,
72
+ "step": 18
73
+ },
74
+ {
75
+ "epoch": 0.01169077889814409,
76
+ "grad_norm": 4.90625,
77
+ "learning_rate": 3.898635477582846e-06,
78
+ "loss": 4.9845,
79
+ "step": 20
80
+ },
81
+ {
82
+ "epoch": 0.012859856787958498,
83
+ "grad_norm": 4.6875,
84
+ "learning_rate": 4.2884990253411305e-06,
85
+ "loss": 4.9933,
86
+ "step": 22
87
+ },
88
+ {
89
+ "epoch": 0.014028934677772907,
90
+ "grad_norm": 4.625,
91
+ "learning_rate": 4.678362573099415e-06,
92
+ "loss": 5.0822,
93
+ "step": 24
94
+ },
95
+ {
96
+ "epoch": 0.015198012567587316,
97
+ "grad_norm": 4.71875,
98
+ "learning_rate": 5.0682261208576995e-06,
99
+ "loss": 4.9392,
100
+ "step": 26
101
+ },
102
+ {
103
+ "epoch": 0.016367090457401725,
104
+ "grad_norm": 4.65625,
105
+ "learning_rate": 5.458089668615984e-06,
106
+ "loss": 5.0441,
107
+ "step": 28
108
+ },
109
+ {
110
+ "epoch": 0.017536168347216132,
111
+ "grad_norm": 4.6875,
112
+ "learning_rate": 5.8479532163742686e-06,
113
+ "loss": 4.8865,
114
+ "step": 30
115
+ },
116
+ {
117
+ "epoch": 0.018705246237030543,
118
+ "grad_norm": 4.59375,
119
+ "learning_rate": 6.237816764132554e-06,
120
+ "loss": 4.9227,
121
+ "step": 32
122
+ },
123
+ {
124
+ "epoch": 0.01987432412684495,
125
+ "grad_norm": 4.65625,
126
+ "learning_rate": 6.6276803118908384e-06,
127
+ "loss": 4.9035,
128
+ "step": 34
129
+ },
130
+ {
131
+ "epoch": 0.02104340201665936,
132
+ "grad_norm": 4.59375,
133
+ "learning_rate": 7.017543859649123e-06,
134
+ "loss": 4.863,
135
+ "step": 36
136
+ },
137
+ {
138
+ "epoch": 0.022212479906473768,
139
+ "grad_norm": 4.71875,
140
+ "learning_rate": 7.4074074074074075e-06,
141
+ "loss": 4.7735,
142
+ "step": 38
143
+ },
144
+ {
145
+ "epoch": 0.02338155779628818,
146
+ "grad_norm": 4.84375,
147
+ "learning_rate": 7.797270955165692e-06,
148
+ "loss": 4.7932,
149
+ "step": 40
150
+ },
151
+ {
152
+ "epoch": 0.024550635686102586,
153
+ "grad_norm": 5.3125,
154
+ "learning_rate": 8.187134502923977e-06,
155
+ "loss": 4.6933,
156
+ "step": 42
157
+ },
158
+ {
159
+ "epoch": 0.025719713575916996,
160
+ "grad_norm": 7.15625,
161
+ "learning_rate": 8.576998050682261e-06,
162
+ "loss": 4.5906,
163
+ "step": 44
164
+ },
165
+ {
166
+ "epoch": 0.026888791465731404,
167
+ "grad_norm": 8.8125,
168
+ "learning_rate": 8.966861598440546e-06,
169
+ "loss": 4.4291,
170
+ "step": 46
171
+ },
172
+ {
173
+ "epoch": 0.028057869355545814,
174
+ "grad_norm": 9.6875,
175
+ "learning_rate": 9.35672514619883e-06,
176
+ "loss": 4.221,
177
+ "step": 48
178
+ },
179
+ {
180
+ "epoch": 0.02922694724536022,
181
+ "grad_norm": 8.8125,
182
+ "learning_rate": 9.746588693957115e-06,
183
+ "loss": 4.032,
184
+ "step": 50
185
+ },
186
+ {
187
+ "epoch": 0.030396025135174632,
188
+ "grad_norm": 9.0,
189
+ "learning_rate": 1.0136452241715399e-05,
190
+ "loss": 3.8603,
191
+ "step": 52
192
+ },
193
+ {
194
+ "epoch": 0.03156510302498904,
195
+ "grad_norm": 7.46875,
196
+ "learning_rate": 1.0526315789473684e-05,
197
+ "loss": 3.6522,
198
+ "step": 54
199
+ },
200
+ {
201
+ "epoch": 0.03273418091480345,
202
+ "grad_norm": 8.375,
203
+ "learning_rate": 1.0916179337231968e-05,
204
+ "loss": 3.4394,
205
+ "step": 56
206
+ },
207
+ {
208
+ "epoch": 0.03390325880461786,
209
+ "grad_norm": 6.4375,
210
+ "learning_rate": 1.1306042884990253e-05,
211
+ "loss": 3.2265,
212
+ "step": 58
213
+ },
214
+ {
215
+ "epoch": 0.035072336694432264,
216
+ "grad_norm": 5.75,
217
+ "learning_rate": 1.1695906432748537e-05,
218
+ "loss": 3.091,
219
+ "step": 60
220
+ },
221
+ {
222
+ "epoch": 0.03624141458424668,
223
+ "grad_norm": 4.8125,
224
+ "learning_rate": 1.2085769980506823e-05,
225
+ "loss": 2.8856,
226
+ "step": 62
227
+ },
228
+ {
229
+ "epoch": 0.037410492474061086,
230
+ "grad_norm": 3.8125,
231
+ "learning_rate": 1.2475633528265108e-05,
232
+ "loss": 2.6299,
233
+ "step": 64
234
+ },
235
+ {
236
+ "epoch": 0.03857957036387549,
237
+ "grad_norm": 3.234375,
238
+ "learning_rate": 1.2865497076023392e-05,
239
+ "loss": 2.5917,
240
+ "step": 66
241
+ },
242
+ {
243
+ "epoch": 0.0397486482536899,
244
+ "grad_norm": 4.0625,
245
+ "learning_rate": 1.3255360623781677e-05,
246
+ "loss": 2.3757,
247
+ "step": 68
248
+ },
249
+ {
250
+ "epoch": 0.040917726143504314,
251
+ "grad_norm": 2.453125,
252
+ "learning_rate": 1.364522417153996e-05,
253
+ "loss": 2.308,
254
+ "step": 70
255
+ },
256
+ {
257
+ "epoch": 0.04208680403331872,
258
+ "grad_norm": 2.390625,
259
+ "learning_rate": 1.4035087719298246e-05,
260
+ "loss": 2.3042,
261
+ "step": 72
262
+ },
263
+ {
264
+ "epoch": 0.04325588192313313,
265
+ "grad_norm": 2.109375,
266
+ "learning_rate": 1.442495126705653e-05,
267
+ "loss": 2.2149,
268
+ "step": 74
269
+ },
270
+ {
271
+ "epoch": 0.044424959812947536,
272
+ "grad_norm": 2.71875,
273
+ "learning_rate": 1.4814814814814815e-05,
274
+ "loss": 2.0738,
275
+ "step": 76
276
+ },
277
+ {
278
+ "epoch": 0.04559403770276195,
279
+ "grad_norm": 2.46875,
280
+ "learning_rate": 1.5204678362573099e-05,
281
+ "loss": 1.9663,
282
+ "step": 78
283
+ },
284
+ {
285
+ "epoch": 0.04676311559257636,
286
+ "grad_norm": 2.640625,
287
+ "learning_rate": 1.5594541910331384e-05,
288
+ "loss": 1.9384,
289
+ "step": 80
290
+ },
291
+ {
292
+ "epoch": 0.047932193482390764,
293
+ "grad_norm": 2.0,
294
+ "learning_rate": 1.5984405458089668e-05,
295
+ "loss": 1.894,
296
+ "step": 82
297
+ },
298
+ {
299
+ "epoch": 0.04910127137220517,
300
+ "grad_norm": 2.078125,
301
+ "learning_rate": 1.6374269005847955e-05,
302
+ "loss": 1.8204,
303
+ "step": 84
304
+ },
305
+ {
306
+ "epoch": 0.050270349262019585,
307
+ "grad_norm": 2.953125,
308
+ "learning_rate": 1.676413255360624e-05,
309
+ "loss": 1.8051,
310
+ "step": 86
311
+ },
312
+ {
313
+ "epoch": 0.05143942715183399,
314
+ "grad_norm": 2.046875,
315
+ "learning_rate": 1.7153996101364522e-05,
316
+ "loss": 1.5832,
317
+ "step": 88
318
+ },
319
+ {
320
+ "epoch": 0.0526085050416484,
321
+ "grad_norm": 1.75,
322
+ "learning_rate": 1.7543859649122806e-05,
323
+ "loss": 1.5298,
324
+ "step": 90
325
+ },
326
+ {
327
+ "epoch": 0.05377758293146281,
328
+ "grad_norm": 1.71875,
329
+ "learning_rate": 1.7933723196881093e-05,
330
+ "loss": 1.4964,
331
+ "step": 92
332
+ },
333
+ {
334
+ "epoch": 0.054946660821277214,
335
+ "grad_norm": 1.3515625,
336
+ "learning_rate": 1.8323586744639376e-05,
337
+ "loss": 1.3987,
338
+ "step": 94
339
+ },
340
+ {
341
+ "epoch": 0.05611573871109163,
342
+ "grad_norm": 1.9140625,
343
+ "learning_rate": 1.871345029239766e-05,
344
+ "loss": 1.3959,
345
+ "step": 96
346
+ },
347
+ {
348
+ "epoch": 0.057284816600906036,
349
+ "grad_norm": 1.34375,
350
+ "learning_rate": 1.9103313840155944e-05,
351
+ "loss": 1.3673,
352
+ "step": 98
353
+ },
354
+ {
355
+ "epoch": 0.05845389449072044,
356
+ "grad_norm": 1.09375,
357
+ "learning_rate": 1.949317738791423e-05,
358
+ "loss": 1.2864,
359
+ "step": 100
360
+ },
361
+ {
362
+ "epoch": 0.05962297238053485,
363
+ "grad_norm": 1.1328125,
364
+ "learning_rate": 1.9883040935672515e-05,
365
+ "loss": 1.2535,
366
+ "step": 102
367
+ },
368
+ {
369
+ "epoch": 0.060792050270349264,
370
+ "grad_norm": 1.2109375,
371
+ "learning_rate": 2.0272904483430798e-05,
372
+ "loss": 1.2242,
373
+ "step": 104
374
+ },
375
+ {
376
+ "epoch": 0.06196112816016367,
377
+ "grad_norm": 2.59375,
378
+ "learning_rate": 2.0662768031189085e-05,
379
+ "loss": 1.1936,
380
+ "step": 106
381
+ },
382
+ {
383
+ "epoch": 0.06313020604997809,
384
+ "grad_norm": 1.953125,
385
+ "learning_rate": 2.105263157894737e-05,
386
+ "loss": 1.1858,
387
+ "step": 108
388
+ },
389
+ {
390
+ "epoch": 0.06429928393979249,
391
+ "grad_norm": 2.140625,
392
+ "learning_rate": 2.1442495126705653e-05,
393
+ "loss": 1.1438,
394
+ "step": 110
395
+ },
396
+ {
397
+ "epoch": 0.0654683618296069,
398
+ "grad_norm": 1.09375,
399
+ "learning_rate": 2.1832358674463936e-05,
400
+ "loss": 1.1217,
401
+ "step": 112
402
+ },
403
+ {
404
+ "epoch": 0.06663743971942131,
405
+ "grad_norm": 1.8671875,
406
+ "learning_rate": 2.2222222222222223e-05,
407
+ "loss": 1.1508,
408
+ "step": 114
409
+ },
410
+ {
411
+ "epoch": 0.06780651760923571,
412
+ "grad_norm": 1.046875,
413
+ "learning_rate": 2.2612085769980507e-05,
414
+ "loss": 1.0976,
415
+ "step": 116
416
+ },
417
+ {
418
+ "epoch": 0.06897559549905012,
419
+ "grad_norm": 1.234375,
420
+ "learning_rate": 2.300194931773879e-05,
421
+ "loss": 1.0716,
422
+ "step": 118
423
+ },
424
+ {
425
+ "epoch": 0.07014467338886453,
426
+ "grad_norm": 1.6171875,
427
+ "learning_rate": 2.3391812865497074e-05,
428
+ "loss": 1.0152,
429
+ "step": 120
430
+ },
431
+ {
432
+ "epoch": 0.07131375127867894,
433
+ "grad_norm": 3.125,
434
+ "learning_rate": 2.378167641325536e-05,
435
+ "loss": 1.0265,
436
+ "step": 122
437
+ },
438
+ {
439
+ "epoch": 0.07248282916849336,
440
+ "grad_norm": 3.546875,
441
+ "learning_rate": 2.4171539961013645e-05,
442
+ "loss": 1.2355,
443
+ "step": 124
444
+ },
445
+ {
446
+ "epoch": 0.07365190705830776,
447
+ "grad_norm": 1.1796875,
448
+ "learning_rate": 2.456140350877193e-05,
449
+ "loss": 1.0111,
450
+ "step": 126
451
+ },
452
+ {
453
+ "epoch": 0.07482098494812217,
454
+ "grad_norm": 2.453125,
455
+ "learning_rate": 2.4951267056530216e-05,
456
+ "loss": 1.2317,
457
+ "step": 128
458
+ },
459
+ {
460
+ "epoch": 0.07599006283793658,
461
+ "grad_norm": 1.09375,
462
+ "learning_rate": 2.53411306042885e-05,
463
+ "loss": 0.9988,
464
+ "step": 130
465
+ },
466
+ {
467
+ "epoch": 0.07715914072775099,
468
+ "grad_norm": 1.09375,
469
+ "learning_rate": 2.5730994152046783e-05,
470
+ "loss": 1.0076,
471
+ "step": 132
472
+ },
473
+ {
474
+ "epoch": 0.07832821861756539,
475
+ "grad_norm": 1.4375,
476
+ "learning_rate": 2.6120857699805067e-05,
477
+ "loss": 1.0155,
478
+ "step": 134
479
+ },
480
+ {
481
+ "epoch": 0.0794972965073798,
482
+ "grad_norm": 1.796875,
483
+ "learning_rate": 2.6510721247563354e-05,
484
+ "loss": 1.1216,
485
+ "step": 136
486
+ },
487
+ {
488
+ "epoch": 0.08066637439719421,
489
+ "grad_norm": 0.9921875,
490
+ "learning_rate": 2.6900584795321637e-05,
491
+ "loss": 0.9788,
492
+ "step": 138
493
+ },
494
+ {
495
+ "epoch": 0.08183545228700863,
496
+ "grad_norm": 1.328125,
497
+ "learning_rate": 2.729044834307992e-05,
498
+ "loss": 0.9372,
499
+ "step": 140
500
+ },
501
+ {
502
+ "epoch": 0.08300453017682304,
503
+ "grad_norm": 1.296875,
504
+ "learning_rate": 2.7680311890838205e-05,
505
+ "loss": 0.897,
506
+ "step": 142
507
+ },
508
+ {
509
+ "epoch": 0.08417360806663744,
510
+ "grad_norm": 1.0703125,
511
+ "learning_rate": 2.8070175438596492e-05,
512
+ "loss": 0.9766,
513
+ "step": 144
514
+ },
515
+ {
516
+ "epoch": 0.08534268595645185,
517
+ "grad_norm": 0.84765625,
518
+ "learning_rate": 2.8460038986354776e-05,
519
+ "loss": 0.9315,
520
+ "step": 146
521
+ },
522
+ {
523
+ "epoch": 0.08651176384626626,
524
+ "grad_norm": 0.9140625,
525
+ "learning_rate": 2.884990253411306e-05,
526
+ "loss": 0.9425,
527
+ "step": 148
528
+ },
529
+ {
530
+ "epoch": 0.08768084173608066,
531
+ "grad_norm": 0.99609375,
532
+ "learning_rate": 2.9239766081871346e-05,
533
+ "loss": 0.9242,
534
+ "step": 150
535
+ },
536
+ {
537
+ "epoch": 0.08884991962589507,
538
+ "grad_norm": 1.8046875,
539
+ "learning_rate": 2.962962962962963e-05,
540
+ "loss": 0.9396,
541
+ "step": 152
542
+ },
543
+ {
544
+ "epoch": 0.09001899751570948,
545
+ "grad_norm": 0.92578125,
546
+ "learning_rate": 3.0019493177387914e-05,
547
+ "loss": 0.8932,
548
+ "step": 154
549
+ },
550
+ {
551
+ "epoch": 0.0911880754055239,
552
+ "grad_norm": 72.5,
553
+ "learning_rate": 3.0409356725146197e-05,
554
+ "loss": 0.9062,
555
+ "step": 156
556
+ },
557
+ {
558
+ "epoch": 0.0923571532953383,
559
+ "grad_norm": 1.703125,
560
+ "learning_rate": 3.0799220272904484e-05,
561
+ "loss": 0.9534,
562
+ "step": 158
563
+ },
564
+ {
565
+ "epoch": 0.09352623118515271,
566
+ "grad_norm": 1.453125,
567
+ "learning_rate": 3.118908382066277e-05,
568
+ "loss": 0.8715,
569
+ "step": 160
570
+ },
571
+ {
572
+ "epoch": 0.09469530907496712,
573
+ "grad_norm": 1.015625,
574
+ "learning_rate": 3.157894736842105e-05,
575
+ "loss": 0.9029,
576
+ "step": 162
577
+ },
578
+ {
579
+ "epoch": 0.09586438696478153,
580
+ "grad_norm": 1.15625,
581
+ "learning_rate": 3.1968810916179335e-05,
582
+ "loss": 0.8968,
583
+ "step": 164
584
+ },
585
+ {
586
+ "epoch": 0.09703346485459594,
587
+ "grad_norm": 1.4296875,
588
+ "learning_rate": 3.235867446393762e-05,
589
+ "loss": 0.9784,
590
+ "step": 166
591
+ },
592
+ {
593
+ "epoch": 0.09820254274441034,
594
+ "grad_norm": 1.0078125,
595
+ "learning_rate": 3.274853801169591e-05,
596
+ "loss": 0.9012,
597
+ "step": 168
598
+ },
599
+ {
600
+ "epoch": 0.09937162063422475,
601
+ "grad_norm": 0.8359375,
602
+ "learning_rate": 3.313840155945419e-05,
603
+ "loss": 0.8895,
604
+ "step": 170
605
+ },
606
+ {
607
+ "epoch": 0.10054069852403917,
608
+ "grad_norm": 0.8046875,
609
+ "learning_rate": 3.352826510721248e-05,
610
+ "loss": 0.8858,
611
+ "step": 172
612
+ },
613
+ {
614
+ "epoch": 0.10170977641385358,
615
+ "grad_norm": 0.83203125,
616
+ "learning_rate": 3.391812865497076e-05,
617
+ "loss": 0.8408,
618
+ "step": 174
619
+ },
620
+ {
621
+ "epoch": 0.10287885430366799,
622
+ "grad_norm": 0.8671875,
623
+ "learning_rate": 3.4307992202729044e-05,
624
+ "loss": 0.8896,
625
+ "step": 176
626
+ },
627
+ {
628
+ "epoch": 0.10404793219348239,
629
+ "grad_norm": 1.78125,
630
+ "learning_rate": 3.469785575048733e-05,
631
+ "loss": 0.8941,
632
+ "step": 178
633
+ },
634
+ {
635
+ "epoch": 0.1052170100832968,
636
+ "grad_norm": 0.8984375,
637
+ "learning_rate": 3.508771929824561e-05,
638
+ "loss": 0.9013,
639
+ "step": 180
640
+ },
641
+ {
642
+ "epoch": 0.1063860879731112,
643
+ "grad_norm": 1.2734375,
644
+ "learning_rate": 3.5477582846003895e-05,
645
+ "loss": 0.8585,
646
+ "step": 182
647
+ },
648
+ {
649
+ "epoch": 0.10755516586292561,
650
+ "grad_norm": 2.015625,
651
+ "learning_rate": 3.5867446393762186e-05,
652
+ "loss": 1.0231,
653
+ "step": 184
654
+ },
655
+ {
656
+ "epoch": 0.10872424375274002,
657
+ "grad_norm": 1.7578125,
658
+ "learning_rate": 3.625730994152047e-05,
659
+ "loss": 0.8835,
660
+ "step": 186
661
+ },
662
+ {
663
+ "epoch": 0.10989332164255443,
664
+ "grad_norm": 1.078125,
665
+ "learning_rate": 3.664717348927875e-05,
666
+ "loss": 0.8465,
667
+ "step": 188
668
+ },
669
+ {
670
+ "epoch": 0.11106239953236885,
671
+ "grad_norm": 2.75,
672
+ "learning_rate": 3.7037037037037037e-05,
673
+ "loss": 1.1014,
674
+ "step": 190
675
+ },
676
+ {
677
+ "epoch": 0.11223147742218326,
678
+ "grad_norm": 2.09375,
679
+ "learning_rate": 3.742690058479532e-05,
680
+ "loss": 0.8691,
681
+ "step": 192
682
+ },
683
+ {
684
+ "epoch": 0.11340055531199766,
685
+ "grad_norm": 0.96875,
686
+ "learning_rate": 3.7816764132553604e-05,
687
+ "loss": 0.8547,
688
+ "step": 194
689
+ },
690
+ {
691
+ "epoch": 0.11456963320181207,
692
+ "grad_norm": 1.5859375,
693
+ "learning_rate": 3.820662768031189e-05,
694
+ "loss": 0.8725,
695
+ "step": 196
696
+ },
697
+ {
698
+ "epoch": 0.11573871109162648,
699
+ "grad_norm": 0.88671875,
700
+ "learning_rate": 3.859649122807018e-05,
701
+ "loss": 0.8393,
702
+ "step": 198
703
+ },
704
+ {
705
+ "epoch": 0.11690778898144089,
706
+ "grad_norm": 1.015625,
707
+ "learning_rate": 3.898635477582846e-05,
708
+ "loss": 0.8345,
709
+ "step": 200
710
+ },
711
+ {
712
+ "epoch": 0.11807686687125529,
713
+ "grad_norm": 0.82421875,
714
+ "learning_rate": 3.9376218323586745e-05,
715
+ "loss": 0.8966,
716
+ "step": 202
717
+ },
718
+ {
719
+ "epoch": 0.1192459447610697,
720
+ "grad_norm": 0.79296875,
721
+ "learning_rate": 3.976608187134503e-05,
722
+ "loss": 0.8291,
723
+ "step": 204
724
+ },
725
+ {
726
+ "epoch": 0.12041502265088412,
727
+ "grad_norm": 0.83203125,
728
+ "learning_rate": 4.015594541910331e-05,
729
+ "loss": 0.8175,
730
+ "step": 206
731
+ },
732
+ {
733
+ "epoch": 0.12158410054069853,
734
+ "grad_norm": 0.74609375,
735
+ "learning_rate": 4.0545808966861596e-05,
736
+ "loss": 0.8353,
737
+ "step": 208
738
+ },
739
+ {
740
+ "epoch": 0.12275317843051294,
741
+ "grad_norm": 1.796875,
742
+ "learning_rate": 4.093567251461988e-05,
743
+ "loss": 1.0084,
744
+ "step": 210
745
+ },
746
+ {
747
+ "epoch": 0.12392225632032734,
748
+ "grad_norm": 59.75,
749
+ "learning_rate": 4.132553606237817e-05,
750
+ "loss": 0.8478,
751
+ "step": 212
752
+ },
753
+ {
754
+ "epoch": 0.12509133421014176,
755
+ "grad_norm": 1.265625,
756
+ "learning_rate": 4.1715399610136454e-05,
757
+ "loss": 0.8267,
758
+ "step": 214
759
+ },
760
+ {
761
+ "epoch": 0.12626041209995617,
762
+ "grad_norm": 1.0625,
763
+ "learning_rate": 4.210526315789474e-05,
764
+ "loss": 0.836,
765
+ "step": 216
766
+ },
767
+ {
768
+ "epoch": 0.12742948998977058,
769
+ "grad_norm": 1.0078125,
770
+ "learning_rate": 4.249512670565302e-05,
771
+ "loss": 0.857,
772
+ "step": 218
773
+ },
774
+ {
775
+ "epoch": 0.12859856787958499,
776
+ "grad_norm": 0.92578125,
777
+ "learning_rate": 4.2884990253411305e-05,
778
+ "loss": 0.8478,
779
+ "step": 220
780
+ },
781
+ {
782
+ "epoch": 0.1297676457693994,
783
+ "grad_norm": 0.953125,
784
+ "learning_rate": 4.327485380116959e-05,
785
+ "loss": 0.844,
786
+ "step": 222
787
+ },
788
+ {
789
+ "epoch": 0.1309367236592138,
790
+ "grad_norm": 0.796875,
791
+ "learning_rate": 4.366471734892787e-05,
792
+ "loss": 0.8128,
793
+ "step": 224
794
+ },
795
+ {
796
+ "epoch": 0.1321058015490282,
797
+ "grad_norm": 1.28125,
798
+ "learning_rate": 4.4054580896686156e-05,
799
+ "loss": 0.8276,
800
+ "step": 226
801
+ },
802
+ {
803
+ "epoch": 0.13327487943884261,
804
+ "grad_norm": 0.94921875,
805
+ "learning_rate": 4.4444444444444447e-05,
806
+ "loss": 0.8478,
807
+ "step": 228
808
+ },
809
+ {
810
+ "epoch": 0.13444395732865702,
811
+ "grad_norm": 1.0703125,
812
+ "learning_rate": 4.483430799220273e-05,
813
+ "loss": 0.814,
814
+ "step": 230
815
+ },
816
+ {
817
+ "epoch": 0.13561303521847143,
818
+ "grad_norm": 1.1328125,
819
+ "learning_rate": 4.5224171539961014e-05,
820
+ "loss": 0.765,
821
+ "step": 232
822
+ },
823
+ {
824
+ "epoch": 0.13678211310828584,
825
+ "grad_norm": 0.87109375,
826
+ "learning_rate": 4.56140350877193e-05,
827
+ "loss": 0.8277,
828
+ "step": 234
829
+ },
830
+ {
831
+ "epoch": 0.13795119099810024,
832
+ "grad_norm": 0.921875,
833
+ "learning_rate": 4.600389863547758e-05,
834
+ "loss": 0.8103,
835
+ "step": 236
836
+ },
837
+ {
838
+ "epoch": 0.13912026888791465,
839
+ "grad_norm": 1.4921875,
840
+ "learning_rate": 4.6393762183235865e-05,
841
+ "loss": 0.7897,
842
+ "step": 238
843
+ },
844
+ {
845
+ "epoch": 0.14028934677772906,
846
+ "grad_norm": 0.7734375,
847
+ "learning_rate": 4.678362573099415e-05,
848
+ "loss": 0.7798,
849
+ "step": 240
850
+ },
851
+ {
852
+ "epoch": 0.14145842466754346,
853
+ "grad_norm": 0.98828125,
854
+ "learning_rate": 4.717348927875244e-05,
855
+ "loss": 0.8442,
856
+ "step": 242
857
+ },
858
+ {
859
+ "epoch": 0.14262750255735787,
860
+ "grad_norm": 0.91796875,
861
+ "learning_rate": 4.756335282651072e-05,
862
+ "loss": 0.7936,
863
+ "step": 244
864
+ },
865
+ {
866
+ "epoch": 0.1437965804471723,
867
+ "grad_norm": 0.859375,
868
+ "learning_rate": 4.7953216374269006e-05,
869
+ "loss": 0.7726,
870
+ "step": 246
871
+ },
872
+ {
873
+ "epoch": 0.1449656583369867,
874
+ "grad_norm": 1.3671875,
875
+ "learning_rate": 4.834307992202729e-05,
876
+ "loss": 0.9626,
877
+ "step": 248
878
+ },
879
+ {
880
+ "epoch": 0.14613473622680112,
881
+ "grad_norm": 0.921875,
882
+ "learning_rate": 4.8732943469785574e-05,
883
+ "loss": 0.7929,
884
+ "step": 250
885
+ },
886
+ {
887
+ "epoch": 0.14730381411661553,
888
+ "grad_norm": 0.92578125,
889
+ "learning_rate": 4.912280701754386e-05,
890
+ "loss": 0.8589,
891
+ "step": 252
892
+ },
893
+ {
894
+ "epoch": 0.14847289200642994,
895
+ "grad_norm": 0.83203125,
896
+ "learning_rate": 4.951267056530214e-05,
897
+ "loss": 0.7383,
898
+ "step": 254
899
+ },
900
+ {
901
+ "epoch": 0.14964196989624434,
902
+ "grad_norm": 1.046875,
903
+ "learning_rate": 4.990253411306043e-05,
904
+ "loss": 0.8011,
905
+ "step": 256
906
+ },
907
+ {
908
+ "epoch": 0.15081104778605875,
909
+ "grad_norm": 1.3359375,
910
+ "learning_rate": 5.0292397660818715e-05,
911
+ "loss": 0.8404,
912
+ "step": 258
913
+ },
914
+ {
915
+ "epoch": 0.15198012567587316,
916
+ "grad_norm": 1.21875,
917
+ "learning_rate": 5.0682261208577e-05,
918
+ "loss": 0.8301,
919
+ "step": 260
920
+ },
921
+ {
922
+ "epoch": 0.15314920356568756,
923
+ "grad_norm": 0.93359375,
924
+ "learning_rate": 5.107212475633528e-05,
925
+ "loss": 0.8157,
926
+ "step": 262
927
+ },
928
+ {
929
+ "epoch": 0.15431828145550197,
930
+ "grad_norm": 1.3203125,
931
+ "learning_rate": 5.1461988304093566e-05,
932
+ "loss": 0.7724,
933
+ "step": 264
934
+ },
935
+ {
936
+ "epoch": 0.15548735934531638,
937
+ "grad_norm": 1.09375,
938
+ "learning_rate": 5.185185185185185e-05,
939
+ "loss": 0.7586,
940
+ "step": 266
941
+ },
942
+ {
943
+ "epoch": 0.15665643723513079,
944
+ "grad_norm": 1.1484375,
945
+ "learning_rate": 5.2241715399610133e-05,
946
+ "loss": 0.7812,
947
+ "step": 268
948
+ },
949
+ {
950
+ "epoch": 0.1578255151249452,
951
+ "grad_norm": 0.74609375,
952
+ "learning_rate": 5.2631578947368424e-05,
953
+ "loss": 0.7486,
954
+ "step": 270
955
+ },
956
+ {
957
+ "epoch": 0.1589945930147596,
958
+ "grad_norm": 0.81640625,
959
+ "learning_rate": 5.302144249512671e-05,
960
+ "loss": 0.7339,
961
+ "step": 272
962
+ },
963
+ {
964
+ "epoch": 0.160163670904574,
965
+ "grad_norm": 1.046875,
966
+ "learning_rate": 5.341130604288499e-05,
967
+ "loss": 0.8381,
968
+ "step": 274
969
+ },
970
+ {
971
+ "epoch": 0.16133274879438841,
972
+ "grad_norm": 0.88671875,
973
+ "learning_rate": 5.3801169590643275e-05,
974
+ "loss": 0.774,
975
+ "step": 276
976
+ },
977
+ {
978
+ "epoch": 0.16250182668420282,
979
+ "grad_norm": 0.796875,
980
+ "learning_rate": 5.419103313840156e-05,
981
+ "loss": 0.8114,
982
+ "step": 278
983
+ },
984
+ {
985
+ "epoch": 0.16367090457401726,
986
+ "grad_norm": 0.9140625,
987
+ "learning_rate": 5.458089668615984e-05,
988
+ "loss": 0.7549,
989
+ "step": 280
990
+ },
991
+ {
992
+ "epoch": 0.16483998246383166,
993
+ "grad_norm": 1.0625,
994
+ "learning_rate": 5.4970760233918126e-05,
995
+ "loss": 0.7393,
996
+ "step": 282
997
+ },
998
+ {
999
+ "epoch": 0.16600906035364607,
1000
+ "grad_norm": 0.88671875,
1001
+ "learning_rate": 5.536062378167641e-05,
1002
+ "loss": 0.7607,
1003
+ "step": 284
1004
+ },
1005
+ {
1006
+ "epoch": 0.16717813824346048,
1007
+ "grad_norm": 1.046875,
1008
+ "learning_rate": 5.57504873294347e-05,
1009
+ "loss": 0.8158,
1010
+ "step": 286
1011
+ },
1012
+ {
1013
+ "epoch": 0.16834721613327489,
1014
+ "grad_norm": 1.734375,
1015
+ "learning_rate": 5.6140350877192984e-05,
1016
+ "loss": 0.7403,
1017
+ "step": 288
1018
+ },
1019
+ {
1020
+ "epoch": 0.1695162940230893,
1021
+ "grad_norm": 1.1015625,
1022
+ "learning_rate": 5.653021442495127e-05,
1023
+ "loss": 0.7278,
1024
+ "step": 290
1025
+ },
1026
+ {
1027
+ "epoch": 0.1706853719129037,
1028
+ "grad_norm": 1.7578125,
1029
+ "learning_rate": 5.692007797270955e-05,
1030
+ "loss": 0.821,
1031
+ "step": 292
1032
+ },
1033
+ {
1034
+ "epoch": 0.1718544498027181,
1035
+ "grad_norm": 3.1875,
1036
+ "learning_rate": 5.7309941520467835e-05,
1037
+ "loss": 0.8276,
1038
+ "step": 294
1039
+ },
1040
+ {
1041
+ "epoch": 0.1730235276925325,
1042
+ "grad_norm": 1.1484375,
1043
+ "learning_rate": 5.769980506822612e-05,
1044
+ "loss": 0.8505,
1045
+ "step": 296
1046
+ },
1047
+ {
1048
+ "epoch": 0.17419260558234692,
1049
+ "grad_norm": 1.2265625,
1050
+ "learning_rate": 5.80896686159844e-05,
1051
+ "loss": 0.7941,
1052
+ "step": 298
1053
+ },
1054
+ {
1055
+ "epoch": 0.17536168347216133,
1056
+ "grad_norm": 1.4765625,
1057
+ "learning_rate": 5.847953216374269e-05,
1058
+ "loss": 0.8621,
1059
+ "step": 300
1060
+ },
1061
+ {
1062
+ "epoch": 0.17653076136197574,
1063
+ "grad_norm": 2.25,
1064
+ "learning_rate": 5.8869395711500976e-05,
1065
+ "loss": 0.8071,
1066
+ "step": 302
1067
+ },
1068
+ {
1069
+ "epoch": 0.17769983925179014,
1070
+ "grad_norm": 0.92578125,
1071
+ "learning_rate": 5.925925925925926e-05,
1072
+ "loss": 0.803,
1073
+ "step": 304
1074
+ },
1075
+ {
1076
+ "epoch": 0.17886891714160455,
1077
+ "grad_norm": 0.9921875,
1078
+ "learning_rate": 5.9649122807017544e-05,
1079
+ "loss": 0.7569,
1080
+ "step": 306
1081
+ },
1082
+ {
1083
+ "epoch": 0.18003799503141896,
1084
+ "grad_norm": 1.09375,
1085
+ "learning_rate": 6.003898635477583e-05,
1086
+ "loss": 0.7722,
1087
+ "step": 308
1088
+ },
1089
+ {
1090
+ "epoch": 0.18120707292123336,
1091
+ "grad_norm": 0.984375,
1092
+ "learning_rate": 6.042884990253411e-05,
1093
+ "loss": 0.7746,
1094
+ "step": 310
1095
+ },
1096
+ {
1097
+ "epoch": 0.1823761508110478,
1098
+ "grad_norm": 2.296875,
1099
+ "learning_rate": 6.0818713450292395e-05,
1100
+ "loss": 0.7292,
1101
+ "step": 312
1102
+ },
1103
+ {
1104
+ "epoch": 0.1835452287008622,
1105
+ "grad_norm": 1.140625,
1106
+ "learning_rate": 6.120857699805068e-05,
1107
+ "loss": 0.7751,
1108
+ "step": 314
1109
+ },
1110
+ {
1111
+ "epoch": 0.1847143065906766,
1112
+ "grad_norm": 0.8203125,
1113
+ "learning_rate": 6.159844054580897e-05,
1114
+ "loss": 0.7473,
1115
+ "step": 316
1116
+ },
1117
+ {
1118
+ "epoch": 0.18588338448049102,
1119
+ "grad_norm": 1.0625,
1120
+ "learning_rate": 6.198830409356725e-05,
1121
+ "loss": 0.7718,
1122
+ "step": 318
1123
+ },
1124
+ {
1125
+ "epoch": 0.18705246237030543,
1126
+ "grad_norm": 1.0390625,
1127
+ "learning_rate": 6.237816764132554e-05,
1128
+ "loss": 0.7812,
1129
+ "step": 320
1130
+ },
1131
+ {
1132
+ "epoch": 0.18822154026011984,
1133
+ "grad_norm": 0.859375,
1134
+ "learning_rate": 6.276803118908382e-05,
1135
+ "loss": 0.7552,
1136
+ "step": 322
1137
+ },
1138
+ {
1139
+ "epoch": 0.18939061814993424,
1140
+ "grad_norm": 0.765625,
1141
+ "learning_rate": 6.31578947368421e-05,
1142
+ "loss": 0.7244,
1143
+ "step": 324
1144
+ },
1145
+ {
1146
+ "epoch": 0.19055969603974865,
1147
+ "grad_norm": 1.2734375,
1148
+ "learning_rate": 6.354775828460039e-05,
1149
+ "loss": 0.8182,
1150
+ "step": 326
1151
+ },
1152
+ {
1153
+ "epoch": 0.19172877392956306,
1154
+ "grad_norm": 0.93359375,
1155
+ "learning_rate": 6.393762183235867e-05,
1156
+ "loss": 0.7578,
1157
+ "step": 328
1158
+ },
1159
+ {
1160
+ "epoch": 0.19289785181937746,
1161
+ "grad_norm": 0.875,
1162
+ "learning_rate": 6.432748538011695e-05,
1163
+ "loss": 0.8755,
1164
+ "step": 330
1165
+ },
1166
+ {
1167
+ "epoch": 0.19406692970919187,
1168
+ "grad_norm": 1.4296875,
1169
+ "learning_rate": 6.471734892787524e-05,
1170
+ "loss": 0.7436,
1171
+ "step": 332
1172
+ },
1173
+ {
1174
+ "epoch": 0.19523600759900628,
1175
+ "grad_norm": 1.859375,
1176
+ "learning_rate": 6.510721247563352e-05,
1177
+ "loss": 0.7415,
1178
+ "step": 334
1179
+ },
1180
+ {
1181
+ "epoch": 0.19640508548882069,
1182
+ "grad_norm": 1.265625,
1183
+ "learning_rate": 6.549707602339182e-05,
1184
+ "loss": 0.8082,
1185
+ "step": 336
1186
+ },
1187
+ {
1188
+ "epoch": 0.1975741633786351,
1189
+ "grad_norm": 1.4453125,
1190
+ "learning_rate": 6.58869395711501e-05,
1191
+ "loss": 0.7226,
1192
+ "step": 338
1193
+ },
1194
+ {
1195
+ "epoch": 0.1987432412684495,
1196
+ "grad_norm": 0.8203125,
1197
+ "learning_rate": 6.627680311890839e-05,
1198
+ "loss": 0.7408,
1199
+ "step": 340
1200
+ },
1201
+ {
1202
+ "epoch": 0.1999123191582639,
1203
+ "grad_norm": 1.1875,
1204
+ "learning_rate": 6.666666666666667e-05,
1205
+ "loss": 0.7752,
1206
+ "step": 342
1207
+ },
1208
+ {
1209
+ "epoch": 0.20108139704807834,
1210
+ "grad_norm": 1.0234375,
1211
+ "learning_rate": 6.705653021442495e-05,
1212
+ "loss": 0.7313,
1213
+ "step": 344
1214
+ },
1215
+ {
1216
+ "epoch": 0.20225047493789275,
1217
+ "grad_norm": 1.1015625,
1218
+ "learning_rate": 6.744639376218324e-05,
1219
+ "loss": 0.7164,
1220
+ "step": 346
1221
+ },
1222
+ {
1223
+ "epoch": 0.20341955282770716,
1224
+ "grad_norm": 0.9296875,
1225
+ "learning_rate": 6.783625730994152e-05,
1226
+ "loss": 0.7836,
1227
+ "step": 348
1228
+ },
1229
+ {
1230
+ "epoch": 0.20458863071752156,
1231
+ "grad_norm": 1.9140625,
1232
+ "learning_rate": 6.82261208576998e-05,
1233
+ "loss": 0.8936,
1234
+ "step": 350
1235
+ },
1236
+ {
1237
+ "epoch": 0.20575770860733597,
1238
+ "grad_norm": 0.8984375,
1239
+ "learning_rate": 6.861598440545809e-05,
1240
+ "loss": 0.7386,
1241
+ "step": 352
1242
+ },
1243
+ {
1244
+ "epoch": 0.20692678649715038,
1245
+ "grad_norm": 0.9609375,
1246
+ "learning_rate": 6.900584795321637e-05,
1247
+ "loss": 0.8893,
1248
+ "step": 354
1249
+ },
1250
+ {
1251
+ "epoch": 0.20809586438696479,
1252
+ "grad_norm": 0.9140625,
1253
+ "learning_rate": 6.939571150097466e-05,
1254
+ "loss": 0.715,
1255
+ "step": 356
1256
+ },
1257
+ {
1258
+ "epoch": 0.2092649422767792,
1259
+ "grad_norm": 0.94921875,
1260
+ "learning_rate": 6.978557504873294e-05,
1261
+ "loss": 0.7087,
1262
+ "step": 358
1263
+ },
1264
+ {
1265
+ "epoch": 0.2104340201665936,
1266
+ "grad_norm": 0.96484375,
1267
+ "learning_rate": 7.017543859649122e-05,
1268
+ "loss": 0.7067,
1269
+ "step": 360
1270
+ },
1271
+ {
1272
+ "epoch": 0.211603098056408,
1273
+ "grad_norm": 0.9375,
1274
+ "learning_rate": 7.05653021442495e-05,
1275
+ "loss": 0.6979,
1276
+ "step": 362
1277
+ },
1278
+ {
1279
+ "epoch": 0.2127721759462224,
1280
+ "grad_norm": 1.8046875,
1281
+ "learning_rate": 7.095516569200779e-05,
1282
+ "loss": 0.8533,
1283
+ "step": 364
1284
+ },
1285
+ {
1286
+ "epoch": 0.21394125383603682,
1287
+ "grad_norm": 0.95703125,
1288
+ "learning_rate": 7.134502923976609e-05,
1289
+ "loss": 0.6971,
1290
+ "step": 366
1291
+ },
1292
+ {
1293
+ "epoch": 0.21511033172585123,
1294
+ "grad_norm": 1.1640625,
1295
+ "learning_rate": 7.173489278752437e-05,
1296
+ "loss": 0.7152,
1297
+ "step": 368
1298
+ },
1299
+ {
1300
+ "epoch": 0.21627940961566564,
1301
+ "grad_norm": 1.3984375,
1302
+ "learning_rate": 7.212475633528265e-05,
1303
+ "loss": 0.7176,
1304
+ "step": 370
1305
+ },
1306
+ {
1307
+ "epoch": 0.21744848750548004,
1308
+ "grad_norm": 0.84375,
1309
+ "learning_rate": 7.251461988304094e-05,
1310
+ "loss": 0.7054,
1311
+ "step": 372
1312
+ },
1313
+ {
1314
+ "epoch": 0.21861756539529445,
1315
+ "grad_norm": 0.859375,
1316
+ "learning_rate": 7.290448343079922e-05,
1317
+ "loss": 0.7157,
1318
+ "step": 374
1319
+ },
1320
+ {
1321
+ "epoch": 0.21978664328510886,
1322
+ "grad_norm": 0.99609375,
1323
+ "learning_rate": 7.32943469785575e-05,
1324
+ "loss": 0.742,
1325
+ "step": 376
1326
+ },
1327
+ {
1328
+ "epoch": 0.2209557211749233,
1329
+ "grad_norm": 1.1171875,
1330
+ "learning_rate": 7.368421052631579e-05,
1331
+ "loss": 0.7016,
1332
+ "step": 378
1333
+ },
1334
+ {
1335
+ "epoch": 0.2221247990647377,
1336
+ "grad_norm": 1.3359375,
1337
+ "learning_rate": 7.407407407407407e-05,
1338
+ "loss": 0.7263,
1339
+ "step": 380
1340
+ },
1341
+ {
1342
+ "epoch": 0.2232938769545521,
1343
+ "grad_norm": 0.8515625,
1344
+ "learning_rate": 7.446393762183236e-05,
1345
+ "loss": 0.7386,
1346
+ "step": 382
1347
+ },
1348
+ {
1349
+ "epoch": 0.2244629548443665,
1350
+ "grad_norm": 1.3359375,
1351
+ "learning_rate": 7.485380116959064e-05,
1352
+ "loss": 0.7488,
1353
+ "step": 384
1354
+ },
1355
+ {
1356
+ "epoch": 0.22563203273418092,
1357
+ "grad_norm": 0.98828125,
1358
+ "learning_rate": 7.524366471734892e-05,
1359
+ "loss": 0.7304,
1360
+ "step": 386
1361
+ },
1362
+ {
1363
+ "epoch": 0.22680111062399533,
1364
+ "grad_norm": 0.90625,
1365
+ "learning_rate": 7.563352826510721e-05,
1366
+ "loss": 0.6926,
1367
+ "step": 388
1368
+ },
1369
+ {
1370
+ "epoch": 0.22797018851380974,
1371
+ "grad_norm": 1.578125,
1372
+ "learning_rate": 7.602339181286549e-05,
1373
+ "loss": 0.7258,
1374
+ "step": 390
1375
+ },
1376
+ {
1377
+ "epoch": 0.22913926640362414,
1378
+ "grad_norm": 0.82421875,
1379
+ "learning_rate": 7.641325536062378e-05,
1380
+ "loss": 0.7169,
1381
+ "step": 392
1382
+ },
1383
+ {
1384
+ "epoch": 0.23030834429343855,
1385
+ "grad_norm": 0.78125,
1386
+ "learning_rate": 7.680311890838207e-05,
1387
+ "loss": 0.6962,
1388
+ "step": 394
1389
+ },
1390
+ {
1391
+ "epoch": 0.23147742218325296,
1392
+ "grad_norm": 1.0703125,
1393
+ "learning_rate": 7.719298245614036e-05,
1394
+ "loss": 0.718,
1395
+ "step": 396
1396
+ },
1397
+ {
1398
+ "epoch": 0.23264650007306736,
1399
+ "grad_norm": 0.84375,
1400
+ "learning_rate": 7.758284600389864e-05,
1401
+ "loss": 0.7106,
1402
+ "step": 398
1403
+ },
1404
+ {
1405
+ "epoch": 0.23381557796288177,
1406
+ "grad_norm": 1.109375,
1407
+ "learning_rate": 7.797270955165692e-05,
1408
+ "loss": 0.739,
1409
+ "step": 400
1410
+ },
1411
+ {
1412
+ "epoch": 0.23498465585269618,
1413
+ "grad_norm": 0.9140625,
1414
+ "learning_rate": 7.836257309941521e-05,
1415
+ "loss": 0.7037,
1416
+ "step": 402
1417
+ },
1418
+ {
1419
+ "epoch": 0.23615373374251059,
1420
+ "grad_norm": 0.76953125,
1421
+ "learning_rate": 7.875243664717349e-05,
1422
+ "loss": 0.717,
1423
+ "step": 404
1424
+ },
1425
+ {
1426
+ "epoch": 0.237322811632325,
1427
+ "grad_norm": 1.0546875,
1428
+ "learning_rate": 7.914230019493177e-05,
1429
+ "loss": 0.7041,
1430
+ "step": 406
1431
+ },
1432
+ {
1433
+ "epoch": 0.2384918895221394,
1434
+ "grad_norm": 0.921875,
1435
+ "learning_rate": 7.953216374269006e-05,
1436
+ "loss": 0.7016,
1437
+ "step": 408
1438
+ },
1439
+ {
1440
+ "epoch": 0.23966096741195383,
1441
+ "grad_norm": 0.8515625,
1442
+ "learning_rate": 7.992202729044834e-05,
1443
+ "loss": 0.7423,
1444
+ "step": 410
1445
+ },
1446
+ {
1447
+ "epoch": 0.24083004530176824,
1448
+ "grad_norm": 0.88671875,
1449
+ "learning_rate": 8.031189083820663e-05,
1450
+ "loss": 0.7128,
1451
+ "step": 412
1452
+ },
1453
+ {
1454
+ "epoch": 0.24199912319158265,
1455
+ "grad_norm": 0.9453125,
1456
+ "learning_rate": 8.070175438596491e-05,
1457
+ "loss": 0.7144,
1458
+ "step": 414
1459
+ },
1460
+ {
1461
+ "epoch": 0.24316820108139706,
1462
+ "grad_norm": 1.1328125,
1463
+ "learning_rate": 8.109161793372319e-05,
1464
+ "loss": 0.6987,
1465
+ "step": 416
1466
+ },
1467
+ {
1468
+ "epoch": 0.24433727897121146,
1469
+ "grad_norm": 0.9296875,
1470
+ "learning_rate": 8.148148148148148e-05,
1471
+ "loss": 0.6946,
1472
+ "step": 418
1473
+ },
1474
+ {
1475
+ "epoch": 0.24550635686102587,
1476
+ "grad_norm": 0.9296875,
1477
+ "learning_rate": 8.187134502923976e-05,
1478
+ "loss": 0.6954,
1479
+ "step": 420
1480
+ },
1481
+ {
1482
+ "epoch": 0.24667543475084028,
1483
+ "grad_norm": 0.84765625,
1484
+ "learning_rate": 8.226120857699804e-05,
1485
+ "loss": 0.7507,
1486
+ "step": 422
1487
+ },
1488
+ {
1489
+ "epoch": 0.24784451264065469,
1490
+ "grad_norm": 2.171875,
1491
+ "learning_rate": 8.265107212475634e-05,
1492
+ "loss": 0.9267,
1493
+ "step": 424
1494
+ },
1495
+ {
1496
+ "epoch": 0.2490135905304691,
1497
+ "grad_norm": 0.81640625,
1498
+ "learning_rate": 8.304093567251462e-05,
1499
+ "loss": 0.7222,
1500
+ "step": 426
1501
+ },
1502
+ {
1503
+ "epoch": 0.2501826684202835,
1504
+ "grad_norm": 1.7421875,
1505
+ "learning_rate": 8.343079922027291e-05,
1506
+ "loss": 0.8919,
1507
+ "step": 428
1508
+ },
1509
+ {
1510
+ "epoch": 0.25135174631009793,
1511
+ "grad_norm": 1.140625,
1512
+ "learning_rate": 8.382066276803119e-05,
1513
+ "loss": 0.7451,
1514
+ "step": 430
1515
+ },
1516
+ {
1517
+ "epoch": 0.25252082419991234,
1518
+ "grad_norm": 0.81640625,
1519
+ "learning_rate": 8.421052631578948e-05,
1520
+ "loss": 0.7027,
1521
+ "step": 432
1522
+ },
1523
+ {
1524
+ "epoch": 0.25368990208972675,
1525
+ "grad_norm": 0.79296875,
1526
+ "learning_rate": 8.460038986354776e-05,
1527
+ "loss": 0.712,
1528
+ "step": 434
1529
+ },
1530
+ {
1531
+ "epoch": 0.25485897997954116,
1532
+ "grad_norm": 0.8984375,
1533
+ "learning_rate": 8.499025341130604e-05,
1534
+ "loss": 0.7122,
1535
+ "step": 436
1536
+ },
1537
+ {
1538
+ "epoch": 0.25602805786935556,
1539
+ "grad_norm": 0.72265625,
1540
+ "learning_rate": 8.538011695906433e-05,
1541
+ "loss": 0.6871,
1542
+ "step": 438
1543
+ },
1544
+ {
1545
+ "epoch": 0.25719713575916997,
1546
+ "grad_norm": 0.77734375,
1547
+ "learning_rate": 8.576998050682261e-05,
1548
+ "loss": 0.7278,
1549
+ "step": 440
1550
+ },
1551
+ {
1552
+ "epoch": 0.2583662136489844,
1553
+ "grad_norm": 1.1171875,
1554
+ "learning_rate": 8.61598440545809e-05,
1555
+ "loss": 0.7023,
1556
+ "step": 442
1557
+ },
1558
+ {
1559
+ "epoch": 0.2595352915387988,
1560
+ "grad_norm": 0.8359375,
1561
+ "learning_rate": 8.654970760233918e-05,
1562
+ "loss": 0.7167,
1563
+ "step": 444
1564
+ },
1565
+ {
1566
+ "epoch": 0.2607043694286132,
1567
+ "grad_norm": 0.8203125,
1568
+ "learning_rate": 8.693957115009746e-05,
1569
+ "loss": 0.7346,
1570
+ "step": 446
1571
+ },
1572
+ {
1573
+ "epoch": 0.2618734473184276,
1574
+ "grad_norm": 0.9765625,
1575
+ "learning_rate": 8.732943469785574e-05,
1576
+ "loss": 0.7223,
1577
+ "step": 448
1578
+ },
1579
+ {
1580
+ "epoch": 0.263042525208242,
1581
+ "grad_norm": 0.96875,
1582
+ "learning_rate": 8.771929824561403e-05,
1583
+ "loss": 0.7032,
1584
+ "step": 450
1585
+ },
1586
+ {
1587
+ "epoch": 0.2642116030980564,
1588
+ "grad_norm": 0.9609375,
1589
+ "learning_rate": 8.810916179337231e-05,
1590
+ "loss": 0.6804,
1591
+ "step": 452
1592
+ },
1593
+ {
1594
+ "epoch": 0.2653806809878708,
1595
+ "grad_norm": 0.875,
1596
+ "learning_rate": 8.849902534113061e-05,
1597
+ "loss": 0.7224,
1598
+ "step": 454
1599
+ },
1600
+ {
1601
+ "epoch": 0.26654975887768523,
1602
+ "grad_norm": 0.8359375,
1603
+ "learning_rate": 8.888888888888889e-05,
1604
+ "loss": 0.7138,
1605
+ "step": 456
1606
+ },
1607
+ {
1608
+ "epoch": 0.26771883676749963,
1609
+ "grad_norm": 0.859375,
1610
+ "learning_rate": 8.927875243664718e-05,
1611
+ "loss": 0.7065,
1612
+ "step": 458
1613
+ },
1614
+ {
1615
+ "epoch": 0.26888791465731404,
1616
+ "grad_norm": 0.96484375,
1617
+ "learning_rate": 8.966861598440546e-05,
1618
+ "loss": 0.7289,
1619
+ "step": 460
1620
+ },
1621
+ {
1622
+ "epoch": 0.27005699254712845,
1623
+ "grad_norm": 1.0703125,
1624
+ "learning_rate": 9.005847953216374e-05,
1625
+ "loss": 0.6917,
1626
+ "step": 462
1627
+ },
1628
+ {
1629
+ "epoch": 0.27122607043694286,
1630
+ "grad_norm": 1.0078125,
1631
+ "learning_rate": 9.044834307992203e-05,
1632
+ "loss": 0.7493,
1633
+ "step": 464
1634
+ },
1635
+ {
1636
+ "epoch": 0.27239514832675726,
1637
+ "grad_norm": 1.0546875,
1638
+ "learning_rate": 9.083820662768031e-05,
1639
+ "loss": 0.7075,
1640
+ "step": 466
1641
+ },
1642
+ {
1643
+ "epoch": 0.27356422621657167,
1644
+ "grad_norm": 0.88671875,
1645
+ "learning_rate": 9.12280701754386e-05,
1646
+ "loss": 0.6926,
1647
+ "step": 468
1648
+ },
1649
+ {
1650
+ "epoch": 0.2747333041063861,
1651
+ "grad_norm": 0.734375,
1652
+ "learning_rate": 9.161793372319688e-05,
1653
+ "loss": 0.6714,
1654
+ "step": 470
1655
+ },
1656
+ {
1657
+ "epoch": 0.2759023819962005,
1658
+ "grad_norm": 0.73046875,
1659
+ "learning_rate": 9.200779727095516e-05,
1660
+ "loss": 0.6973,
1661
+ "step": 472
1662
+ },
1663
+ {
1664
+ "epoch": 0.2770714598860149,
1665
+ "grad_norm": 0.7578125,
1666
+ "learning_rate": 9.239766081871345e-05,
1667
+ "loss": 0.7289,
1668
+ "step": 474
1669
+ },
1670
+ {
1671
+ "epoch": 0.2782405377758293,
1672
+ "grad_norm": 0.84765625,
1673
+ "learning_rate": 9.278752436647173e-05,
1674
+ "loss": 0.7296,
1675
+ "step": 476
1676
+ },
1677
+ {
1678
+ "epoch": 0.2794096156656437,
1679
+ "grad_norm": 0.99609375,
1680
+ "learning_rate": 9.317738791423001e-05,
1681
+ "loss": 0.7229,
1682
+ "step": 478
1683
+ },
1684
+ {
1685
+ "epoch": 0.2805786935554581,
1686
+ "grad_norm": 1.0859375,
1687
+ "learning_rate": 9.35672514619883e-05,
1688
+ "loss": 0.7158,
1689
+ "step": 480
1690
+ },
1691
+ {
1692
+ "epoch": 0.2817477714452725,
1693
+ "grad_norm": 0.8671875,
1694
+ "learning_rate": 9.39571150097466e-05,
1695
+ "loss": 0.6862,
1696
+ "step": 482
1697
+ },
1698
+ {
1699
+ "epoch": 0.28291684933508693,
1700
+ "grad_norm": 0.77734375,
1701
+ "learning_rate": 9.434697855750488e-05,
1702
+ "loss": 0.6535,
1703
+ "step": 484
1704
+ },
1705
+ {
1706
+ "epoch": 0.28408592722490134,
1707
+ "grad_norm": 0.90234375,
1708
+ "learning_rate": 9.473684210526316e-05,
1709
+ "loss": 0.703,
1710
+ "step": 486
1711
+ },
1712
+ {
1713
+ "epoch": 0.28525500511471574,
1714
+ "grad_norm": 0.80078125,
1715
+ "learning_rate": 9.512670565302145e-05,
1716
+ "loss": 0.7296,
1717
+ "step": 488
1718
+ },
1719
+ {
1720
+ "epoch": 0.28642408300453015,
1721
+ "grad_norm": 1.3203125,
1722
+ "learning_rate": 9.551656920077973e-05,
1723
+ "loss": 0.7061,
1724
+ "step": 490
1725
+ },
1726
+ {
1727
+ "epoch": 0.2875931608943446,
1728
+ "grad_norm": 0.76953125,
1729
+ "learning_rate": 9.590643274853801e-05,
1730
+ "loss": 0.6731,
1731
+ "step": 492
1732
+ },
1733
+ {
1734
+ "epoch": 0.288762238784159,
1735
+ "grad_norm": 1.0234375,
1736
+ "learning_rate": 9.62962962962963e-05,
1737
+ "loss": 0.7141,
1738
+ "step": 494
1739
+ },
1740
+ {
1741
+ "epoch": 0.2899313166739734,
1742
+ "grad_norm": 1.8359375,
1743
+ "learning_rate": 9.668615984405458e-05,
1744
+ "loss": 0.7874,
1745
+ "step": 496
1746
+ },
1747
+ {
1748
+ "epoch": 0.29110039456378783,
1749
+ "grad_norm": 0.97265625,
1750
+ "learning_rate": 9.707602339181286e-05,
1751
+ "loss": 0.7179,
1752
+ "step": 498
1753
+ },
1754
+ {
1755
+ "epoch": 0.29226947245360224,
1756
+ "grad_norm": 0.8984375,
1757
+ "learning_rate": 9.746588693957115e-05,
1758
+ "loss": 0.6819,
1759
+ "step": 500
1760
+ },
1761
+ {
1762
+ "epoch": 0.29343855034341665,
1763
+ "grad_norm": 0.8359375,
1764
+ "learning_rate": 9.785575048732943e-05,
1765
+ "loss": 0.688,
1766
+ "step": 502
1767
+ },
1768
+ {
1769
+ "epoch": 0.29460762823323106,
1770
+ "grad_norm": 2.21875,
1771
+ "learning_rate": 9.824561403508771e-05,
1772
+ "loss": 0.8356,
1773
+ "step": 504
1774
+ },
1775
+ {
1776
+ "epoch": 0.29577670612304546,
1777
+ "grad_norm": 2.171875,
1778
+ "learning_rate": 9.8635477582846e-05,
1779
+ "loss": 0.7733,
1780
+ "step": 506
1781
+ },
1782
+ {
1783
+ "epoch": 0.29694578401285987,
1784
+ "grad_norm": 0.8203125,
1785
+ "learning_rate": 9.902534113060428e-05,
1786
+ "loss": 0.7087,
1787
+ "step": 508
1788
+ },
1789
+ {
1790
+ "epoch": 0.2981148619026743,
1791
+ "grad_norm": 1.2109375,
1792
+ "learning_rate": 9.941520467836257e-05,
1793
+ "loss": 0.7023,
1794
+ "step": 510
1795
+ },
1796
+ {
1797
+ "epoch": 0.2992839397924887,
1798
+ "grad_norm": 1.28125,
1799
+ "learning_rate": 9.980506822612086e-05,
1800
+ "loss": 0.7087,
1801
+ "step": 512
1802
+ },
1803
+ {
1804
+ "epoch": 0.3004530176823031,
1805
+ "grad_norm": 1.328125,
1806
+ "learning_rate": 0.00010019493177387915,
1807
+ "loss": 0.7002,
1808
+ "step": 514
1809
+ },
1810
+ {
1811
+ "epoch": 0.3016220955721175,
1812
+ "grad_norm": 0.8359375,
1813
+ "learning_rate": 0.00010058479532163743,
1814
+ "loss": 0.7355,
1815
+ "step": 516
1816
+ },
1817
+ {
1818
+ "epoch": 0.3027911734619319,
1819
+ "grad_norm": 1.21875,
1820
+ "learning_rate": 0.00010097465886939573,
1821
+ "loss": 0.6983,
1822
+ "step": 518
1823
+ },
1824
+ {
1825
+ "epoch": 0.3039602513517463,
1826
+ "grad_norm": 1.65625,
1827
+ "learning_rate": 0.000101364522417154,
1828
+ "loss": 0.7016,
1829
+ "step": 520
1830
+ },
1831
+ {
1832
+ "epoch": 0.3051293292415607,
1833
+ "grad_norm": 0.765625,
1834
+ "learning_rate": 0.0001017543859649123,
1835
+ "loss": 0.7096,
1836
+ "step": 522
1837
+ },
1838
+ {
1839
+ "epoch": 0.3062984071313751,
1840
+ "grad_norm": 1.46875,
1841
+ "learning_rate": 0.00010214424951267056,
1842
+ "loss": 0.7099,
1843
+ "step": 524
1844
+ },
1845
+ {
1846
+ "epoch": 0.30746748502118953,
1847
+ "grad_norm": 1.4375,
1848
+ "learning_rate": 0.00010253411306042886,
1849
+ "loss": 0.6931,
1850
+ "step": 526
1851
+ },
1852
+ {
1853
+ "epoch": 0.30863656291100394,
1854
+ "grad_norm": 0.96875,
1855
+ "learning_rate": 0.00010292397660818713,
1856
+ "loss": 0.7427,
1857
+ "step": 528
1858
+ },
1859
+ {
1860
+ "epoch": 0.30980564080081835,
1861
+ "grad_norm": 1.234375,
1862
+ "learning_rate": 0.00010331384015594543,
1863
+ "loss": 0.6609,
1864
+ "step": 530
1865
+ },
1866
+ {
1867
+ "epoch": 0.31097471869063276,
1868
+ "grad_norm": 1.0703125,
1869
+ "learning_rate": 0.0001037037037037037,
1870
+ "loss": 0.7195,
1871
+ "step": 532
1872
+ },
1873
+ {
1874
+ "epoch": 0.31214379658044716,
1875
+ "grad_norm": 0.7109375,
1876
+ "learning_rate": 0.000104093567251462,
1877
+ "loss": 0.6735,
1878
+ "step": 534
1879
+ },
1880
+ {
1881
+ "epoch": 0.31331287447026157,
1882
+ "grad_norm": 0.82421875,
1883
+ "learning_rate": 0.00010448343079922027,
1884
+ "loss": 0.7455,
1885
+ "step": 536
1886
+ },
1887
+ {
1888
+ "epoch": 0.314481952360076,
1889
+ "grad_norm": 0.91015625,
1890
+ "learning_rate": 0.00010487329434697856,
1891
+ "loss": 0.6683,
1892
+ "step": 538
1893
+ },
1894
+ {
1895
+ "epoch": 0.3156510302498904,
1896
+ "grad_norm": 0.7265625,
1897
+ "learning_rate": 0.00010526315789473685,
1898
+ "loss": 0.6981,
1899
+ "step": 540
1900
+ },
1901
+ {
1902
+ "epoch": 0.3168201081397048,
1903
+ "grad_norm": 0.69921875,
1904
+ "learning_rate": 0.00010565302144249513,
1905
+ "loss": 0.65,
1906
+ "step": 542
1907
+ },
1908
+ {
1909
+ "epoch": 0.3179891860295192,
1910
+ "grad_norm": 2.203125,
1911
+ "learning_rate": 0.00010604288499025342,
1912
+ "loss": 0.7997,
1913
+ "step": 544
1914
+ },
1915
+ {
1916
+ "epoch": 0.3191582639193336,
1917
+ "grad_norm": 0.890625,
1918
+ "learning_rate": 0.00010643274853801171,
1919
+ "loss": 0.7143,
1920
+ "step": 546
1921
+ },
1922
+ {
1923
+ "epoch": 0.320327341809148,
1924
+ "grad_norm": 0.98046875,
1925
+ "learning_rate": 0.00010682261208576998,
1926
+ "loss": 0.6619,
1927
+ "step": 548
1928
+ },
1929
+ {
1930
+ "epoch": 0.3214964196989624,
1931
+ "grad_norm": 0.97265625,
1932
+ "learning_rate": 0.00010721247563352828,
1933
+ "loss": 0.6632,
1934
+ "step": 550
1935
+ },
1936
+ {
1937
+ "epoch": 0.32266549758877683,
1938
+ "grad_norm": 0.765625,
1939
+ "learning_rate": 0.00010760233918128655,
1940
+ "loss": 0.6844,
1941
+ "step": 552
1942
+ },
1943
+ {
1944
+ "epoch": 0.32383457547859124,
1945
+ "grad_norm": 0.828125,
1946
+ "learning_rate": 0.00010799220272904485,
1947
+ "loss": 0.698,
1948
+ "step": 554
1949
+ },
1950
+ {
1951
+ "epoch": 0.32500365336840564,
1952
+ "grad_norm": 0.94140625,
1953
+ "learning_rate": 0.00010838206627680312,
1954
+ "loss": 0.6132,
1955
+ "step": 556
1956
+ },
1957
+ {
1958
+ "epoch": 0.3261727312582201,
1959
+ "grad_norm": 0.765625,
1960
+ "learning_rate": 0.00010877192982456141,
1961
+ "loss": 0.6507,
1962
+ "step": 558
1963
+ },
1964
+ {
1965
+ "epoch": 0.3273418091480345,
1966
+ "grad_norm": 1.03125,
1967
+ "learning_rate": 0.00010916179337231968,
1968
+ "loss": 0.696,
1969
+ "step": 560
1970
+ },
1971
+ {
1972
+ "epoch": 0.3285108870378489,
1973
+ "grad_norm": 3.0,
1974
+ "learning_rate": 0.00010955165692007798,
1975
+ "loss": 0.926,
1976
+ "step": 562
1977
+ },
1978
+ {
1979
+ "epoch": 0.3296799649276633,
1980
+ "grad_norm": 1.453125,
1981
+ "learning_rate": 0.00010994152046783625,
1982
+ "loss": 0.7797,
1983
+ "step": 564
1984
+ },
1985
+ {
1986
+ "epoch": 0.33084904281747773,
1987
+ "grad_norm": 1.71875,
1988
+ "learning_rate": 0.00011033138401559455,
1989
+ "loss": 0.7379,
1990
+ "step": 566
1991
+ },
1992
+ {
1993
+ "epoch": 0.33201812070729214,
1994
+ "grad_norm": 0.9921875,
1995
+ "learning_rate": 0.00011072124756335282,
1996
+ "loss": 0.679,
1997
+ "step": 568
1998
+ },
1999
+ {
2000
+ "epoch": 0.33318719859710655,
2001
+ "grad_norm": 0.77734375,
2002
+ "learning_rate": 0.00011111111111111112,
2003
+ "loss": 0.6668,
2004
+ "step": 570
2005
+ },
2006
+ {
2007
+ "epoch": 0.33435627648692096,
2008
+ "grad_norm": 0.796875,
2009
+ "learning_rate": 0.0001115009746588694,
2010
+ "loss": 0.7378,
2011
+ "step": 572
2012
+ },
2013
+ {
2014
+ "epoch": 0.33552535437673536,
2015
+ "grad_norm": 0.8046875,
2016
+ "learning_rate": 0.0001118908382066277,
2017
+ "loss": 0.7341,
2018
+ "step": 574
2019
+ },
2020
+ {
2021
+ "epoch": 0.33669443226654977,
2022
+ "grad_norm": 0.71875,
2023
+ "learning_rate": 0.00011228070175438597,
2024
+ "loss": 0.6793,
2025
+ "step": 576
2026
+ },
2027
+ {
2028
+ "epoch": 0.3378635101563642,
2029
+ "grad_norm": 0.82421875,
2030
+ "learning_rate": 0.00011267056530214426,
2031
+ "loss": 0.6926,
2032
+ "step": 578
2033
+ },
2034
+ {
2035
+ "epoch": 0.3390325880461786,
2036
+ "grad_norm": 0.89453125,
2037
+ "learning_rate": 0.00011306042884990253,
2038
+ "loss": 0.6393,
2039
+ "step": 580
2040
+ },
2041
+ {
2042
+ "epoch": 0.340201665935993,
2043
+ "grad_norm": 0.9375,
2044
+ "learning_rate": 0.00011345029239766083,
2045
+ "loss": 0.7207,
2046
+ "step": 582
2047
+ },
2048
+ {
2049
+ "epoch": 0.3413707438258074,
2050
+ "grad_norm": 0.75390625,
2051
+ "learning_rate": 0.0001138401559454191,
2052
+ "loss": 0.6168,
2053
+ "step": 584
2054
+ },
2055
+ {
2056
+ "epoch": 0.3425398217156218,
2057
+ "grad_norm": 0.7265625,
2058
+ "learning_rate": 0.0001142300194931774,
2059
+ "loss": 0.712,
2060
+ "step": 586
2061
+ },
2062
+ {
2063
+ "epoch": 0.3437088996054362,
2064
+ "grad_norm": 0.66796875,
2065
+ "learning_rate": 0.00011461988304093567,
2066
+ "loss": 0.6532,
2067
+ "step": 588
2068
+ },
2069
+ {
2070
+ "epoch": 0.3448779774952506,
2071
+ "grad_norm": 0.734375,
2072
+ "learning_rate": 0.00011500974658869397,
2073
+ "loss": 0.6766,
2074
+ "step": 590
2075
+ },
2076
+ {
2077
+ "epoch": 0.346047055385065,
2078
+ "grad_norm": 0.72265625,
2079
+ "learning_rate": 0.00011539961013645224,
2080
+ "loss": 0.6986,
2081
+ "step": 592
2082
+ },
2083
+ {
2084
+ "epoch": 0.34721613327487943,
2085
+ "grad_norm": 0.7109375,
2086
+ "learning_rate": 0.00011578947368421053,
2087
+ "loss": 0.6867,
2088
+ "step": 594
2089
+ },
2090
+ {
2091
+ "epoch": 0.34838521116469384,
2092
+ "grad_norm": 1.484375,
2093
+ "learning_rate": 0.0001161793372319688,
2094
+ "loss": 0.8175,
2095
+ "step": 596
2096
+ },
2097
+ {
2098
+ "epoch": 0.34955428905450825,
2099
+ "grad_norm": 0.8828125,
2100
+ "learning_rate": 0.0001165692007797271,
2101
+ "loss": 0.8122,
2102
+ "step": 598
2103
+ },
2104
+ {
2105
+ "epoch": 0.35072336694432266,
2106
+ "grad_norm": 0.96875,
2107
+ "learning_rate": 0.00011695906432748539,
2108
+ "loss": 0.7642,
2109
+ "step": 600
2110
+ },
2111
+ {
2112
+ "epoch": 0.35189244483413706,
2113
+ "grad_norm": 0.91796875,
2114
+ "learning_rate": 0.00011734892787524367,
2115
+ "loss": 0.703,
2116
+ "step": 602
2117
+ },
2118
+ {
2119
+ "epoch": 0.35306152272395147,
2120
+ "grad_norm": 0.7890625,
2121
+ "learning_rate": 0.00011773879142300195,
2122
+ "loss": 0.7031,
2123
+ "step": 604
2124
+ },
2125
+ {
2126
+ "epoch": 0.3542306006137659,
2127
+ "grad_norm": 0.98828125,
2128
+ "learning_rate": 0.00011812865497076025,
2129
+ "loss": 0.686,
2130
+ "step": 606
2131
+ },
2132
+ {
2133
+ "epoch": 0.3553996785035803,
2134
+ "grad_norm": 0.84375,
2135
+ "learning_rate": 0.00011851851851851852,
2136
+ "loss": 0.6708,
2137
+ "step": 608
2138
+ },
2139
+ {
2140
+ "epoch": 0.3565687563933947,
2141
+ "grad_norm": 0.8515625,
2142
+ "learning_rate": 0.00011890838206627682,
2143
+ "loss": 0.8084,
2144
+ "step": 610
2145
+ },
2146
+ {
2147
+ "epoch": 0.3577378342832091,
2148
+ "grad_norm": 0.75,
2149
+ "learning_rate": 0.00011929824561403509,
2150
+ "loss": 0.6705,
2151
+ "step": 612
2152
+ },
2153
+ {
2154
+ "epoch": 0.3589069121730235,
2155
+ "grad_norm": 0.6875,
2156
+ "learning_rate": 0.00011968810916179338,
2157
+ "loss": 0.6409,
2158
+ "step": 614
2159
+ },
2160
+ {
2161
+ "epoch": 0.3600759900628379,
2162
+ "grad_norm": 0.9375,
2163
+ "learning_rate": 0.00012007797270955165,
2164
+ "loss": 0.6805,
2165
+ "step": 616
2166
+ },
2167
+ {
2168
+ "epoch": 0.3612450679526523,
2169
+ "grad_norm": 1.59375,
2170
+ "learning_rate": 0.00012046783625730995,
2171
+ "loss": 0.7129,
2172
+ "step": 618
2173
+ },
2174
+ {
2175
+ "epoch": 0.36241414584246673,
2176
+ "grad_norm": 0.9609375,
2177
+ "learning_rate": 0.00012085769980506822,
2178
+ "loss": 0.6819,
2179
+ "step": 620
2180
+ },
2181
+ {
2182
+ "epoch": 0.36358322373228114,
2183
+ "grad_norm": 0.828125,
2184
+ "learning_rate": 0.00012124756335282652,
2185
+ "loss": 0.6424,
2186
+ "step": 622
2187
+ },
2188
+ {
2189
+ "epoch": 0.3647523016220956,
2190
+ "grad_norm": 0.7890625,
2191
+ "learning_rate": 0.00012163742690058479,
2192
+ "loss": 0.704,
2193
+ "step": 624
2194
+ },
2195
+ {
2196
+ "epoch": 0.36592137951191,
2197
+ "grad_norm": 1.3359375,
2198
+ "learning_rate": 0.00012202729044834309,
2199
+ "loss": 0.8172,
2200
+ "step": 626
2201
+ },
2202
+ {
2203
+ "epoch": 0.3670904574017244,
2204
+ "grad_norm": 0.92578125,
2205
+ "learning_rate": 0.00012241715399610137,
2206
+ "loss": 0.6486,
2207
+ "step": 628
2208
+ },
2209
+ {
2210
+ "epoch": 0.3682595352915388,
2211
+ "grad_norm": 1.3046875,
2212
+ "learning_rate": 0.00012280701754385965,
2213
+ "loss": 0.6772,
2214
+ "step": 630
2215
+ },
2216
+ {
2217
+ "epoch": 0.3694286131813532,
2218
+ "grad_norm": 1.1171875,
2219
+ "learning_rate": 0.00012319688109161794,
2220
+ "loss": 0.6997,
2221
+ "step": 632
2222
+ },
2223
+ {
2224
+ "epoch": 0.37059769107116763,
2225
+ "grad_norm": 0.94921875,
2226
+ "learning_rate": 0.00012358674463937622,
2227
+ "loss": 0.6984,
2228
+ "step": 634
2229
+ },
2230
+ {
2231
+ "epoch": 0.37176676896098204,
2232
+ "grad_norm": 0.72265625,
2233
+ "learning_rate": 0.0001239766081871345,
2234
+ "loss": 0.6796,
2235
+ "step": 636
2236
+ },
2237
+ {
2238
+ "epoch": 0.37293584685079645,
2239
+ "grad_norm": 0.92578125,
2240
+ "learning_rate": 0.0001243664717348928,
2241
+ "loss": 0.8194,
2242
+ "step": 638
2243
+ },
2244
+ {
2245
+ "epoch": 0.37410492474061086,
2246
+ "grad_norm": 0.84375,
2247
+ "learning_rate": 0.00012475633528265107,
2248
+ "loss": 0.7189,
2249
+ "step": 640
2250
+ },
2251
+ {
2252
+ "epoch": 0.37527400263042526,
2253
+ "grad_norm": 0.93359375,
2254
+ "learning_rate": 0.00012514619883040936,
2255
+ "loss": 0.6934,
2256
+ "step": 642
2257
+ },
2258
+ {
2259
+ "epoch": 0.37644308052023967,
2260
+ "grad_norm": 0.93359375,
2261
+ "learning_rate": 0.00012553606237816764,
2262
+ "loss": 0.6885,
2263
+ "step": 644
2264
+ },
2265
+ {
2266
+ "epoch": 0.3776121584100541,
2267
+ "grad_norm": 0.96875,
2268
+ "learning_rate": 0.00012592592592592592,
2269
+ "loss": 0.671,
2270
+ "step": 646
2271
+ },
2272
+ {
2273
+ "epoch": 0.3787812362998685,
2274
+ "grad_norm": 0.74609375,
2275
+ "learning_rate": 0.0001263157894736842,
2276
+ "loss": 0.6515,
2277
+ "step": 648
2278
+ },
2279
+ {
2280
+ "epoch": 0.3799503141896829,
2281
+ "grad_norm": 0.765625,
2282
+ "learning_rate": 0.00012670565302144252,
2283
+ "loss": 0.6833,
2284
+ "step": 650
2285
+ },
2286
+ {
2287
+ "epoch": 0.3811193920794973,
2288
+ "grad_norm": 0.76953125,
2289
+ "learning_rate": 0.00012709551656920077,
2290
+ "loss": 0.6363,
2291
+ "step": 652
2292
+ },
2293
+ {
2294
+ "epoch": 0.3822884699693117,
2295
+ "grad_norm": 0.70703125,
2296
+ "learning_rate": 0.00012748538011695908,
2297
+ "loss": 0.6848,
2298
+ "step": 654
2299
+ },
2300
+ {
2301
+ "epoch": 0.3834575478591261,
2302
+ "grad_norm": 0.93359375,
2303
+ "learning_rate": 0.00012787524366471734,
2304
+ "loss": 0.7383,
2305
+ "step": 656
2306
+ },
2307
+ {
2308
+ "epoch": 0.3846266257489405,
2309
+ "grad_norm": 0.92578125,
2310
+ "learning_rate": 0.00012826510721247565,
2311
+ "loss": 0.7057,
2312
+ "step": 658
2313
+ },
2314
+ {
2315
+ "epoch": 0.3857957036387549,
2316
+ "grad_norm": 0.734375,
2317
+ "learning_rate": 0.0001286549707602339,
2318
+ "loss": 0.7039,
2319
+ "step": 660
2320
+ },
2321
+ {
2322
+ "epoch": 0.38696478152856933,
2323
+ "grad_norm": 0.86328125,
2324
+ "learning_rate": 0.00012904483430799222,
2325
+ "loss": 0.727,
2326
+ "step": 662
2327
+ },
2328
+ {
2329
+ "epoch": 0.38813385941838374,
2330
+ "grad_norm": 1.2265625,
2331
+ "learning_rate": 0.00012943469785575048,
2332
+ "loss": 0.7128,
2333
+ "step": 664
2334
+ },
2335
+ {
2336
+ "epoch": 0.38930293730819815,
2337
+ "grad_norm": 0.671875,
2338
+ "learning_rate": 0.0001298245614035088,
2339
+ "loss": 0.6446,
2340
+ "step": 666
2341
+ },
2342
+ {
2343
+ "epoch": 0.39047201519801256,
2344
+ "grad_norm": 0.65234375,
2345
+ "learning_rate": 0.00013021442495126704,
2346
+ "loss": 0.6434,
2347
+ "step": 668
2348
+ },
2349
+ {
2350
+ "epoch": 0.39164109308782696,
2351
+ "grad_norm": 0.671875,
2352
+ "learning_rate": 0.00013060428849902535,
2353
+ "loss": 0.642,
2354
+ "step": 670
2355
+ },
2356
+ {
2357
+ "epoch": 0.39281017097764137,
2358
+ "grad_norm": 0.9609375,
2359
+ "learning_rate": 0.00013099415204678364,
2360
+ "loss": 0.6805,
2361
+ "step": 672
2362
+ },
2363
+ {
2364
+ "epoch": 0.3939792488674558,
2365
+ "grad_norm": 0.83984375,
2366
+ "learning_rate": 0.00013138401559454192,
2367
+ "loss": 0.6923,
2368
+ "step": 674
2369
+ },
2370
+ {
2371
+ "epoch": 0.3951483267572702,
2372
+ "grad_norm": 0.921875,
2373
+ "learning_rate": 0.0001317738791423002,
2374
+ "loss": 0.7944,
2375
+ "step": 676
2376
+ },
2377
+ {
2378
+ "epoch": 0.3963174046470846,
2379
+ "grad_norm": 0.94921875,
2380
+ "learning_rate": 0.0001321637426900585,
2381
+ "loss": 0.6876,
2382
+ "step": 678
2383
+ },
2384
+ {
2385
+ "epoch": 0.397486482536899,
2386
+ "grad_norm": 0.73046875,
2387
+ "learning_rate": 0.00013255360623781677,
2388
+ "loss": 0.6857,
2389
+ "step": 680
2390
+ },
2391
+ {
2392
+ "epoch": 0.3986555604267134,
2393
+ "grad_norm": 0.83984375,
2394
+ "learning_rate": 0.00013294346978557506,
2395
+ "loss": 0.6796,
2396
+ "step": 682
2397
+ },
2398
+ {
2399
+ "epoch": 0.3998246383165278,
2400
+ "grad_norm": 0.65234375,
2401
+ "learning_rate": 0.00013333333333333334,
2402
+ "loss": 0.6604,
2403
+ "step": 684
2404
+ },
2405
+ {
2406
+ "epoch": 0.4009937162063422,
2407
+ "grad_norm": 0.703125,
2408
+ "learning_rate": 0.00013372319688109162,
2409
+ "loss": 0.6999,
2410
+ "step": 686
2411
+ },
2412
+ {
2413
+ "epoch": 0.4021627940961567,
2414
+ "grad_norm": 1.0234375,
2415
+ "learning_rate": 0.0001341130604288499,
2416
+ "loss": 0.663,
2417
+ "step": 688
2418
+ },
2419
+ {
2420
+ "epoch": 0.4033318719859711,
2421
+ "grad_norm": 0.91796875,
2422
+ "learning_rate": 0.0001345029239766082,
2423
+ "loss": 0.6553,
2424
+ "step": 690
2425
+ },
2426
+ {
2427
+ "epoch": 0.4045009498757855,
2428
+ "grad_norm": 0.77734375,
2429
+ "learning_rate": 0.00013489278752436647,
2430
+ "loss": 0.6892,
2431
+ "step": 692
2432
+ },
2433
+ {
2434
+ "epoch": 0.4056700277655999,
2435
+ "grad_norm": 0.609375,
2436
+ "learning_rate": 0.00013528265107212476,
2437
+ "loss": 0.6529,
2438
+ "step": 694
2439
+ },
2440
+ {
2441
+ "epoch": 0.4068391056554143,
2442
+ "grad_norm": 0.7109375,
2443
+ "learning_rate": 0.00013567251461988304,
2444
+ "loss": 0.6728,
2445
+ "step": 696
2446
+ },
2447
+ {
2448
+ "epoch": 0.4080081835452287,
2449
+ "grad_norm": 0.671875,
2450
+ "learning_rate": 0.00013606237816764133,
2451
+ "loss": 0.6847,
2452
+ "step": 698
2453
+ },
2454
+ {
2455
+ "epoch": 0.4091772614350431,
2456
+ "grad_norm": 0.63671875,
2457
+ "learning_rate": 0.0001364522417153996,
2458
+ "loss": 0.7475,
2459
+ "step": 700
2460
+ },
2461
+ {
2462
+ "epoch": 0.41034633932485753,
2463
+ "grad_norm": 0.76171875,
2464
+ "learning_rate": 0.0001368421052631579,
2465
+ "loss": 0.6805,
2466
+ "step": 702
2467
+ },
2468
+ {
2469
+ "epoch": 0.41151541721467194,
2470
+ "grad_norm": 0.875,
2471
+ "learning_rate": 0.00013723196881091618,
2472
+ "loss": 0.8327,
2473
+ "step": 704
2474
+ },
2475
+ {
2476
+ "epoch": 0.41268449510448635,
2477
+ "grad_norm": 0.94921875,
2478
+ "learning_rate": 0.0001376218323586745,
2479
+ "loss": 0.663,
2480
+ "step": 706
2481
+ },
2482
+ {
2483
+ "epoch": 0.41385357299430076,
2484
+ "grad_norm": 0.6640625,
2485
+ "learning_rate": 0.00013801169590643274,
2486
+ "loss": 0.6675,
2487
+ "step": 708
2488
+ },
2489
+ {
2490
+ "epoch": 0.41502265088411516,
2491
+ "grad_norm": 0.95703125,
2492
+ "learning_rate": 0.00013840155945419105,
2493
+ "loss": 0.7458,
2494
+ "step": 710
2495
+ },
2496
+ {
2497
+ "epoch": 0.41619172877392957,
2498
+ "grad_norm": 0.66015625,
2499
+ "learning_rate": 0.0001387914230019493,
2500
+ "loss": 0.6385,
2501
+ "step": 712
2502
+ },
2503
+ {
2504
+ "epoch": 0.417360806663744,
2505
+ "grad_norm": 0.99609375,
2506
+ "learning_rate": 0.00013918128654970762,
2507
+ "loss": 0.7979,
2508
+ "step": 714
2509
+ },
2510
+ {
2511
+ "epoch": 0.4185298845535584,
2512
+ "grad_norm": 0.92578125,
2513
+ "learning_rate": 0.00013957115009746588,
2514
+ "loss": 0.6949,
2515
+ "step": 716
2516
+ },
2517
+ {
2518
+ "epoch": 0.4196989624433728,
2519
+ "grad_norm": 0.6953125,
2520
+ "learning_rate": 0.0001399610136452242,
2521
+ "loss": 0.6713,
2522
+ "step": 718
2523
+ },
2524
+ {
2525
+ "epoch": 0.4208680403331872,
2526
+ "grad_norm": 0.64453125,
2527
+ "learning_rate": 0.00014035087719298245,
2528
+ "loss": 0.6503,
2529
+ "step": 720
2530
+ },
2531
+ {
2532
+ "epoch": 0.4220371182230016,
2533
+ "grad_norm": 0.6640625,
2534
+ "learning_rate": 0.00014074074074074076,
2535
+ "loss": 0.6641,
2536
+ "step": 722
2537
+ },
2538
+ {
2539
+ "epoch": 0.423206196112816,
2540
+ "grad_norm": 0.6640625,
2541
+ "learning_rate": 0.000141130604288499,
2542
+ "loss": 0.7159,
2543
+ "step": 724
2544
+ },
2545
+ {
2546
+ "epoch": 0.4243752740026304,
2547
+ "grad_norm": 0.69140625,
2548
+ "learning_rate": 0.00014152046783625732,
2549
+ "loss": 0.786,
2550
+ "step": 726
2551
+ },
2552
+ {
2553
+ "epoch": 0.4255443518924448,
2554
+ "grad_norm": 0.92578125,
2555
+ "learning_rate": 0.00014191033138401558,
2556
+ "loss": 0.6511,
2557
+ "step": 728
2558
+ },
2559
+ {
2560
+ "epoch": 0.42671342978225923,
2561
+ "grad_norm": 0.890625,
2562
+ "learning_rate": 0.0001423001949317739,
2563
+ "loss": 0.8375,
2564
+ "step": 730
2565
+ },
2566
+ {
2567
+ "epoch": 0.42788250767207364,
2568
+ "grad_norm": 1.609375,
2569
+ "learning_rate": 0.00014269005847953217,
2570
+ "loss": 0.7475,
2571
+ "step": 732
2572
+ },
2573
+ {
2574
+ "epoch": 0.42905158556188805,
2575
+ "grad_norm": 1.3046875,
2576
+ "learning_rate": 0.00014307992202729046,
2577
+ "loss": 0.6989,
2578
+ "step": 734
2579
+ },
2580
+ {
2581
+ "epoch": 0.43022066345170246,
2582
+ "grad_norm": 0.9453125,
2583
+ "learning_rate": 0.00014346978557504874,
2584
+ "loss": 0.6792,
2585
+ "step": 736
2586
+ },
2587
+ {
2588
+ "epoch": 0.43138974134151686,
2589
+ "grad_norm": 0.734375,
2590
+ "learning_rate": 0.00014385964912280703,
2591
+ "loss": 0.7084,
2592
+ "step": 738
2593
+ },
2594
+ {
2595
+ "epoch": 0.43255881923133127,
2596
+ "grad_norm": 0.83203125,
2597
+ "learning_rate": 0.0001442495126705653,
2598
+ "loss": 0.7263,
2599
+ "step": 740
2600
+ },
2601
+ {
2602
+ "epoch": 0.4337278971211457,
2603
+ "grad_norm": 0.78515625,
2604
+ "learning_rate": 0.0001446393762183236,
2605
+ "loss": 0.706,
2606
+ "step": 742
2607
+ },
2608
+ {
2609
+ "epoch": 0.4348969750109601,
2610
+ "grad_norm": 0.671875,
2611
+ "learning_rate": 0.00014502923976608188,
2612
+ "loss": 0.6989,
2613
+ "step": 744
2614
+ },
2615
+ {
2616
+ "epoch": 0.4360660529007745,
2617
+ "grad_norm": 0.77734375,
2618
+ "learning_rate": 0.00014541910331384016,
2619
+ "loss": 0.7097,
2620
+ "step": 746
2621
+ },
2622
+ {
2623
+ "epoch": 0.4372351307905889,
2624
+ "grad_norm": 0.58984375,
2625
+ "learning_rate": 0.00014580896686159844,
2626
+ "loss": 0.6767,
2627
+ "step": 748
2628
+ },
2629
+ {
2630
+ "epoch": 0.4384042086804033,
2631
+ "grad_norm": 0.6796875,
2632
+ "learning_rate": 0.00014619883040935673,
2633
+ "loss": 0.7045,
2634
+ "step": 750
2635
+ },
2636
+ {
2637
+ "epoch": 0.4395732865702177,
2638
+ "grad_norm": 0.828125,
2639
+ "learning_rate": 0.000146588693957115,
2640
+ "loss": 0.6492,
2641
+ "step": 752
2642
+ },
2643
+ {
2644
+ "epoch": 0.4407423644600322,
2645
+ "grad_norm": 0.609375,
2646
+ "learning_rate": 0.0001469785575048733,
2647
+ "loss": 0.6776,
2648
+ "step": 754
2649
+ },
2650
+ {
2651
+ "epoch": 0.4419114423498466,
2652
+ "grad_norm": 0.77734375,
2653
+ "learning_rate": 0.00014736842105263158,
2654
+ "loss": 0.8008,
2655
+ "step": 756
2656
+ },
2657
+ {
2658
+ "epoch": 0.443080520239661,
2659
+ "grad_norm": 0.76953125,
2660
+ "learning_rate": 0.00014775828460038986,
2661
+ "loss": 0.6677,
2662
+ "step": 758
2663
+ },
2664
+ {
2665
+ "epoch": 0.4442495981294754,
2666
+ "grad_norm": 0.6796875,
2667
+ "learning_rate": 0.00014814814814814815,
2668
+ "loss": 0.6799,
2669
+ "step": 760
2670
+ },
2671
+ {
2672
+ "epoch": 0.4454186760192898,
2673
+ "grad_norm": 0.796875,
2674
+ "learning_rate": 0.00014853801169590643,
2675
+ "loss": 0.6626,
2676
+ "step": 762
2677
+ },
2678
+ {
2679
+ "epoch": 0.4465877539091042,
2680
+ "grad_norm": 0.8828125,
2681
+ "learning_rate": 0.0001489278752436647,
2682
+ "loss": 0.6814,
2683
+ "step": 764
2684
+ },
2685
+ {
2686
+ "epoch": 0.4477568317989186,
2687
+ "grad_norm": 0.671875,
2688
+ "learning_rate": 0.00014931773879142302,
2689
+ "loss": 0.7181,
2690
+ "step": 766
2691
+ },
2692
+ {
2693
+ "epoch": 0.448925909688733,
2694
+ "grad_norm": 0.8046875,
2695
+ "learning_rate": 0.00014970760233918128,
2696
+ "loss": 0.681,
2697
+ "step": 768
2698
+ },
2699
+ {
2700
+ "epoch": 0.45009498757854743,
2701
+ "grad_norm": 0.734375,
2702
+ "learning_rate": 0.0001500974658869396,
2703
+ "loss": 0.6625,
2704
+ "step": 770
2705
+ },
2706
+ {
2707
+ "epoch": 0.45126406546836184,
2708
+ "grad_norm": 0.69140625,
2709
+ "learning_rate": 0.00015048732943469785,
2710
+ "loss": 0.6558,
2711
+ "step": 772
2712
+ },
2713
+ {
2714
+ "epoch": 0.45243314335817625,
2715
+ "grad_norm": 0.66796875,
2716
+ "learning_rate": 0.00015087719298245616,
2717
+ "loss": 0.6968,
2718
+ "step": 774
2719
+ },
2720
+ {
2721
+ "epoch": 0.45360222124799066,
2722
+ "grad_norm": 1.171875,
2723
+ "learning_rate": 0.00015126705653021442,
2724
+ "loss": 0.7073,
2725
+ "step": 776
2726
+ },
2727
+ {
2728
+ "epoch": 0.45477129913780506,
2729
+ "grad_norm": 0.6875,
2730
+ "learning_rate": 0.00015165692007797273,
2731
+ "loss": 0.7151,
2732
+ "step": 778
2733
+ },
2734
+ {
2735
+ "epoch": 0.45594037702761947,
2736
+ "grad_norm": 0.66796875,
2737
+ "learning_rate": 0.00015204678362573098,
2738
+ "loss": 0.7192,
2739
+ "step": 780
2740
+ },
2741
+ {
2742
+ "epoch": 0.4571094549174339,
2743
+ "grad_norm": 0.70703125,
2744
+ "learning_rate": 0.0001524366471734893,
2745
+ "loss": 0.6937,
2746
+ "step": 782
2747
+ },
2748
+ {
2749
+ "epoch": 0.4582785328072483,
2750
+ "grad_norm": 0.76171875,
2751
+ "learning_rate": 0.00015282651072124755,
2752
+ "loss": 0.6859,
2753
+ "step": 784
2754
+ },
2755
+ {
2756
+ "epoch": 0.4594476106970627,
2757
+ "grad_norm": 0.7265625,
2758
+ "learning_rate": 0.00015321637426900586,
2759
+ "loss": 0.6491,
2760
+ "step": 786
2761
+ },
2762
+ {
2763
+ "epoch": 0.4606166885868771,
2764
+ "grad_norm": 0.6953125,
2765
+ "learning_rate": 0.00015360623781676414,
2766
+ "loss": 0.719,
2767
+ "step": 788
2768
+ },
2769
+ {
2770
+ "epoch": 0.4617857664766915,
2771
+ "grad_norm": 0.65625,
2772
+ "learning_rate": 0.00015399610136452243,
2773
+ "loss": 0.6906,
2774
+ "step": 790
2775
+ },
2776
+ {
2777
+ "epoch": 0.4629548443665059,
2778
+ "grad_norm": 0.72265625,
2779
+ "learning_rate": 0.0001543859649122807,
2780
+ "loss": 0.6957,
2781
+ "step": 792
2782
+ },
2783
+ {
2784
+ "epoch": 0.4641239222563203,
2785
+ "grad_norm": 0.59375,
2786
+ "learning_rate": 0.000154775828460039,
2787
+ "loss": 0.6863,
2788
+ "step": 794
2789
+ },
2790
+ {
2791
+ "epoch": 0.4652930001461347,
2792
+ "grad_norm": 0.69921875,
2793
+ "learning_rate": 0.00015516569200779728,
2794
+ "loss": 0.6845,
2795
+ "step": 796
2796
+ },
2797
+ {
2798
+ "epoch": 0.46646207803594913,
2799
+ "grad_norm": 0.625,
2800
+ "learning_rate": 0.00015555555555555556,
2801
+ "loss": 0.6393,
2802
+ "step": 798
2803
+ },
2804
+ {
2805
+ "epoch": 0.46763115592576354,
2806
+ "grad_norm": 0.67578125,
2807
+ "learning_rate": 0.00015594541910331385,
2808
+ "loss": 0.6739,
2809
+ "step": 800
2810
+ },
2811
+ {
2812
+ "epoch": 0.46880023381557795,
2813
+ "grad_norm": 0.62890625,
2814
+ "learning_rate": 0.00015633528265107213,
2815
+ "loss": 0.6919,
2816
+ "step": 802
2817
+ },
2818
+ {
2819
+ "epoch": 0.46996931170539236,
2820
+ "grad_norm": 0.6171875,
2821
+ "learning_rate": 0.00015672514619883041,
2822
+ "loss": 0.6885,
2823
+ "step": 804
2824
+ },
2825
+ {
2826
+ "epoch": 0.47113838959520676,
2827
+ "grad_norm": 0.64453125,
2828
+ "learning_rate": 0.0001571150097465887,
2829
+ "loss": 0.7075,
2830
+ "step": 806
2831
+ },
2832
+ {
2833
+ "epoch": 0.47230746748502117,
2834
+ "grad_norm": 0.6484375,
2835
+ "learning_rate": 0.00015750487329434698,
2836
+ "loss": 0.6716,
2837
+ "step": 808
2838
+ },
2839
+ {
2840
+ "epoch": 0.4734765453748356,
2841
+ "grad_norm": 0.5859375,
2842
+ "learning_rate": 0.00015789473684210527,
2843
+ "loss": 0.6595,
2844
+ "step": 810
2845
+ },
2846
+ {
2847
+ "epoch": 0.47464562326465,
2848
+ "grad_norm": 0.5390625,
2849
+ "learning_rate": 0.00015828460038986355,
2850
+ "loss": 0.6272,
2851
+ "step": 812
2852
+ },
2853
+ {
2854
+ "epoch": 0.4758147011544644,
2855
+ "grad_norm": 0.65234375,
2856
+ "learning_rate": 0.00015867446393762183,
2857
+ "loss": 0.6849,
2858
+ "step": 814
2859
+ },
2860
+ {
2861
+ "epoch": 0.4769837790442788,
2862
+ "grad_norm": 0.6953125,
2863
+ "learning_rate": 0.00015906432748538012,
2864
+ "loss": 0.6523,
2865
+ "step": 816
2866
+ },
2867
+ {
2868
+ "epoch": 0.4781528569340932,
2869
+ "grad_norm": 1.40625,
2870
+ "learning_rate": 0.0001594541910331384,
2871
+ "loss": 0.7112,
2872
+ "step": 818
2873
+ },
2874
+ {
2875
+ "epoch": 0.47932193482390767,
2876
+ "grad_norm": 0.625,
2877
+ "learning_rate": 0.00015984405458089668,
2878
+ "loss": 0.7073,
2879
+ "step": 820
2880
+ },
2881
+ {
2882
+ "epoch": 0.4804910127137221,
2883
+ "grad_norm": 0.5703125,
2884
+ "learning_rate": 0.000160233918128655,
2885
+ "loss": 0.6436,
2886
+ "step": 822
2887
+ },
2888
+ {
2889
+ "epoch": 0.4816600906035365,
2890
+ "grad_norm": 0.58203125,
2891
+ "learning_rate": 0.00016062378167641325,
2892
+ "loss": 0.669,
2893
+ "step": 824
2894
+ },
2895
+ {
2896
+ "epoch": 0.4828291684933509,
2897
+ "grad_norm": 0.94921875,
2898
+ "learning_rate": 0.00016101364522417156,
2899
+ "loss": 0.8224,
2900
+ "step": 826
2901
+ },
2902
+ {
2903
+ "epoch": 0.4839982463831653,
2904
+ "grad_norm": 0.75390625,
2905
+ "learning_rate": 0.00016140350877192982,
2906
+ "loss": 0.7526,
2907
+ "step": 828
2908
+ },
2909
+ {
2910
+ "epoch": 0.4851673242729797,
2911
+ "grad_norm": 0.64453125,
2912
+ "learning_rate": 0.00016179337231968813,
2913
+ "loss": 0.6986,
2914
+ "step": 830
2915
+ },
2916
+ {
2917
+ "epoch": 0.4863364021627941,
2918
+ "grad_norm": 0.625,
2919
+ "learning_rate": 0.00016218323586744639,
2920
+ "loss": 0.6496,
2921
+ "step": 832
2922
+ },
2923
+ {
2924
+ "epoch": 0.4875054800526085,
2925
+ "grad_norm": 0.6953125,
2926
+ "learning_rate": 0.0001625730994152047,
2927
+ "loss": 0.7165,
2928
+ "step": 834
2929
+ },
2930
+ {
2931
+ "epoch": 0.4886745579424229,
2932
+ "grad_norm": 0.58203125,
2933
+ "learning_rate": 0.00016296296296296295,
2934
+ "loss": 0.6453,
2935
+ "step": 836
2936
+ },
2937
+ {
2938
+ "epoch": 0.48984363583223733,
2939
+ "grad_norm": 0.61328125,
2940
+ "learning_rate": 0.00016335282651072126,
2941
+ "loss": 0.617,
2942
+ "step": 838
2943
+ },
2944
+ {
2945
+ "epoch": 0.49101271372205174,
2946
+ "grad_norm": 0.60546875,
2947
+ "learning_rate": 0.00016374269005847952,
2948
+ "loss": 0.6841,
2949
+ "step": 840
2950
+ },
2951
+ {
2952
+ "epoch": 0.49218179161186615,
2953
+ "grad_norm": 0.671875,
2954
+ "learning_rate": 0.00016413255360623783,
2955
+ "loss": 0.6668,
2956
+ "step": 842
2957
+ },
2958
+ {
2959
+ "epoch": 0.49335086950168056,
2960
+ "grad_norm": 0.71875,
2961
+ "learning_rate": 0.0001645224171539961,
2962
+ "loss": 0.7092,
2963
+ "step": 844
2964
+ },
2965
+ {
2966
+ "epoch": 0.49451994739149496,
2967
+ "grad_norm": 0.71484375,
2968
+ "learning_rate": 0.0001649122807017544,
2969
+ "loss": 0.6479,
2970
+ "step": 846
2971
+ },
2972
+ {
2973
+ "epoch": 0.49568902528130937,
2974
+ "grad_norm": 0.81640625,
2975
+ "learning_rate": 0.00016530214424951268,
2976
+ "loss": 0.6687,
2977
+ "step": 848
2978
+ },
2979
+ {
2980
+ "epoch": 0.4968581031711238,
2981
+ "grad_norm": 0.70703125,
2982
+ "learning_rate": 0.00016569200779727097,
2983
+ "loss": 0.6704,
2984
+ "step": 850
2985
+ },
2986
+ {
2987
+ "epoch": 0.4980271810609382,
2988
+ "grad_norm": 0.65234375,
2989
+ "learning_rate": 0.00016608187134502925,
2990
+ "loss": 0.6678,
2991
+ "step": 852
2992
+ },
2993
+ {
2994
+ "epoch": 0.4991962589507526,
2995
+ "grad_norm": 0.69140625,
2996
+ "learning_rate": 0.00016647173489278753,
2997
+ "loss": 0.6744,
2998
+ "step": 854
2999
+ },
3000
+ {
3001
+ "epoch": 0.500365336840567,
3002
+ "grad_norm": 0.79296875,
3003
+ "learning_rate": 0.00016686159844054582,
3004
+ "loss": 0.6565,
3005
+ "step": 856
3006
+ },
3007
+ {
3008
+ "epoch": 0.5015344147303814,
3009
+ "grad_norm": 0.62109375,
3010
+ "learning_rate": 0.0001672514619883041,
3011
+ "loss": 0.6499,
3012
+ "step": 858
3013
+ },
3014
+ {
3015
+ "epoch": 0.5027034926201959,
3016
+ "grad_norm": 0.84375,
3017
+ "learning_rate": 0.00016764132553606238,
3018
+ "loss": 0.6619,
3019
+ "step": 860
3020
+ },
3021
+ {
3022
+ "epoch": 0.5038725705100102,
3023
+ "grad_norm": 1.703125,
3024
+ "learning_rate": 0.00016803118908382067,
3025
+ "loss": 0.8945,
3026
+ "step": 862
3027
+ },
3028
+ {
3029
+ "epoch": 0.5050416483998247,
3030
+ "grad_norm": 0.9296875,
3031
+ "learning_rate": 0.00016842105263157895,
3032
+ "loss": 0.7597,
3033
+ "step": 864
3034
+ },
3035
+ {
3036
+ "epoch": 0.506210726289639,
3037
+ "grad_norm": 0.80078125,
3038
+ "learning_rate": 0.00016881091617933723,
3039
+ "loss": 0.6698,
3040
+ "step": 866
3041
+ },
3042
+ {
3043
+ "epoch": 0.5073798041794535,
3044
+ "grad_norm": 1.1953125,
3045
+ "learning_rate": 0.00016920077972709552,
3046
+ "loss": 0.7135,
3047
+ "step": 868
3048
+ },
3049
+ {
3050
+ "epoch": 0.5085488820692678,
3051
+ "grad_norm": 1.0859375,
3052
+ "learning_rate": 0.0001695906432748538,
3053
+ "loss": 0.8298,
3054
+ "step": 870
3055
+ },
3056
+ {
3057
+ "epoch": 0.5097179599590823,
3058
+ "grad_norm": 0.75390625,
3059
+ "learning_rate": 0.00016998050682261209,
3060
+ "loss": 0.6749,
3061
+ "step": 872
3062
+ },
3063
+ {
3064
+ "epoch": 0.5108870378488967,
3065
+ "grad_norm": 1.09375,
3066
+ "learning_rate": 0.00017037037037037037,
3067
+ "loss": 0.8318,
3068
+ "step": 874
3069
+ },
3070
+ {
3071
+ "epoch": 0.5120561157387111,
3072
+ "grad_norm": 0.7109375,
3073
+ "learning_rate": 0.00017076023391812865,
3074
+ "loss": 0.7329,
3075
+ "step": 876
3076
+ },
3077
+ {
3078
+ "epoch": 0.5132251936285255,
3079
+ "grad_norm": 0.87109375,
3080
+ "learning_rate": 0.00017115009746588694,
3081
+ "loss": 0.7127,
3082
+ "step": 878
3083
+ },
3084
+ {
3085
+ "epoch": 0.5143942715183399,
3086
+ "grad_norm": 9.375,
3087
+ "learning_rate": 0.00017153996101364522,
3088
+ "loss": 0.7969,
3089
+ "step": 880
3090
+ },
3091
+ {
3092
+ "epoch": 0.5155633494081543,
3093
+ "grad_norm": 2.015625,
3094
+ "learning_rate": 0.00017192982456140353,
3095
+ "loss": 0.8313,
3096
+ "step": 882
3097
+ },
3098
+ {
3099
+ "epoch": 0.5167324272979688,
3100
+ "grad_norm": 0.71875,
3101
+ "learning_rate": 0.0001723196881091618,
3102
+ "loss": 0.7693,
3103
+ "step": 884
3104
+ },
3105
+ {
3106
+ "epoch": 0.5179015051877831,
3107
+ "grad_norm": 0.625,
3108
+ "learning_rate": 0.0001727095516569201,
3109
+ "loss": 0.7127,
3110
+ "step": 886
3111
+ },
3112
+ {
3113
+ "epoch": 0.5190705830775976,
3114
+ "grad_norm": 0.72265625,
3115
+ "learning_rate": 0.00017309941520467836,
3116
+ "loss": 0.7283,
3117
+ "step": 888
3118
+ },
3119
+ {
3120
+ "epoch": 0.5202396609674119,
3121
+ "grad_norm": 0.70703125,
3122
+ "learning_rate": 0.00017348927875243667,
3123
+ "loss": 0.6591,
3124
+ "step": 890
3125
+ },
3126
+ {
3127
+ "epoch": 0.5214087388572264,
3128
+ "grad_norm": 0.66015625,
3129
+ "learning_rate": 0.00017387914230019492,
3130
+ "loss": 0.6794,
3131
+ "step": 892
3132
+ },
3133
+ {
3134
+ "epoch": 0.5225778167470407,
3135
+ "grad_norm": 0.57421875,
3136
+ "learning_rate": 0.00017426900584795323,
3137
+ "loss": 0.6299,
3138
+ "step": 894
3139
+ },
3140
+ {
3141
+ "epoch": 0.5237468946368552,
3142
+ "grad_norm": 0.92578125,
3143
+ "learning_rate": 0.0001746588693957115,
3144
+ "loss": 0.73,
3145
+ "step": 896
3146
+ },
3147
+ {
3148
+ "epoch": 0.5249159725266695,
3149
+ "grad_norm": 0.65625,
3150
+ "learning_rate": 0.0001750487329434698,
3151
+ "loss": 0.6538,
3152
+ "step": 898
3153
+ },
3154
+ {
3155
+ "epoch": 0.526085050416484,
3156
+ "grad_norm": 0.60546875,
3157
+ "learning_rate": 0.00017543859649122806,
3158
+ "loss": 0.7013,
3159
+ "step": 900
3160
+ },
3161
+ {
3162
+ "epoch": 0.5272541283062984,
3163
+ "grad_norm": 0.66015625,
3164
+ "learning_rate": 0.00017582846003898637,
3165
+ "loss": 0.6633,
3166
+ "step": 902
3167
+ },
3168
+ {
3169
+ "epoch": 0.5284232061961128,
3170
+ "grad_norm": 0.5234375,
3171
+ "learning_rate": 0.00017621832358674462,
3172
+ "loss": 0.6638,
3173
+ "step": 904
3174
+ },
3175
+ {
3176
+ "epoch": 0.5295922840859272,
3177
+ "grad_norm": 0.65234375,
3178
+ "learning_rate": 0.00017660818713450294,
3179
+ "loss": 0.6616,
3180
+ "step": 906
3181
+ },
3182
+ {
3183
+ "epoch": 0.5307613619757416,
3184
+ "grad_norm": 0.765625,
3185
+ "learning_rate": 0.00017699805068226122,
3186
+ "loss": 0.6935,
3187
+ "step": 908
3188
+ },
3189
+ {
3190
+ "epoch": 0.531930439865556,
3191
+ "grad_norm": 0.65625,
3192
+ "learning_rate": 0.0001773879142300195,
3193
+ "loss": 0.6795,
3194
+ "step": 910
3195
+ },
3196
+ {
3197
+ "epoch": 0.5330995177553705,
3198
+ "grad_norm": 0.87890625,
3199
+ "learning_rate": 0.00017777777777777779,
3200
+ "loss": 0.6797,
3201
+ "step": 912
3202
+ },
3203
+ {
3204
+ "epoch": 0.5342685956451848,
3205
+ "grad_norm": 0.75,
3206
+ "learning_rate": 0.00017816764132553607,
3207
+ "loss": 0.6583,
3208
+ "step": 914
3209
+ },
3210
+ {
3211
+ "epoch": 0.5354376735349993,
3212
+ "grad_norm": 0.640625,
3213
+ "learning_rate": 0.00017855750487329435,
3214
+ "loss": 0.6592,
3215
+ "step": 916
3216
+ },
3217
+ {
3218
+ "epoch": 0.5366067514248137,
3219
+ "grad_norm": 0.6328125,
3220
+ "learning_rate": 0.00017894736842105264,
3221
+ "loss": 0.7534,
3222
+ "step": 918
3223
+ },
3224
+ {
3225
+ "epoch": 0.5377758293146281,
3226
+ "grad_norm": 0.66796875,
3227
+ "learning_rate": 0.00017933723196881092,
3228
+ "loss": 0.699,
3229
+ "step": 920
3230
+ },
3231
+ {
3232
+ "epoch": 0.5389449072044425,
3233
+ "grad_norm": 0.7109375,
3234
+ "learning_rate": 0.0001797270955165692,
3235
+ "loss": 0.6759,
3236
+ "step": 922
3237
+ },
3238
+ {
3239
+ "epoch": 0.5401139850942569,
3240
+ "grad_norm": 0.58984375,
3241
+ "learning_rate": 0.0001801169590643275,
3242
+ "loss": 0.6676,
3243
+ "step": 924
3244
+ },
3245
+ {
3246
+ "epoch": 0.5412830629840714,
3247
+ "grad_norm": 0.6328125,
3248
+ "learning_rate": 0.00018050682261208577,
3249
+ "loss": 0.6734,
3250
+ "step": 926
3251
+ },
3252
+ {
3253
+ "epoch": 0.5424521408738857,
3254
+ "grad_norm": 0.72265625,
3255
+ "learning_rate": 0.00018089668615984406,
3256
+ "loss": 0.6545,
3257
+ "step": 928
3258
+ },
3259
+ {
3260
+ "epoch": 0.5436212187637002,
3261
+ "grad_norm": 0.65234375,
3262
+ "learning_rate": 0.00018128654970760234,
3263
+ "loss": 0.6863,
3264
+ "step": 930
3265
+ },
3266
+ {
3267
+ "epoch": 0.5447902966535145,
3268
+ "grad_norm": 0.54296875,
3269
+ "learning_rate": 0.00018167641325536062,
3270
+ "loss": 0.6536,
3271
+ "step": 932
3272
+ },
3273
+ {
3274
+ "epoch": 0.545959374543329,
3275
+ "grad_norm": 0.77734375,
3276
+ "learning_rate": 0.0001820662768031189,
3277
+ "loss": 0.6825,
3278
+ "step": 934
3279
+ },
3280
+ {
3281
+ "epoch": 0.5471284524331433,
3282
+ "grad_norm": 0.69921875,
3283
+ "learning_rate": 0.0001824561403508772,
3284
+ "loss": 0.7075,
3285
+ "step": 936
3286
+ },
3287
+ {
3288
+ "epoch": 0.5482975303229578,
3289
+ "grad_norm": 1.265625,
3290
+ "learning_rate": 0.00018284600389863547,
3291
+ "loss": 0.8103,
3292
+ "step": 938
3293
+ },
3294
+ {
3295
+ "epoch": 0.5494666082127722,
3296
+ "grad_norm": 1.4765625,
3297
+ "learning_rate": 0.00018323586744639376,
3298
+ "loss": 0.7081,
3299
+ "step": 940
3300
+ },
3301
+ {
3302
+ "epoch": 0.5506356861025866,
3303
+ "grad_norm": 0.734375,
3304
+ "learning_rate": 0.00018362573099415207,
3305
+ "loss": 0.6647,
3306
+ "step": 942
3307
+ },
3308
+ {
3309
+ "epoch": 0.551804763992401,
3310
+ "grad_norm": 0.98828125,
3311
+ "learning_rate": 0.00018401559454191032,
3312
+ "loss": 0.7202,
3313
+ "step": 944
3314
+ },
3315
+ {
3316
+ "epoch": 0.5529738418822154,
3317
+ "grad_norm": 1.09375,
3318
+ "learning_rate": 0.00018440545808966864,
3319
+ "loss": 0.7734,
3320
+ "step": 946
3321
+ },
3322
+ {
3323
+ "epoch": 0.5541429197720298,
3324
+ "grad_norm": 0.64453125,
3325
+ "learning_rate": 0.0001847953216374269,
3326
+ "loss": 0.6381,
3327
+ "step": 948
3328
+ },
3329
+ {
3330
+ "epoch": 0.5553119976618442,
3331
+ "grad_norm": 0.6796875,
3332
+ "learning_rate": 0.0001851851851851852,
3333
+ "loss": 0.7235,
3334
+ "step": 950
3335
+ },
3336
+ {
3337
+ "epoch": 0.5564810755516586,
3338
+ "grad_norm": 0.66796875,
3339
+ "learning_rate": 0.00018557504873294346,
3340
+ "loss": 0.8296,
3341
+ "step": 952
3342
+ },
3343
+ {
3344
+ "epoch": 0.5576501534414731,
3345
+ "grad_norm": 0.65234375,
3346
+ "learning_rate": 0.00018596491228070177,
3347
+ "loss": 0.6901,
3348
+ "step": 954
3349
+ },
3350
+ {
3351
+ "epoch": 0.5588192313312874,
3352
+ "grad_norm": 0.6796875,
3353
+ "learning_rate": 0.00018635477582846003,
3354
+ "loss": 0.6628,
3355
+ "step": 956
3356
+ },
3357
+ {
3358
+ "epoch": 0.5599883092211019,
3359
+ "grad_norm": 0.671875,
3360
+ "learning_rate": 0.00018674463937621834,
3361
+ "loss": 0.6689,
3362
+ "step": 958
3363
+ },
3364
+ {
3365
+ "epoch": 0.5611573871109162,
3366
+ "grad_norm": 0.6875,
3367
+ "learning_rate": 0.0001871345029239766,
3368
+ "loss": 0.7377,
3369
+ "step": 960
3370
+ },
3371
+ {
3372
+ "epoch": 0.5623264650007307,
3373
+ "grad_norm": 0.62890625,
3374
+ "learning_rate": 0.0001875243664717349,
3375
+ "loss": 0.7217,
3376
+ "step": 962
3377
+ },
3378
+ {
3379
+ "epoch": 0.563495542890545,
3380
+ "grad_norm": 0.65625,
3381
+ "learning_rate": 0.0001879142300194932,
3382
+ "loss": 0.6776,
3383
+ "step": 964
3384
+ },
3385
+ {
3386
+ "epoch": 0.5646646207803595,
3387
+ "grad_norm": 0.74609375,
3388
+ "learning_rate": 0.00018830409356725147,
3389
+ "loss": 0.694,
3390
+ "step": 966
3391
+ },
3392
+ {
3393
+ "epoch": 0.5658336986701739,
3394
+ "grad_norm": 0.5859375,
3395
+ "learning_rate": 0.00018869395711500976,
3396
+ "loss": 0.7305,
3397
+ "step": 968
3398
+ },
3399
+ {
3400
+ "epoch": 0.5670027765599883,
3401
+ "grad_norm": 0.671875,
3402
+ "learning_rate": 0.00018908382066276804,
3403
+ "loss": 0.6659,
3404
+ "step": 970
3405
+ },
3406
+ {
3407
+ "epoch": 0.5681718544498027,
3408
+ "grad_norm": 0.7578125,
3409
+ "learning_rate": 0.00018947368421052632,
3410
+ "loss": 0.705,
3411
+ "step": 972
3412
+ },
3413
+ {
3414
+ "epoch": 0.5693409323396171,
3415
+ "grad_norm": 0.64453125,
3416
+ "learning_rate": 0.0001898635477582846,
3417
+ "loss": 0.8348,
3418
+ "step": 974
3419
+ },
3420
+ {
3421
+ "epoch": 0.5705100102294315,
3422
+ "grad_norm": 0.65625,
3423
+ "learning_rate": 0.0001902534113060429,
3424
+ "loss": 0.7383,
3425
+ "step": 976
3426
+ },
3427
+ {
3428
+ "epoch": 0.571679088119246,
3429
+ "grad_norm": 0.609375,
3430
+ "learning_rate": 0.00019064327485380117,
3431
+ "loss": 0.7039,
3432
+ "step": 978
3433
+ },
3434
+ {
3435
+ "epoch": 0.5728481660090603,
3436
+ "grad_norm": 0.69921875,
3437
+ "learning_rate": 0.00019103313840155946,
3438
+ "loss": 0.7163,
3439
+ "step": 980
3440
+ },
3441
+ {
3442
+ "epoch": 0.5740172438988748,
3443
+ "grad_norm": 0.69140625,
3444
+ "learning_rate": 0.00019142300194931774,
3445
+ "loss": 0.6648,
3446
+ "step": 982
3447
+ },
3448
+ {
3449
+ "epoch": 0.5751863217886892,
3450
+ "grad_norm": 0.54296875,
3451
+ "learning_rate": 0.00019181286549707603,
3452
+ "loss": 0.6896,
3453
+ "step": 984
3454
+ },
3455
+ {
3456
+ "epoch": 0.5763553996785036,
3457
+ "grad_norm": 0.70703125,
3458
+ "learning_rate": 0.0001922027290448343,
3459
+ "loss": 0.7127,
3460
+ "step": 986
3461
+ },
3462
+ {
3463
+ "epoch": 0.577524477568318,
3464
+ "grad_norm": 0.8828125,
3465
+ "learning_rate": 0.0001925925925925926,
3466
+ "loss": 0.6549,
3467
+ "step": 988
3468
+ },
3469
+ {
3470
+ "epoch": 0.5786935554581324,
3471
+ "grad_norm": 0.77734375,
3472
+ "learning_rate": 0.00019298245614035088,
3473
+ "loss": 0.6581,
3474
+ "step": 990
3475
+ },
3476
+ {
3477
+ "epoch": 0.5798626333479469,
3478
+ "grad_norm": 0.578125,
3479
+ "learning_rate": 0.00019337231968810916,
3480
+ "loss": 0.6802,
3481
+ "step": 992
3482
+ },
3483
+ {
3484
+ "epoch": 0.5810317112377612,
3485
+ "grad_norm": 1.03125,
3486
+ "learning_rate": 0.00019376218323586744,
3487
+ "loss": 0.6906,
3488
+ "step": 994
3489
+ },
3490
+ {
3491
+ "epoch": 0.5822007891275757,
3492
+ "grad_norm": 0.7265625,
3493
+ "learning_rate": 0.00019415204678362573,
3494
+ "loss": 0.7383,
3495
+ "step": 996
3496
+ },
3497
+ {
3498
+ "epoch": 0.58336986701739,
3499
+ "grad_norm": 0.65234375,
3500
+ "learning_rate": 0.00019454191033138404,
3501
+ "loss": 0.7141,
3502
+ "step": 998
3503
+ },
3504
+ {
3505
+ "epoch": 0.5845389449072045,
3506
+ "grad_norm": 0.58984375,
3507
+ "learning_rate": 0.0001949317738791423,
3508
+ "loss": 0.6729,
3509
+ "step": 1000
3510
+ }
3511
+ ],
3512
+ "logging_steps": 2,
3513
+ "max_steps": 5130,
3514
+ "num_input_tokens_seen": 0,
3515
+ "num_train_epochs": 3,
3516
+ "save_steps": 1000,
3517
+ "stateful_callbacks": {
3518
+ "TrainerControl": {
3519
+ "args": {
3520
+ "should_epoch_stop": false,
3521
+ "should_evaluate": false,
3522
+ "should_log": false,
3523
+ "should_save": true,
3524
+ "should_training_stop": false
3525
+ },
3526
+ "attributes": {}
3527
+ }
3528
+ },
3529
+ "total_flos": 2.57623446257664e+17,
3530
+ "train_batch_size": 1,
3531
+ "trial_name": null,
3532
+ "trial_params": null
3533
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76e9814eadbb1390f5aa76cd6f56acdb6c21c135f8242b5f7bc6e0aab4d1414f
3
+ size 5432
checkpoint-1000/vocab.json ADDED
The diff for this file is too large to render. See raw diff