Hugging Face's logo

Varine
/
opus-mt-zh-en-model

Text Generation
Transformers
TensorBoard
Safetensors
Chinese
English
marian
text2text-generation
Model card Files
xet
Metrics Community
Varine commited on
Commit
c73aa4a
·
1 Parent(s): c122ee7

Add model files

Browse files
Files changed (50) hide show
  1. checkpoint-233865/config.json +61 -0
  2. checkpoint-233865/generation_config.json +16 -0
  3. checkpoint-233865/model.safetensors +3 -0
  4. checkpoint-233865/optimizer.pt +3 -0
  5. checkpoint-233865/rng_state.pth +3 -0
  6. checkpoint-233865/scheduler.pt +3 -0
  7. checkpoint-233865/source.spm +0 -0
  8. checkpoint-233865/special_tokens_map.json +5 -0
  9. checkpoint-233865/target.spm +0 -0
  10. checkpoint-233865/tokenizer_config.json +38 -0
  11. checkpoint-233865/trainer_state.json +3362 -0
  12. checkpoint-233865/training_args.bin +3 -0
  13. checkpoint-233865/vocab.json +0 -0
  14. checkpoint-25985/config.json +61 -0
  15. checkpoint-25985/generation_config.json +16 -0
  16. checkpoint-25985/model.safetensors +3 -0
  17. checkpoint-25985/optimizer.pt +3 -0
  18. checkpoint-25985/rng_state.pth +3 -0
  19. checkpoint-25985/scheduler.pt +3 -0
  20. checkpoint-25985/source.spm +0 -0
  21. checkpoint-25985/special_tokens_map.json +5 -0
  22. checkpoint-25985/target.spm +0 -0
  23. checkpoint-25985/tokenizer_config.json +38 -0
  24. checkpoint-25985/trainer_state.json +386 -0
  25. checkpoint-25985/training_args.bin +3 -0
  26. checkpoint-25985/vocab.json +0 -0
  27. checkpoint-259850/config.json +61 -0
  28. checkpoint-259850/generation_config.json +16 -0
  29. checkpoint-259850/model.safetensors +3 -0
  30. checkpoint-259850/optimizer.pt +3 -0
  31. checkpoint-259850/rng_state.pth +3 -0
  32. checkpoint-259850/scheduler.pt +3 -0
  33. checkpoint-259850/source.spm +0 -0
  34. checkpoint-259850/special_tokens_map.json +5 -0
  35. checkpoint-259850/target.spm +0 -0
  36. checkpoint-259850/tokenizer_config.json +38 -0
  37. checkpoint-259850/trainer_state.json +3734 -0
  38. checkpoint-259850/training_args.bin +3 -0
  39. checkpoint-259850/vocab.json +0 -0
  40. config.json +61 -0
  41. generation_config.json +16 -0
  42. model.safetensors +3 -0
  43. runs/Apr16_07-58-43_129-213-63-124/events.out.tfevents.1713254333.129-213-63-124.2528.3 +3 -0
  44. runs/Apr16_12-36-45_129-213-89-192/events.out.tfevents.1713271010.129-213-89-192.2749.0 +3 -0
  45. runs/Apr16_12-38-25_129-213-89-192/events.out.tfevents.1713271109.129-213-89-192.2749.1 +3 -0
  46. source.spm +0 -0
  47. special_tokens_map.json +5 -0
  48. target.spm +0 -0
  49. tokenizer_config.json +38 -0
  50. vocab.json +0 -0
checkpoint-233865/config.json ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-51970",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "swish",
5
+ "add_bias_logits": false,
6
+ "add_final_layer_norm": false,
7
+ "architectures": [
8
+ "MarianMTModel"
9
+ ],
10
+ "attention_dropout": 0.0,
11
+ "bad_words_ids": [
12
+ [
13
+ 65000
14
+ ]
15
+ ],
16
+ "bos_token_id": 0,
17
+ "classif_dropout": 0.0,
18
+ "classifier_dropout": 0.0,
19
+ "d_model": 512,
20
+ "decoder_attention_heads": 8,
21
+ "decoder_ffn_dim": 2048,
22
+ "decoder_layerdrop": 0.0,
23
+ "decoder_layers": 6,
24
+ "decoder_start_token_id": 65000,
25
+ "decoder_vocab_size": 65001,
26
+ "dropout": 0.1,
27
+ "encoder_attention_heads": 8,
28
+ "encoder_ffn_dim": 2048,
29
+ "encoder_layerdrop": 0.0,
30
+ "encoder_layers": 6,
31
+ "eos_token_id": 0,
32
+ "extra_pos_embeddings": 65001,
33
+ "forced_eos_token_id": 0,
34
+ "id2label": {
35
+ "0": "LABEL_0",
36
+ "1": "LABEL_1",
37
+ "2": "LABEL_2"
38
+ },
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "label2id": {
42
+ "LABEL_0": 0,
43
+ "LABEL_1": 1,
44
+ "LABEL_2": 2
45
+ },
46
+ "max_length": 512,
47
+ "max_position_embeddings": 512,
48
+ "model_type": "marian",
49
+ "normalize_before": false,
50
+ "normalize_embedding": false,
51
+ "num_beams": 6,
52
+ "num_hidden_layers": 6,
53
+ "pad_token_id": 65000,
54
+ "scale_embedding": true,
55
+ "share_encoder_decoder_embeddings": true,
56
+ "static_position_embeddings": true,
57
+ "torch_dtype": "float32",
58
+ "transformers_version": "4.39.3",
59
+ "use_cache": true,
60
+ "vocab_size": 65001
61
+ }
checkpoint-233865/generation_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bad_words_ids": [
3
+ [
4
+ 65000
5
+ ]
6
+ ],
7
+ "bos_token_id": 0,
8
+ "decoder_start_token_id": 65000,
9
+ "eos_token_id": 0,
10
+ "forced_eos_token_id": 0,
11
+ "max_length": 512,
12
+ "num_beams": 6,
13
+ "pad_token_id": 65000,
14
+ "renormalize_logits": true,
15
+ "transformers_version": "4.39.3"
16
+ }
checkpoint-233865/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f25814585811b09ed58ff5ad66875fad5572bb1b964effb5e817e53fb9dc129
3
+ size 309965092
checkpoint-233865/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7bc487bde9a556db591357ce22da0ad23f4db233a9ec0e022cc6d22e76e07ef8
3
+ size 619563141
checkpoint-233865/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:da3fc7a3c42bfd6089d11062a3f808cb320afc3013c78f64637c90c3f2375747
3
+ size 14575
checkpoint-233865/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57a592895bc9126b54efe398b7e89ea20d4d6ed124a500fdc45d0b7541b97627
3
+ size 627
checkpoint-233865/source.spm ADDED
Binary file (805 kB). View file
 
checkpoint-233865/special_tokens_map.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "eos_token": "</s>",
3
+ "pad_token": "<pad>",
4
+ "unk_token": "<unk>"
5
+ }
checkpoint-233865/target.spm ADDED
Binary file (807 kB). View file
 
checkpoint-233865/tokenizer_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "</s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<unk>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "65000": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ }
27
+ },
28
+ "clean_up_tokenization_spaces": true,
29
+ "eos_token": "</s>",
30
+ "model_max_length": 512,
31
+ "pad_token": "<pad>",
32
+ "separate_vocabs": false,
33
+ "source_lang": "zho",
34
+ "sp_model_kwargs": {},
35
+ "target_lang": "eng",
36
+ "tokenizer_class": "MarianTokenizer",
37
+ "unk_token": "<unk>"
38
+ }
checkpoint-233865/trainer_state.json ADDED
@@ -0,0 +1,3362 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.5939216613769531,
3
+ "best_model_checkpoint": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-25985",
4
+ "epoch": 9.0,
5
+ "eval_steps": 500,
6
+ "global_step": 233865,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "grad_norm": 2.777989149093628,
14
+ "learning_rate": 1.9961593226861654e-05,
15
+ "loss": 0.608,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.04,
20
+ "grad_norm": 2.571350336074829,
21
+ "learning_rate": 1.9923109486242066e-05,
22
+ "loss": 0.6031,
23
+ "step": 1000
24
+ },
25
+ {
26
+ "epoch": 0.06,
27
+ "grad_norm": 6.669718265533447,
28
+ "learning_rate": 1.9884625745622478e-05,
29
+ "loss": 0.5823,
30
+ "step": 1500
31
+ },
32
+ {
33
+ "epoch": 0.08,
34
+ "grad_norm": 3.766803503036499,
35
+ "learning_rate": 1.984614200500289e-05,
36
+ "loss": 0.625,
37
+ "step": 2000
38
+ },
39
+ {
40
+ "epoch": 0.1,
41
+ "grad_norm": 2.617687225341797,
42
+ "learning_rate": 1.9807735231864538e-05,
43
+ "loss": 0.6324,
44
+ "step": 2500
45
+ },
46
+ {
47
+ "epoch": 0.12,
48
+ "grad_norm": 4.26535177230835,
49
+ "learning_rate": 1.976932845872619e-05,
50
+ "loss": 0.6184,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 0.13,
55
+ "grad_norm": 5.035365581512451,
56
+ "learning_rate": 1.9730844718106602e-05,
57
+ "loss": 0.6277,
58
+ "step": 3500
59
+ },
60
+ {
61
+ "epoch": 0.15,
62
+ "grad_norm": 6.626140594482422,
63
+ "learning_rate": 1.9692360977487014e-05,
64
+ "loss": 0.6557,
65
+ "step": 4000
66
+ },
67
+ {
68
+ "epoch": 0.17,
69
+ "grad_norm": 3.718008279800415,
70
+ "learning_rate": 1.9653877236867426e-05,
71
+ "loss": 0.6623,
72
+ "step": 4500
73
+ },
74
+ {
75
+ "epoch": 0.19,
76
+ "grad_norm": 4.26861047744751,
77
+ "learning_rate": 1.9615470463729075e-05,
78
+ "loss": 0.6719,
79
+ "step": 5000
80
+ },
81
+ {
82
+ "epoch": 0.21,
83
+ "grad_norm": 5.266055107116699,
84
+ "learning_rate": 1.9576986723109487e-05,
85
+ "loss": 0.636,
86
+ "step": 5500
87
+ },
88
+ {
89
+ "epoch": 0.23,
90
+ "grad_norm": 6.086512565612793,
91
+ "learning_rate": 1.95385029824899e-05,
92
+ "loss": 0.6585,
93
+ "step": 6000
94
+ },
95
+ {
96
+ "epoch": 0.25,
97
+ "grad_norm": 4.0940327644348145,
98
+ "learning_rate": 1.950001924187031e-05,
99
+ "loss": 0.6341,
100
+ "step": 6500
101
+ },
102
+ {
103
+ "epoch": 0.27,
104
+ "grad_norm": 7.52503776550293,
105
+ "learning_rate": 1.9461535501250722e-05,
106
+ "loss": 0.6589,
107
+ "step": 7000
108
+ },
109
+ {
110
+ "epoch": 0.29,
111
+ "grad_norm": 3.5400032997131348,
112
+ "learning_rate": 1.9423051760631134e-05,
113
+ "loss": 0.6627,
114
+ "step": 7500
115
+ },
116
+ {
117
+ "epoch": 0.31,
118
+ "grad_norm": 3.525592565536499,
119
+ "learning_rate": 1.9384568020011546e-05,
120
+ "loss": 0.639,
121
+ "step": 8000
122
+ },
123
+ {
124
+ "epoch": 0.33,
125
+ "grad_norm": 3.1940386295318604,
126
+ "learning_rate": 1.9346084279391958e-05,
127
+ "loss": 0.6433,
128
+ "step": 8500
129
+ },
130
+ {
131
+ "epoch": 0.35,
132
+ "grad_norm": 5.362303733825684,
133
+ "learning_rate": 1.930767750625361e-05,
134
+ "loss": 0.6714,
135
+ "step": 9000
136
+ },
137
+ {
138
+ "epoch": 0.37,
139
+ "grad_norm": 4.885318279266357,
140
+ "learning_rate": 1.9269193765634022e-05,
141
+ "loss": 0.6334,
142
+ "step": 9500
143
+ },
144
+ {
145
+ "epoch": 0.38,
146
+ "grad_norm": 7.104074001312256,
147
+ "learning_rate": 1.9230710025014434e-05,
148
+ "loss": 0.6496,
149
+ "step": 10000
150
+ },
151
+ {
152
+ "epoch": 0.4,
153
+ "grad_norm": 5.086849689483643,
154
+ "learning_rate": 1.9192226284394843e-05,
155
+ "loss": 0.6647,
156
+ "step": 10500
157
+ },
158
+ {
159
+ "epoch": 0.42,
160
+ "grad_norm": 10.493572235107422,
161
+ "learning_rate": 1.9153819511256495e-05,
162
+ "loss": 0.6609,
163
+ "step": 11000
164
+ },
165
+ {
166
+ "epoch": 0.44,
167
+ "grad_norm": 10.206048011779785,
168
+ "learning_rate": 1.9115335770636907e-05,
169
+ "loss": 0.6806,
170
+ "step": 11500
171
+ },
172
+ {
173
+ "epoch": 0.46,
174
+ "grad_norm": 9.489506721496582,
175
+ "learning_rate": 1.907685203001732e-05,
176
+ "loss": 0.641,
177
+ "step": 12000
178
+ },
179
+ {
180
+ "epoch": 0.48,
181
+ "grad_norm": 4.525002479553223,
182
+ "learning_rate": 1.903844525687897e-05,
183
+ "loss": 0.6863,
184
+ "step": 12500
185
+ },
186
+ {
187
+ "epoch": 0.5,
188
+ "grad_norm": 7.507730007171631,
189
+ "learning_rate": 1.9000038483740623e-05,
190
+ "loss": 0.6829,
191
+ "step": 13000
192
+ },
193
+ {
194
+ "epoch": 0.52,
195
+ "grad_norm": 4.827548980712891,
196
+ "learning_rate": 1.8961554743121035e-05,
197
+ "loss": 0.6597,
198
+ "step": 13500
199
+ },
200
+ {
201
+ "epoch": 0.54,
202
+ "grad_norm": 6.409604072570801,
203
+ "learning_rate": 1.8923071002501447e-05,
204
+ "loss": 0.6445,
205
+ "step": 14000
206
+ },
207
+ {
208
+ "epoch": 0.56,
209
+ "grad_norm": 9.57291316986084,
210
+ "learning_rate": 1.8884664229363095e-05,
211
+ "loss": 0.6882,
212
+ "step": 14500
213
+ },
214
+ {
215
+ "epoch": 0.58,
216
+ "grad_norm": 9.129105567932129,
217
+ "learning_rate": 1.8846180488743507e-05,
218
+ "loss": 0.6719,
219
+ "step": 15000
220
+ },
221
+ {
222
+ "epoch": 0.6,
223
+ "grad_norm": 8.071757316589355,
224
+ "learning_rate": 1.880769674812392e-05,
225
+ "loss": 0.6668,
226
+ "step": 15500
227
+ },
228
+ {
229
+ "epoch": 0.62,
230
+ "grad_norm": 3.636784791946411,
231
+ "learning_rate": 1.876921300750433e-05,
232
+ "loss": 0.6773,
233
+ "step": 16000
234
+ },
235
+ {
236
+ "epoch": 0.63,
237
+ "grad_norm": 6.6433515548706055,
238
+ "learning_rate": 1.8730729266884743e-05,
239
+ "loss": 0.6776,
240
+ "step": 16500
241
+ },
242
+ {
243
+ "epoch": 0.65,
244
+ "grad_norm": 3.0176546573638916,
245
+ "learning_rate": 1.8692245526265155e-05,
246
+ "loss": 0.6587,
247
+ "step": 17000
248
+ },
249
+ {
250
+ "epoch": 0.67,
251
+ "grad_norm": 7.913615703582764,
252
+ "learning_rate": 1.8653761785645567e-05,
253
+ "loss": 0.6665,
254
+ "step": 17500
255
+ },
256
+ {
257
+ "epoch": 0.69,
258
+ "grad_norm": 5.335203647613525,
259
+ "learning_rate": 1.861527804502598e-05,
260
+ "loss": 0.6684,
261
+ "step": 18000
262
+ },
263
+ {
264
+ "epoch": 0.71,
265
+ "grad_norm": 5.51749324798584,
266
+ "learning_rate": 1.857679430440639e-05,
267
+ "loss": 0.6577,
268
+ "step": 18500
269
+ },
270
+ {
271
+ "epoch": 0.73,
272
+ "grad_norm": 4.254025936126709,
273
+ "learning_rate": 1.8538310563786802e-05,
274
+ "loss": 0.6776,
275
+ "step": 19000
276
+ },
277
+ {
278
+ "epoch": 0.75,
279
+ "grad_norm": 3.3869874477386475,
280
+ "learning_rate": 1.8499826823167214e-05,
281
+ "loss": 0.6351,
282
+ "step": 19500
283
+ },
284
+ {
285
+ "epoch": 0.77,
286
+ "grad_norm": 12.704727172851562,
287
+ "learning_rate": 1.8461343082547626e-05,
288
+ "loss": 0.6904,
289
+ "step": 20000
290
+ },
291
+ {
292
+ "epoch": 0.79,
293
+ "grad_norm": 3.478996515274048,
294
+ "learning_rate": 1.8422859341928038e-05,
295
+ "loss": 0.6829,
296
+ "step": 20500
297
+ },
298
+ {
299
+ "epoch": 0.81,
300
+ "grad_norm": 6.513260841369629,
301
+ "learning_rate": 1.838437560130845e-05,
302
+ "loss": 0.6615,
303
+ "step": 21000
304
+ },
305
+ {
306
+ "epoch": 0.83,
307
+ "grad_norm": 5.242546081542969,
308
+ "learning_rate": 1.8345891860688862e-05,
309
+ "loss": 0.6859,
310
+ "step": 21500
311
+ },
312
+ {
313
+ "epoch": 0.85,
314
+ "grad_norm": 6.638239860534668,
315
+ "learning_rate": 1.8307408120069274e-05,
316
+ "loss": 0.6485,
317
+ "step": 22000
318
+ },
319
+ {
320
+ "epoch": 0.87,
321
+ "grad_norm": 8.32187557220459,
322
+ "learning_rate": 1.8269001346930923e-05,
323
+ "loss": 0.6567,
324
+ "step": 22500
325
+ },
326
+ {
327
+ "epoch": 0.89,
328
+ "grad_norm": 3.9929823875427246,
329
+ "learning_rate": 1.8230517606311334e-05,
330
+ "loss": 0.6996,
331
+ "step": 23000
332
+ },
333
+ {
334
+ "epoch": 0.9,
335
+ "grad_norm": 5.426717281341553,
336
+ "learning_rate": 1.8192110833172987e-05,
337
+ "loss": 0.6529,
338
+ "step": 23500
339
+ },
340
+ {
341
+ "epoch": 0.92,
342
+ "grad_norm": 7.328232288360596,
343
+ "learning_rate": 1.81536270925534e-05,
344
+ "loss": 0.6711,
345
+ "step": 24000
346
+ },
347
+ {
348
+ "epoch": 0.94,
349
+ "grad_norm": 6.648090362548828,
350
+ "learning_rate": 1.811514335193381e-05,
351
+ "loss": 0.7026,
352
+ "step": 24500
353
+ },
354
+ {
355
+ "epoch": 0.96,
356
+ "grad_norm": 4.047011375427246,
357
+ "learning_rate": 1.8076659611314222e-05,
358
+ "loss": 0.6867,
359
+ "step": 25000
360
+ },
361
+ {
362
+ "epoch": 0.98,
363
+ "grad_norm": 10.354930877685547,
364
+ "learning_rate": 1.803825283817587e-05,
365
+ "loss": 0.6718,
366
+ "step": 25500
367
+ },
368
+ {
369
+ "epoch": 1.0,
370
+ "eval_loss": 1.5939216613769531,
371
+ "eval_runtime": 11.3331,
372
+ "eval_samples_per_second": 351.271,
373
+ "eval_steps_per_second": 87.884,
374
+ "step": 25985
375
+ },
376
+ {
377
+ "epoch": 1.0,
378
+ "grad_norm": 4.946779251098633,
379
+ "learning_rate": 1.7999846065037523e-05,
380
+ "loss": 0.6928,
381
+ "step": 26000
382
+ },
383
+ {
384
+ "epoch": 1.02,
385
+ "grad_norm": 4.127852916717529,
386
+ "learning_rate": 1.7961362324417935e-05,
387
+ "loss": 0.5471,
388
+ "step": 26500
389
+ },
390
+ {
391
+ "epoch": 1.04,
392
+ "grad_norm": 8.648958206176758,
393
+ "learning_rate": 1.7922878583798347e-05,
394
+ "loss": 0.5216,
395
+ "step": 27000
396
+ },
397
+ {
398
+ "epoch": 1.06,
399
+ "grad_norm": 3.198063611984253,
400
+ "learning_rate": 1.788439484317876e-05,
401
+ "loss": 0.5649,
402
+ "step": 27500
403
+ },
404
+ {
405
+ "epoch": 1.08,
406
+ "grad_norm": 7.709962368011475,
407
+ "learning_rate": 1.784591110255917e-05,
408
+ "loss": 0.5739,
409
+ "step": 28000
410
+ },
411
+ {
412
+ "epoch": 1.1,
413
+ "grad_norm": 3.456827402114868,
414
+ "learning_rate": 1.7807427361939583e-05,
415
+ "loss": 0.549,
416
+ "step": 28500
417
+ },
418
+ {
419
+ "epoch": 1.12,
420
+ "grad_norm": 5.343823432922363,
421
+ "learning_rate": 1.7768943621319995e-05,
422
+ "loss": 0.5624,
423
+ "step": 29000
424
+ },
425
+ {
426
+ "epoch": 1.14,
427
+ "grad_norm": 6.258249759674072,
428
+ "learning_rate": 1.7730459880700407e-05,
429
+ "loss": 0.5497,
430
+ "step": 29500
431
+ },
432
+ {
433
+ "epoch": 1.15,
434
+ "grad_norm": 3.7241768836975098,
435
+ "learning_rate": 1.769197614008082e-05,
436
+ "loss": 0.5422,
437
+ "step": 30000
438
+ },
439
+ {
440
+ "epoch": 1.17,
441
+ "grad_norm": 4.565840244293213,
442
+ "learning_rate": 1.765349239946123e-05,
443
+ "loss": 0.5805,
444
+ "step": 30500
445
+ },
446
+ {
447
+ "epoch": 1.19,
448
+ "grad_norm": 6.31953763961792,
449
+ "learning_rate": 1.7615008658841642e-05,
450
+ "loss": 0.5889,
451
+ "step": 31000
452
+ },
453
+ {
454
+ "epoch": 1.21,
455
+ "grad_norm": 1.6492664813995361,
456
+ "learning_rate": 1.7576524918222054e-05,
457
+ "loss": 0.5732,
458
+ "step": 31500
459
+ },
460
+ {
461
+ "epoch": 1.23,
462
+ "grad_norm": 6.656198978424072,
463
+ "learning_rate": 1.7538041177602466e-05,
464
+ "loss": 0.5499,
465
+ "step": 32000
466
+ },
467
+ {
468
+ "epoch": 1.25,
469
+ "grad_norm": 4.865792274475098,
470
+ "learning_rate": 1.7499557436982878e-05,
471
+ "loss": 0.5519,
472
+ "step": 32500
473
+ },
474
+ {
475
+ "epoch": 1.27,
476
+ "grad_norm": 3.8847908973693848,
477
+ "learning_rate": 1.746107369636329e-05,
478
+ "loss": 0.5695,
479
+ "step": 33000
480
+ },
481
+ {
482
+ "epoch": 1.29,
483
+ "grad_norm": 9.86506462097168,
484
+ "learning_rate": 1.74225899557437e-05,
485
+ "loss": 0.5877,
486
+ "step": 33500
487
+ },
488
+ {
489
+ "epoch": 1.31,
490
+ "grad_norm": 5.732994079589844,
491
+ "learning_rate": 1.738418318260535e-05,
492
+ "loss": 0.5629,
493
+ "step": 34000
494
+ },
495
+ {
496
+ "epoch": 1.33,
497
+ "grad_norm": 2.4092957973480225,
498
+ "learning_rate": 1.7345699441985762e-05,
499
+ "loss": 0.5734,
500
+ "step": 34500
501
+ },
502
+ {
503
+ "epoch": 1.35,
504
+ "grad_norm": 4.129636764526367,
505
+ "learning_rate": 1.7307215701366174e-05,
506
+ "loss": 0.5931,
507
+ "step": 35000
508
+ },
509
+ {
510
+ "epoch": 1.37,
511
+ "grad_norm": 4.32247257232666,
512
+ "learning_rate": 1.7268731960746586e-05,
513
+ "loss": 0.5703,
514
+ "step": 35500
515
+ },
516
+ {
517
+ "epoch": 1.39,
518
+ "grad_norm": 4.780704975128174,
519
+ "learning_rate": 1.7230325187608238e-05,
520
+ "loss": 0.5978,
521
+ "step": 36000
522
+ },
523
+ {
524
+ "epoch": 1.4,
525
+ "grad_norm": 5.10893440246582,
526
+ "learning_rate": 1.719184144698865e-05,
527
+ "loss": 0.57,
528
+ "step": 36500
529
+ },
530
+ {
531
+ "epoch": 1.42,
532
+ "grad_norm": 3.899763584136963,
533
+ "learning_rate": 1.71534346738503e-05,
534
+ "loss": 0.5778,
535
+ "step": 37000
536
+ },
537
+ {
538
+ "epoch": 1.44,
539
+ "grad_norm": 6.0712480545043945,
540
+ "learning_rate": 1.711495093323071e-05,
541
+ "loss": 0.5744,
542
+ "step": 37500
543
+ },
544
+ {
545
+ "epoch": 1.46,
546
+ "grad_norm": 9.512677192687988,
547
+ "learning_rate": 1.7076467192611123e-05,
548
+ "loss": 0.5825,
549
+ "step": 38000
550
+ },
551
+ {
552
+ "epoch": 1.48,
553
+ "grad_norm": 5.049200534820557,
554
+ "learning_rate": 1.7037983451991535e-05,
555
+ "loss": 0.6039,
556
+ "step": 38500
557
+ },
558
+ {
559
+ "epoch": 1.5,
560
+ "grad_norm": 5.021922588348389,
561
+ "learning_rate": 1.6999499711371947e-05,
562
+ "loss": 0.5708,
563
+ "step": 39000
564
+ },
565
+ {
566
+ "epoch": 1.52,
567
+ "grad_norm": 3.546837568283081,
568
+ "learning_rate": 1.696101597075236e-05,
569
+ "loss": 0.5808,
570
+ "step": 39500
571
+ },
572
+ {
573
+ "epoch": 1.54,
574
+ "grad_norm": 4.707235813140869,
575
+ "learning_rate": 1.692253223013277e-05,
576
+ "loss": 0.5894,
577
+ "step": 40000
578
+ },
579
+ {
580
+ "epoch": 1.56,
581
+ "grad_norm": 8.543360710144043,
582
+ "learning_rate": 1.688412545699442e-05,
583
+ "loss": 0.5828,
584
+ "step": 40500
585
+ },
586
+ {
587
+ "epoch": 1.58,
588
+ "grad_norm": 5.515990257263184,
589
+ "learning_rate": 1.684564171637483e-05,
590
+ "loss": 0.6009,
591
+ "step": 41000
592
+ },
593
+ {
594
+ "epoch": 1.6,
595
+ "grad_norm": 8.573718070983887,
596
+ "learning_rate": 1.6807234943236483e-05,
597
+ "loss": 0.5796,
598
+ "step": 41500
599
+ },
600
+ {
601
+ "epoch": 1.62,
602
+ "grad_norm": 3.3873181343078613,
603
+ "learning_rate": 1.6768751202616895e-05,
604
+ "loss": 0.6052,
605
+ "step": 42000
606
+ },
607
+ {
608
+ "epoch": 1.64,
609
+ "grad_norm": 5.747949123382568,
610
+ "learning_rate": 1.6730344429478547e-05,
611
+ "loss": 0.5998,
612
+ "step": 42500
613
+ },
614
+ {
615
+ "epoch": 1.65,
616
+ "grad_norm": 4.767616271972656,
617
+ "learning_rate": 1.669186068885896e-05,
618
+ "loss": 0.631,
619
+ "step": 43000
620
+ },
621
+ {
622
+ "epoch": 1.67,
623
+ "grad_norm": 5.872443199157715,
624
+ "learning_rate": 1.665337694823937e-05,
625
+ "loss": 0.5903,
626
+ "step": 43500
627
+ },
628
+ {
629
+ "epoch": 1.69,
630
+ "grad_norm": 5.379058837890625,
631
+ "learning_rate": 1.6614893207619783e-05,
632
+ "loss": 0.6141,
633
+ "step": 44000
634
+ },
635
+ {
636
+ "epoch": 1.71,
637
+ "grad_norm": 7.023738384246826,
638
+ "learning_rate": 1.6576409467000195e-05,
639
+ "loss": 0.601,
640
+ "step": 44500
641
+ },
642
+ {
643
+ "epoch": 1.73,
644
+ "grad_norm": 7.2762932777404785,
645
+ "learning_rate": 1.6537925726380607e-05,
646
+ "loss": 0.614,
647
+ "step": 45000
648
+ },
649
+ {
650
+ "epoch": 1.75,
651
+ "grad_norm": 5.506533145904541,
652
+ "learning_rate": 1.649944198576102e-05,
653
+ "loss": 0.6156,
654
+ "step": 45500
655
+ },
656
+ {
657
+ "epoch": 1.77,
658
+ "grad_norm": 4.782810688018799,
659
+ "learning_rate": 1.646095824514143e-05,
660
+ "loss": 0.5983,
661
+ "step": 46000
662
+ },
663
+ {
664
+ "epoch": 1.79,
665
+ "grad_norm": 8.239398002624512,
666
+ "learning_rate": 1.6422474504521842e-05,
667
+ "loss": 0.6115,
668
+ "step": 46500
669
+ },
670
+ {
671
+ "epoch": 1.81,
672
+ "grad_norm": 5.413235664367676,
673
+ "learning_rate": 1.6383990763902254e-05,
674
+ "loss": 0.6157,
675
+ "step": 47000
676
+ },
677
+ {
678
+ "epoch": 1.83,
679
+ "grad_norm": 5.1087446212768555,
680
+ "learning_rate": 1.6345507023282666e-05,
681
+ "loss": 0.6225,
682
+ "step": 47500
683
+ },
684
+ {
685
+ "epoch": 1.85,
686
+ "grad_norm": 10.661508560180664,
687
+ "learning_rate": 1.6307023282663078e-05,
688
+ "loss": 0.6185,
689
+ "step": 48000
690
+ },
691
+ {
692
+ "epoch": 1.87,
693
+ "grad_norm": 5.0764312744140625,
694
+ "learning_rate": 1.626853954204349e-05,
695
+ "loss": 0.5878,
696
+ "step": 48500
697
+ },
698
+ {
699
+ "epoch": 1.89,
700
+ "grad_norm": 6.266598224639893,
701
+ "learning_rate": 1.6230055801423902e-05,
702
+ "loss": 0.6057,
703
+ "step": 49000
704
+ },
705
+ {
706
+ "epoch": 1.9,
707
+ "grad_norm": 5.597410678863525,
708
+ "learning_rate": 1.6191572060804314e-05,
709
+ "loss": 0.591,
710
+ "step": 49500
711
+ },
712
+ {
713
+ "epoch": 1.92,
714
+ "grad_norm": 5.9798970222473145,
715
+ "learning_rate": 1.6153088320184722e-05,
716
+ "loss": 0.6114,
717
+ "step": 50000
718
+ },
719
+ {
720
+ "epoch": 1.94,
721
+ "grad_norm": 5.151573181152344,
722
+ "learning_rate": 1.6114604579565134e-05,
723
+ "loss": 0.5875,
724
+ "step": 50500
725
+ },
726
+ {
727
+ "epoch": 1.96,
728
+ "grad_norm": 3.2873849868774414,
729
+ "learning_rate": 1.6076197806426786e-05,
730
+ "loss": 0.6393,
731
+ "step": 51000
732
+ },
733
+ {
734
+ "epoch": 1.98,
735
+ "grad_norm": 3.8436408042907715,
736
+ "learning_rate": 1.6037714065807198e-05,
737
+ "loss": 0.6202,
738
+ "step": 51500
739
+ },
740
+ {
741
+ "epoch": 2.0,
742
+ "eval_loss": 1.6332776546478271,
743
+ "eval_runtime": 11.3148,
744
+ "eval_samples_per_second": 351.839,
745
+ "eval_steps_per_second": 88.026,
746
+ "step": 51970
747
+ },
748
+ {
749
+ "epoch": 2.0,
750
+ "grad_norm": 11.219087600708008,
751
+ "learning_rate": 1.599923032518761e-05,
752
+ "loss": 0.6083,
753
+ "step": 52000
754
+ },
755
+ {
756
+ "epoch": 2.02,
757
+ "grad_norm": 5.6387038230896,
758
+ "learning_rate": 1.5960746584568022e-05,
759
+ "loss": 0.5929,
760
+ "step": 52500
761
+ },
762
+ {
763
+ "epoch": 2.04,
764
+ "grad_norm": 10.761116981506348,
765
+ "learning_rate": 1.592241677891091e-05,
766
+ "loss": 0.5913,
767
+ "step": 53000
768
+ },
769
+ {
770
+ "epoch": 2.06,
771
+ "grad_norm": 4.182568073272705,
772
+ "learning_rate": 1.5883933038291323e-05,
773
+ "loss": 0.5983,
774
+ "step": 53500
775
+ },
776
+ {
777
+ "epoch": 2.08,
778
+ "grad_norm": 1.4628002643585205,
779
+ "learning_rate": 1.5845449297671735e-05,
780
+ "loss": 0.5789,
781
+ "step": 54000
782
+ },
783
+ {
784
+ "epoch": 2.1,
785
+ "grad_norm": 6.6901535987854,
786
+ "learning_rate": 1.5806965557052147e-05,
787
+ "loss": 0.5827,
788
+ "step": 54500
789
+ },
790
+ {
791
+ "epoch": 2.12,
792
+ "grad_norm": 4.275672912597656,
793
+ "learning_rate": 1.576848181643256e-05,
794
+ "loss": 0.593,
795
+ "step": 55000
796
+ },
797
+ {
798
+ "epoch": 2.14,
799
+ "grad_norm": 6.460977554321289,
800
+ "learning_rate": 1.572999807581297e-05,
801
+ "loss": 0.6012,
802
+ "step": 55500
803
+ },
804
+ {
805
+ "epoch": 2.16,
806
+ "grad_norm": 6.66805362701416,
807
+ "learning_rate": 1.5691514335193382e-05,
808
+ "loss": 0.6148,
809
+ "step": 56000
810
+ },
811
+ {
812
+ "epoch": 2.17,
813
+ "grad_norm": 9.386066436767578,
814
+ "learning_rate": 1.5653030594573794e-05,
815
+ "loss": 0.6147,
816
+ "step": 56500
817
+ },
818
+ {
819
+ "epoch": 2.19,
820
+ "grad_norm": 4.505888938903809,
821
+ "learning_rate": 1.5614700788916683e-05,
822
+ "loss": 0.6,
823
+ "step": 57000
824
+ },
825
+ {
826
+ "epoch": 2.21,
827
+ "grad_norm": 6.781282424926758,
828
+ "learning_rate": 1.5576217048297095e-05,
829
+ "loss": 0.5938,
830
+ "step": 57500
831
+ },
832
+ {
833
+ "epoch": 2.23,
834
+ "grad_norm": 4.20095157623291,
835
+ "learning_rate": 1.5537810275158747e-05,
836
+ "loss": 0.6118,
837
+ "step": 58000
838
+ },
839
+ {
840
+ "epoch": 2.25,
841
+ "grad_norm": 6.635080337524414,
842
+ "learning_rate": 1.549932653453916e-05,
843
+ "loss": 0.6375,
844
+ "step": 58500
845
+ },
846
+ {
847
+ "epoch": 2.27,
848
+ "grad_norm": 5.1331329345703125,
849
+ "learning_rate": 1.546084279391957e-05,
850
+ "loss": 0.625,
851
+ "step": 59000
852
+ },
853
+ {
854
+ "epoch": 2.29,
855
+ "grad_norm": 4.725217342376709,
856
+ "learning_rate": 1.5422359053299983e-05,
857
+ "loss": 0.6443,
858
+ "step": 59500
859
+ },
860
+ {
861
+ "epoch": 2.31,
862
+ "grad_norm": 6.601204872131348,
863
+ "learning_rate": 1.5383875312680395e-05,
864
+ "loss": 0.6206,
865
+ "step": 60000
866
+ },
867
+ {
868
+ "epoch": 2.33,
869
+ "grad_norm": 16.34359359741211,
870
+ "learning_rate": 1.5345391572060807e-05,
871
+ "loss": 0.6144,
872
+ "step": 60500
873
+ },
874
+ {
875
+ "epoch": 2.35,
876
+ "grad_norm": 6.883621692657471,
877
+ "learning_rate": 1.530690783144122e-05,
878
+ "loss": 0.6112,
879
+ "step": 61000
880
+ },
881
+ {
882
+ "epoch": 2.37,
883
+ "grad_norm": 5.42087984085083,
884
+ "learning_rate": 1.526842409082163e-05,
885
+ "loss": 0.6286,
886
+ "step": 61500
887
+ },
888
+ {
889
+ "epoch": 2.39,
890
+ "grad_norm": 7.191906452178955,
891
+ "learning_rate": 1.522994035020204e-05,
892
+ "loss": 0.6103,
893
+ "step": 62000
894
+ },
895
+ {
896
+ "epoch": 2.41,
897
+ "grad_norm": 8.114423751831055,
898
+ "learning_rate": 1.5191456609582453e-05,
899
+ "loss": 0.6464,
900
+ "step": 62500
901
+ },
902
+ {
903
+ "epoch": 2.42,
904
+ "grad_norm": 6.188645839691162,
905
+ "learning_rate": 1.5152972868962865e-05,
906
+ "loss": 0.5874,
907
+ "step": 63000
908
+ },
909
+ {
910
+ "epoch": 2.44,
911
+ "grad_norm": 5.572692394256592,
912
+ "learning_rate": 1.5114489128343276e-05,
913
+ "loss": 0.6235,
914
+ "step": 63500
915
+ },
916
+ {
917
+ "epoch": 2.46,
918
+ "grad_norm": 8.618267059326172,
919
+ "learning_rate": 1.5076082355204927e-05,
920
+ "loss": 0.6144,
921
+ "step": 64000
922
+ },
923
+ {
924
+ "epoch": 2.48,
925
+ "grad_norm": 6.580616474151611,
926
+ "learning_rate": 1.5037598614585339e-05,
927
+ "loss": 0.6544,
928
+ "step": 64500
929
+ },
930
+ {
931
+ "epoch": 2.5,
932
+ "grad_norm": 7.99381160736084,
933
+ "learning_rate": 1.499919184144699e-05,
934
+ "loss": 0.6281,
935
+ "step": 65000
936
+ },
937
+ {
938
+ "epoch": 2.52,
939
+ "grad_norm": 4.514256954193115,
940
+ "learning_rate": 1.4960708100827401e-05,
941
+ "loss": 0.6295,
942
+ "step": 65500
943
+ },
944
+ {
945
+ "epoch": 2.54,
946
+ "grad_norm": 8.09423828125,
947
+ "learning_rate": 1.4922224360207813e-05,
948
+ "loss": 0.6332,
949
+ "step": 66000
950
+ },
951
+ {
952
+ "epoch": 2.56,
953
+ "grad_norm": 5.654535293579102,
954
+ "learning_rate": 1.4883740619588225e-05,
955
+ "loss": 0.6573,
956
+ "step": 66500
957
+ },
958
+ {
959
+ "epoch": 2.58,
960
+ "grad_norm": 8.867722511291504,
961
+ "learning_rate": 1.4845256878968637e-05,
962
+ "loss": 0.6331,
963
+ "step": 67000
964
+ },
965
+ {
966
+ "epoch": 2.6,
967
+ "grad_norm": 11.65078353881836,
968
+ "learning_rate": 1.4806773138349049e-05,
969
+ "loss": 0.6522,
970
+ "step": 67500
971
+ },
972
+ {
973
+ "epoch": 2.62,
974
+ "grad_norm": 4.735153675079346,
975
+ "learning_rate": 1.476828939772946e-05,
976
+ "loss": 0.628,
977
+ "step": 68000
978
+ },
979
+ {
980
+ "epoch": 2.64,
981
+ "grad_norm": 10.033109664916992,
982
+ "learning_rate": 1.4729805657109873e-05,
983
+ "loss": 0.6588,
984
+ "step": 68500
985
+ },
986
+ {
987
+ "epoch": 2.66,
988
+ "grad_norm": 5.765382289886475,
989
+ "learning_rate": 1.4691398883971523e-05,
990
+ "loss": 0.657,
991
+ "step": 69000
992
+ },
993
+ {
994
+ "epoch": 2.67,
995
+ "grad_norm": 8.563407897949219,
996
+ "learning_rate": 1.4652915143351935e-05,
997
+ "loss": 0.6476,
998
+ "step": 69500
999
+ },
1000
+ {
1001
+ "epoch": 2.69,
1002
+ "grad_norm": 5.050264358520508,
1003
+ "learning_rate": 1.4614431402732347e-05,
1004
+ "loss": 0.643,
1005
+ "step": 70000
1006
+ },
1007
+ {
1008
+ "epoch": 2.71,
1009
+ "grad_norm": 4.829985618591309,
1010
+ "learning_rate": 1.4575947662112759e-05,
1011
+ "loss": 0.6637,
1012
+ "step": 70500
1013
+ },
1014
+ {
1015
+ "epoch": 2.73,
1016
+ "grad_norm": 10.406055450439453,
1017
+ "learning_rate": 1.453754088897441e-05,
1018
+ "loss": 0.6588,
1019
+ "step": 71000
1020
+ },
1021
+ {
1022
+ "epoch": 2.75,
1023
+ "grad_norm": 5.833386421203613,
1024
+ "learning_rate": 1.4499057148354821e-05,
1025
+ "loss": 0.6443,
1026
+ "step": 71500
1027
+ },
1028
+ {
1029
+ "epoch": 2.77,
1030
+ "grad_norm": 5.603066921234131,
1031
+ "learning_rate": 1.4460573407735233e-05,
1032
+ "loss": 0.6382,
1033
+ "step": 72000
1034
+ },
1035
+ {
1036
+ "epoch": 2.79,
1037
+ "grad_norm": 6.070736885070801,
1038
+ "learning_rate": 1.4422166634596883e-05,
1039
+ "loss": 0.654,
1040
+ "step": 72500
1041
+ },
1042
+ {
1043
+ "epoch": 2.81,
1044
+ "grad_norm": 4.568108081817627,
1045
+ "learning_rate": 1.4383682893977295e-05,
1046
+ "loss": 0.6496,
1047
+ "step": 73000
1048
+ },
1049
+ {
1050
+ "epoch": 2.83,
1051
+ "grad_norm": 5.071454048156738,
1052
+ "learning_rate": 1.4345199153357707e-05,
1053
+ "loss": 0.6412,
1054
+ "step": 73500
1055
+ },
1056
+ {
1057
+ "epoch": 2.85,
1058
+ "grad_norm": 9.910784721374512,
1059
+ "learning_rate": 1.4306715412738119e-05,
1060
+ "loss": 0.6329,
1061
+ "step": 74000
1062
+ },
1063
+ {
1064
+ "epoch": 2.87,
1065
+ "grad_norm": 4.771623134613037,
1066
+ "learning_rate": 1.4268231672118531e-05,
1067
+ "loss": 0.6661,
1068
+ "step": 74500
1069
+ },
1070
+ {
1071
+ "epoch": 2.89,
1072
+ "grad_norm": 12.325662612915039,
1073
+ "learning_rate": 1.4229747931498943e-05,
1074
+ "loss": 0.6389,
1075
+ "step": 75000
1076
+ },
1077
+ {
1078
+ "epoch": 2.91,
1079
+ "grad_norm": 4.688502311706543,
1080
+ "learning_rate": 1.4191341158360595e-05,
1081
+ "loss": 0.6475,
1082
+ "step": 75500
1083
+ },
1084
+ {
1085
+ "epoch": 2.92,
1086
+ "grad_norm": 7.340636730194092,
1087
+ "learning_rate": 1.4152857417741005e-05,
1088
+ "loss": 0.678,
1089
+ "step": 76000
1090
+ },
1091
+ {
1092
+ "epoch": 2.94,
1093
+ "grad_norm": 11.585808753967285,
1094
+ "learning_rate": 1.4114373677121417e-05,
1095
+ "loss": 0.6422,
1096
+ "step": 76500
1097
+ },
1098
+ {
1099
+ "epoch": 2.96,
1100
+ "grad_norm": 4.971041679382324,
1101
+ "learning_rate": 1.4075889936501829e-05,
1102
+ "loss": 0.6493,
1103
+ "step": 77000
1104
+ },
1105
+ {
1106
+ "epoch": 2.98,
1107
+ "grad_norm": 8.936264991760254,
1108
+ "learning_rate": 1.4037406195882241e-05,
1109
+ "loss": 0.6719,
1110
+ "step": 77500
1111
+ },
1112
+ {
1113
+ "epoch": 3.0,
1114
+ "eval_loss": 1.634682297706604,
1115
+ "eval_runtime": 11.4139,
1116
+ "eval_samples_per_second": 348.784,
1117
+ "eval_steps_per_second": 87.262,
1118
+ "step": 77955
1119
+ },
1120
+ {
1121
+ "epoch": 3.0,
1122
+ "grad_norm": 4.788780212402344,
1123
+ "learning_rate": 1.3998922455262653e-05,
1124
+ "loss": 0.6236,
1125
+ "step": 78000
1126
+ },
1127
+ {
1128
+ "epoch": 3.02,
1129
+ "grad_norm": 8.352863311767578,
1130
+ "learning_rate": 1.3960438714643065e-05,
1131
+ "loss": 0.5519,
1132
+ "step": 78500
1133
+ },
1134
+ {
1135
+ "epoch": 3.04,
1136
+ "grad_norm": 9.03128433227539,
1137
+ "learning_rate": 1.3922031941504715e-05,
1138
+ "loss": 0.5617,
1139
+ "step": 79000
1140
+ },
1141
+ {
1142
+ "epoch": 3.06,
1143
+ "grad_norm": 8.761636734008789,
1144
+ "learning_rate": 1.3883548200885127e-05,
1145
+ "loss": 0.554,
1146
+ "step": 79500
1147
+ },
1148
+ {
1149
+ "epoch": 3.08,
1150
+ "grad_norm": 4.592207908630371,
1151
+ "learning_rate": 1.3845064460265539e-05,
1152
+ "loss": 0.5218,
1153
+ "step": 80000
1154
+ },
1155
+ {
1156
+ "epoch": 3.1,
1157
+ "grad_norm": 5.602917671203613,
1158
+ "learning_rate": 1.380658071964595e-05,
1159
+ "loss": 0.5432,
1160
+ "step": 80500
1161
+ },
1162
+ {
1163
+ "epoch": 3.12,
1164
+ "grad_norm": 10.146926879882812,
1165
+ "learning_rate": 1.3768096979026363e-05,
1166
+ "loss": 0.5332,
1167
+ "step": 81000
1168
+ },
1169
+ {
1170
+ "epoch": 3.14,
1171
+ "grad_norm": 7.7499494552612305,
1172
+ "learning_rate": 1.3729613238406773e-05,
1173
+ "loss": 0.5459,
1174
+ "step": 81500
1175
+ },
1176
+ {
1177
+ "epoch": 3.16,
1178
+ "grad_norm": 10.009146690368652,
1179
+ "learning_rate": 1.3691129497787185e-05,
1180
+ "loss": 0.5721,
1181
+ "step": 82000
1182
+ },
1183
+ {
1184
+ "epoch": 3.17,
1185
+ "grad_norm": 5.544896602630615,
1186
+ "learning_rate": 1.3652645757167597e-05,
1187
+ "loss": 0.5407,
1188
+ "step": 82500
1189
+ },
1190
+ {
1191
+ "epoch": 3.19,
1192
+ "grad_norm": 11.661575317382812,
1193
+ "learning_rate": 1.3614238984029249e-05,
1194
+ "loss": 0.5678,
1195
+ "step": 83000
1196
+ },
1197
+ {
1198
+ "epoch": 3.21,
1199
+ "grad_norm": 2.23122501373291,
1200
+ "learning_rate": 1.357575524340966e-05,
1201
+ "loss": 0.5577,
1202
+ "step": 83500
1203
+ },
1204
+ {
1205
+ "epoch": 3.23,
1206
+ "grad_norm": 2.9503915309906006,
1207
+ "learning_rate": 1.3537271502790073e-05,
1208
+ "loss": 0.568,
1209
+ "step": 84000
1210
+ },
1211
+ {
1212
+ "epoch": 3.25,
1213
+ "grad_norm": 5.630256175994873,
1214
+ "learning_rate": 1.3498787762170485e-05,
1215
+ "loss": 0.5524,
1216
+ "step": 84500
1217
+ },
1218
+ {
1219
+ "epoch": 3.27,
1220
+ "grad_norm": 5.076592922210693,
1221
+ "learning_rate": 1.3460304021550896e-05,
1222
+ "loss": 0.574,
1223
+ "step": 85000
1224
+ },
1225
+ {
1226
+ "epoch": 3.29,
1227
+ "grad_norm": 7.131458759307861,
1228
+ "learning_rate": 1.3421974215893786e-05,
1229
+ "loss": 0.5718,
1230
+ "step": 85500
1231
+ },
1232
+ {
1233
+ "epoch": 3.31,
1234
+ "grad_norm": 16.959537506103516,
1235
+ "learning_rate": 1.3383490475274197e-05,
1236
+ "loss": 0.5743,
1237
+ "step": 86000
1238
+ },
1239
+ {
1240
+ "epoch": 3.33,
1241
+ "grad_norm": 8.548674583435059,
1242
+ "learning_rate": 1.334500673465461e-05,
1243
+ "loss": 0.5747,
1244
+ "step": 86500
1245
+ },
1246
+ {
1247
+ "epoch": 3.35,
1248
+ "grad_norm": 16.18918800354004,
1249
+ "learning_rate": 1.3306522994035021e-05,
1250
+ "loss": 0.594,
1251
+ "step": 87000
1252
+ },
1253
+ {
1254
+ "epoch": 3.37,
1255
+ "grad_norm": 3.274156093597412,
1256
+ "learning_rate": 1.3268039253415433e-05,
1257
+ "loss": 0.5758,
1258
+ "step": 87500
1259
+ },
1260
+ {
1261
+ "epoch": 3.39,
1262
+ "grad_norm": 5.2989959716796875,
1263
+ "learning_rate": 1.3229555512795845e-05,
1264
+ "loss": 0.5658,
1265
+ "step": 88000
1266
+ },
1267
+ {
1268
+ "epoch": 3.41,
1269
+ "grad_norm": 3.634488821029663,
1270
+ "learning_rate": 1.3191148739657495e-05,
1271
+ "loss": 0.5572,
1272
+ "step": 88500
1273
+ },
1274
+ {
1275
+ "epoch": 3.43,
1276
+ "grad_norm": 4.335933208465576,
1277
+ "learning_rate": 1.3152664999037907e-05,
1278
+ "loss": 0.5654,
1279
+ "step": 89000
1280
+ },
1281
+ {
1282
+ "epoch": 3.44,
1283
+ "grad_norm": 7.001922607421875,
1284
+ "learning_rate": 1.311418125841832e-05,
1285
+ "loss": 0.5741,
1286
+ "step": 89500
1287
+ },
1288
+ {
1289
+ "epoch": 3.46,
1290
+ "grad_norm": 9.779496192932129,
1291
+ "learning_rate": 1.3075697517798731e-05,
1292
+ "loss": 0.5782,
1293
+ "step": 90000
1294
+ },
1295
+ {
1296
+ "epoch": 3.48,
1297
+ "grad_norm": 7.241322994232178,
1298
+ "learning_rate": 1.3037213777179143e-05,
1299
+ "loss": 0.585,
1300
+ "step": 90500
1301
+ },
1302
+ {
1303
+ "epoch": 3.5,
1304
+ "grad_norm": 10.485682487487793,
1305
+ "learning_rate": 1.2998730036559555e-05,
1306
+ "loss": 0.5743,
1307
+ "step": 91000
1308
+ },
1309
+ {
1310
+ "epoch": 3.52,
1311
+ "grad_norm": 5.369238376617432,
1312
+ "learning_rate": 1.2960246295939967e-05,
1313
+ "loss": 0.5743,
1314
+ "step": 91500
1315
+ },
1316
+ {
1317
+ "epoch": 3.54,
1318
+ "grad_norm": 2.732351064682007,
1319
+ "learning_rate": 1.2921762555320379e-05,
1320
+ "loss": 0.5753,
1321
+ "step": 92000
1322
+ },
1323
+ {
1324
+ "epoch": 3.56,
1325
+ "grad_norm": 5.905006408691406,
1326
+ "learning_rate": 1.288335578218203e-05,
1327
+ "loss": 0.5816,
1328
+ "step": 92500
1329
+ },
1330
+ {
1331
+ "epoch": 3.58,
1332
+ "grad_norm": 6.421899795532227,
1333
+ "learning_rate": 1.2844872041562441e-05,
1334
+ "loss": 0.5907,
1335
+ "step": 93000
1336
+ },
1337
+ {
1338
+ "epoch": 3.6,
1339
+ "grad_norm": 8.367085456848145,
1340
+ "learning_rate": 1.2806388300942853e-05,
1341
+ "loss": 0.6026,
1342
+ "step": 93500
1343
+ },
1344
+ {
1345
+ "epoch": 3.62,
1346
+ "grad_norm": 5.6347784996032715,
1347
+ "learning_rate": 1.2767904560323265e-05,
1348
+ "loss": 0.6028,
1349
+ "step": 94000
1350
+ },
1351
+ {
1352
+ "epoch": 3.64,
1353
+ "grad_norm": 4.885034084320068,
1354
+ "learning_rate": 1.2729497787184915e-05,
1355
+ "loss": 0.6062,
1356
+ "step": 94500
1357
+ },
1358
+ {
1359
+ "epoch": 3.66,
1360
+ "grad_norm": 9.47564697265625,
1361
+ "learning_rate": 1.2691014046565327e-05,
1362
+ "loss": 0.6116,
1363
+ "step": 95000
1364
+ },
1365
+ {
1366
+ "epoch": 3.68,
1367
+ "grad_norm": 4.893156051635742,
1368
+ "learning_rate": 1.2652530305945739e-05,
1369
+ "loss": 0.5769,
1370
+ "step": 95500
1371
+ },
1372
+ {
1373
+ "epoch": 3.69,
1374
+ "grad_norm": 7.7681074142456055,
1375
+ "learning_rate": 1.261404656532615e-05,
1376
+ "loss": 0.5922,
1377
+ "step": 96000
1378
+ },
1379
+ {
1380
+ "epoch": 3.71,
1381
+ "grad_norm": 9.778617858886719,
1382
+ "learning_rate": 1.2575639792187801e-05,
1383
+ "loss": 0.623,
1384
+ "step": 96500
1385
+ },
1386
+ {
1387
+ "epoch": 3.73,
1388
+ "grad_norm": 10.256739616394043,
1389
+ "learning_rate": 1.2537156051568213e-05,
1390
+ "loss": 0.5552,
1391
+ "step": 97000
1392
+ },
1393
+ {
1394
+ "epoch": 3.75,
1395
+ "grad_norm": 6.800660133361816,
1396
+ "learning_rate": 1.2498672310948625e-05,
1397
+ "loss": 0.5961,
1398
+ "step": 97500
1399
+ },
1400
+ {
1401
+ "epoch": 3.77,
1402
+ "grad_norm": 11.549945831298828,
1403
+ "learning_rate": 1.2460188570329037e-05,
1404
+ "loss": 0.5983,
1405
+ "step": 98000
1406
+ },
1407
+ {
1408
+ "epoch": 3.79,
1409
+ "grad_norm": 7.519701957702637,
1410
+ "learning_rate": 1.2421704829709449e-05,
1411
+ "loss": 0.5769,
1412
+ "step": 98500
1413
+ },
1414
+ {
1415
+ "epoch": 3.81,
1416
+ "grad_norm": 7.1441168785095215,
1417
+ "learning_rate": 1.23832980565711e-05,
1418
+ "loss": 0.6006,
1419
+ "step": 99000
1420
+ },
1421
+ {
1422
+ "epoch": 3.83,
1423
+ "grad_norm": 4.6727375984191895,
1424
+ "learning_rate": 1.2344814315951511e-05,
1425
+ "loss": 0.6096,
1426
+ "step": 99500
1427
+ },
1428
+ {
1429
+ "epoch": 3.85,
1430
+ "grad_norm": 10.61463737487793,
1431
+ "learning_rate": 1.2306407542813164e-05,
1432
+ "loss": 0.6088,
1433
+ "step": 100000
1434
+ },
1435
+ {
1436
+ "epoch": 3.87,
1437
+ "grad_norm": 11.35522747039795,
1438
+ "learning_rate": 1.2267923802193574e-05,
1439
+ "loss": 0.603,
1440
+ "step": 100500
1441
+ },
1442
+ {
1443
+ "epoch": 3.89,
1444
+ "grad_norm": 4.605931282043457,
1445
+ "learning_rate": 1.2229440061573986e-05,
1446
+ "loss": 0.5742,
1447
+ "step": 101000
1448
+ },
1449
+ {
1450
+ "epoch": 3.91,
1451
+ "grad_norm": 5.55601692199707,
1452
+ "learning_rate": 1.2190956320954398e-05,
1453
+ "loss": 0.6005,
1454
+ "step": 101500
1455
+ },
1456
+ {
1457
+ "epoch": 3.93,
1458
+ "grad_norm": 4.776731014251709,
1459
+ "learning_rate": 1.215247258033481e-05,
1460
+ "loss": 0.6167,
1461
+ "step": 102000
1462
+ },
1463
+ {
1464
+ "epoch": 3.94,
1465
+ "grad_norm": 6.725343227386475,
1466
+ "learning_rate": 1.2113988839715221e-05,
1467
+ "loss": 0.5845,
1468
+ "step": 102500
1469
+ },
1470
+ {
1471
+ "epoch": 3.96,
1472
+ "grad_norm": 2.1615917682647705,
1473
+ "learning_rate": 1.2075505099095633e-05,
1474
+ "loss": 0.6123,
1475
+ "step": 103000
1476
+ },
1477
+ {
1478
+ "epoch": 3.98,
1479
+ "grad_norm": 3.8668465614318848,
1480
+ "learning_rate": 1.2037021358476043e-05,
1481
+ "loss": 0.5945,
1482
+ "step": 103500
1483
+ },
1484
+ {
1485
+ "epoch": 4.0,
1486
+ "eval_loss": 1.6611100435256958,
1487
+ "eval_runtime": 11.217,
1488
+ "eval_samples_per_second": 354.908,
1489
+ "eval_steps_per_second": 88.794,
1490
+ "step": 103940
1491
+ },
1492
+ {
1493
+ "epoch": 4.0,
1494
+ "grad_norm": 5.890031337738037,
1495
+ "learning_rate": 1.1998614585337696e-05,
1496
+ "loss": 0.5796,
1497
+ "step": 104000
1498
+ },
1499
+ {
1500
+ "epoch": 4.02,
1501
+ "grad_norm": 8.080361366271973,
1502
+ "learning_rate": 1.1960130844718107e-05,
1503
+ "loss": 0.5026,
1504
+ "step": 104500
1505
+ },
1506
+ {
1507
+ "epoch": 4.04,
1508
+ "grad_norm": 7.6675496101379395,
1509
+ "learning_rate": 1.192164710409852e-05,
1510
+ "loss": 0.522,
1511
+ "step": 105000
1512
+ },
1513
+ {
1514
+ "epoch": 4.06,
1515
+ "grad_norm": 6.038026332855225,
1516
+ "learning_rate": 1.1883163363478931e-05,
1517
+ "loss": 0.491,
1518
+ "step": 105500
1519
+ },
1520
+ {
1521
+ "epoch": 4.08,
1522
+ "grad_norm": 2.374932050704956,
1523
+ "learning_rate": 1.1844756590340582e-05,
1524
+ "loss": 0.5123,
1525
+ "step": 106000
1526
+ },
1527
+ {
1528
+ "epoch": 4.1,
1529
+ "grad_norm": 1.8449337482452393,
1530
+ "learning_rate": 1.1806272849720994e-05,
1531
+ "loss": 0.5104,
1532
+ "step": 106500
1533
+ },
1534
+ {
1535
+ "epoch": 4.12,
1536
+ "grad_norm": 7.582018852233887,
1537
+ "learning_rate": 1.1767789109101406e-05,
1538
+ "loss": 0.525,
1539
+ "step": 107000
1540
+ },
1541
+ {
1542
+ "epoch": 4.14,
1543
+ "grad_norm": 6.123476028442383,
1544
+ "learning_rate": 1.1729305368481817e-05,
1545
+ "loss": 0.5276,
1546
+ "step": 107500
1547
+ },
1548
+ {
1549
+ "epoch": 4.16,
1550
+ "grad_norm": 6.717319011688232,
1551
+ "learning_rate": 1.1690898595343468e-05,
1552
+ "loss": 0.5204,
1553
+ "step": 108000
1554
+ },
1555
+ {
1556
+ "epoch": 4.18,
1557
+ "grad_norm": 5.3222336769104,
1558
+ "learning_rate": 1.165241485472388e-05,
1559
+ "loss": 0.5039,
1560
+ "step": 108500
1561
+ },
1562
+ {
1563
+ "epoch": 4.19,
1564
+ "grad_norm": 7.88124942779541,
1565
+ "learning_rate": 1.1613931114104292e-05,
1566
+ "loss": 0.5177,
1567
+ "step": 109000
1568
+ },
1569
+ {
1570
+ "epoch": 4.21,
1571
+ "grad_norm": 13.466960906982422,
1572
+ "learning_rate": 1.1575447373484704e-05,
1573
+ "loss": 0.5076,
1574
+ "step": 109500
1575
+ },
1576
+ {
1577
+ "epoch": 4.23,
1578
+ "grad_norm": 3.5181362628936768,
1579
+ "learning_rate": 1.1536963632865115e-05,
1580
+ "loss": 0.5227,
1581
+ "step": 110000
1582
+ },
1583
+ {
1584
+ "epoch": 4.25,
1585
+ "grad_norm": 9.861627578735352,
1586
+ "learning_rate": 1.1498479892245527e-05,
1587
+ "loss": 0.5297,
1588
+ "step": 110500
1589
+ },
1590
+ {
1591
+ "epoch": 4.27,
1592
+ "grad_norm": 9.91634464263916,
1593
+ "learning_rate": 1.1460073119107178e-05,
1594
+ "loss": 0.5425,
1595
+ "step": 111000
1596
+ },
1597
+ {
1598
+ "epoch": 4.29,
1599
+ "grad_norm": 8.588738441467285,
1600
+ "learning_rate": 1.142158937848759e-05,
1601
+ "loss": 0.5298,
1602
+ "step": 111500
1603
+ },
1604
+ {
1605
+ "epoch": 4.31,
1606
+ "grad_norm": 4.605578899383545,
1607
+ "learning_rate": 1.1383105637868002e-05,
1608
+ "loss": 0.5202,
1609
+ "step": 112000
1610
+ },
1611
+ {
1612
+ "epoch": 4.33,
1613
+ "grad_norm": 4.7749481201171875,
1614
+ "learning_rate": 1.1344621897248413e-05,
1615
+ "loss": 0.5262,
1616
+ "step": 112500
1617
+ },
1618
+ {
1619
+ "epoch": 4.35,
1620
+ "grad_norm": 6.725896835327148,
1621
+ "learning_rate": 1.1306215124110066e-05,
1622
+ "loss": 0.5012,
1623
+ "step": 113000
1624
+ },
1625
+ {
1626
+ "epoch": 4.37,
1627
+ "grad_norm": 5.475027561187744,
1628
+ "learning_rate": 1.1267731383490476e-05,
1629
+ "loss": 0.5077,
1630
+ "step": 113500
1631
+ },
1632
+ {
1633
+ "epoch": 4.39,
1634
+ "grad_norm": 6.180666923522949,
1635
+ "learning_rate": 1.1229247642870888e-05,
1636
+ "loss": 0.514,
1637
+ "step": 114000
1638
+ },
1639
+ {
1640
+ "epoch": 4.41,
1641
+ "grad_norm": 6.444517135620117,
1642
+ "learning_rate": 1.119084086973254e-05,
1643
+ "loss": 0.5538,
1644
+ "step": 114500
1645
+ },
1646
+ {
1647
+ "epoch": 4.43,
1648
+ "grad_norm": 6.281551361083984,
1649
+ "learning_rate": 1.1152357129112952e-05,
1650
+ "loss": 0.5441,
1651
+ "step": 115000
1652
+ },
1653
+ {
1654
+ "epoch": 4.44,
1655
+ "grad_norm": 9.76223373413086,
1656
+ "learning_rate": 1.1113873388493362e-05,
1657
+ "loss": 0.5512,
1658
+ "step": 115500
1659
+ },
1660
+ {
1661
+ "epoch": 4.46,
1662
+ "grad_norm": 5.260191440582275,
1663
+ "learning_rate": 1.1075389647873774e-05,
1664
+ "loss": 0.5178,
1665
+ "step": 116000
1666
+ },
1667
+ {
1668
+ "epoch": 4.48,
1669
+ "grad_norm": 7.131434917449951,
1670
+ "learning_rate": 1.1036905907254186e-05,
1671
+ "loss": 0.5017,
1672
+ "step": 116500
1673
+ },
1674
+ {
1675
+ "epoch": 4.5,
1676
+ "grad_norm": 6.9372992515563965,
1677
+ "learning_rate": 1.0998422166634598e-05,
1678
+ "loss": 0.5323,
1679
+ "step": 117000
1680
+ },
1681
+ {
1682
+ "epoch": 4.52,
1683
+ "grad_norm": 3.6769461631774902,
1684
+ "learning_rate": 1.095993842601501e-05,
1685
+ "loss": 0.5472,
1686
+ "step": 117500
1687
+ },
1688
+ {
1689
+ "epoch": 4.54,
1690
+ "grad_norm": 4.933724403381348,
1691
+ "learning_rate": 1.0921454685395421e-05,
1692
+ "loss": 0.5275,
1693
+ "step": 118000
1694
+ },
1695
+ {
1696
+ "epoch": 4.56,
1697
+ "grad_norm": 8.011011123657227,
1698
+ "learning_rate": 1.0882970944775832e-05,
1699
+ "loss": 0.5754,
1700
+ "step": 118500
1701
+ },
1702
+ {
1703
+ "epoch": 4.58,
1704
+ "grad_norm": 5.938791275024414,
1705
+ "learning_rate": 1.0844564171637484e-05,
1706
+ "loss": 0.5244,
1707
+ "step": 119000
1708
+ },
1709
+ {
1710
+ "epoch": 4.6,
1711
+ "grad_norm": 5.401825904846191,
1712
+ "learning_rate": 1.0806080431017896e-05,
1713
+ "loss": 0.5601,
1714
+ "step": 119500
1715
+ },
1716
+ {
1717
+ "epoch": 4.62,
1718
+ "grad_norm": 3.8994340896606445,
1719
+ "learning_rate": 1.0767596690398308e-05,
1720
+ "loss": 0.5399,
1721
+ "step": 120000
1722
+ },
1723
+ {
1724
+ "epoch": 4.64,
1725
+ "grad_norm": 7.102384090423584,
1726
+ "learning_rate": 1.072911294977872e-05,
1727
+ "loss": 0.5426,
1728
+ "step": 120500
1729
+ },
1730
+ {
1731
+ "epoch": 4.66,
1732
+ "grad_norm": 4.785587787628174,
1733
+ "learning_rate": 1.069070617664037e-05,
1734
+ "loss": 0.541,
1735
+ "step": 121000
1736
+ },
1737
+ {
1738
+ "epoch": 4.68,
1739
+ "grad_norm": 3.6323719024658203,
1740
+ "learning_rate": 1.0652222436020782e-05,
1741
+ "loss": 0.5327,
1742
+ "step": 121500
1743
+ },
1744
+ {
1745
+ "epoch": 4.7,
1746
+ "grad_norm": 3.2487778663635254,
1747
+ "learning_rate": 1.0613738695401194e-05,
1748
+ "loss": 0.5437,
1749
+ "step": 122000
1750
+ },
1751
+ {
1752
+ "epoch": 4.71,
1753
+ "grad_norm": 5.607531547546387,
1754
+ "learning_rate": 1.0575254954781606e-05,
1755
+ "loss": 0.5691,
1756
+ "step": 122500
1757
+ },
1758
+ {
1759
+ "epoch": 4.73,
1760
+ "grad_norm": 4.990538597106934,
1761
+ "learning_rate": 1.0536848181643256e-05,
1762
+ "loss": 0.5544,
1763
+ "step": 123000
1764
+ },
1765
+ {
1766
+ "epoch": 4.75,
1767
+ "grad_norm": 11.409914016723633,
1768
+ "learning_rate": 1.0498364441023668e-05,
1769
+ "loss": 0.5229,
1770
+ "step": 123500
1771
+ },
1772
+ {
1773
+ "epoch": 4.77,
1774
+ "grad_norm": 5.91348123550415,
1775
+ "learning_rate": 1.045988070040408e-05,
1776
+ "loss": 0.5609,
1777
+ "step": 124000
1778
+ },
1779
+ {
1780
+ "epoch": 4.79,
1781
+ "grad_norm": 5.885928630828857,
1782
+ "learning_rate": 1.0421396959784492e-05,
1783
+ "loss": 0.5503,
1784
+ "step": 124500
1785
+ },
1786
+ {
1787
+ "epoch": 4.81,
1788
+ "grad_norm": 3.7525486946105957,
1789
+ "learning_rate": 1.0382913219164904e-05,
1790
+ "loss": 0.5452,
1791
+ "step": 125000
1792
+ },
1793
+ {
1794
+ "epoch": 4.83,
1795
+ "grad_norm": 4.8420305252075195,
1796
+ "learning_rate": 1.0344506446026554e-05,
1797
+ "loss": 0.566,
1798
+ "step": 125500
1799
+ },
1800
+ {
1801
+ "epoch": 4.85,
1802
+ "grad_norm": 5.222369194030762,
1803
+ "learning_rate": 1.0306022705406966e-05,
1804
+ "loss": 0.5481,
1805
+ "step": 126000
1806
+ },
1807
+ {
1808
+ "epoch": 4.87,
1809
+ "grad_norm": 6.775496006011963,
1810
+ "learning_rate": 1.0267538964787378e-05,
1811
+ "loss": 0.5353,
1812
+ "step": 126500
1813
+ },
1814
+ {
1815
+ "epoch": 4.89,
1816
+ "grad_norm": 13.353194236755371,
1817
+ "learning_rate": 1.022905522416779e-05,
1818
+ "loss": 0.5574,
1819
+ "step": 127000
1820
+ },
1821
+ {
1822
+ "epoch": 4.91,
1823
+ "grad_norm": 6.453900337219238,
1824
+ "learning_rate": 1.0190648451029442e-05,
1825
+ "loss": 0.5715,
1826
+ "step": 127500
1827
+ },
1828
+ {
1829
+ "epoch": 4.93,
1830
+ "grad_norm": 9.484274864196777,
1831
+ "learning_rate": 1.0152241677891092e-05,
1832
+ "loss": 0.5678,
1833
+ "step": 128000
1834
+ },
1835
+ {
1836
+ "epoch": 4.95,
1837
+ "grad_norm": 5.4346160888671875,
1838
+ "learning_rate": 1.0113757937271504e-05,
1839
+ "loss": 0.5661,
1840
+ "step": 128500
1841
+ },
1842
+ {
1843
+ "epoch": 4.96,
1844
+ "grad_norm": 7.368044376373291,
1845
+ "learning_rate": 1.0075274196651916e-05,
1846
+ "loss": 0.5524,
1847
+ "step": 129000
1848
+ },
1849
+ {
1850
+ "epoch": 4.98,
1851
+ "grad_norm": 5.8485941886901855,
1852
+ "learning_rate": 1.0036790456032328e-05,
1853
+ "loss": 0.5472,
1854
+ "step": 129500
1855
+ },
1856
+ {
1857
+ "epoch": 5.0,
1858
+ "eval_loss": 1.6937280893325806,
1859
+ "eval_runtime": 11.3508,
1860
+ "eval_samples_per_second": 350.726,
1861
+ "eval_steps_per_second": 87.747,
1862
+ "step": 129925
1863
+ },
1864
+ {
1865
+ "epoch": 5.0,
1866
+ "grad_norm": 3.9238245487213135,
1867
+ "learning_rate": 9.998306715412738e-06,
1868
+ "loss": 0.5324,
1869
+ "step": 130000
1870
+ },
1871
+ {
1872
+ "epoch": 5.02,
1873
+ "grad_norm": 3.3814291954040527,
1874
+ "learning_rate": 9.95982297479315e-06,
1875
+ "loss": 0.4778,
1876
+ "step": 130500
1877
+ },
1878
+ {
1879
+ "epoch": 5.04,
1880
+ "grad_norm": 3.6909124851226807,
1881
+ "learning_rate": 9.921339234173562e-06,
1882
+ "loss": 0.4872,
1883
+ "step": 131000
1884
+ },
1885
+ {
1886
+ "epoch": 5.06,
1887
+ "grad_norm": 3.4291248321533203,
1888
+ "learning_rate": 9.882855493553974e-06,
1889
+ "loss": 0.4572,
1890
+ "step": 131500
1891
+ },
1892
+ {
1893
+ "epoch": 5.08,
1894
+ "grad_norm": 7.226670742034912,
1895
+ "learning_rate": 9.844448720415625e-06,
1896
+ "loss": 0.4887,
1897
+ "step": 132000
1898
+ },
1899
+ {
1900
+ "epoch": 5.1,
1901
+ "grad_norm": 3.1274802684783936,
1902
+ "learning_rate": 9.805964979796036e-06,
1903
+ "loss": 0.4687,
1904
+ "step": 132500
1905
+ },
1906
+ {
1907
+ "epoch": 5.12,
1908
+ "grad_norm": 6.210909366607666,
1909
+ "learning_rate": 9.767558206657689e-06,
1910
+ "loss": 0.4695,
1911
+ "step": 133000
1912
+ },
1913
+ {
1914
+ "epoch": 5.14,
1915
+ "grad_norm": 4.919899940490723,
1916
+ "learning_rate": 9.7290744660381e-06,
1917
+ "loss": 0.4697,
1918
+ "step": 133500
1919
+ },
1920
+ {
1921
+ "epoch": 5.16,
1922
+ "grad_norm": 5.735771179199219,
1923
+ "learning_rate": 9.690590725418512e-06,
1924
+ "loss": 0.4663,
1925
+ "step": 134000
1926
+ },
1927
+ {
1928
+ "epoch": 5.18,
1929
+ "grad_norm": 4.188859462738037,
1930
+ "learning_rate": 9.652106984798924e-06,
1931
+ "loss": 0.4997,
1932
+ "step": 134500
1933
+ },
1934
+ {
1935
+ "epoch": 5.2,
1936
+ "grad_norm": 6.6499786376953125,
1937
+ "learning_rate": 9.613623244179336e-06,
1938
+ "loss": 0.4713,
1939
+ "step": 135000
1940
+ },
1941
+ {
1942
+ "epoch": 5.21,
1943
+ "grad_norm": 5.279160499572754,
1944
+ "learning_rate": 9.575139503559748e-06,
1945
+ "loss": 0.4742,
1946
+ "step": 135500
1947
+ },
1948
+ {
1949
+ "epoch": 5.23,
1950
+ "grad_norm": 19.4254093170166,
1951
+ "learning_rate": 9.536655762940158e-06,
1952
+ "loss": 0.4906,
1953
+ "step": 136000
1954
+ },
1955
+ {
1956
+ "epoch": 5.25,
1957
+ "grad_norm": 4.88319206237793,
1958
+ "learning_rate": 9.49817202232057e-06,
1959
+ "loss": 0.466,
1960
+ "step": 136500
1961
+ },
1962
+ {
1963
+ "epoch": 5.27,
1964
+ "grad_norm": 3.700117826461792,
1965
+ "learning_rate": 9.459688281700982e-06,
1966
+ "loss": 0.5211,
1967
+ "step": 137000
1968
+ },
1969
+ {
1970
+ "epoch": 5.29,
1971
+ "grad_norm": 5.820217609405518,
1972
+ "learning_rate": 9.421281508562632e-06,
1973
+ "loss": 0.5012,
1974
+ "step": 137500
1975
+ },
1976
+ {
1977
+ "epoch": 5.31,
1978
+ "grad_norm": 4.963629245758057,
1979
+ "learning_rate": 9.382797767943044e-06,
1980
+ "loss": 0.4893,
1981
+ "step": 138000
1982
+ },
1983
+ {
1984
+ "epoch": 5.33,
1985
+ "grad_norm": 3.2575323581695557,
1986
+ "learning_rate": 9.344314027323456e-06,
1987
+ "loss": 0.4881,
1988
+ "step": 138500
1989
+ },
1990
+ {
1991
+ "epoch": 5.35,
1992
+ "grad_norm": 4.5853424072265625,
1993
+ "learning_rate": 9.305830286703868e-06,
1994
+ "loss": 0.497,
1995
+ "step": 139000
1996
+ },
1997
+ {
1998
+ "epoch": 5.37,
1999
+ "grad_norm": 9.991304397583008,
2000
+ "learning_rate": 9.26734654608428e-06,
2001
+ "loss": 0.5029,
2002
+ "step": 139500
2003
+ },
2004
+ {
2005
+ "epoch": 5.39,
2006
+ "grad_norm": 7.186006546020508,
2007
+ "learning_rate": 9.228862805464692e-06,
2008
+ "loss": 0.5146,
2009
+ "step": 140000
2010
+ },
2011
+ {
2012
+ "epoch": 5.41,
2013
+ "grad_norm": 4.827492713928223,
2014
+ "learning_rate": 9.190456032326342e-06,
2015
+ "loss": 0.5155,
2016
+ "step": 140500
2017
+ },
2018
+ {
2019
+ "epoch": 5.43,
2020
+ "grad_norm": 6.840179443359375,
2021
+ "learning_rate": 9.151972291706754e-06,
2022
+ "loss": 0.5204,
2023
+ "step": 141000
2024
+ },
2025
+ {
2026
+ "epoch": 5.45,
2027
+ "grad_norm": 3.26387357711792,
2028
+ "learning_rate": 9.113488551087166e-06,
2029
+ "loss": 0.4909,
2030
+ "step": 141500
2031
+ },
2032
+ {
2033
+ "epoch": 5.46,
2034
+ "grad_norm": 8.286077499389648,
2035
+ "learning_rate": 9.075004810467578e-06,
2036
+ "loss": 0.4973,
2037
+ "step": 142000
2038
+ },
2039
+ {
2040
+ "epoch": 5.48,
2041
+ "grad_norm": 6.248400688171387,
2042
+ "learning_rate": 9.03659803732923e-06,
2043
+ "loss": 0.5213,
2044
+ "step": 142500
2045
+ },
2046
+ {
2047
+ "epoch": 5.5,
2048
+ "grad_norm": 4.121056079864502,
2049
+ "learning_rate": 8.998114296709642e-06,
2050
+ "loss": 0.5176,
2051
+ "step": 143000
2052
+ },
2053
+ {
2054
+ "epoch": 5.52,
2055
+ "grad_norm": 5.463682651519775,
2056
+ "learning_rate": 8.959630556090052e-06,
2057
+ "loss": 0.4707,
2058
+ "step": 143500
2059
+ },
2060
+ {
2061
+ "epoch": 5.54,
2062
+ "grad_norm": 7.710265159606934,
2063
+ "learning_rate": 8.921146815470464e-06,
2064
+ "loss": 0.4728,
2065
+ "step": 144000
2066
+ },
2067
+ {
2068
+ "epoch": 5.56,
2069
+ "grad_norm": 6.872325420379639,
2070
+ "learning_rate": 8.882740042332115e-06,
2071
+ "loss": 0.5183,
2072
+ "step": 144500
2073
+ },
2074
+ {
2075
+ "epoch": 5.58,
2076
+ "grad_norm": 7.6797614097595215,
2077
+ "learning_rate": 8.844256301712527e-06,
2078
+ "loss": 0.5065,
2079
+ "step": 145000
2080
+ },
2081
+ {
2082
+ "epoch": 5.6,
2083
+ "grad_norm": 4.729648590087891,
2084
+ "learning_rate": 8.805772561092938e-06,
2085
+ "loss": 0.5117,
2086
+ "step": 145500
2087
+ },
2088
+ {
2089
+ "epoch": 5.62,
2090
+ "grad_norm": 6.701267719268799,
2091
+ "learning_rate": 8.76728882047335e-06,
2092
+ "loss": 0.5427,
2093
+ "step": 146000
2094
+ },
2095
+ {
2096
+ "epoch": 5.64,
2097
+ "grad_norm": 8.45148754119873,
2098
+ "learning_rate": 8.728805079853762e-06,
2099
+ "loss": 0.4782,
2100
+ "step": 146500
2101
+ },
2102
+ {
2103
+ "epoch": 5.66,
2104
+ "grad_norm": 4.278749465942383,
2105
+ "learning_rate": 8.690398306715413e-06,
2106
+ "loss": 0.5093,
2107
+ "step": 147000
2108
+ },
2109
+ {
2110
+ "epoch": 5.68,
2111
+ "grad_norm": 6.3357343673706055,
2112
+ "learning_rate": 8.651914566095825e-06,
2113
+ "loss": 0.5328,
2114
+ "step": 147500
2115
+ },
2116
+ {
2117
+ "epoch": 5.7,
2118
+ "grad_norm": 9.23112678527832,
2119
+ "learning_rate": 8.613430825476237e-06,
2120
+ "loss": 0.526,
2121
+ "step": 148000
2122
+ },
2123
+ {
2124
+ "epoch": 5.71,
2125
+ "grad_norm": 2.7398366928100586,
2126
+ "learning_rate": 8.574947084856648e-06,
2127
+ "loss": 0.5004,
2128
+ "step": 148500
2129
+ },
2130
+ {
2131
+ "epoch": 5.73,
2132
+ "grad_norm": 7.174339771270752,
2133
+ "learning_rate": 8.5365403117183e-06,
2134
+ "loss": 0.4947,
2135
+ "step": 149000
2136
+ },
2137
+ {
2138
+ "epoch": 5.75,
2139
+ "grad_norm": 6.2081427574157715,
2140
+ "learning_rate": 8.498056571098712e-06,
2141
+ "loss": 0.4961,
2142
+ "step": 149500
2143
+ },
2144
+ {
2145
+ "epoch": 5.77,
2146
+ "grad_norm": 5.464582443237305,
2147
+ "learning_rate": 8.459572830479124e-06,
2148
+ "loss": 0.4984,
2149
+ "step": 150000
2150
+ },
2151
+ {
2152
+ "epoch": 5.79,
2153
+ "grad_norm": 4.864925384521484,
2154
+ "learning_rate": 8.421089089859536e-06,
2155
+ "loss": 0.5101,
2156
+ "step": 150500
2157
+ },
2158
+ {
2159
+ "epoch": 5.81,
2160
+ "grad_norm": 10.367297172546387,
2161
+ "learning_rate": 8.382682316721185e-06,
2162
+ "loss": 0.5004,
2163
+ "step": 151000
2164
+ },
2165
+ {
2166
+ "epoch": 5.83,
2167
+ "grad_norm": 4.001099109649658,
2168
+ "learning_rate": 8.344198576101597e-06,
2169
+ "loss": 0.5109,
2170
+ "step": 151500
2171
+ },
2172
+ {
2173
+ "epoch": 5.85,
2174
+ "grad_norm": 8.188924789428711,
2175
+ "learning_rate": 8.305714835482009e-06,
2176
+ "loss": 0.4988,
2177
+ "step": 152000
2178
+ },
2179
+ {
2180
+ "epoch": 5.87,
2181
+ "grad_norm": 5.423083782196045,
2182
+ "learning_rate": 8.26723109486242e-06,
2183
+ "loss": 0.4957,
2184
+ "step": 152500
2185
+ },
2186
+ {
2187
+ "epoch": 5.89,
2188
+ "grad_norm": 5.513723373413086,
2189
+ "learning_rate": 8.228824321724073e-06,
2190
+ "loss": 0.5113,
2191
+ "step": 153000
2192
+ },
2193
+ {
2194
+ "epoch": 5.91,
2195
+ "grad_norm": 5.753915309906006,
2196
+ "learning_rate": 8.190340581104485e-06,
2197
+ "loss": 0.5088,
2198
+ "step": 153500
2199
+ },
2200
+ {
2201
+ "epoch": 5.93,
2202
+ "grad_norm": 5.672900676727295,
2203
+ "learning_rate": 8.151933807966135e-06,
2204
+ "loss": 0.5117,
2205
+ "step": 154000
2206
+ },
2207
+ {
2208
+ "epoch": 5.95,
2209
+ "grad_norm": 6.9963202476501465,
2210
+ "learning_rate": 8.113450067346547e-06,
2211
+ "loss": 0.5474,
2212
+ "step": 154500
2213
+ },
2214
+ {
2215
+ "epoch": 5.96,
2216
+ "grad_norm": 6.640045166015625,
2217
+ "learning_rate": 8.074966326726959e-06,
2218
+ "loss": 0.5135,
2219
+ "step": 155000
2220
+ },
2221
+ {
2222
+ "epoch": 5.98,
2223
+ "grad_norm": 5.076375961303711,
2224
+ "learning_rate": 8.036482586107371e-06,
2225
+ "loss": 0.5147,
2226
+ "step": 155500
2227
+ },
2228
+ {
2229
+ "epoch": 6.0,
2230
+ "eval_loss": 1.7142382860183716,
2231
+ "eval_runtime": 11.2855,
2232
+ "eval_samples_per_second": 352.753,
2233
+ "eval_steps_per_second": 88.255,
2234
+ "step": 155910
2235
+ },
2236
+ {
2237
+ "epoch": 6.0,
2238
+ "grad_norm": 6.40647554397583,
2239
+ "learning_rate": 7.997998845487783e-06,
2240
+ "loss": 0.467,
2241
+ "step": 156000
2242
+ },
2243
+ {
2244
+ "epoch": 6.02,
2245
+ "grad_norm": 3.021810293197632,
2246
+ "learning_rate": 7.959515104868195e-06,
2247
+ "loss": 0.4311,
2248
+ "step": 156500
2249
+ },
2250
+ {
2251
+ "epoch": 6.04,
2252
+ "grad_norm": 11.23317813873291,
2253
+ "learning_rate": 7.921031364248607e-06,
2254
+ "loss": 0.4569,
2255
+ "step": 157000
2256
+ },
2257
+ {
2258
+ "epoch": 6.06,
2259
+ "grad_norm": 4.276322841644287,
2260
+ "learning_rate": 7.882547623629019e-06,
2261
+ "loss": 0.4523,
2262
+ "step": 157500
2263
+ },
2264
+ {
2265
+ "epoch": 6.08,
2266
+ "grad_norm": 5.726410388946533,
2267
+ "learning_rate": 7.844140850490667e-06,
2268
+ "loss": 0.4601,
2269
+ "step": 158000
2270
+ },
2271
+ {
2272
+ "epoch": 6.1,
2273
+ "grad_norm": 8.776612281799316,
2274
+ "learning_rate": 7.80565710987108e-06,
2275
+ "loss": 0.4254,
2276
+ "step": 158500
2277
+ },
2278
+ {
2279
+ "epoch": 6.12,
2280
+ "grad_norm": 9.005372047424316,
2281
+ "learning_rate": 7.767173369251491e-06,
2282
+ "loss": 0.4328,
2283
+ "step": 159000
2284
+ },
2285
+ {
2286
+ "epoch": 6.14,
2287
+ "grad_norm": 9.345296859741211,
2288
+ "learning_rate": 7.728689628631903e-06,
2289
+ "loss": 0.4619,
2290
+ "step": 159500
2291
+ },
2292
+ {
2293
+ "epoch": 6.16,
2294
+ "grad_norm": 5.089489936828613,
2295
+ "learning_rate": 7.690282855493555e-06,
2296
+ "loss": 0.4546,
2297
+ "step": 160000
2298
+ },
2299
+ {
2300
+ "epoch": 6.18,
2301
+ "grad_norm": 5.524288177490234,
2302
+ "learning_rate": 7.651799114873967e-06,
2303
+ "loss": 0.4444,
2304
+ "step": 160500
2305
+ },
2306
+ {
2307
+ "epoch": 6.2,
2308
+ "grad_norm": 5.413343906402588,
2309
+ "learning_rate": 7.613315374254379e-06,
2310
+ "loss": 0.4672,
2311
+ "step": 161000
2312
+ },
2313
+ {
2314
+ "epoch": 6.22,
2315
+ "grad_norm": 2.5175459384918213,
2316
+ "learning_rate": 7.5749086011160285e-06,
2317
+ "loss": 0.4534,
2318
+ "step": 161500
2319
+ },
2320
+ {
2321
+ "epoch": 6.23,
2322
+ "grad_norm": 9.147507667541504,
2323
+ "learning_rate": 7.5364248604964404e-06,
2324
+ "loss": 0.4642,
2325
+ "step": 162000
2326
+ },
2327
+ {
2328
+ "epoch": 6.25,
2329
+ "grad_norm": 4.600451946258545,
2330
+ "learning_rate": 7.497941119876852e-06,
2331
+ "loss": 0.4448,
2332
+ "step": 162500
2333
+ },
2334
+ {
2335
+ "epoch": 6.27,
2336
+ "grad_norm": 5.7864556312561035,
2337
+ "learning_rate": 7.459457379257264e-06,
2338
+ "loss": 0.4463,
2339
+ "step": 163000
2340
+ },
2341
+ {
2342
+ "epoch": 6.29,
2343
+ "grad_norm": 6.185606002807617,
2344
+ "learning_rate": 7.420973638637676e-06,
2345
+ "loss": 0.4608,
2346
+ "step": 163500
2347
+ },
2348
+ {
2349
+ "epoch": 6.31,
2350
+ "grad_norm": 4.961802959442139,
2351
+ "learning_rate": 7.382489898018088e-06,
2352
+ "loss": 0.4609,
2353
+ "step": 164000
2354
+ },
2355
+ {
2356
+ "epoch": 6.33,
2357
+ "grad_norm": 8.758057594299316,
2358
+ "learning_rate": 7.3440061573985e-06,
2359
+ "loss": 0.459,
2360
+ "step": 164500
2361
+ },
2362
+ {
2363
+ "epoch": 6.35,
2364
+ "grad_norm": 13.342981338500977,
2365
+ "learning_rate": 7.305522416778912e-06,
2366
+ "loss": 0.4684,
2367
+ "step": 165000
2368
+ },
2369
+ {
2370
+ "epoch": 6.37,
2371
+ "grad_norm": 3.678190231323242,
2372
+ "learning_rate": 7.267038676159324e-06,
2373
+ "loss": 0.4482,
2374
+ "step": 165500
2375
+ },
2376
+ {
2377
+ "epoch": 6.39,
2378
+ "grad_norm": 5.763678550720215,
2379
+ "learning_rate": 7.2285549355397356e-06,
2380
+ "loss": 0.4828,
2381
+ "step": 166000
2382
+ },
2383
+ {
2384
+ "epoch": 6.41,
2385
+ "grad_norm": 2.4365394115448,
2386
+ "learning_rate": 7.1900711949201475e-06,
2387
+ "loss": 0.4519,
2388
+ "step": 166500
2389
+ },
2390
+ {
2391
+ "epoch": 6.43,
2392
+ "grad_norm": 7.325679779052734,
2393
+ "learning_rate": 7.1515874543005585e-06,
2394
+ "loss": 0.4725,
2395
+ "step": 167000
2396
+ },
2397
+ {
2398
+ "epoch": 6.45,
2399
+ "grad_norm": 7.1001811027526855,
2400
+ "learning_rate": 7.11310371368097e-06,
2401
+ "loss": 0.4552,
2402
+ "step": 167500
2403
+ },
2404
+ {
2405
+ "epoch": 6.47,
2406
+ "grad_norm": 4.369767189025879,
2407
+ "learning_rate": 7.0746199730613814e-06,
2408
+ "loss": 0.4615,
2409
+ "step": 168000
2410
+ },
2411
+ {
2412
+ "epoch": 6.48,
2413
+ "grad_norm": 5.81917667388916,
2414
+ "learning_rate": 7.036213199923034e-06,
2415
+ "loss": 0.4518,
2416
+ "step": 168500
2417
+ },
2418
+ {
2419
+ "epoch": 6.5,
2420
+ "grad_norm": 4.713387966156006,
2421
+ "learning_rate": 6.997729459303445e-06,
2422
+ "loss": 0.473,
2423
+ "step": 169000
2424
+ },
2425
+ {
2426
+ "epoch": 6.52,
2427
+ "grad_norm": 7.274464130401611,
2428
+ "learning_rate": 6.9592457186838565e-06,
2429
+ "loss": 0.4729,
2430
+ "step": 169500
2431
+ },
2432
+ {
2433
+ "epoch": 6.54,
2434
+ "grad_norm": 8.085463523864746,
2435
+ "learning_rate": 6.9207619780642684e-06,
2436
+ "loss": 0.4723,
2437
+ "step": 170000
2438
+ },
2439
+ {
2440
+ "epoch": 6.56,
2441
+ "grad_norm": 9.212237358093262,
2442
+ "learning_rate": 6.88227823744468e-06,
2443
+ "loss": 0.4571,
2444
+ "step": 170500
2445
+ },
2446
+ {
2447
+ "epoch": 6.58,
2448
+ "grad_norm": 3.6301112174987793,
2449
+ "learning_rate": 6.843871464306331e-06,
2450
+ "loss": 0.4726,
2451
+ "step": 171000
2452
+ },
2453
+ {
2454
+ "epoch": 6.6,
2455
+ "grad_norm": 5.698800563812256,
2456
+ "learning_rate": 6.805464691167982e-06,
2457
+ "loss": 0.4831,
2458
+ "step": 171500
2459
+ },
2460
+ {
2461
+ "epoch": 6.62,
2462
+ "grad_norm": 5.532044887542725,
2463
+ "learning_rate": 6.766980950548394e-06,
2464
+ "loss": 0.4708,
2465
+ "step": 172000
2466
+ },
2467
+ {
2468
+ "epoch": 6.64,
2469
+ "grad_norm": 7.939385414123535,
2470
+ "learning_rate": 6.728497209928806e-06,
2471
+ "loss": 0.468,
2472
+ "step": 172500
2473
+ },
2474
+ {
2475
+ "epoch": 6.66,
2476
+ "grad_norm": 3.2672829627990723,
2477
+ "learning_rate": 6.690013469309218e-06,
2478
+ "loss": 0.4935,
2479
+ "step": 173000
2480
+ },
2481
+ {
2482
+ "epoch": 6.68,
2483
+ "grad_norm": 11.197509765625,
2484
+ "learning_rate": 6.65152972868963e-06,
2485
+ "loss": 0.4777,
2486
+ "step": 173500
2487
+ },
2488
+ {
2489
+ "epoch": 6.7,
2490
+ "grad_norm": 2.946389675140381,
2491
+ "learning_rate": 6.6130459880700416e-06,
2492
+ "loss": 0.4786,
2493
+ "step": 174000
2494
+ },
2495
+ {
2496
+ "epoch": 6.72,
2497
+ "grad_norm": 1.725356936454773,
2498
+ "learning_rate": 6.574562247450453e-06,
2499
+ "loss": 0.4471,
2500
+ "step": 174500
2501
+ },
2502
+ {
2503
+ "epoch": 6.73,
2504
+ "grad_norm": 3.4278807640075684,
2505
+ "learning_rate": 6.5360785068308645e-06,
2506
+ "loss": 0.472,
2507
+ "step": 175000
2508
+ },
2509
+ {
2510
+ "epoch": 6.75,
2511
+ "grad_norm": 7.619284629821777,
2512
+ "learning_rate": 6.4975947662112756e-06,
2513
+ "loss": 0.4566,
2514
+ "step": 175500
2515
+ },
2516
+ {
2517
+ "epoch": 6.77,
2518
+ "grad_norm": 3.8815529346466064,
2519
+ "learning_rate": 6.459187993072928e-06,
2520
+ "loss": 0.4738,
2521
+ "step": 176000
2522
+ },
2523
+ {
2524
+ "epoch": 6.79,
2525
+ "grad_norm": 4.85827112197876,
2526
+ "learning_rate": 6.420704252453339e-06,
2527
+ "loss": 0.4637,
2528
+ "step": 176500
2529
+ },
2530
+ {
2531
+ "epoch": 6.81,
2532
+ "grad_norm": 5.672398090362549,
2533
+ "learning_rate": 6.382220511833751e-06,
2534
+ "loss": 0.4646,
2535
+ "step": 177000
2536
+ },
2537
+ {
2538
+ "epoch": 6.83,
2539
+ "grad_norm": 3.5175821781158447,
2540
+ "learning_rate": 6.3437367712141626e-06,
2541
+ "loss": 0.4949,
2542
+ "step": 177500
2543
+ },
2544
+ {
2545
+ "epoch": 6.85,
2546
+ "grad_norm": 12.251681327819824,
2547
+ "learning_rate": 6.3052530305945744e-06,
2548
+ "loss": 0.4764,
2549
+ "step": 178000
2550
+ },
2551
+ {
2552
+ "epoch": 6.87,
2553
+ "grad_norm": 7.721076488494873,
2554
+ "learning_rate": 6.266846257456225e-06,
2555
+ "loss": 0.4982,
2556
+ "step": 178500
2557
+ },
2558
+ {
2559
+ "epoch": 6.89,
2560
+ "grad_norm": 3.9712600708007812,
2561
+ "learning_rate": 6.228362516836637e-06,
2562
+ "loss": 0.4859,
2563
+ "step": 179000
2564
+ },
2565
+ {
2566
+ "epoch": 6.91,
2567
+ "grad_norm": 5.2524213790893555,
2568
+ "learning_rate": 6.189878776217049e-06,
2569
+ "loss": 0.4697,
2570
+ "step": 179500
2571
+ },
2572
+ {
2573
+ "epoch": 6.93,
2574
+ "grad_norm": 4.3481340408325195,
2575
+ "learning_rate": 6.151395035597461e-06,
2576
+ "loss": 0.4916,
2577
+ "step": 180000
2578
+ },
2579
+ {
2580
+ "epoch": 6.95,
2581
+ "grad_norm": 5.635584354400635,
2582
+ "learning_rate": 6.1129112949778725e-06,
2583
+ "loss": 0.476,
2584
+ "step": 180500
2585
+ },
2586
+ {
2587
+ "epoch": 6.97,
2588
+ "grad_norm": 4.938673973083496,
2589
+ "learning_rate": 6.074504521839524e-06,
2590
+ "loss": 0.4842,
2591
+ "step": 181000
2592
+ },
2593
+ {
2594
+ "epoch": 6.98,
2595
+ "grad_norm": 3.648951530456543,
2596
+ "learning_rate": 6.036097748701174e-06,
2597
+ "loss": 0.4961,
2598
+ "step": 181500
2599
+ },
2600
+ {
2601
+ "epoch": 7.0,
2602
+ "eval_loss": 1.7367790937423706,
2603
+ "eval_runtime": 11.3431,
2604
+ "eval_samples_per_second": 350.961,
2605
+ "eval_steps_per_second": 87.806,
2606
+ "step": 181895
2607
+ },
2608
+ {
2609
+ "epoch": 7.0,
2610
+ "grad_norm": 6.272878646850586,
2611
+ "learning_rate": 5.997614008081586e-06,
2612
+ "loss": 0.4793,
2613
+ "step": 182000
2614
+ },
2615
+ {
2616
+ "epoch": 7.02,
2617
+ "grad_norm": 8.145177841186523,
2618
+ "learning_rate": 5.959130267461998e-06,
2619
+ "loss": 0.4003,
2620
+ "step": 182500
2621
+ },
2622
+ {
2623
+ "epoch": 7.04,
2624
+ "grad_norm": 3.661898374557495,
2625
+ "learning_rate": 5.92064652684241e-06,
2626
+ "loss": 0.4182,
2627
+ "step": 183000
2628
+ },
2629
+ {
2630
+ "epoch": 7.06,
2631
+ "grad_norm": 4.206645488739014,
2632
+ "learning_rate": 5.882162786222822e-06,
2633
+ "loss": 0.4278,
2634
+ "step": 183500
2635
+ },
2636
+ {
2637
+ "epoch": 7.08,
2638
+ "grad_norm": 2.098565101623535,
2639
+ "learning_rate": 5.843679045603233e-06,
2640
+ "loss": 0.4281,
2641
+ "step": 184000
2642
+ },
2643
+ {
2644
+ "epoch": 7.1,
2645
+ "grad_norm": 5.96292781829834,
2646
+ "learning_rate": 5.805195304983645e-06,
2647
+ "loss": 0.4374,
2648
+ "step": 184500
2649
+ },
2650
+ {
2651
+ "epoch": 7.12,
2652
+ "grad_norm": 2.972966194152832,
2653
+ "learning_rate": 5.766711564364057e-06,
2654
+ "loss": 0.4143,
2655
+ "step": 185000
2656
+ },
2657
+ {
2658
+ "epoch": 7.14,
2659
+ "grad_norm": 9.22352123260498,
2660
+ "learning_rate": 5.7282278237444686e-06,
2661
+ "loss": 0.401,
2662
+ "step": 185500
2663
+ },
2664
+ {
2665
+ "epoch": 7.16,
2666
+ "grad_norm": 8.188728332519531,
2667
+ "learning_rate": 5.689821050606119e-06,
2668
+ "loss": 0.4134,
2669
+ "step": 186000
2670
+ },
2671
+ {
2672
+ "epoch": 7.18,
2673
+ "grad_norm": 8.370118141174316,
2674
+ "learning_rate": 5.65141427746777e-06,
2675
+ "loss": 0.445,
2676
+ "step": 186500
2677
+ },
2678
+ {
2679
+ "epoch": 7.2,
2680
+ "grad_norm": 4.1989922523498535,
2681
+ "learning_rate": 5.612930536848182e-06,
2682
+ "loss": 0.4428,
2683
+ "step": 187000
2684
+ },
2685
+ {
2686
+ "epoch": 7.22,
2687
+ "grad_norm": 8.339373588562012,
2688
+ "learning_rate": 5.574446796228594e-06,
2689
+ "loss": 0.4091,
2690
+ "step": 187500
2691
+ },
2692
+ {
2693
+ "epoch": 7.23,
2694
+ "grad_norm": 5.580036640167236,
2695
+ "learning_rate": 5.535963055609006e-06,
2696
+ "loss": 0.4169,
2697
+ "step": 188000
2698
+ },
2699
+ {
2700
+ "epoch": 7.25,
2701
+ "grad_norm": 5.285128116607666,
2702
+ "learning_rate": 5.4975562824706565e-06,
2703
+ "loss": 0.4282,
2704
+ "step": 188500
2705
+ },
2706
+ {
2707
+ "epoch": 7.27,
2708
+ "grad_norm": 4.649808406829834,
2709
+ "learning_rate": 5.459072541851068e-06,
2710
+ "loss": 0.4199,
2711
+ "step": 189000
2712
+ },
2713
+ {
2714
+ "epoch": 7.29,
2715
+ "grad_norm": 7.711951732635498,
2716
+ "learning_rate": 5.42058880123148e-06,
2717
+ "loss": 0.4331,
2718
+ "step": 189500
2719
+ },
2720
+ {
2721
+ "epoch": 7.31,
2722
+ "grad_norm": 3.5178699493408203,
2723
+ "learning_rate": 5.382105060611892e-06,
2724
+ "loss": 0.4362,
2725
+ "step": 190000
2726
+ },
2727
+ {
2728
+ "epoch": 7.33,
2729
+ "grad_norm": 6.855963230133057,
2730
+ "learning_rate": 5.343621319992304e-06,
2731
+ "loss": 0.449,
2732
+ "step": 190500
2733
+ },
2734
+ {
2735
+ "epoch": 7.35,
2736
+ "grad_norm": 5.831274509429932,
2737
+ "learning_rate": 5.305137579372716e-06,
2738
+ "loss": 0.4286,
2739
+ "step": 191000
2740
+ },
2741
+ {
2742
+ "epoch": 7.37,
2743
+ "grad_norm": 5.696512699127197,
2744
+ "learning_rate": 5.266653838753127e-06,
2745
+ "loss": 0.4388,
2746
+ "step": 191500
2747
+ },
2748
+ {
2749
+ "epoch": 7.39,
2750
+ "grad_norm": 6.7039055824279785,
2751
+ "learning_rate": 5.228170098133539e-06,
2752
+ "loss": 0.4395,
2753
+ "step": 192000
2754
+ },
2755
+ {
2756
+ "epoch": 7.41,
2757
+ "grad_norm": 7.356657028198242,
2758
+ "learning_rate": 5.189686357513951e-06,
2759
+ "loss": 0.4488,
2760
+ "step": 192500
2761
+ },
2762
+ {
2763
+ "epoch": 7.43,
2764
+ "grad_norm": 4.289228439331055,
2765
+ "learning_rate": 5.151279584375601e-06,
2766
+ "loss": 0.4367,
2767
+ "step": 193000
2768
+ },
2769
+ {
2770
+ "epoch": 7.45,
2771
+ "grad_norm": 6.5351948738098145,
2772
+ "learning_rate": 5.112795843756013e-06,
2773
+ "loss": 0.438,
2774
+ "step": 193500
2775
+ },
2776
+ {
2777
+ "epoch": 7.47,
2778
+ "grad_norm": 6.228538990020752,
2779
+ "learning_rate": 5.0743890706176645e-06,
2780
+ "loss": 0.4419,
2781
+ "step": 194000
2782
+ },
2783
+ {
2784
+ "epoch": 7.49,
2785
+ "grad_norm": 8.515239715576172,
2786
+ "learning_rate": 5.035905329998076e-06,
2787
+ "loss": 0.4501,
2788
+ "step": 194500
2789
+ },
2790
+ {
2791
+ "epoch": 7.5,
2792
+ "grad_norm": 6.8445725440979,
2793
+ "learning_rate": 4.9974215893784874e-06,
2794
+ "loss": 0.4257,
2795
+ "step": 195000
2796
+ },
2797
+ {
2798
+ "epoch": 7.52,
2799
+ "grad_norm": 9.162801742553711,
2800
+ "learning_rate": 4.958937848758899e-06,
2801
+ "loss": 0.4349,
2802
+ "step": 195500
2803
+ },
2804
+ {
2805
+ "epoch": 7.54,
2806
+ "grad_norm": 10.447943687438965,
2807
+ "learning_rate": 4.920454108139311e-06,
2808
+ "loss": 0.4364,
2809
+ "step": 196000
2810
+ },
2811
+ {
2812
+ "epoch": 7.56,
2813
+ "grad_norm": 6.493594169616699,
2814
+ "learning_rate": 4.881970367519723e-06,
2815
+ "loss": 0.4436,
2816
+ "step": 196500
2817
+ },
2818
+ {
2819
+ "epoch": 7.58,
2820
+ "grad_norm": 6.790309429168701,
2821
+ "learning_rate": 4.843486626900135e-06,
2822
+ "loss": 0.4535,
2823
+ "step": 197000
2824
+ },
2825
+ {
2826
+ "epoch": 7.6,
2827
+ "grad_norm": 4.3611860275268555,
2828
+ "learning_rate": 4.805002886280547e-06,
2829
+ "loss": 0.443,
2830
+ "step": 197500
2831
+ },
2832
+ {
2833
+ "epoch": 7.62,
2834
+ "grad_norm": 7.639214992523193,
2835
+ "learning_rate": 4.766519145660959e-06,
2836
+ "loss": 0.4433,
2837
+ "step": 198000
2838
+ },
2839
+ {
2840
+ "epoch": 7.64,
2841
+ "grad_norm": 4.556090831756592,
2842
+ "learning_rate": 4.72811237252261e-06,
2843
+ "loss": 0.4433,
2844
+ "step": 198500
2845
+ },
2846
+ {
2847
+ "epoch": 7.66,
2848
+ "grad_norm": 6.17908239364624,
2849
+ "learning_rate": 4.689628631903021e-06,
2850
+ "loss": 0.4389,
2851
+ "step": 199000
2852
+ },
2853
+ {
2854
+ "epoch": 7.68,
2855
+ "grad_norm": 5.435972213745117,
2856
+ "learning_rate": 4.651144891283433e-06,
2857
+ "loss": 0.4413,
2858
+ "step": 199500
2859
+ },
2860
+ {
2861
+ "epoch": 7.7,
2862
+ "grad_norm": 8.422209739685059,
2863
+ "learning_rate": 4.612661150663845e-06,
2864
+ "loss": 0.4444,
2865
+ "step": 200000
2866
+ },
2867
+ {
2868
+ "epoch": 7.72,
2869
+ "grad_norm": 9.431715965270996,
2870
+ "learning_rate": 4.574254377525496e-06,
2871
+ "loss": 0.4495,
2872
+ "step": 200500
2873
+ },
2874
+ {
2875
+ "epoch": 7.74,
2876
+ "grad_norm": 2.9533908367156982,
2877
+ "learning_rate": 4.535847604387147e-06,
2878
+ "loss": 0.4417,
2879
+ "step": 201000
2880
+ },
2881
+ {
2882
+ "epoch": 7.75,
2883
+ "grad_norm": 4.222201824188232,
2884
+ "learning_rate": 4.497363863767559e-06,
2885
+ "loss": 0.4456,
2886
+ "step": 201500
2887
+ },
2888
+ {
2889
+ "epoch": 7.77,
2890
+ "grad_norm": 8.314460754394531,
2891
+ "learning_rate": 4.4588801231479705e-06,
2892
+ "loss": 0.4297,
2893
+ "step": 202000
2894
+ },
2895
+ {
2896
+ "epoch": 7.79,
2897
+ "grad_norm": 8.237698554992676,
2898
+ "learning_rate": 4.4203963825283815e-06,
2899
+ "loss": 0.4637,
2900
+ "step": 202500
2901
+ },
2902
+ {
2903
+ "epoch": 7.81,
2904
+ "grad_norm": 7.0175933837890625,
2905
+ "learning_rate": 4.3819126419087934e-06,
2906
+ "loss": 0.4253,
2907
+ "step": 203000
2908
+ },
2909
+ {
2910
+ "epoch": 7.83,
2911
+ "grad_norm": 9.906578063964844,
2912
+ "learning_rate": 4.343428901289205e-06,
2913
+ "loss": 0.4261,
2914
+ "step": 203500
2915
+ },
2916
+ {
2917
+ "epoch": 7.85,
2918
+ "grad_norm": 2.473310947418213,
2919
+ "learning_rate": 4.304945160669617e-06,
2920
+ "loss": 0.4576,
2921
+ "step": 204000
2922
+ },
2923
+ {
2924
+ "epoch": 7.87,
2925
+ "grad_norm": 7.461137771606445,
2926
+ "learning_rate": 4.266461420050029e-06,
2927
+ "loss": 0.458,
2928
+ "step": 204500
2929
+ },
2930
+ {
2931
+ "epoch": 7.89,
2932
+ "grad_norm": 4.642632007598877,
2933
+ "learning_rate": 4.227977679430441e-06,
2934
+ "loss": 0.4477,
2935
+ "step": 205000
2936
+ },
2937
+ {
2938
+ "epoch": 7.91,
2939
+ "grad_norm": 4.437539100646973,
2940
+ "learning_rate": 4.189570906292092e-06,
2941
+ "loss": 0.4441,
2942
+ "step": 205500
2943
+ },
2944
+ {
2945
+ "epoch": 7.93,
2946
+ "grad_norm": 6.345538139343262,
2947
+ "learning_rate": 4.151087165672504e-06,
2948
+ "loss": 0.4549,
2949
+ "step": 206000
2950
+ },
2951
+ {
2952
+ "epoch": 7.95,
2953
+ "grad_norm": 5.00540828704834,
2954
+ "learning_rate": 4.112603425052915e-06,
2955
+ "loss": 0.4563,
2956
+ "step": 206500
2957
+ },
2958
+ {
2959
+ "epoch": 7.97,
2960
+ "grad_norm": 2.3221421241760254,
2961
+ "learning_rate": 4.074119684433327e-06,
2962
+ "loss": 0.4462,
2963
+ "step": 207000
2964
+ },
2965
+ {
2966
+ "epoch": 7.99,
2967
+ "grad_norm": 4.091787815093994,
2968
+ "learning_rate": 4.0357129112949785e-06,
2969
+ "loss": 0.4588,
2970
+ "step": 207500
2971
+ },
2972
+ {
2973
+ "epoch": 8.0,
2974
+ "eval_loss": 1.752930998802185,
2975
+ "eval_runtime": 11.2624,
2976
+ "eval_samples_per_second": 353.477,
2977
+ "eval_steps_per_second": 88.436,
2978
+ "step": 207880
2979
+ },
2980
+ {
2981
+ "epoch": 8.0,
2982
+ "grad_norm": 10.46442985534668,
2983
+ "learning_rate": 3.99722917067539e-06,
2984
+ "loss": 0.4772,
2985
+ "step": 208000
2986
+ },
2987
+ {
2988
+ "epoch": 8.02,
2989
+ "grad_norm": 2.7908215522766113,
2990
+ "learning_rate": 3.958745430055801e-06,
2991
+ "loss": 0.4165,
2992
+ "step": 208500
2993
+ },
2994
+ {
2995
+ "epoch": 8.04,
2996
+ "grad_norm": 7.39349365234375,
2997
+ "learning_rate": 3.920261689436213e-06,
2998
+ "loss": 0.4127,
2999
+ "step": 209000
3000
+ },
3001
+ {
3002
+ "epoch": 8.06,
3003
+ "grad_norm": 7.740577697753906,
3004
+ "learning_rate": 3.881777948816625e-06,
3005
+ "loss": 0.4235,
3006
+ "step": 209500
3007
+ },
3008
+ {
3009
+ "epoch": 8.08,
3010
+ "grad_norm": 3.9219789505004883,
3011
+ "learning_rate": 3.843294208197037e-06,
3012
+ "loss": 0.4272,
3013
+ "step": 210000
3014
+ },
3015
+ {
3016
+ "epoch": 8.1,
3017
+ "grad_norm": 6.718163967132568,
3018
+ "learning_rate": 3.804887435058688e-06,
3019
+ "loss": 0.4062,
3020
+ "step": 210500
3021
+ },
3022
+ {
3023
+ "epoch": 8.12,
3024
+ "grad_norm": 4.927985191345215,
3025
+ "learning_rate": 3.7664036944391e-06,
3026
+ "loss": 0.4,
3027
+ "step": 211000
3028
+ },
3029
+ {
3030
+ "epoch": 8.14,
3031
+ "grad_norm": 6.605733871459961,
3032
+ "learning_rate": 3.7279199538195113e-06,
3033
+ "loss": 0.4224,
3034
+ "step": 211500
3035
+ },
3036
+ {
3037
+ "epoch": 8.16,
3038
+ "grad_norm": 5.900667190551758,
3039
+ "learning_rate": 3.6894362131999232e-06,
3040
+ "loss": 0.401,
3041
+ "step": 212000
3042
+ },
3043
+ {
3044
+ "epoch": 8.18,
3045
+ "grad_norm": 3.047395706176758,
3046
+ "learning_rate": 3.651029440061574e-06,
3047
+ "loss": 0.4099,
3048
+ "step": 212500
3049
+ },
3050
+ {
3051
+ "epoch": 8.2,
3052
+ "grad_norm": 10.504358291625977,
3053
+ "learning_rate": 3.612545699441986e-06,
3054
+ "loss": 0.4094,
3055
+ "step": 213000
3056
+ },
3057
+ {
3058
+ "epoch": 8.22,
3059
+ "grad_norm": 2.9013960361480713,
3060
+ "learning_rate": 3.574061958822398e-06,
3061
+ "loss": 0.4225,
3062
+ "step": 213500
3063
+ },
3064
+ {
3065
+ "epoch": 8.24,
3066
+ "grad_norm": 1.0684086084365845,
3067
+ "learning_rate": 3.53557821820281e-06,
3068
+ "loss": 0.4333,
3069
+ "step": 214000
3070
+ },
3071
+ {
3072
+ "epoch": 8.25,
3073
+ "grad_norm": 1.7722007036209106,
3074
+ "learning_rate": 3.4970944775832217e-06,
3075
+ "loss": 0.419,
3076
+ "step": 214500
3077
+ },
3078
+ {
3079
+ "epoch": 8.27,
3080
+ "grad_norm": 5.168225288391113,
3081
+ "learning_rate": 3.4586107369636327e-06,
3082
+ "loss": 0.429,
3083
+ "step": 215000
3084
+ },
3085
+ {
3086
+ "epoch": 8.29,
3087
+ "grad_norm": 9.540655136108398,
3088
+ "learning_rate": 3.4201269963440446e-06,
3089
+ "loss": 0.4024,
3090
+ "step": 215500
3091
+ },
3092
+ {
3093
+ "epoch": 8.31,
3094
+ "grad_norm": 7.926085472106934,
3095
+ "learning_rate": 3.3817202232056955e-06,
3096
+ "loss": 0.4081,
3097
+ "step": 216000
3098
+ },
3099
+ {
3100
+ "epoch": 8.33,
3101
+ "grad_norm": 3.9807188510894775,
3102
+ "learning_rate": 3.3432364825861074e-06,
3103
+ "loss": 0.4098,
3104
+ "step": 216500
3105
+ },
3106
+ {
3107
+ "epoch": 8.35,
3108
+ "grad_norm": 4.447396755218506,
3109
+ "learning_rate": 3.3047527419665193e-06,
3110
+ "loss": 0.4045,
3111
+ "step": 217000
3112
+ },
3113
+ {
3114
+ "epoch": 8.37,
3115
+ "grad_norm": 5.585531711578369,
3116
+ "learning_rate": 3.266269001346931e-06,
3117
+ "loss": 0.409,
3118
+ "step": 217500
3119
+ },
3120
+ {
3121
+ "epoch": 8.39,
3122
+ "grad_norm": 4.316734790802002,
3123
+ "learning_rate": 3.227862228208582e-06,
3124
+ "loss": 0.4239,
3125
+ "step": 218000
3126
+ },
3127
+ {
3128
+ "epoch": 8.41,
3129
+ "grad_norm": 7.370926380157471,
3130
+ "learning_rate": 3.189378487588994e-06,
3131
+ "loss": 0.4104,
3132
+ "step": 218500
3133
+ },
3134
+ {
3135
+ "epoch": 8.43,
3136
+ "grad_norm": 6.9890522956848145,
3137
+ "learning_rate": 3.1508947469694055e-06,
3138
+ "loss": 0.4024,
3139
+ "step": 219000
3140
+ },
3141
+ {
3142
+ "epoch": 8.45,
3143
+ "grad_norm": 5.685979843139648,
3144
+ "learning_rate": 3.1124110063498174e-06,
3145
+ "loss": 0.4167,
3146
+ "step": 219500
3147
+ },
3148
+ {
3149
+ "epoch": 8.47,
3150
+ "grad_norm": 5.407013416290283,
3151
+ "learning_rate": 3.0739272657302293e-06,
3152
+ "loss": 0.3967,
3153
+ "step": 220000
3154
+ },
3155
+ {
3156
+ "epoch": 8.49,
3157
+ "grad_norm": 3.3302626609802246,
3158
+ "learning_rate": 3.03552049259188e-06,
3159
+ "loss": 0.4303,
3160
+ "step": 220500
3161
+ },
3162
+ {
3163
+ "epoch": 8.5,
3164
+ "grad_norm": 9.727193832397461,
3165
+ "learning_rate": 2.997113719453531e-06,
3166
+ "loss": 0.4108,
3167
+ "step": 221000
3168
+ },
3169
+ {
3170
+ "epoch": 8.52,
3171
+ "grad_norm": 5.004329681396484,
3172
+ "learning_rate": 2.958629978833943e-06,
3173
+ "loss": 0.4135,
3174
+ "step": 221500
3175
+ },
3176
+ {
3177
+ "epoch": 8.54,
3178
+ "grad_norm": 5.899045467376709,
3179
+ "learning_rate": 2.920146238214355e-06,
3180
+ "loss": 0.4214,
3181
+ "step": 222000
3182
+ },
3183
+ {
3184
+ "epoch": 8.56,
3185
+ "grad_norm": 9.472413063049316,
3186
+ "learning_rate": 2.8816624975947667e-06,
3187
+ "loss": 0.4063,
3188
+ "step": 222500
3189
+ },
3190
+ {
3191
+ "epoch": 8.58,
3192
+ "grad_norm": 3.2998526096343994,
3193
+ "learning_rate": 2.8431787569751786e-06,
3194
+ "loss": 0.4152,
3195
+ "step": 223000
3196
+ },
3197
+ {
3198
+ "epoch": 8.6,
3199
+ "grad_norm": 9.45439338684082,
3200
+ "learning_rate": 2.8046950163555897e-06,
3201
+ "loss": 0.4156,
3202
+ "step": 223500
3203
+ },
3204
+ {
3205
+ "epoch": 8.62,
3206
+ "grad_norm": 10.916293144226074,
3207
+ "learning_rate": 2.7662112757360016e-06,
3208
+ "loss": 0.4177,
3209
+ "step": 224000
3210
+ },
3211
+ {
3212
+ "epoch": 8.64,
3213
+ "grad_norm": 9.067635536193848,
3214
+ "learning_rate": 2.7277275351164134e-06,
3215
+ "loss": 0.4107,
3216
+ "step": 224500
3217
+ },
3218
+ {
3219
+ "epoch": 8.66,
3220
+ "grad_norm": 8.393898010253906,
3221
+ "learning_rate": 2.6892437944968253e-06,
3222
+ "loss": 0.4167,
3223
+ "step": 225000
3224
+ },
3225
+ {
3226
+ "epoch": 8.68,
3227
+ "grad_norm": 4.151984691619873,
3228
+ "learning_rate": 2.6508370213584762e-06,
3229
+ "loss": 0.415,
3230
+ "step": 225500
3231
+ },
3232
+ {
3233
+ "epoch": 8.7,
3234
+ "grad_norm": 5.011842250823975,
3235
+ "learning_rate": 2.612353280738888e-06,
3236
+ "loss": 0.4358,
3237
+ "step": 226000
3238
+ },
3239
+ {
3240
+ "epoch": 8.72,
3241
+ "grad_norm": 2.5369720458984375,
3242
+ "learning_rate": 2.5738695401192996e-06,
3243
+ "loss": 0.4114,
3244
+ "step": 226500
3245
+ },
3246
+ {
3247
+ "epoch": 8.74,
3248
+ "grad_norm": 6.167263507843018,
3249
+ "learning_rate": 2.5353857994997115e-06,
3250
+ "loss": 0.4117,
3251
+ "step": 227000
3252
+ },
3253
+ {
3254
+ "epoch": 8.76,
3255
+ "grad_norm": 3.8160266876220703,
3256
+ "learning_rate": 2.4969020588801234e-06,
3257
+ "loss": 0.4195,
3258
+ "step": 227500
3259
+ },
3260
+ {
3261
+ "epoch": 8.77,
3262
+ "grad_norm": 6.264139175415039,
3263
+ "learning_rate": 2.4584952857417743e-06,
3264
+ "loss": 0.4182,
3265
+ "step": 228000
3266
+ },
3267
+ {
3268
+ "epoch": 8.79,
3269
+ "grad_norm": 2.6473159790039062,
3270
+ "learning_rate": 2.420011545122186e-06,
3271
+ "loss": 0.4109,
3272
+ "step": 228500
3273
+ },
3274
+ {
3275
+ "epoch": 8.81,
3276
+ "grad_norm": 5.942705154418945,
3277
+ "learning_rate": 2.3815278045025976e-06,
3278
+ "loss": 0.4385,
3279
+ "step": 229000
3280
+ },
3281
+ {
3282
+ "epoch": 8.83,
3283
+ "grad_norm": 6.846287250518799,
3284
+ "learning_rate": 2.3430440638830095e-06,
3285
+ "loss": 0.4406,
3286
+ "step": 229500
3287
+ },
3288
+ {
3289
+ "epoch": 8.85,
3290
+ "grad_norm": 3.9785611629486084,
3291
+ "learning_rate": 2.3045603232634214e-06,
3292
+ "loss": 0.4154,
3293
+ "step": 230000
3294
+ },
3295
+ {
3296
+ "epoch": 8.87,
3297
+ "grad_norm": 4.567147254943848,
3298
+ "learning_rate": 2.2661535501250723e-06,
3299
+ "loss": 0.4214,
3300
+ "step": 230500
3301
+ },
3302
+ {
3303
+ "epoch": 8.89,
3304
+ "grad_norm": 7.9429240226745605,
3305
+ "learning_rate": 2.227669809505484e-06,
3306
+ "loss": 0.446,
3307
+ "step": 231000
3308
+ },
3309
+ {
3310
+ "epoch": 8.91,
3311
+ "grad_norm": 6.396302223205566,
3312
+ "learning_rate": 2.189186068885896e-06,
3313
+ "loss": 0.4306,
3314
+ "step": 231500
3315
+ },
3316
+ {
3317
+ "epoch": 8.93,
3318
+ "grad_norm": 7.975338935852051,
3319
+ "learning_rate": 2.1507023282663076e-06,
3320
+ "loss": 0.419,
3321
+ "step": 232000
3322
+ },
3323
+ {
3324
+ "epoch": 8.95,
3325
+ "grad_norm": 6.531688690185547,
3326
+ "learning_rate": 2.1122955551279585e-06,
3327
+ "loss": 0.4144,
3328
+ "step": 232500
3329
+ },
3330
+ {
3331
+ "epoch": 8.97,
3332
+ "grad_norm": 4.969785213470459,
3333
+ "learning_rate": 2.0738118145083704e-06,
3334
+ "loss": 0.4283,
3335
+ "step": 233000
3336
+ },
3337
+ {
3338
+ "epoch": 8.99,
3339
+ "grad_norm": 7.7668538093566895,
3340
+ "learning_rate": 2.0353280738887822e-06,
3341
+ "loss": 0.4025,
3342
+ "step": 233500
3343
+ },
3344
+ {
3345
+ "epoch": 9.0,
3346
+ "eval_loss": 1.7583389282226562,
3347
+ "eval_runtime": 11.3764,
3348
+ "eval_samples_per_second": 349.935,
3349
+ "eval_steps_per_second": 87.55,
3350
+ "step": 233865
3351
+ }
3352
+ ],
3353
+ "logging_steps": 500,
3354
+ "max_steps": 259850,
3355
+ "num_input_tokens_seen": 0,
3356
+ "num_train_epochs": 10,
3357
+ "save_steps": 500,
3358
+ "total_flos": 9268381298589696.0,
3359
+ "train_batch_size": 4,
3360
+ "trial_name": null,
3361
+ "trial_params": null
3362
+ }
checkpoint-233865/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f10481b97b85db779f30bdfe4858bc0eccd200747cd750186b7f3c88a54eeb56
3
+ size 4667
checkpoint-233865/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-25985/config.json ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-51970",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "swish",
5
+ "add_bias_logits": false,
6
+ "add_final_layer_norm": false,
7
+ "architectures": [
8
+ "MarianMTModel"
9
+ ],
10
+ "attention_dropout": 0.0,
11
+ "bad_words_ids": [
12
+ [
13
+ 65000
14
+ ]
15
+ ],
16
+ "bos_token_id": 0,
17
+ "classif_dropout": 0.0,
18
+ "classifier_dropout": 0.0,
19
+ "d_model": 512,
20
+ "decoder_attention_heads": 8,
21
+ "decoder_ffn_dim": 2048,
22
+ "decoder_layerdrop": 0.0,
23
+ "decoder_layers": 6,
24
+ "decoder_start_token_id": 65000,
25
+ "decoder_vocab_size": 65001,
26
+ "dropout": 0.1,
27
+ "encoder_attention_heads": 8,
28
+ "encoder_ffn_dim": 2048,
29
+ "encoder_layerdrop": 0.0,
30
+ "encoder_layers": 6,
31
+ "eos_token_id": 0,
32
+ "extra_pos_embeddings": 65001,
33
+ "forced_eos_token_id": 0,
34
+ "id2label": {
35
+ "0": "LABEL_0",
36
+ "1": "LABEL_1",
37
+ "2": "LABEL_2"
38
+ },
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "label2id": {
42
+ "LABEL_0": 0,
43
+ "LABEL_1": 1,
44
+ "LABEL_2": 2
45
+ },
46
+ "max_length": 512,
47
+ "max_position_embeddings": 512,
48
+ "model_type": "marian",
49
+ "normalize_before": false,
50
+ "normalize_embedding": false,
51
+ "num_beams": 6,
52
+ "num_hidden_layers": 6,
53
+ "pad_token_id": 65000,
54
+ "scale_embedding": true,
55
+ "share_encoder_decoder_embeddings": true,
56
+ "static_position_embeddings": true,
57
+ "torch_dtype": "float32",
58
+ "transformers_version": "4.39.3",
59
+ "use_cache": true,
60
+ "vocab_size": 65001
61
+ }
checkpoint-25985/generation_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bad_words_ids": [
3
+ [
4
+ 65000
5
+ ]
6
+ ],
7
+ "bos_token_id": 0,
8
+ "decoder_start_token_id": 65000,
9
+ "eos_token_id": 0,
10
+ "forced_eos_token_id": 0,
11
+ "max_length": 512,
12
+ "num_beams": 6,
13
+ "pad_token_id": 65000,
14
+ "renormalize_logits": true,
15
+ "transformers_version": "4.39.3"
16
+ }
checkpoint-25985/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca3579c434713fc47d13e38d92ff2f0470237dc5bdcf686cdd8e18a6dc4ead80
3
+ size 309965092
checkpoint-25985/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99cc9db327b05d97345edc26269bf4df4bf129b0d21d0e8812ba41511bfe45f7
3
+ size 619563141
checkpoint-25985/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:00be5de164b6a7df1a41edb9efd3bb49283f4a66cd75b61751be54dbe0251cf8
3
+ size 14575
checkpoint-25985/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18a26713a8cdd617f1ed3254573655f777e06f9bc58b216d58b535b5dda60056
3
+ size 627
checkpoint-25985/source.spm ADDED
Binary file (805 kB). View file
 
checkpoint-25985/special_tokens_map.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "eos_token": "</s>",
3
+ "pad_token": "<pad>",
4
+ "unk_token": "<unk>"
5
+ }
checkpoint-25985/target.spm ADDED
Binary file (807 kB). View file
 
checkpoint-25985/tokenizer_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "</s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<unk>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "65000": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ }
27
+ },
28
+ "clean_up_tokenization_spaces": true,
29
+ "eos_token": "</s>",
30
+ "model_max_length": 512,
31
+ "pad_token": "<pad>",
32
+ "separate_vocabs": false,
33
+ "source_lang": "zho",
34
+ "sp_model_kwargs": {},
35
+ "target_lang": "eng",
36
+ "tokenizer_class": "MarianTokenizer",
37
+ "unk_token": "<unk>"
38
+ }
checkpoint-25985/trainer_state.json ADDED
@@ -0,0 +1,386 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.5939216613769531,
3
+ "best_model_checkpoint": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-25985",
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 25985,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "grad_norm": 2.777989149093628,
14
+ "learning_rate": 1.9961593226861654e-05,
15
+ "loss": 0.608,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.04,
20
+ "grad_norm": 2.571350336074829,
21
+ "learning_rate": 1.9923109486242066e-05,
22
+ "loss": 0.6031,
23
+ "step": 1000
24
+ },
25
+ {
26
+ "epoch": 0.06,
27
+ "grad_norm": 6.669718265533447,
28
+ "learning_rate": 1.9884625745622478e-05,
29
+ "loss": 0.5823,
30
+ "step": 1500
31
+ },
32
+ {
33
+ "epoch": 0.08,
34
+ "grad_norm": 3.766803503036499,
35
+ "learning_rate": 1.984614200500289e-05,
36
+ "loss": 0.625,
37
+ "step": 2000
38
+ },
39
+ {
40
+ "epoch": 0.1,
41
+ "grad_norm": 2.617687225341797,
42
+ "learning_rate": 1.9807735231864538e-05,
43
+ "loss": 0.6324,
44
+ "step": 2500
45
+ },
46
+ {
47
+ "epoch": 0.12,
48
+ "grad_norm": 4.26535177230835,
49
+ "learning_rate": 1.976932845872619e-05,
50
+ "loss": 0.6184,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 0.13,
55
+ "grad_norm": 5.035365581512451,
56
+ "learning_rate": 1.9730844718106602e-05,
57
+ "loss": 0.6277,
58
+ "step": 3500
59
+ },
60
+ {
61
+ "epoch": 0.15,
62
+ "grad_norm": 6.626140594482422,
63
+ "learning_rate": 1.9692360977487014e-05,
64
+ "loss": 0.6557,
65
+ "step": 4000
66
+ },
67
+ {
68
+ "epoch": 0.17,
69
+ "grad_norm": 3.718008279800415,
70
+ "learning_rate": 1.9653877236867426e-05,
71
+ "loss": 0.6623,
72
+ "step": 4500
73
+ },
74
+ {
75
+ "epoch": 0.19,
76
+ "grad_norm": 4.26861047744751,
77
+ "learning_rate": 1.9615470463729075e-05,
78
+ "loss": 0.6719,
79
+ "step": 5000
80
+ },
81
+ {
82
+ "epoch": 0.21,
83
+ "grad_norm": 5.266055107116699,
84
+ "learning_rate": 1.9576986723109487e-05,
85
+ "loss": 0.636,
86
+ "step": 5500
87
+ },
88
+ {
89
+ "epoch": 0.23,
90
+ "grad_norm": 6.086512565612793,
91
+ "learning_rate": 1.95385029824899e-05,
92
+ "loss": 0.6585,
93
+ "step": 6000
94
+ },
95
+ {
96
+ "epoch": 0.25,
97
+ "grad_norm": 4.0940327644348145,
98
+ "learning_rate": 1.950001924187031e-05,
99
+ "loss": 0.6341,
100
+ "step": 6500
101
+ },
102
+ {
103
+ "epoch": 0.27,
104
+ "grad_norm": 7.52503776550293,
105
+ "learning_rate": 1.9461535501250722e-05,
106
+ "loss": 0.6589,
107
+ "step": 7000
108
+ },
109
+ {
110
+ "epoch": 0.29,
111
+ "grad_norm": 3.5400032997131348,
112
+ "learning_rate": 1.9423051760631134e-05,
113
+ "loss": 0.6627,
114
+ "step": 7500
115
+ },
116
+ {
117
+ "epoch": 0.31,
118
+ "grad_norm": 3.525592565536499,
119
+ "learning_rate": 1.9384568020011546e-05,
120
+ "loss": 0.639,
121
+ "step": 8000
122
+ },
123
+ {
124
+ "epoch": 0.33,
125
+ "grad_norm": 3.1940386295318604,
126
+ "learning_rate": 1.9346084279391958e-05,
127
+ "loss": 0.6433,
128
+ "step": 8500
129
+ },
130
+ {
131
+ "epoch": 0.35,
132
+ "grad_norm": 5.362303733825684,
133
+ "learning_rate": 1.930767750625361e-05,
134
+ "loss": 0.6714,
135
+ "step": 9000
136
+ },
137
+ {
138
+ "epoch": 0.37,
139
+ "grad_norm": 4.885318279266357,
140
+ "learning_rate": 1.9269193765634022e-05,
141
+ "loss": 0.6334,
142
+ "step": 9500
143
+ },
144
+ {
145
+ "epoch": 0.38,
146
+ "grad_norm": 7.104074001312256,
147
+ "learning_rate": 1.9230710025014434e-05,
148
+ "loss": 0.6496,
149
+ "step": 10000
150
+ },
151
+ {
152
+ "epoch": 0.4,
153
+ "grad_norm": 5.086849689483643,
154
+ "learning_rate": 1.9192226284394843e-05,
155
+ "loss": 0.6647,
156
+ "step": 10500
157
+ },
158
+ {
159
+ "epoch": 0.42,
160
+ "grad_norm": 10.493572235107422,
161
+ "learning_rate": 1.9153819511256495e-05,
162
+ "loss": 0.6609,
163
+ "step": 11000
164
+ },
165
+ {
166
+ "epoch": 0.44,
167
+ "grad_norm": 10.206048011779785,
168
+ "learning_rate": 1.9115335770636907e-05,
169
+ "loss": 0.6806,
170
+ "step": 11500
171
+ },
172
+ {
173
+ "epoch": 0.46,
174
+ "grad_norm": 9.489506721496582,
175
+ "learning_rate": 1.907685203001732e-05,
176
+ "loss": 0.641,
177
+ "step": 12000
178
+ },
179
+ {
180
+ "epoch": 0.48,
181
+ "grad_norm": 4.525002479553223,
182
+ "learning_rate": 1.903844525687897e-05,
183
+ "loss": 0.6863,
184
+ "step": 12500
185
+ },
186
+ {
187
+ "epoch": 0.5,
188
+ "grad_norm": 7.507730007171631,
189
+ "learning_rate": 1.9000038483740623e-05,
190
+ "loss": 0.6829,
191
+ "step": 13000
192
+ },
193
+ {
194
+ "epoch": 0.52,
195
+ "grad_norm": 4.827548980712891,
196
+ "learning_rate": 1.8961554743121035e-05,
197
+ "loss": 0.6597,
198
+ "step": 13500
199
+ },
200
+ {
201
+ "epoch": 0.54,
202
+ "grad_norm": 6.409604072570801,
203
+ "learning_rate": 1.8923071002501447e-05,
204
+ "loss": 0.6445,
205
+ "step": 14000
206
+ },
207
+ {
208
+ "epoch": 0.56,
209
+ "grad_norm": 9.57291316986084,
210
+ "learning_rate": 1.8884664229363095e-05,
211
+ "loss": 0.6882,
212
+ "step": 14500
213
+ },
214
+ {
215
+ "epoch": 0.58,
216
+ "grad_norm": 9.129105567932129,
217
+ "learning_rate": 1.8846180488743507e-05,
218
+ "loss": 0.6719,
219
+ "step": 15000
220
+ },
221
+ {
222
+ "epoch": 0.6,
223
+ "grad_norm": 8.071757316589355,
224
+ "learning_rate": 1.880769674812392e-05,
225
+ "loss": 0.6668,
226
+ "step": 15500
227
+ },
228
+ {
229
+ "epoch": 0.62,
230
+ "grad_norm": 3.636784791946411,
231
+ "learning_rate": 1.876921300750433e-05,
232
+ "loss": 0.6773,
233
+ "step": 16000
234
+ },
235
+ {
236
+ "epoch": 0.63,
237
+ "grad_norm": 6.6433515548706055,
238
+ "learning_rate": 1.8730729266884743e-05,
239
+ "loss": 0.6776,
240
+ "step": 16500
241
+ },
242
+ {
243
+ "epoch": 0.65,
244
+ "grad_norm": 3.0176546573638916,
245
+ "learning_rate": 1.8692245526265155e-05,
246
+ "loss": 0.6587,
247
+ "step": 17000
248
+ },
249
+ {
250
+ "epoch": 0.67,
251
+ "grad_norm": 7.913615703582764,
252
+ "learning_rate": 1.8653761785645567e-05,
253
+ "loss": 0.6665,
254
+ "step": 17500
255
+ },
256
+ {
257
+ "epoch": 0.69,
258
+ "grad_norm": 5.335203647613525,
259
+ "learning_rate": 1.861527804502598e-05,
260
+ "loss": 0.6684,
261
+ "step": 18000
262
+ },
263
+ {
264
+ "epoch": 0.71,
265
+ "grad_norm": 5.51749324798584,
266
+ "learning_rate": 1.857679430440639e-05,
267
+ "loss": 0.6577,
268
+ "step": 18500
269
+ },
270
+ {
271
+ "epoch": 0.73,
272
+ "grad_norm": 4.254025936126709,
273
+ "learning_rate": 1.8538310563786802e-05,
274
+ "loss": 0.6776,
275
+ "step": 19000
276
+ },
277
+ {
278
+ "epoch": 0.75,
279
+ "grad_norm": 3.3869874477386475,
280
+ "learning_rate": 1.8499826823167214e-05,
281
+ "loss": 0.6351,
282
+ "step": 19500
283
+ },
284
+ {
285
+ "epoch": 0.77,
286
+ "grad_norm": 12.704727172851562,
287
+ "learning_rate": 1.8461343082547626e-05,
288
+ "loss": 0.6904,
289
+ "step": 20000
290
+ },
291
+ {
292
+ "epoch": 0.79,
293
+ "grad_norm": 3.478996515274048,
294
+ "learning_rate": 1.8422859341928038e-05,
295
+ "loss": 0.6829,
296
+ "step": 20500
297
+ },
298
+ {
299
+ "epoch": 0.81,
300
+ "grad_norm": 6.513260841369629,
301
+ "learning_rate": 1.838437560130845e-05,
302
+ "loss": 0.6615,
303
+ "step": 21000
304
+ },
305
+ {
306
+ "epoch": 0.83,
307
+ "grad_norm": 5.242546081542969,
308
+ "learning_rate": 1.8345891860688862e-05,
309
+ "loss": 0.6859,
310
+ "step": 21500
311
+ },
312
+ {
313
+ "epoch": 0.85,
314
+ "grad_norm": 6.638239860534668,
315
+ "learning_rate": 1.8307408120069274e-05,
316
+ "loss": 0.6485,
317
+ "step": 22000
318
+ },
319
+ {
320
+ "epoch": 0.87,
321
+ "grad_norm": 8.32187557220459,
322
+ "learning_rate": 1.8269001346930923e-05,
323
+ "loss": 0.6567,
324
+ "step": 22500
325
+ },
326
+ {
327
+ "epoch": 0.89,
328
+ "grad_norm": 3.9929823875427246,
329
+ "learning_rate": 1.8230517606311334e-05,
330
+ "loss": 0.6996,
331
+ "step": 23000
332
+ },
333
+ {
334
+ "epoch": 0.9,
335
+ "grad_norm": 5.426717281341553,
336
+ "learning_rate": 1.8192110833172987e-05,
337
+ "loss": 0.6529,
338
+ "step": 23500
339
+ },
340
+ {
341
+ "epoch": 0.92,
342
+ "grad_norm": 7.328232288360596,
343
+ "learning_rate": 1.81536270925534e-05,
344
+ "loss": 0.6711,
345
+ "step": 24000
346
+ },
347
+ {
348
+ "epoch": 0.94,
349
+ "grad_norm": 6.648090362548828,
350
+ "learning_rate": 1.811514335193381e-05,
351
+ "loss": 0.7026,
352
+ "step": 24500
353
+ },
354
+ {
355
+ "epoch": 0.96,
356
+ "grad_norm": 4.047011375427246,
357
+ "learning_rate": 1.8076659611314222e-05,
358
+ "loss": 0.6867,
359
+ "step": 25000
360
+ },
361
+ {
362
+ "epoch": 0.98,
363
+ "grad_norm": 10.354930877685547,
364
+ "learning_rate": 1.803825283817587e-05,
365
+ "loss": 0.6718,
366
+ "step": 25500
367
+ },
368
+ {
369
+ "epoch": 1.0,
370
+ "eval_loss": 1.5939216613769531,
371
+ "eval_runtime": 11.3331,
372
+ "eval_samples_per_second": 351.271,
373
+ "eval_steps_per_second": 87.884,
374
+ "step": 25985
375
+ }
376
+ ],
377
+ "logging_steps": 500,
378
+ "max_steps": 259850,
379
+ "num_input_tokens_seen": 0,
380
+ "num_train_epochs": 10,
381
+ "save_steps": 500,
382
+ "total_flos": 1029890059665408.0,
383
+ "train_batch_size": 4,
384
+ "trial_name": null,
385
+ "trial_params": null
386
+ }
checkpoint-25985/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f10481b97b85db779f30bdfe4858bc0eccd200747cd750186b7f3c88a54eeb56
3
+ size 4667
checkpoint-25985/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-259850/config.json ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-51970",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "swish",
5
+ "add_bias_logits": false,
6
+ "add_final_layer_norm": false,
7
+ "architectures": [
8
+ "MarianMTModel"
9
+ ],
10
+ "attention_dropout": 0.0,
11
+ "bad_words_ids": [
12
+ [
13
+ 65000
14
+ ]
15
+ ],
16
+ "bos_token_id": 0,
17
+ "classif_dropout": 0.0,
18
+ "classifier_dropout": 0.0,
19
+ "d_model": 512,
20
+ "decoder_attention_heads": 8,
21
+ "decoder_ffn_dim": 2048,
22
+ "decoder_layerdrop": 0.0,
23
+ "decoder_layers": 6,
24
+ "decoder_start_token_id": 65000,
25
+ "decoder_vocab_size": 65001,
26
+ "dropout": 0.1,
27
+ "encoder_attention_heads": 8,
28
+ "encoder_ffn_dim": 2048,
29
+ "encoder_layerdrop": 0.0,
30
+ "encoder_layers": 6,
31
+ "eos_token_id": 0,
32
+ "extra_pos_embeddings": 65001,
33
+ "forced_eos_token_id": 0,
34
+ "id2label": {
35
+ "0": "LABEL_0",
36
+ "1": "LABEL_1",
37
+ "2": "LABEL_2"
38
+ },
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "label2id": {
42
+ "LABEL_0": 0,
43
+ "LABEL_1": 1,
44
+ "LABEL_2": 2
45
+ },
46
+ "max_length": 512,
47
+ "max_position_embeddings": 512,
48
+ "model_type": "marian",
49
+ "normalize_before": false,
50
+ "normalize_embedding": false,
51
+ "num_beams": 6,
52
+ "num_hidden_layers": 6,
53
+ "pad_token_id": 65000,
54
+ "scale_embedding": true,
55
+ "share_encoder_decoder_embeddings": true,
56
+ "static_position_embeddings": true,
57
+ "torch_dtype": "float32",
58
+ "transformers_version": "4.39.3",
59
+ "use_cache": true,
60
+ "vocab_size": 65001
61
+ }
checkpoint-259850/generation_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bad_words_ids": [
3
+ [
4
+ 65000
5
+ ]
6
+ ],
7
+ "bos_token_id": 0,
8
+ "decoder_start_token_id": 65000,
9
+ "eos_token_id": 0,
10
+ "forced_eos_token_id": 0,
11
+ "max_length": 512,
12
+ "num_beams": 6,
13
+ "pad_token_id": 65000,
14
+ "renormalize_logits": true,
15
+ "transformers_version": "4.39.3"
16
+ }
checkpoint-259850/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7d773ce339040f4b4c9f280b61c270e7169d31a0275d009ff8d548c3baf7dbf
3
+ size 309965092
checkpoint-259850/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21a179fd3cc782ae0d17eb267ae7240e3a3558c7e4759d31a665d97d6d9c9cdc
3
+ size 619563141
checkpoint-259850/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:170f28d6c5e6a11c65c90c81bf55e315064ac3556f75a34b8ba7b481f3126019
3
+ size 14575
checkpoint-259850/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdf215eadd633ae255a8266c86f161f50f4278de185282c0cf0c1fb3a1a009c1
3
+ size 627
checkpoint-259850/source.spm ADDED
Binary file (805 kB). View file
 
checkpoint-259850/special_tokens_map.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "eos_token": "</s>",
3
+ "pad_token": "<pad>",
4
+ "unk_token": "<unk>"
5
+ }
checkpoint-259850/target.spm ADDED
Binary file (807 kB). View file
 
checkpoint-259850/tokenizer_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "</s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<unk>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "65000": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ }
27
+ },
28
+ "clean_up_tokenization_spaces": true,
29
+ "eos_token": "</s>",
30
+ "model_max_length": 512,
31
+ "pad_token": "<pad>",
32
+ "separate_vocabs": false,
33
+ "source_lang": "zho",
34
+ "sp_model_kwargs": {},
35
+ "target_lang": "eng",
36
+ "tokenizer_class": "MarianTokenizer",
37
+ "unk_token": "<unk>"
38
+ }
checkpoint-259850/trainer_state.json ADDED
@@ -0,0 +1,3734 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.5939216613769531,
3
+ "best_model_checkpoint": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-25985",
4
+ "epoch": 10.0,
5
+ "eval_steps": 500,
6
+ "global_step": 259850,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "grad_norm": 2.777989149093628,
14
+ "learning_rate": 1.9961593226861654e-05,
15
+ "loss": 0.608,
16
+ "step": 500
17
+ },
18
+ {
19
+ "epoch": 0.04,
20
+ "grad_norm": 2.571350336074829,
21
+ "learning_rate": 1.9923109486242066e-05,
22
+ "loss": 0.6031,
23
+ "step": 1000
24
+ },
25
+ {
26
+ "epoch": 0.06,
27
+ "grad_norm": 6.669718265533447,
28
+ "learning_rate": 1.9884625745622478e-05,
29
+ "loss": 0.5823,
30
+ "step": 1500
31
+ },
32
+ {
33
+ "epoch": 0.08,
34
+ "grad_norm": 3.766803503036499,
35
+ "learning_rate": 1.984614200500289e-05,
36
+ "loss": 0.625,
37
+ "step": 2000
38
+ },
39
+ {
40
+ "epoch": 0.1,
41
+ "grad_norm": 2.617687225341797,
42
+ "learning_rate": 1.9807735231864538e-05,
43
+ "loss": 0.6324,
44
+ "step": 2500
45
+ },
46
+ {
47
+ "epoch": 0.12,
48
+ "grad_norm": 4.26535177230835,
49
+ "learning_rate": 1.976932845872619e-05,
50
+ "loss": 0.6184,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 0.13,
55
+ "grad_norm": 5.035365581512451,
56
+ "learning_rate": 1.9730844718106602e-05,
57
+ "loss": 0.6277,
58
+ "step": 3500
59
+ },
60
+ {
61
+ "epoch": 0.15,
62
+ "grad_norm": 6.626140594482422,
63
+ "learning_rate": 1.9692360977487014e-05,
64
+ "loss": 0.6557,
65
+ "step": 4000
66
+ },
67
+ {
68
+ "epoch": 0.17,
69
+ "grad_norm": 3.718008279800415,
70
+ "learning_rate": 1.9653877236867426e-05,
71
+ "loss": 0.6623,
72
+ "step": 4500
73
+ },
74
+ {
75
+ "epoch": 0.19,
76
+ "grad_norm": 4.26861047744751,
77
+ "learning_rate": 1.9615470463729075e-05,
78
+ "loss": 0.6719,
79
+ "step": 5000
80
+ },
81
+ {
82
+ "epoch": 0.21,
83
+ "grad_norm": 5.266055107116699,
84
+ "learning_rate": 1.9576986723109487e-05,
85
+ "loss": 0.636,
86
+ "step": 5500
87
+ },
88
+ {
89
+ "epoch": 0.23,
90
+ "grad_norm": 6.086512565612793,
91
+ "learning_rate": 1.95385029824899e-05,
92
+ "loss": 0.6585,
93
+ "step": 6000
94
+ },
95
+ {
96
+ "epoch": 0.25,
97
+ "grad_norm": 4.0940327644348145,
98
+ "learning_rate": 1.950001924187031e-05,
99
+ "loss": 0.6341,
100
+ "step": 6500
101
+ },
102
+ {
103
+ "epoch": 0.27,
104
+ "grad_norm": 7.52503776550293,
105
+ "learning_rate": 1.9461535501250722e-05,
106
+ "loss": 0.6589,
107
+ "step": 7000
108
+ },
109
+ {
110
+ "epoch": 0.29,
111
+ "grad_norm": 3.5400032997131348,
112
+ "learning_rate": 1.9423051760631134e-05,
113
+ "loss": 0.6627,
114
+ "step": 7500
115
+ },
116
+ {
117
+ "epoch": 0.31,
118
+ "grad_norm": 3.525592565536499,
119
+ "learning_rate": 1.9384568020011546e-05,
120
+ "loss": 0.639,
121
+ "step": 8000
122
+ },
123
+ {
124
+ "epoch": 0.33,
125
+ "grad_norm": 3.1940386295318604,
126
+ "learning_rate": 1.9346084279391958e-05,
127
+ "loss": 0.6433,
128
+ "step": 8500
129
+ },
130
+ {
131
+ "epoch": 0.35,
132
+ "grad_norm": 5.362303733825684,
133
+ "learning_rate": 1.930767750625361e-05,
134
+ "loss": 0.6714,
135
+ "step": 9000
136
+ },
137
+ {
138
+ "epoch": 0.37,
139
+ "grad_norm": 4.885318279266357,
140
+ "learning_rate": 1.9269193765634022e-05,
141
+ "loss": 0.6334,
142
+ "step": 9500
143
+ },
144
+ {
145
+ "epoch": 0.38,
146
+ "grad_norm": 7.104074001312256,
147
+ "learning_rate": 1.9230710025014434e-05,
148
+ "loss": 0.6496,
149
+ "step": 10000
150
+ },
151
+ {
152
+ "epoch": 0.4,
153
+ "grad_norm": 5.086849689483643,
154
+ "learning_rate": 1.9192226284394843e-05,
155
+ "loss": 0.6647,
156
+ "step": 10500
157
+ },
158
+ {
159
+ "epoch": 0.42,
160
+ "grad_norm": 10.493572235107422,
161
+ "learning_rate": 1.9153819511256495e-05,
162
+ "loss": 0.6609,
163
+ "step": 11000
164
+ },
165
+ {
166
+ "epoch": 0.44,
167
+ "grad_norm": 10.206048011779785,
168
+ "learning_rate": 1.9115335770636907e-05,
169
+ "loss": 0.6806,
170
+ "step": 11500
171
+ },
172
+ {
173
+ "epoch": 0.46,
174
+ "grad_norm": 9.489506721496582,
175
+ "learning_rate": 1.907685203001732e-05,
176
+ "loss": 0.641,
177
+ "step": 12000
178
+ },
179
+ {
180
+ "epoch": 0.48,
181
+ "grad_norm": 4.525002479553223,
182
+ "learning_rate": 1.903844525687897e-05,
183
+ "loss": 0.6863,
184
+ "step": 12500
185
+ },
186
+ {
187
+ "epoch": 0.5,
188
+ "grad_norm": 7.507730007171631,
189
+ "learning_rate": 1.9000038483740623e-05,
190
+ "loss": 0.6829,
191
+ "step": 13000
192
+ },
193
+ {
194
+ "epoch": 0.52,
195
+ "grad_norm": 4.827548980712891,
196
+ "learning_rate": 1.8961554743121035e-05,
197
+ "loss": 0.6597,
198
+ "step": 13500
199
+ },
200
+ {
201
+ "epoch": 0.54,
202
+ "grad_norm": 6.409604072570801,
203
+ "learning_rate": 1.8923071002501447e-05,
204
+ "loss": 0.6445,
205
+ "step": 14000
206
+ },
207
+ {
208
+ "epoch": 0.56,
209
+ "grad_norm": 9.57291316986084,
210
+ "learning_rate": 1.8884664229363095e-05,
211
+ "loss": 0.6882,
212
+ "step": 14500
213
+ },
214
+ {
215
+ "epoch": 0.58,
216
+ "grad_norm": 9.129105567932129,
217
+ "learning_rate": 1.8846180488743507e-05,
218
+ "loss": 0.6719,
219
+ "step": 15000
220
+ },
221
+ {
222
+ "epoch": 0.6,
223
+ "grad_norm": 8.071757316589355,
224
+ "learning_rate": 1.880769674812392e-05,
225
+ "loss": 0.6668,
226
+ "step": 15500
227
+ },
228
+ {
229
+ "epoch": 0.62,
230
+ "grad_norm": 3.636784791946411,
231
+ "learning_rate": 1.876921300750433e-05,
232
+ "loss": 0.6773,
233
+ "step": 16000
234
+ },
235
+ {
236
+ "epoch": 0.63,
237
+ "grad_norm": 6.6433515548706055,
238
+ "learning_rate": 1.8730729266884743e-05,
239
+ "loss": 0.6776,
240
+ "step": 16500
241
+ },
242
+ {
243
+ "epoch": 0.65,
244
+ "grad_norm": 3.0176546573638916,
245
+ "learning_rate": 1.8692245526265155e-05,
246
+ "loss": 0.6587,
247
+ "step": 17000
248
+ },
249
+ {
250
+ "epoch": 0.67,
251
+ "grad_norm": 7.913615703582764,
252
+ "learning_rate": 1.8653761785645567e-05,
253
+ "loss": 0.6665,
254
+ "step": 17500
255
+ },
256
+ {
257
+ "epoch": 0.69,
258
+ "grad_norm": 5.335203647613525,
259
+ "learning_rate": 1.861527804502598e-05,
260
+ "loss": 0.6684,
261
+ "step": 18000
262
+ },
263
+ {
264
+ "epoch": 0.71,
265
+ "grad_norm": 5.51749324798584,
266
+ "learning_rate": 1.857679430440639e-05,
267
+ "loss": 0.6577,
268
+ "step": 18500
269
+ },
270
+ {
271
+ "epoch": 0.73,
272
+ "grad_norm": 4.254025936126709,
273
+ "learning_rate": 1.8538310563786802e-05,
274
+ "loss": 0.6776,
275
+ "step": 19000
276
+ },
277
+ {
278
+ "epoch": 0.75,
279
+ "grad_norm": 3.3869874477386475,
280
+ "learning_rate": 1.8499826823167214e-05,
281
+ "loss": 0.6351,
282
+ "step": 19500
283
+ },
284
+ {
285
+ "epoch": 0.77,
286
+ "grad_norm": 12.704727172851562,
287
+ "learning_rate": 1.8461343082547626e-05,
288
+ "loss": 0.6904,
289
+ "step": 20000
290
+ },
291
+ {
292
+ "epoch": 0.79,
293
+ "grad_norm": 3.478996515274048,
294
+ "learning_rate": 1.8422859341928038e-05,
295
+ "loss": 0.6829,
296
+ "step": 20500
297
+ },
298
+ {
299
+ "epoch": 0.81,
300
+ "grad_norm": 6.513260841369629,
301
+ "learning_rate": 1.838437560130845e-05,
302
+ "loss": 0.6615,
303
+ "step": 21000
304
+ },
305
+ {
306
+ "epoch": 0.83,
307
+ "grad_norm": 5.242546081542969,
308
+ "learning_rate": 1.8345891860688862e-05,
309
+ "loss": 0.6859,
310
+ "step": 21500
311
+ },
312
+ {
313
+ "epoch": 0.85,
314
+ "grad_norm": 6.638239860534668,
315
+ "learning_rate": 1.8307408120069274e-05,
316
+ "loss": 0.6485,
317
+ "step": 22000
318
+ },
319
+ {
320
+ "epoch": 0.87,
321
+ "grad_norm": 8.32187557220459,
322
+ "learning_rate": 1.8269001346930923e-05,
323
+ "loss": 0.6567,
324
+ "step": 22500
325
+ },
326
+ {
327
+ "epoch": 0.89,
328
+ "grad_norm": 3.9929823875427246,
329
+ "learning_rate": 1.8230517606311334e-05,
330
+ "loss": 0.6996,
331
+ "step": 23000
332
+ },
333
+ {
334
+ "epoch": 0.9,
335
+ "grad_norm": 5.426717281341553,
336
+ "learning_rate": 1.8192110833172987e-05,
337
+ "loss": 0.6529,
338
+ "step": 23500
339
+ },
340
+ {
341
+ "epoch": 0.92,
342
+ "grad_norm": 7.328232288360596,
343
+ "learning_rate": 1.81536270925534e-05,
344
+ "loss": 0.6711,
345
+ "step": 24000
346
+ },
347
+ {
348
+ "epoch": 0.94,
349
+ "grad_norm": 6.648090362548828,
350
+ "learning_rate": 1.811514335193381e-05,
351
+ "loss": 0.7026,
352
+ "step": 24500
353
+ },
354
+ {
355
+ "epoch": 0.96,
356
+ "grad_norm": 4.047011375427246,
357
+ "learning_rate": 1.8076659611314222e-05,
358
+ "loss": 0.6867,
359
+ "step": 25000
360
+ },
361
+ {
362
+ "epoch": 0.98,
363
+ "grad_norm": 10.354930877685547,
364
+ "learning_rate": 1.803825283817587e-05,
365
+ "loss": 0.6718,
366
+ "step": 25500
367
+ },
368
+ {
369
+ "epoch": 1.0,
370
+ "eval_loss": 1.5939216613769531,
371
+ "eval_runtime": 11.3331,
372
+ "eval_samples_per_second": 351.271,
373
+ "eval_steps_per_second": 87.884,
374
+ "step": 25985
375
+ },
376
+ {
377
+ "epoch": 1.0,
378
+ "grad_norm": 4.946779251098633,
379
+ "learning_rate": 1.7999846065037523e-05,
380
+ "loss": 0.6928,
381
+ "step": 26000
382
+ },
383
+ {
384
+ "epoch": 1.02,
385
+ "grad_norm": 4.127852916717529,
386
+ "learning_rate": 1.7961362324417935e-05,
387
+ "loss": 0.5471,
388
+ "step": 26500
389
+ },
390
+ {
391
+ "epoch": 1.04,
392
+ "grad_norm": 8.648958206176758,
393
+ "learning_rate": 1.7922878583798347e-05,
394
+ "loss": 0.5216,
395
+ "step": 27000
396
+ },
397
+ {
398
+ "epoch": 1.06,
399
+ "grad_norm": 3.198063611984253,
400
+ "learning_rate": 1.788439484317876e-05,
401
+ "loss": 0.5649,
402
+ "step": 27500
403
+ },
404
+ {
405
+ "epoch": 1.08,
406
+ "grad_norm": 7.709962368011475,
407
+ "learning_rate": 1.784591110255917e-05,
408
+ "loss": 0.5739,
409
+ "step": 28000
410
+ },
411
+ {
412
+ "epoch": 1.1,
413
+ "grad_norm": 3.456827402114868,
414
+ "learning_rate": 1.7807427361939583e-05,
415
+ "loss": 0.549,
416
+ "step": 28500
417
+ },
418
+ {
419
+ "epoch": 1.12,
420
+ "grad_norm": 5.343823432922363,
421
+ "learning_rate": 1.7768943621319995e-05,
422
+ "loss": 0.5624,
423
+ "step": 29000
424
+ },
425
+ {
426
+ "epoch": 1.14,
427
+ "grad_norm": 6.258249759674072,
428
+ "learning_rate": 1.7730459880700407e-05,
429
+ "loss": 0.5497,
430
+ "step": 29500
431
+ },
432
+ {
433
+ "epoch": 1.15,
434
+ "grad_norm": 3.7241768836975098,
435
+ "learning_rate": 1.769197614008082e-05,
436
+ "loss": 0.5422,
437
+ "step": 30000
438
+ },
439
+ {
440
+ "epoch": 1.17,
441
+ "grad_norm": 4.565840244293213,
442
+ "learning_rate": 1.765349239946123e-05,
443
+ "loss": 0.5805,
444
+ "step": 30500
445
+ },
446
+ {
447
+ "epoch": 1.19,
448
+ "grad_norm": 6.31953763961792,
449
+ "learning_rate": 1.7615008658841642e-05,
450
+ "loss": 0.5889,
451
+ "step": 31000
452
+ },
453
+ {
454
+ "epoch": 1.21,
455
+ "grad_norm": 1.6492664813995361,
456
+ "learning_rate": 1.7576524918222054e-05,
457
+ "loss": 0.5732,
458
+ "step": 31500
459
+ },
460
+ {
461
+ "epoch": 1.23,
462
+ "grad_norm": 6.656198978424072,
463
+ "learning_rate": 1.7538041177602466e-05,
464
+ "loss": 0.5499,
465
+ "step": 32000
466
+ },
467
+ {
468
+ "epoch": 1.25,
469
+ "grad_norm": 4.865792274475098,
470
+ "learning_rate": 1.7499557436982878e-05,
471
+ "loss": 0.5519,
472
+ "step": 32500
473
+ },
474
+ {
475
+ "epoch": 1.27,
476
+ "grad_norm": 3.8847908973693848,
477
+ "learning_rate": 1.746107369636329e-05,
478
+ "loss": 0.5695,
479
+ "step": 33000
480
+ },
481
+ {
482
+ "epoch": 1.29,
483
+ "grad_norm": 9.86506462097168,
484
+ "learning_rate": 1.74225899557437e-05,
485
+ "loss": 0.5877,
486
+ "step": 33500
487
+ },
488
+ {
489
+ "epoch": 1.31,
490
+ "grad_norm": 5.732994079589844,
491
+ "learning_rate": 1.738418318260535e-05,
492
+ "loss": 0.5629,
493
+ "step": 34000
494
+ },
495
+ {
496
+ "epoch": 1.33,
497
+ "grad_norm": 2.4092957973480225,
498
+ "learning_rate": 1.7345699441985762e-05,
499
+ "loss": 0.5734,
500
+ "step": 34500
501
+ },
502
+ {
503
+ "epoch": 1.35,
504
+ "grad_norm": 4.129636764526367,
505
+ "learning_rate": 1.7307215701366174e-05,
506
+ "loss": 0.5931,
507
+ "step": 35000
508
+ },
509
+ {
510
+ "epoch": 1.37,
511
+ "grad_norm": 4.32247257232666,
512
+ "learning_rate": 1.7268731960746586e-05,
513
+ "loss": 0.5703,
514
+ "step": 35500
515
+ },
516
+ {
517
+ "epoch": 1.39,
518
+ "grad_norm": 4.780704975128174,
519
+ "learning_rate": 1.7230325187608238e-05,
520
+ "loss": 0.5978,
521
+ "step": 36000
522
+ },
523
+ {
524
+ "epoch": 1.4,
525
+ "grad_norm": 5.10893440246582,
526
+ "learning_rate": 1.719184144698865e-05,
527
+ "loss": 0.57,
528
+ "step": 36500
529
+ },
530
+ {
531
+ "epoch": 1.42,
532
+ "grad_norm": 3.899763584136963,
533
+ "learning_rate": 1.71534346738503e-05,
534
+ "loss": 0.5778,
535
+ "step": 37000
536
+ },
537
+ {
538
+ "epoch": 1.44,
539
+ "grad_norm": 6.0712480545043945,
540
+ "learning_rate": 1.711495093323071e-05,
541
+ "loss": 0.5744,
542
+ "step": 37500
543
+ },
544
+ {
545
+ "epoch": 1.46,
546
+ "grad_norm": 9.512677192687988,
547
+ "learning_rate": 1.7076467192611123e-05,
548
+ "loss": 0.5825,
549
+ "step": 38000
550
+ },
551
+ {
552
+ "epoch": 1.48,
553
+ "grad_norm": 5.049200534820557,
554
+ "learning_rate": 1.7037983451991535e-05,
555
+ "loss": 0.6039,
556
+ "step": 38500
557
+ },
558
+ {
559
+ "epoch": 1.5,
560
+ "grad_norm": 5.021922588348389,
561
+ "learning_rate": 1.6999499711371947e-05,
562
+ "loss": 0.5708,
563
+ "step": 39000
564
+ },
565
+ {
566
+ "epoch": 1.52,
567
+ "grad_norm": 3.546837568283081,
568
+ "learning_rate": 1.696101597075236e-05,
569
+ "loss": 0.5808,
570
+ "step": 39500
571
+ },
572
+ {
573
+ "epoch": 1.54,
574
+ "grad_norm": 4.707235813140869,
575
+ "learning_rate": 1.692253223013277e-05,
576
+ "loss": 0.5894,
577
+ "step": 40000
578
+ },
579
+ {
580
+ "epoch": 1.56,
581
+ "grad_norm": 8.543360710144043,
582
+ "learning_rate": 1.688412545699442e-05,
583
+ "loss": 0.5828,
584
+ "step": 40500
585
+ },
586
+ {
587
+ "epoch": 1.58,
588
+ "grad_norm": 5.515990257263184,
589
+ "learning_rate": 1.684564171637483e-05,
590
+ "loss": 0.6009,
591
+ "step": 41000
592
+ },
593
+ {
594
+ "epoch": 1.6,
595
+ "grad_norm": 8.573718070983887,
596
+ "learning_rate": 1.6807234943236483e-05,
597
+ "loss": 0.5796,
598
+ "step": 41500
599
+ },
600
+ {
601
+ "epoch": 1.62,
602
+ "grad_norm": 3.3873181343078613,
603
+ "learning_rate": 1.6768751202616895e-05,
604
+ "loss": 0.6052,
605
+ "step": 42000
606
+ },
607
+ {
608
+ "epoch": 1.64,
609
+ "grad_norm": 5.747949123382568,
610
+ "learning_rate": 1.6730344429478547e-05,
611
+ "loss": 0.5998,
612
+ "step": 42500
613
+ },
614
+ {
615
+ "epoch": 1.65,
616
+ "grad_norm": 4.767616271972656,
617
+ "learning_rate": 1.669186068885896e-05,
618
+ "loss": 0.631,
619
+ "step": 43000
620
+ },
621
+ {
622
+ "epoch": 1.67,
623
+ "grad_norm": 5.872443199157715,
624
+ "learning_rate": 1.665337694823937e-05,
625
+ "loss": 0.5903,
626
+ "step": 43500
627
+ },
628
+ {
629
+ "epoch": 1.69,
630
+ "grad_norm": 5.379058837890625,
631
+ "learning_rate": 1.6614893207619783e-05,
632
+ "loss": 0.6141,
633
+ "step": 44000
634
+ },
635
+ {
636
+ "epoch": 1.71,
637
+ "grad_norm": 7.023738384246826,
638
+ "learning_rate": 1.6576409467000195e-05,
639
+ "loss": 0.601,
640
+ "step": 44500
641
+ },
642
+ {
643
+ "epoch": 1.73,
644
+ "grad_norm": 7.2762932777404785,
645
+ "learning_rate": 1.6537925726380607e-05,
646
+ "loss": 0.614,
647
+ "step": 45000
648
+ },
649
+ {
650
+ "epoch": 1.75,
651
+ "grad_norm": 5.506533145904541,
652
+ "learning_rate": 1.649944198576102e-05,
653
+ "loss": 0.6156,
654
+ "step": 45500
655
+ },
656
+ {
657
+ "epoch": 1.77,
658
+ "grad_norm": 4.782810688018799,
659
+ "learning_rate": 1.646095824514143e-05,
660
+ "loss": 0.5983,
661
+ "step": 46000
662
+ },
663
+ {
664
+ "epoch": 1.79,
665
+ "grad_norm": 8.239398002624512,
666
+ "learning_rate": 1.6422474504521842e-05,
667
+ "loss": 0.6115,
668
+ "step": 46500
669
+ },
670
+ {
671
+ "epoch": 1.81,
672
+ "grad_norm": 5.413235664367676,
673
+ "learning_rate": 1.6383990763902254e-05,
674
+ "loss": 0.6157,
675
+ "step": 47000
676
+ },
677
+ {
678
+ "epoch": 1.83,
679
+ "grad_norm": 5.1087446212768555,
680
+ "learning_rate": 1.6345507023282666e-05,
681
+ "loss": 0.6225,
682
+ "step": 47500
683
+ },
684
+ {
685
+ "epoch": 1.85,
686
+ "grad_norm": 10.661508560180664,
687
+ "learning_rate": 1.6307023282663078e-05,
688
+ "loss": 0.6185,
689
+ "step": 48000
690
+ },
691
+ {
692
+ "epoch": 1.87,
693
+ "grad_norm": 5.0764312744140625,
694
+ "learning_rate": 1.626853954204349e-05,
695
+ "loss": 0.5878,
696
+ "step": 48500
697
+ },
698
+ {
699
+ "epoch": 1.89,
700
+ "grad_norm": 6.266598224639893,
701
+ "learning_rate": 1.6230055801423902e-05,
702
+ "loss": 0.6057,
703
+ "step": 49000
704
+ },
705
+ {
706
+ "epoch": 1.9,
707
+ "grad_norm": 5.597410678863525,
708
+ "learning_rate": 1.6191572060804314e-05,
709
+ "loss": 0.591,
710
+ "step": 49500
711
+ },
712
+ {
713
+ "epoch": 1.92,
714
+ "grad_norm": 5.9798970222473145,
715
+ "learning_rate": 1.6153088320184722e-05,
716
+ "loss": 0.6114,
717
+ "step": 50000
718
+ },
719
+ {
720
+ "epoch": 1.94,
721
+ "grad_norm": 5.151573181152344,
722
+ "learning_rate": 1.6114604579565134e-05,
723
+ "loss": 0.5875,
724
+ "step": 50500
725
+ },
726
+ {
727
+ "epoch": 1.96,
728
+ "grad_norm": 3.2873849868774414,
729
+ "learning_rate": 1.6076197806426786e-05,
730
+ "loss": 0.6393,
731
+ "step": 51000
732
+ },
733
+ {
734
+ "epoch": 1.98,
735
+ "grad_norm": 3.8436408042907715,
736
+ "learning_rate": 1.6037714065807198e-05,
737
+ "loss": 0.6202,
738
+ "step": 51500
739
+ },
740
+ {
741
+ "epoch": 2.0,
742
+ "eval_loss": 1.6332776546478271,
743
+ "eval_runtime": 11.3148,
744
+ "eval_samples_per_second": 351.839,
745
+ "eval_steps_per_second": 88.026,
746
+ "step": 51970
747
+ },
748
+ {
749
+ "epoch": 2.0,
750
+ "grad_norm": 11.219087600708008,
751
+ "learning_rate": 1.599923032518761e-05,
752
+ "loss": 0.6083,
753
+ "step": 52000
754
+ },
755
+ {
756
+ "epoch": 2.02,
757
+ "grad_norm": 5.6387038230896,
758
+ "learning_rate": 1.5960746584568022e-05,
759
+ "loss": 0.5929,
760
+ "step": 52500
761
+ },
762
+ {
763
+ "epoch": 2.04,
764
+ "grad_norm": 10.761116981506348,
765
+ "learning_rate": 1.592241677891091e-05,
766
+ "loss": 0.5913,
767
+ "step": 53000
768
+ },
769
+ {
770
+ "epoch": 2.06,
771
+ "grad_norm": 4.182568073272705,
772
+ "learning_rate": 1.5883933038291323e-05,
773
+ "loss": 0.5983,
774
+ "step": 53500
775
+ },
776
+ {
777
+ "epoch": 2.08,
778
+ "grad_norm": 1.4628002643585205,
779
+ "learning_rate": 1.5845449297671735e-05,
780
+ "loss": 0.5789,
781
+ "step": 54000
782
+ },
783
+ {
784
+ "epoch": 2.1,
785
+ "grad_norm": 6.6901535987854,
786
+ "learning_rate": 1.5806965557052147e-05,
787
+ "loss": 0.5827,
788
+ "step": 54500
789
+ },
790
+ {
791
+ "epoch": 2.12,
792
+ "grad_norm": 4.275672912597656,
793
+ "learning_rate": 1.576848181643256e-05,
794
+ "loss": 0.593,
795
+ "step": 55000
796
+ },
797
+ {
798
+ "epoch": 2.14,
799
+ "grad_norm": 6.460977554321289,
800
+ "learning_rate": 1.572999807581297e-05,
801
+ "loss": 0.6012,
802
+ "step": 55500
803
+ },
804
+ {
805
+ "epoch": 2.16,
806
+ "grad_norm": 6.66805362701416,
807
+ "learning_rate": 1.5691514335193382e-05,
808
+ "loss": 0.6148,
809
+ "step": 56000
810
+ },
811
+ {
812
+ "epoch": 2.17,
813
+ "grad_norm": 9.386066436767578,
814
+ "learning_rate": 1.5653030594573794e-05,
815
+ "loss": 0.6147,
816
+ "step": 56500
817
+ },
818
+ {
819
+ "epoch": 2.19,
820
+ "grad_norm": 4.505888938903809,
821
+ "learning_rate": 1.5614700788916683e-05,
822
+ "loss": 0.6,
823
+ "step": 57000
824
+ },
825
+ {
826
+ "epoch": 2.21,
827
+ "grad_norm": 6.781282424926758,
828
+ "learning_rate": 1.5576217048297095e-05,
829
+ "loss": 0.5938,
830
+ "step": 57500
831
+ },
832
+ {
833
+ "epoch": 2.23,
834
+ "grad_norm": 4.20095157623291,
835
+ "learning_rate": 1.5537810275158747e-05,
836
+ "loss": 0.6118,
837
+ "step": 58000
838
+ },
839
+ {
840
+ "epoch": 2.25,
841
+ "grad_norm": 6.635080337524414,
842
+ "learning_rate": 1.549932653453916e-05,
843
+ "loss": 0.6375,
844
+ "step": 58500
845
+ },
846
+ {
847
+ "epoch": 2.27,
848
+ "grad_norm": 5.1331329345703125,
849
+ "learning_rate": 1.546084279391957e-05,
850
+ "loss": 0.625,
851
+ "step": 59000
852
+ },
853
+ {
854
+ "epoch": 2.29,
855
+ "grad_norm": 4.725217342376709,
856
+ "learning_rate": 1.5422359053299983e-05,
857
+ "loss": 0.6443,
858
+ "step": 59500
859
+ },
860
+ {
861
+ "epoch": 2.31,
862
+ "grad_norm": 6.601204872131348,
863
+ "learning_rate": 1.5383875312680395e-05,
864
+ "loss": 0.6206,
865
+ "step": 60000
866
+ },
867
+ {
868
+ "epoch": 2.33,
869
+ "grad_norm": 16.34359359741211,
870
+ "learning_rate": 1.5345391572060807e-05,
871
+ "loss": 0.6144,
872
+ "step": 60500
873
+ },
874
+ {
875
+ "epoch": 2.35,
876
+ "grad_norm": 6.883621692657471,
877
+ "learning_rate": 1.530690783144122e-05,
878
+ "loss": 0.6112,
879
+ "step": 61000
880
+ },
881
+ {
882
+ "epoch": 2.37,
883
+ "grad_norm": 5.42087984085083,
884
+ "learning_rate": 1.526842409082163e-05,
885
+ "loss": 0.6286,
886
+ "step": 61500
887
+ },
888
+ {
889
+ "epoch": 2.39,
890
+ "grad_norm": 7.191906452178955,
891
+ "learning_rate": 1.522994035020204e-05,
892
+ "loss": 0.6103,
893
+ "step": 62000
894
+ },
895
+ {
896
+ "epoch": 2.41,
897
+ "grad_norm": 8.114423751831055,
898
+ "learning_rate": 1.5191456609582453e-05,
899
+ "loss": 0.6464,
900
+ "step": 62500
901
+ },
902
+ {
903
+ "epoch": 2.42,
904
+ "grad_norm": 6.188645839691162,
905
+ "learning_rate": 1.5152972868962865e-05,
906
+ "loss": 0.5874,
907
+ "step": 63000
908
+ },
909
+ {
910
+ "epoch": 2.44,
911
+ "grad_norm": 5.572692394256592,
912
+ "learning_rate": 1.5114489128343276e-05,
913
+ "loss": 0.6235,
914
+ "step": 63500
915
+ },
916
+ {
917
+ "epoch": 2.46,
918
+ "grad_norm": 8.618267059326172,
919
+ "learning_rate": 1.5076082355204927e-05,
920
+ "loss": 0.6144,
921
+ "step": 64000
922
+ },
923
+ {
924
+ "epoch": 2.48,
925
+ "grad_norm": 6.580616474151611,
926
+ "learning_rate": 1.5037598614585339e-05,
927
+ "loss": 0.6544,
928
+ "step": 64500
929
+ },
930
+ {
931
+ "epoch": 2.5,
932
+ "grad_norm": 7.99381160736084,
933
+ "learning_rate": 1.499919184144699e-05,
934
+ "loss": 0.6281,
935
+ "step": 65000
936
+ },
937
+ {
938
+ "epoch": 2.52,
939
+ "grad_norm": 4.514256954193115,
940
+ "learning_rate": 1.4960708100827401e-05,
941
+ "loss": 0.6295,
942
+ "step": 65500
943
+ },
944
+ {
945
+ "epoch": 2.54,
946
+ "grad_norm": 8.09423828125,
947
+ "learning_rate": 1.4922224360207813e-05,
948
+ "loss": 0.6332,
949
+ "step": 66000
950
+ },
951
+ {
952
+ "epoch": 2.56,
953
+ "grad_norm": 5.654535293579102,
954
+ "learning_rate": 1.4883740619588225e-05,
955
+ "loss": 0.6573,
956
+ "step": 66500
957
+ },
958
+ {
959
+ "epoch": 2.58,
960
+ "grad_norm": 8.867722511291504,
961
+ "learning_rate": 1.4845256878968637e-05,
962
+ "loss": 0.6331,
963
+ "step": 67000
964
+ },
965
+ {
966
+ "epoch": 2.6,
967
+ "grad_norm": 11.65078353881836,
968
+ "learning_rate": 1.4806773138349049e-05,
969
+ "loss": 0.6522,
970
+ "step": 67500
971
+ },
972
+ {
973
+ "epoch": 2.62,
974
+ "grad_norm": 4.735153675079346,
975
+ "learning_rate": 1.476828939772946e-05,
976
+ "loss": 0.628,
977
+ "step": 68000
978
+ },
979
+ {
980
+ "epoch": 2.64,
981
+ "grad_norm": 10.033109664916992,
982
+ "learning_rate": 1.4729805657109873e-05,
983
+ "loss": 0.6588,
984
+ "step": 68500
985
+ },
986
+ {
987
+ "epoch": 2.66,
988
+ "grad_norm": 5.765382289886475,
989
+ "learning_rate": 1.4691398883971523e-05,
990
+ "loss": 0.657,
991
+ "step": 69000
992
+ },
993
+ {
994
+ "epoch": 2.67,
995
+ "grad_norm": 8.563407897949219,
996
+ "learning_rate": 1.4652915143351935e-05,
997
+ "loss": 0.6476,
998
+ "step": 69500
999
+ },
1000
+ {
1001
+ "epoch": 2.69,
1002
+ "grad_norm": 5.050264358520508,
1003
+ "learning_rate": 1.4614431402732347e-05,
1004
+ "loss": 0.643,
1005
+ "step": 70000
1006
+ },
1007
+ {
1008
+ "epoch": 2.71,
1009
+ "grad_norm": 4.829985618591309,
1010
+ "learning_rate": 1.4575947662112759e-05,
1011
+ "loss": 0.6637,
1012
+ "step": 70500
1013
+ },
1014
+ {
1015
+ "epoch": 2.73,
1016
+ "grad_norm": 10.406055450439453,
1017
+ "learning_rate": 1.453754088897441e-05,
1018
+ "loss": 0.6588,
1019
+ "step": 71000
1020
+ },
1021
+ {
1022
+ "epoch": 2.75,
1023
+ "grad_norm": 5.833386421203613,
1024
+ "learning_rate": 1.4499057148354821e-05,
1025
+ "loss": 0.6443,
1026
+ "step": 71500
1027
+ },
1028
+ {
1029
+ "epoch": 2.77,
1030
+ "grad_norm": 5.603066921234131,
1031
+ "learning_rate": 1.4460573407735233e-05,
1032
+ "loss": 0.6382,
1033
+ "step": 72000
1034
+ },
1035
+ {
1036
+ "epoch": 2.79,
1037
+ "grad_norm": 6.070736885070801,
1038
+ "learning_rate": 1.4422166634596883e-05,
1039
+ "loss": 0.654,
1040
+ "step": 72500
1041
+ },
1042
+ {
1043
+ "epoch": 2.81,
1044
+ "grad_norm": 4.568108081817627,
1045
+ "learning_rate": 1.4383682893977295e-05,
1046
+ "loss": 0.6496,
1047
+ "step": 73000
1048
+ },
1049
+ {
1050
+ "epoch": 2.83,
1051
+ "grad_norm": 5.071454048156738,
1052
+ "learning_rate": 1.4345199153357707e-05,
1053
+ "loss": 0.6412,
1054
+ "step": 73500
1055
+ },
1056
+ {
1057
+ "epoch": 2.85,
1058
+ "grad_norm": 9.910784721374512,
1059
+ "learning_rate": 1.4306715412738119e-05,
1060
+ "loss": 0.6329,
1061
+ "step": 74000
1062
+ },
1063
+ {
1064
+ "epoch": 2.87,
1065
+ "grad_norm": 4.771623134613037,
1066
+ "learning_rate": 1.4268231672118531e-05,
1067
+ "loss": 0.6661,
1068
+ "step": 74500
1069
+ },
1070
+ {
1071
+ "epoch": 2.89,
1072
+ "grad_norm": 12.325662612915039,
1073
+ "learning_rate": 1.4229747931498943e-05,
1074
+ "loss": 0.6389,
1075
+ "step": 75000
1076
+ },
1077
+ {
1078
+ "epoch": 2.91,
1079
+ "grad_norm": 4.688502311706543,
1080
+ "learning_rate": 1.4191341158360595e-05,
1081
+ "loss": 0.6475,
1082
+ "step": 75500
1083
+ },
1084
+ {
1085
+ "epoch": 2.92,
1086
+ "grad_norm": 7.340636730194092,
1087
+ "learning_rate": 1.4152857417741005e-05,
1088
+ "loss": 0.678,
1089
+ "step": 76000
1090
+ },
1091
+ {
1092
+ "epoch": 2.94,
1093
+ "grad_norm": 11.585808753967285,
1094
+ "learning_rate": 1.4114373677121417e-05,
1095
+ "loss": 0.6422,
1096
+ "step": 76500
1097
+ },
1098
+ {
1099
+ "epoch": 2.96,
1100
+ "grad_norm": 4.971041679382324,
1101
+ "learning_rate": 1.4075889936501829e-05,
1102
+ "loss": 0.6493,
1103
+ "step": 77000
1104
+ },
1105
+ {
1106
+ "epoch": 2.98,
1107
+ "grad_norm": 8.936264991760254,
1108
+ "learning_rate": 1.4037406195882241e-05,
1109
+ "loss": 0.6719,
1110
+ "step": 77500
1111
+ },
1112
+ {
1113
+ "epoch": 3.0,
1114
+ "eval_loss": 1.634682297706604,
1115
+ "eval_runtime": 11.4139,
1116
+ "eval_samples_per_second": 348.784,
1117
+ "eval_steps_per_second": 87.262,
1118
+ "step": 77955
1119
+ },
1120
+ {
1121
+ "epoch": 3.0,
1122
+ "grad_norm": 4.788780212402344,
1123
+ "learning_rate": 1.3998922455262653e-05,
1124
+ "loss": 0.6236,
1125
+ "step": 78000
1126
+ },
1127
+ {
1128
+ "epoch": 3.02,
1129
+ "grad_norm": 8.352863311767578,
1130
+ "learning_rate": 1.3960438714643065e-05,
1131
+ "loss": 0.5519,
1132
+ "step": 78500
1133
+ },
1134
+ {
1135
+ "epoch": 3.04,
1136
+ "grad_norm": 9.03128433227539,
1137
+ "learning_rate": 1.3922031941504715e-05,
1138
+ "loss": 0.5617,
1139
+ "step": 79000
1140
+ },
1141
+ {
1142
+ "epoch": 3.06,
1143
+ "grad_norm": 8.761636734008789,
1144
+ "learning_rate": 1.3883548200885127e-05,
1145
+ "loss": 0.554,
1146
+ "step": 79500
1147
+ },
1148
+ {
1149
+ "epoch": 3.08,
1150
+ "grad_norm": 4.592207908630371,
1151
+ "learning_rate": 1.3845064460265539e-05,
1152
+ "loss": 0.5218,
1153
+ "step": 80000
1154
+ },
1155
+ {
1156
+ "epoch": 3.1,
1157
+ "grad_norm": 5.602917671203613,
1158
+ "learning_rate": 1.380658071964595e-05,
1159
+ "loss": 0.5432,
1160
+ "step": 80500
1161
+ },
1162
+ {
1163
+ "epoch": 3.12,
1164
+ "grad_norm": 10.146926879882812,
1165
+ "learning_rate": 1.3768096979026363e-05,
1166
+ "loss": 0.5332,
1167
+ "step": 81000
1168
+ },
1169
+ {
1170
+ "epoch": 3.14,
1171
+ "grad_norm": 7.7499494552612305,
1172
+ "learning_rate": 1.3729613238406773e-05,
1173
+ "loss": 0.5459,
1174
+ "step": 81500
1175
+ },
1176
+ {
1177
+ "epoch": 3.16,
1178
+ "grad_norm": 10.009146690368652,
1179
+ "learning_rate": 1.3691129497787185e-05,
1180
+ "loss": 0.5721,
1181
+ "step": 82000
1182
+ },
1183
+ {
1184
+ "epoch": 3.17,
1185
+ "grad_norm": 5.544896602630615,
1186
+ "learning_rate": 1.3652645757167597e-05,
1187
+ "loss": 0.5407,
1188
+ "step": 82500
1189
+ },
1190
+ {
1191
+ "epoch": 3.19,
1192
+ "grad_norm": 11.661575317382812,
1193
+ "learning_rate": 1.3614238984029249e-05,
1194
+ "loss": 0.5678,
1195
+ "step": 83000
1196
+ },
1197
+ {
1198
+ "epoch": 3.21,
1199
+ "grad_norm": 2.23122501373291,
1200
+ "learning_rate": 1.357575524340966e-05,
1201
+ "loss": 0.5577,
1202
+ "step": 83500
1203
+ },
1204
+ {
1205
+ "epoch": 3.23,
1206
+ "grad_norm": 2.9503915309906006,
1207
+ "learning_rate": 1.3537271502790073e-05,
1208
+ "loss": 0.568,
1209
+ "step": 84000
1210
+ },
1211
+ {
1212
+ "epoch": 3.25,
1213
+ "grad_norm": 5.630256175994873,
1214
+ "learning_rate": 1.3498787762170485e-05,
1215
+ "loss": 0.5524,
1216
+ "step": 84500
1217
+ },
1218
+ {
1219
+ "epoch": 3.27,
1220
+ "grad_norm": 5.076592922210693,
1221
+ "learning_rate": 1.3460304021550896e-05,
1222
+ "loss": 0.574,
1223
+ "step": 85000
1224
+ },
1225
+ {
1226
+ "epoch": 3.29,
1227
+ "grad_norm": 7.131458759307861,
1228
+ "learning_rate": 1.3421974215893786e-05,
1229
+ "loss": 0.5718,
1230
+ "step": 85500
1231
+ },
1232
+ {
1233
+ "epoch": 3.31,
1234
+ "grad_norm": 16.959537506103516,
1235
+ "learning_rate": 1.3383490475274197e-05,
1236
+ "loss": 0.5743,
1237
+ "step": 86000
1238
+ },
1239
+ {
1240
+ "epoch": 3.33,
1241
+ "grad_norm": 8.548674583435059,
1242
+ "learning_rate": 1.334500673465461e-05,
1243
+ "loss": 0.5747,
1244
+ "step": 86500
1245
+ },
1246
+ {
1247
+ "epoch": 3.35,
1248
+ "grad_norm": 16.18918800354004,
1249
+ "learning_rate": 1.3306522994035021e-05,
1250
+ "loss": 0.594,
1251
+ "step": 87000
1252
+ },
1253
+ {
1254
+ "epoch": 3.37,
1255
+ "grad_norm": 3.274156093597412,
1256
+ "learning_rate": 1.3268039253415433e-05,
1257
+ "loss": 0.5758,
1258
+ "step": 87500
1259
+ },
1260
+ {
1261
+ "epoch": 3.39,
1262
+ "grad_norm": 5.2989959716796875,
1263
+ "learning_rate": 1.3229555512795845e-05,
1264
+ "loss": 0.5658,
1265
+ "step": 88000
1266
+ },
1267
+ {
1268
+ "epoch": 3.41,
1269
+ "grad_norm": 3.634488821029663,
1270
+ "learning_rate": 1.3191148739657495e-05,
1271
+ "loss": 0.5572,
1272
+ "step": 88500
1273
+ },
1274
+ {
1275
+ "epoch": 3.43,
1276
+ "grad_norm": 4.335933208465576,
1277
+ "learning_rate": 1.3152664999037907e-05,
1278
+ "loss": 0.5654,
1279
+ "step": 89000
1280
+ },
1281
+ {
1282
+ "epoch": 3.44,
1283
+ "grad_norm": 7.001922607421875,
1284
+ "learning_rate": 1.311418125841832e-05,
1285
+ "loss": 0.5741,
1286
+ "step": 89500
1287
+ },
1288
+ {
1289
+ "epoch": 3.46,
1290
+ "grad_norm": 9.779496192932129,
1291
+ "learning_rate": 1.3075697517798731e-05,
1292
+ "loss": 0.5782,
1293
+ "step": 90000
1294
+ },
1295
+ {
1296
+ "epoch": 3.48,
1297
+ "grad_norm": 7.241322994232178,
1298
+ "learning_rate": 1.3037213777179143e-05,
1299
+ "loss": 0.585,
1300
+ "step": 90500
1301
+ },
1302
+ {
1303
+ "epoch": 3.5,
1304
+ "grad_norm": 10.485682487487793,
1305
+ "learning_rate": 1.2998730036559555e-05,
1306
+ "loss": 0.5743,
1307
+ "step": 91000
1308
+ },
1309
+ {
1310
+ "epoch": 3.52,
1311
+ "grad_norm": 5.369238376617432,
1312
+ "learning_rate": 1.2960246295939967e-05,
1313
+ "loss": 0.5743,
1314
+ "step": 91500
1315
+ },
1316
+ {
1317
+ "epoch": 3.54,
1318
+ "grad_norm": 2.732351064682007,
1319
+ "learning_rate": 1.2921762555320379e-05,
1320
+ "loss": 0.5753,
1321
+ "step": 92000
1322
+ },
1323
+ {
1324
+ "epoch": 3.56,
1325
+ "grad_norm": 5.905006408691406,
1326
+ "learning_rate": 1.288335578218203e-05,
1327
+ "loss": 0.5816,
1328
+ "step": 92500
1329
+ },
1330
+ {
1331
+ "epoch": 3.58,
1332
+ "grad_norm": 6.421899795532227,
1333
+ "learning_rate": 1.2844872041562441e-05,
1334
+ "loss": 0.5907,
1335
+ "step": 93000
1336
+ },
1337
+ {
1338
+ "epoch": 3.6,
1339
+ "grad_norm": 8.367085456848145,
1340
+ "learning_rate": 1.2806388300942853e-05,
1341
+ "loss": 0.6026,
1342
+ "step": 93500
1343
+ },
1344
+ {
1345
+ "epoch": 3.62,
1346
+ "grad_norm": 5.6347784996032715,
1347
+ "learning_rate": 1.2767904560323265e-05,
1348
+ "loss": 0.6028,
1349
+ "step": 94000
1350
+ },
1351
+ {
1352
+ "epoch": 3.64,
1353
+ "grad_norm": 4.885034084320068,
1354
+ "learning_rate": 1.2729497787184915e-05,
1355
+ "loss": 0.6062,
1356
+ "step": 94500
1357
+ },
1358
+ {
1359
+ "epoch": 3.66,
1360
+ "grad_norm": 9.47564697265625,
1361
+ "learning_rate": 1.2691014046565327e-05,
1362
+ "loss": 0.6116,
1363
+ "step": 95000
1364
+ },
1365
+ {
1366
+ "epoch": 3.68,
1367
+ "grad_norm": 4.893156051635742,
1368
+ "learning_rate": 1.2652530305945739e-05,
1369
+ "loss": 0.5769,
1370
+ "step": 95500
1371
+ },
1372
+ {
1373
+ "epoch": 3.69,
1374
+ "grad_norm": 7.7681074142456055,
1375
+ "learning_rate": 1.261404656532615e-05,
1376
+ "loss": 0.5922,
1377
+ "step": 96000
1378
+ },
1379
+ {
1380
+ "epoch": 3.71,
1381
+ "grad_norm": 9.778617858886719,
1382
+ "learning_rate": 1.2575639792187801e-05,
1383
+ "loss": 0.623,
1384
+ "step": 96500
1385
+ },
1386
+ {
1387
+ "epoch": 3.73,
1388
+ "grad_norm": 10.256739616394043,
1389
+ "learning_rate": 1.2537156051568213e-05,
1390
+ "loss": 0.5552,
1391
+ "step": 97000
1392
+ },
1393
+ {
1394
+ "epoch": 3.75,
1395
+ "grad_norm": 6.800660133361816,
1396
+ "learning_rate": 1.2498672310948625e-05,
1397
+ "loss": 0.5961,
1398
+ "step": 97500
1399
+ },
1400
+ {
1401
+ "epoch": 3.77,
1402
+ "grad_norm": 11.549945831298828,
1403
+ "learning_rate": 1.2460188570329037e-05,
1404
+ "loss": 0.5983,
1405
+ "step": 98000
1406
+ },
1407
+ {
1408
+ "epoch": 3.79,
1409
+ "grad_norm": 7.519701957702637,
1410
+ "learning_rate": 1.2421704829709449e-05,
1411
+ "loss": 0.5769,
1412
+ "step": 98500
1413
+ },
1414
+ {
1415
+ "epoch": 3.81,
1416
+ "grad_norm": 7.1441168785095215,
1417
+ "learning_rate": 1.23832980565711e-05,
1418
+ "loss": 0.6006,
1419
+ "step": 99000
1420
+ },
1421
+ {
1422
+ "epoch": 3.83,
1423
+ "grad_norm": 4.6727375984191895,
1424
+ "learning_rate": 1.2344814315951511e-05,
1425
+ "loss": 0.6096,
1426
+ "step": 99500
1427
+ },
1428
+ {
1429
+ "epoch": 3.85,
1430
+ "grad_norm": 10.61463737487793,
1431
+ "learning_rate": 1.2306407542813164e-05,
1432
+ "loss": 0.6088,
1433
+ "step": 100000
1434
+ },
1435
+ {
1436
+ "epoch": 3.87,
1437
+ "grad_norm": 11.35522747039795,
1438
+ "learning_rate": 1.2267923802193574e-05,
1439
+ "loss": 0.603,
1440
+ "step": 100500
1441
+ },
1442
+ {
1443
+ "epoch": 3.89,
1444
+ "grad_norm": 4.605931282043457,
1445
+ "learning_rate": 1.2229440061573986e-05,
1446
+ "loss": 0.5742,
1447
+ "step": 101000
1448
+ },
1449
+ {
1450
+ "epoch": 3.91,
1451
+ "grad_norm": 5.55601692199707,
1452
+ "learning_rate": 1.2190956320954398e-05,
1453
+ "loss": 0.6005,
1454
+ "step": 101500
1455
+ },
1456
+ {
1457
+ "epoch": 3.93,
1458
+ "grad_norm": 4.776731014251709,
1459
+ "learning_rate": 1.215247258033481e-05,
1460
+ "loss": 0.6167,
1461
+ "step": 102000
1462
+ },
1463
+ {
1464
+ "epoch": 3.94,
1465
+ "grad_norm": 6.725343227386475,
1466
+ "learning_rate": 1.2113988839715221e-05,
1467
+ "loss": 0.5845,
1468
+ "step": 102500
1469
+ },
1470
+ {
1471
+ "epoch": 3.96,
1472
+ "grad_norm": 2.1615917682647705,
1473
+ "learning_rate": 1.2075505099095633e-05,
1474
+ "loss": 0.6123,
1475
+ "step": 103000
1476
+ },
1477
+ {
1478
+ "epoch": 3.98,
1479
+ "grad_norm": 3.8668465614318848,
1480
+ "learning_rate": 1.2037021358476043e-05,
1481
+ "loss": 0.5945,
1482
+ "step": 103500
1483
+ },
1484
+ {
1485
+ "epoch": 4.0,
1486
+ "eval_loss": 1.6611100435256958,
1487
+ "eval_runtime": 11.217,
1488
+ "eval_samples_per_second": 354.908,
1489
+ "eval_steps_per_second": 88.794,
1490
+ "step": 103940
1491
+ },
1492
+ {
1493
+ "epoch": 4.0,
1494
+ "grad_norm": 5.890031337738037,
1495
+ "learning_rate": 1.1998614585337696e-05,
1496
+ "loss": 0.5796,
1497
+ "step": 104000
1498
+ },
1499
+ {
1500
+ "epoch": 4.02,
1501
+ "grad_norm": 8.080361366271973,
1502
+ "learning_rate": 1.1960130844718107e-05,
1503
+ "loss": 0.5026,
1504
+ "step": 104500
1505
+ },
1506
+ {
1507
+ "epoch": 4.04,
1508
+ "grad_norm": 7.6675496101379395,
1509
+ "learning_rate": 1.192164710409852e-05,
1510
+ "loss": 0.522,
1511
+ "step": 105000
1512
+ },
1513
+ {
1514
+ "epoch": 4.06,
1515
+ "grad_norm": 6.038026332855225,
1516
+ "learning_rate": 1.1883163363478931e-05,
1517
+ "loss": 0.491,
1518
+ "step": 105500
1519
+ },
1520
+ {
1521
+ "epoch": 4.08,
1522
+ "grad_norm": 2.374932050704956,
1523
+ "learning_rate": 1.1844756590340582e-05,
1524
+ "loss": 0.5123,
1525
+ "step": 106000
1526
+ },
1527
+ {
1528
+ "epoch": 4.1,
1529
+ "grad_norm": 1.8449337482452393,
1530
+ "learning_rate": 1.1806272849720994e-05,
1531
+ "loss": 0.5104,
1532
+ "step": 106500
1533
+ },
1534
+ {
1535
+ "epoch": 4.12,
1536
+ "grad_norm": 7.582018852233887,
1537
+ "learning_rate": 1.1767789109101406e-05,
1538
+ "loss": 0.525,
1539
+ "step": 107000
1540
+ },
1541
+ {
1542
+ "epoch": 4.14,
1543
+ "grad_norm": 6.123476028442383,
1544
+ "learning_rate": 1.1729305368481817e-05,
1545
+ "loss": 0.5276,
1546
+ "step": 107500
1547
+ },
1548
+ {
1549
+ "epoch": 4.16,
1550
+ "grad_norm": 6.717319011688232,
1551
+ "learning_rate": 1.1690898595343468e-05,
1552
+ "loss": 0.5204,
1553
+ "step": 108000
1554
+ },
1555
+ {
1556
+ "epoch": 4.18,
1557
+ "grad_norm": 5.3222336769104,
1558
+ "learning_rate": 1.165241485472388e-05,
1559
+ "loss": 0.5039,
1560
+ "step": 108500
1561
+ },
1562
+ {
1563
+ "epoch": 4.19,
1564
+ "grad_norm": 7.88124942779541,
1565
+ "learning_rate": 1.1613931114104292e-05,
1566
+ "loss": 0.5177,
1567
+ "step": 109000
1568
+ },
1569
+ {
1570
+ "epoch": 4.21,
1571
+ "grad_norm": 13.466960906982422,
1572
+ "learning_rate": 1.1575447373484704e-05,
1573
+ "loss": 0.5076,
1574
+ "step": 109500
1575
+ },
1576
+ {
1577
+ "epoch": 4.23,
1578
+ "grad_norm": 3.5181362628936768,
1579
+ "learning_rate": 1.1536963632865115e-05,
1580
+ "loss": 0.5227,
1581
+ "step": 110000
1582
+ },
1583
+ {
1584
+ "epoch": 4.25,
1585
+ "grad_norm": 9.861627578735352,
1586
+ "learning_rate": 1.1498479892245527e-05,
1587
+ "loss": 0.5297,
1588
+ "step": 110500
1589
+ },
1590
+ {
1591
+ "epoch": 4.27,
1592
+ "grad_norm": 9.91634464263916,
1593
+ "learning_rate": 1.1460073119107178e-05,
1594
+ "loss": 0.5425,
1595
+ "step": 111000
1596
+ },
1597
+ {
1598
+ "epoch": 4.29,
1599
+ "grad_norm": 8.588738441467285,
1600
+ "learning_rate": 1.142158937848759e-05,
1601
+ "loss": 0.5298,
1602
+ "step": 111500
1603
+ },
1604
+ {
1605
+ "epoch": 4.31,
1606
+ "grad_norm": 4.605578899383545,
1607
+ "learning_rate": 1.1383105637868002e-05,
1608
+ "loss": 0.5202,
1609
+ "step": 112000
1610
+ },
1611
+ {
1612
+ "epoch": 4.33,
1613
+ "grad_norm": 4.7749481201171875,
1614
+ "learning_rate": 1.1344621897248413e-05,
1615
+ "loss": 0.5262,
1616
+ "step": 112500
1617
+ },
1618
+ {
1619
+ "epoch": 4.35,
1620
+ "grad_norm": 6.725896835327148,
1621
+ "learning_rate": 1.1306215124110066e-05,
1622
+ "loss": 0.5012,
1623
+ "step": 113000
1624
+ },
1625
+ {
1626
+ "epoch": 4.37,
1627
+ "grad_norm": 5.475027561187744,
1628
+ "learning_rate": 1.1267731383490476e-05,
1629
+ "loss": 0.5077,
1630
+ "step": 113500
1631
+ },
1632
+ {
1633
+ "epoch": 4.39,
1634
+ "grad_norm": 6.180666923522949,
1635
+ "learning_rate": 1.1229247642870888e-05,
1636
+ "loss": 0.514,
1637
+ "step": 114000
1638
+ },
1639
+ {
1640
+ "epoch": 4.41,
1641
+ "grad_norm": 6.444517135620117,
1642
+ "learning_rate": 1.119084086973254e-05,
1643
+ "loss": 0.5538,
1644
+ "step": 114500
1645
+ },
1646
+ {
1647
+ "epoch": 4.43,
1648
+ "grad_norm": 6.281551361083984,
1649
+ "learning_rate": 1.1152357129112952e-05,
1650
+ "loss": 0.5441,
1651
+ "step": 115000
1652
+ },
1653
+ {
1654
+ "epoch": 4.44,
1655
+ "grad_norm": 9.76223373413086,
1656
+ "learning_rate": 1.1113873388493362e-05,
1657
+ "loss": 0.5512,
1658
+ "step": 115500
1659
+ },
1660
+ {
1661
+ "epoch": 4.46,
1662
+ "grad_norm": 5.260191440582275,
1663
+ "learning_rate": 1.1075389647873774e-05,
1664
+ "loss": 0.5178,
1665
+ "step": 116000
1666
+ },
1667
+ {
1668
+ "epoch": 4.48,
1669
+ "grad_norm": 7.131434917449951,
1670
+ "learning_rate": 1.1036905907254186e-05,
1671
+ "loss": 0.5017,
1672
+ "step": 116500
1673
+ },
1674
+ {
1675
+ "epoch": 4.5,
1676
+ "grad_norm": 6.9372992515563965,
1677
+ "learning_rate": 1.0998422166634598e-05,
1678
+ "loss": 0.5323,
1679
+ "step": 117000
1680
+ },
1681
+ {
1682
+ "epoch": 4.52,
1683
+ "grad_norm": 3.6769461631774902,
1684
+ "learning_rate": 1.095993842601501e-05,
1685
+ "loss": 0.5472,
1686
+ "step": 117500
1687
+ },
1688
+ {
1689
+ "epoch": 4.54,
1690
+ "grad_norm": 4.933724403381348,
1691
+ "learning_rate": 1.0921454685395421e-05,
1692
+ "loss": 0.5275,
1693
+ "step": 118000
1694
+ },
1695
+ {
1696
+ "epoch": 4.56,
1697
+ "grad_norm": 8.011011123657227,
1698
+ "learning_rate": 1.0882970944775832e-05,
1699
+ "loss": 0.5754,
1700
+ "step": 118500
1701
+ },
1702
+ {
1703
+ "epoch": 4.58,
1704
+ "grad_norm": 5.938791275024414,
1705
+ "learning_rate": 1.0844564171637484e-05,
1706
+ "loss": 0.5244,
1707
+ "step": 119000
1708
+ },
1709
+ {
1710
+ "epoch": 4.6,
1711
+ "grad_norm": 5.401825904846191,
1712
+ "learning_rate": 1.0806080431017896e-05,
1713
+ "loss": 0.5601,
1714
+ "step": 119500
1715
+ },
1716
+ {
1717
+ "epoch": 4.62,
1718
+ "grad_norm": 3.8994340896606445,
1719
+ "learning_rate": 1.0767596690398308e-05,
1720
+ "loss": 0.5399,
1721
+ "step": 120000
1722
+ },
1723
+ {
1724
+ "epoch": 4.64,
1725
+ "grad_norm": 7.102384090423584,
1726
+ "learning_rate": 1.072911294977872e-05,
1727
+ "loss": 0.5426,
1728
+ "step": 120500
1729
+ },
1730
+ {
1731
+ "epoch": 4.66,
1732
+ "grad_norm": 4.785587787628174,
1733
+ "learning_rate": 1.069070617664037e-05,
1734
+ "loss": 0.541,
1735
+ "step": 121000
1736
+ },
1737
+ {
1738
+ "epoch": 4.68,
1739
+ "grad_norm": 3.6323719024658203,
1740
+ "learning_rate": 1.0652222436020782e-05,
1741
+ "loss": 0.5327,
1742
+ "step": 121500
1743
+ },
1744
+ {
1745
+ "epoch": 4.7,
1746
+ "grad_norm": 3.2487778663635254,
1747
+ "learning_rate": 1.0613738695401194e-05,
1748
+ "loss": 0.5437,
1749
+ "step": 122000
1750
+ },
1751
+ {
1752
+ "epoch": 4.71,
1753
+ "grad_norm": 5.607531547546387,
1754
+ "learning_rate": 1.0575254954781606e-05,
1755
+ "loss": 0.5691,
1756
+ "step": 122500
1757
+ },
1758
+ {
1759
+ "epoch": 4.73,
1760
+ "grad_norm": 4.990538597106934,
1761
+ "learning_rate": 1.0536848181643256e-05,
1762
+ "loss": 0.5544,
1763
+ "step": 123000
1764
+ },
1765
+ {
1766
+ "epoch": 4.75,
1767
+ "grad_norm": 11.409914016723633,
1768
+ "learning_rate": 1.0498364441023668e-05,
1769
+ "loss": 0.5229,
1770
+ "step": 123500
1771
+ },
1772
+ {
1773
+ "epoch": 4.77,
1774
+ "grad_norm": 5.91348123550415,
1775
+ "learning_rate": 1.045988070040408e-05,
1776
+ "loss": 0.5609,
1777
+ "step": 124000
1778
+ },
1779
+ {
1780
+ "epoch": 4.79,
1781
+ "grad_norm": 5.885928630828857,
1782
+ "learning_rate": 1.0421396959784492e-05,
1783
+ "loss": 0.5503,
1784
+ "step": 124500
1785
+ },
1786
+ {
1787
+ "epoch": 4.81,
1788
+ "grad_norm": 3.7525486946105957,
1789
+ "learning_rate": 1.0382913219164904e-05,
1790
+ "loss": 0.5452,
1791
+ "step": 125000
1792
+ },
1793
+ {
1794
+ "epoch": 4.83,
1795
+ "grad_norm": 4.8420305252075195,
1796
+ "learning_rate": 1.0344506446026554e-05,
1797
+ "loss": 0.566,
1798
+ "step": 125500
1799
+ },
1800
+ {
1801
+ "epoch": 4.85,
1802
+ "grad_norm": 5.222369194030762,
1803
+ "learning_rate": 1.0306022705406966e-05,
1804
+ "loss": 0.5481,
1805
+ "step": 126000
1806
+ },
1807
+ {
1808
+ "epoch": 4.87,
1809
+ "grad_norm": 6.775496006011963,
1810
+ "learning_rate": 1.0267538964787378e-05,
1811
+ "loss": 0.5353,
1812
+ "step": 126500
1813
+ },
1814
+ {
1815
+ "epoch": 4.89,
1816
+ "grad_norm": 13.353194236755371,
1817
+ "learning_rate": 1.022905522416779e-05,
1818
+ "loss": 0.5574,
1819
+ "step": 127000
1820
+ },
1821
+ {
1822
+ "epoch": 4.91,
1823
+ "grad_norm": 6.453900337219238,
1824
+ "learning_rate": 1.0190648451029442e-05,
1825
+ "loss": 0.5715,
1826
+ "step": 127500
1827
+ },
1828
+ {
1829
+ "epoch": 4.93,
1830
+ "grad_norm": 9.484274864196777,
1831
+ "learning_rate": 1.0152241677891092e-05,
1832
+ "loss": 0.5678,
1833
+ "step": 128000
1834
+ },
1835
+ {
1836
+ "epoch": 4.95,
1837
+ "grad_norm": 5.4346160888671875,
1838
+ "learning_rate": 1.0113757937271504e-05,
1839
+ "loss": 0.5661,
1840
+ "step": 128500
1841
+ },
1842
+ {
1843
+ "epoch": 4.96,
1844
+ "grad_norm": 7.368044376373291,
1845
+ "learning_rate": 1.0075274196651916e-05,
1846
+ "loss": 0.5524,
1847
+ "step": 129000
1848
+ },
1849
+ {
1850
+ "epoch": 4.98,
1851
+ "grad_norm": 5.8485941886901855,
1852
+ "learning_rate": 1.0036790456032328e-05,
1853
+ "loss": 0.5472,
1854
+ "step": 129500
1855
+ },
1856
+ {
1857
+ "epoch": 5.0,
1858
+ "eval_loss": 1.6937280893325806,
1859
+ "eval_runtime": 11.3508,
1860
+ "eval_samples_per_second": 350.726,
1861
+ "eval_steps_per_second": 87.747,
1862
+ "step": 129925
1863
+ },
1864
+ {
1865
+ "epoch": 5.0,
1866
+ "grad_norm": 3.9238245487213135,
1867
+ "learning_rate": 9.998306715412738e-06,
1868
+ "loss": 0.5324,
1869
+ "step": 130000
1870
+ },
1871
+ {
1872
+ "epoch": 5.02,
1873
+ "grad_norm": 3.3814291954040527,
1874
+ "learning_rate": 9.95982297479315e-06,
1875
+ "loss": 0.4778,
1876
+ "step": 130500
1877
+ },
1878
+ {
1879
+ "epoch": 5.04,
1880
+ "grad_norm": 3.6909124851226807,
1881
+ "learning_rate": 9.921339234173562e-06,
1882
+ "loss": 0.4872,
1883
+ "step": 131000
1884
+ },
1885
+ {
1886
+ "epoch": 5.06,
1887
+ "grad_norm": 3.4291248321533203,
1888
+ "learning_rate": 9.882855493553974e-06,
1889
+ "loss": 0.4572,
1890
+ "step": 131500
1891
+ },
1892
+ {
1893
+ "epoch": 5.08,
1894
+ "grad_norm": 7.226670742034912,
1895
+ "learning_rate": 9.844448720415625e-06,
1896
+ "loss": 0.4887,
1897
+ "step": 132000
1898
+ },
1899
+ {
1900
+ "epoch": 5.1,
1901
+ "grad_norm": 3.1274802684783936,
1902
+ "learning_rate": 9.805964979796036e-06,
1903
+ "loss": 0.4687,
1904
+ "step": 132500
1905
+ },
1906
+ {
1907
+ "epoch": 5.12,
1908
+ "grad_norm": 6.210909366607666,
1909
+ "learning_rate": 9.767558206657689e-06,
1910
+ "loss": 0.4695,
1911
+ "step": 133000
1912
+ },
1913
+ {
1914
+ "epoch": 5.14,
1915
+ "grad_norm": 4.919899940490723,
1916
+ "learning_rate": 9.7290744660381e-06,
1917
+ "loss": 0.4697,
1918
+ "step": 133500
1919
+ },
1920
+ {
1921
+ "epoch": 5.16,
1922
+ "grad_norm": 5.735771179199219,
1923
+ "learning_rate": 9.690590725418512e-06,
1924
+ "loss": 0.4663,
1925
+ "step": 134000
1926
+ },
1927
+ {
1928
+ "epoch": 5.18,
1929
+ "grad_norm": 4.188859462738037,
1930
+ "learning_rate": 9.652106984798924e-06,
1931
+ "loss": 0.4997,
1932
+ "step": 134500
1933
+ },
1934
+ {
1935
+ "epoch": 5.2,
1936
+ "grad_norm": 6.6499786376953125,
1937
+ "learning_rate": 9.613623244179336e-06,
1938
+ "loss": 0.4713,
1939
+ "step": 135000
1940
+ },
1941
+ {
1942
+ "epoch": 5.21,
1943
+ "grad_norm": 5.279160499572754,
1944
+ "learning_rate": 9.575139503559748e-06,
1945
+ "loss": 0.4742,
1946
+ "step": 135500
1947
+ },
1948
+ {
1949
+ "epoch": 5.23,
1950
+ "grad_norm": 19.4254093170166,
1951
+ "learning_rate": 9.536655762940158e-06,
1952
+ "loss": 0.4906,
1953
+ "step": 136000
1954
+ },
1955
+ {
1956
+ "epoch": 5.25,
1957
+ "grad_norm": 4.88319206237793,
1958
+ "learning_rate": 9.49817202232057e-06,
1959
+ "loss": 0.466,
1960
+ "step": 136500
1961
+ },
1962
+ {
1963
+ "epoch": 5.27,
1964
+ "grad_norm": 3.700117826461792,
1965
+ "learning_rate": 9.459688281700982e-06,
1966
+ "loss": 0.5211,
1967
+ "step": 137000
1968
+ },
1969
+ {
1970
+ "epoch": 5.29,
1971
+ "grad_norm": 5.820217609405518,
1972
+ "learning_rate": 9.421281508562632e-06,
1973
+ "loss": 0.5012,
1974
+ "step": 137500
1975
+ },
1976
+ {
1977
+ "epoch": 5.31,
1978
+ "grad_norm": 4.963629245758057,
1979
+ "learning_rate": 9.382797767943044e-06,
1980
+ "loss": 0.4893,
1981
+ "step": 138000
1982
+ },
1983
+ {
1984
+ "epoch": 5.33,
1985
+ "grad_norm": 3.2575323581695557,
1986
+ "learning_rate": 9.344314027323456e-06,
1987
+ "loss": 0.4881,
1988
+ "step": 138500
1989
+ },
1990
+ {
1991
+ "epoch": 5.35,
1992
+ "grad_norm": 4.5853424072265625,
1993
+ "learning_rate": 9.305830286703868e-06,
1994
+ "loss": 0.497,
1995
+ "step": 139000
1996
+ },
1997
+ {
1998
+ "epoch": 5.37,
1999
+ "grad_norm": 9.991304397583008,
2000
+ "learning_rate": 9.26734654608428e-06,
2001
+ "loss": 0.5029,
2002
+ "step": 139500
2003
+ },
2004
+ {
2005
+ "epoch": 5.39,
2006
+ "grad_norm": 7.186006546020508,
2007
+ "learning_rate": 9.228862805464692e-06,
2008
+ "loss": 0.5146,
2009
+ "step": 140000
2010
+ },
2011
+ {
2012
+ "epoch": 5.41,
2013
+ "grad_norm": 4.827492713928223,
2014
+ "learning_rate": 9.190456032326342e-06,
2015
+ "loss": 0.5155,
2016
+ "step": 140500
2017
+ },
2018
+ {
2019
+ "epoch": 5.43,
2020
+ "grad_norm": 6.840179443359375,
2021
+ "learning_rate": 9.151972291706754e-06,
2022
+ "loss": 0.5204,
2023
+ "step": 141000
2024
+ },
2025
+ {
2026
+ "epoch": 5.45,
2027
+ "grad_norm": 3.26387357711792,
2028
+ "learning_rate": 9.113488551087166e-06,
2029
+ "loss": 0.4909,
2030
+ "step": 141500
2031
+ },
2032
+ {
2033
+ "epoch": 5.46,
2034
+ "grad_norm": 8.286077499389648,
2035
+ "learning_rate": 9.075004810467578e-06,
2036
+ "loss": 0.4973,
2037
+ "step": 142000
2038
+ },
2039
+ {
2040
+ "epoch": 5.48,
2041
+ "grad_norm": 6.248400688171387,
2042
+ "learning_rate": 9.03659803732923e-06,
2043
+ "loss": 0.5213,
2044
+ "step": 142500
2045
+ },
2046
+ {
2047
+ "epoch": 5.5,
2048
+ "grad_norm": 4.121056079864502,
2049
+ "learning_rate": 8.998114296709642e-06,
2050
+ "loss": 0.5176,
2051
+ "step": 143000
2052
+ },
2053
+ {
2054
+ "epoch": 5.52,
2055
+ "grad_norm": 5.463682651519775,
2056
+ "learning_rate": 8.959630556090052e-06,
2057
+ "loss": 0.4707,
2058
+ "step": 143500
2059
+ },
2060
+ {
2061
+ "epoch": 5.54,
2062
+ "grad_norm": 7.710265159606934,
2063
+ "learning_rate": 8.921146815470464e-06,
2064
+ "loss": 0.4728,
2065
+ "step": 144000
2066
+ },
2067
+ {
2068
+ "epoch": 5.56,
2069
+ "grad_norm": 6.872325420379639,
2070
+ "learning_rate": 8.882740042332115e-06,
2071
+ "loss": 0.5183,
2072
+ "step": 144500
2073
+ },
2074
+ {
2075
+ "epoch": 5.58,
2076
+ "grad_norm": 7.6797614097595215,
2077
+ "learning_rate": 8.844256301712527e-06,
2078
+ "loss": 0.5065,
2079
+ "step": 145000
2080
+ },
2081
+ {
2082
+ "epoch": 5.6,
2083
+ "grad_norm": 4.729648590087891,
2084
+ "learning_rate": 8.805772561092938e-06,
2085
+ "loss": 0.5117,
2086
+ "step": 145500
2087
+ },
2088
+ {
2089
+ "epoch": 5.62,
2090
+ "grad_norm": 6.701267719268799,
2091
+ "learning_rate": 8.76728882047335e-06,
2092
+ "loss": 0.5427,
2093
+ "step": 146000
2094
+ },
2095
+ {
2096
+ "epoch": 5.64,
2097
+ "grad_norm": 8.45148754119873,
2098
+ "learning_rate": 8.728805079853762e-06,
2099
+ "loss": 0.4782,
2100
+ "step": 146500
2101
+ },
2102
+ {
2103
+ "epoch": 5.66,
2104
+ "grad_norm": 4.278749465942383,
2105
+ "learning_rate": 8.690398306715413e-06,
2106
+ "loss": 0.5093,
2107
+ "step": 147000
2108
+ },
2109
+ {
2110
+ "epoch": 5.68,
2111
+ "grad_norm": 6.3357343673706055,
2112
+ "learning_rate": 8.651914566095825e-06,
2113
+ "loss": 0.5328,
2114
+ "step": 147500
2115
+ },
2116
+ {
2117
+ "epoch": 5.7,
2118
+ "grad_norm": 9.23112678527832,
2119
+ "learning_rate": 8.613430825476237e-06,
2120
+ "loss": 0.526,
2121
+ "step": 148000
2122
+ },
2123
+ {
2124
+ "epoch": 5.71,
2125
+ "grad_norm": 2.7398366928100586,
2126
+ "learning_rate": 8.574947084856648e-06,
2127
+ "loss": 0.5004,
2128
+ "step": 148500
2129
+ },
2130
+ {
2131
+ "epoch": 5.73,
2132
+ "grad_norm": 7.174339771270752,
2133
+ "learning_rate": 8.5365403117183e-06,
2134
+ "loss": 0.4947,
2135
+ "step": 149000
2136
+ },
2137
+ {
2138
+ "epoch": 5.75,
2139
+ "grad_norm": 6.2081427574157715,
2140
+ "learning_rate": 8.498056571098712e-06,
2141
+ "loss": 0.4961,
2142
+ "step": 149500
2143
+ },
2144
+ {
2145
+ "epoch": 5.77,
2146
+ "grad_norm": 5.464582443237305,
2147
+ "learning_rate": 8.459572830479124e-06,
2148
+ "loss": 0.4984,
2149
+ "step": 150000
2150
+ },
2151
+ {
2152
+ "epoch": 5.79,
2153
+ "grad_norm": 4.864925384521484,
2154
+ "learning_rate": 8.421089089859536e-06,
2155
+ "loss": 0.5101,
2156
+ "step": 150500
2157
+ },
2158
+ {
2159
+ "epoch": 5.81,
2160
+ "grad_norm": 10.367297172546387,
2161
+ "learning_rate": 8.382682316721185e-06,
2162
+ "loss": 0.5004,
2163
+ "step": 151000
2164
+ },
2165
+ {
2166
+ "epoch": 5.83,
2167
+ "grad_norm": 4.001099109649658,
2168
+ "learning_rate": 8.344198576101597e-06,
2169
+ "loss": 0.5109,
2170
+ "step": 151500
2171
+ },
2172
+ {
2173
+ "epoch": 5.85,
2174
+ "grad_norm": 8.188924789428711,
2175
+ "learning_rate": 8.305714835482009e-06,
2176
+ "loss": 0.4988,
2177
+ "step": 152000
2178
+ },
2179
+ {
2180
+ "epoch": 5.87,
2181
+ "grad_norm": 5.423083782196045,
2182
+ "learning_rate": 8.26723109486242e-06,
2183
+ "loss": 0.4957,
2184
+ "step": 152500
2185
+ },
2186
+ {
2187
+ "epoch": 5.89,
2188
+ "grad_norm": 5.513723373413086,
2189
+ "learning_rate": 8.228824321724073e-06,
2190
+ "loss": 0.5113,
2191
+ "step": 153000
2192
+ },
2193
+ {
2194
+ "epoch": 5.91,
2195
+ "grad_norm": 5.753915309906006,
2196
+ "learning_rate": 8.190340581104485e-06,
2197
+ "loss": 0.5088,
2198
+ "step": 153500
2199
+ },
2200
+ {
2201
+ "epoch": 5.93,
2202
+ "grad_norm": 5.672900676727295,
2203
+ "learning_rate": 8.151933807966135e-06,
2204
+ "loss": 0.5117,
2205
+ "step": 154000
2206
+ },
2207
+ {
2208
+ "epoch": 5.95,
2209
+ "grad_norm": 6.9963202476501465,
2210
+ "learning_rate": 8.113450067346547e-06,
2211
+ "loss": 0.5474,
2212
+ "step": 154500
2213
+ },
2214
+ {
2215
+ "epoch": 5.96,
2216
+ "grad_norm": 6.640045166015625,
2217
+ "learning_rate": 8.074966326726959e-06,
2218
+ "loss": 0.5135,
2219
+ "step": 155000
2220
+ },
2221
+ {
2222
+ "epoch": 5.98,
2223
+ "grad_norm": 5.076375961303711,
2224
+ "learning_rate": 8.036482586107371e-06,
2225
+ "loss": 0.5147,
2226
+ "step": 155500
2227
+ },
2228
+ {
2229
+ "epoch": 6.0,
2230
+ "eval_loss": 1.7142382860183716,
2231
+ "eval_runtime": 11.2855,
2232
+ "eval_samples_per_second": 352.753,
2233
+ "eval_steps_per_second": 88.255,
2234
+ "step": 155910
2235
+ },
2236
+ {
2237
+ "epoch": 6.0,
2238
+ "grad_norm": 6.40647554397583,
2239
+ "learning_rate": 7.997998845487783e-06,
2240
+ "loss": 0.467,
2241
+ "step": 156000
2242
+ },
2243
+ {
2244
+ "epoch": 6.02,
2245
+ "grad_norm": 3.021810293197632,
2246
+ "learning_rate": 7.959515104868195e-06,
2247
+ "loss": 0.4311,
2248
+ "step": 156500
2249
+ },
2250
+ {
2251
+ "epoch": 6.04,
2252
+ "grad_norm": 11.23317813873291,
2253
+ "learning_rate": 7.921031364248607e-06,
2254
+ "loss": 0.4569,
2255
+ "step": 157000
2256
+ },
2257
+ {
2258
+ "epoch": 6.06,
2259
+ "grad_norm": 4.276322841644287,
2260
+ "learning_rate": 7.882547623629019e-06,
2261
+ "loss": 0.4523,
2262
+ "step": 157500
2263
+ },
2264
+ {
2265
+ "epoch": 6.08,
2266
+ "grad_norm": 5.726410388946533,
2267
+ "learning_rate": 7.844140850490667e-06,
2268
+ "loss": 0.4601,
2269
+ "step": 158000
2270
+ },
2271
+ {
2272
+ "epoch": 6.1,
2273
+ "grad_norm": 8.776612281799316,
2274
+ "learning_rate": 7.80565710987108e-06,
2275
+ "loss": 0.4254,
2276
+ "step": 158500
2277
+ },
2278
+ {
2279
+ "epoch": 6.12,
2280
+ "grad_norm": 9.005372047424316,
2281
+ "learning_rate": 7.767173369251491e-06,
2282
+ "loss": 0.4328,
2283
+ "step": 159000
2284
+ },
2285
+ {
2286
+ "epoch": 6.14,
2287
+ "grad_norm": 9.345296859741211,
2288
+ "learning_rate": 7.728689628631903e-06,
2289
+ "loss": 0.4619,
2290
+ "step": 159500
2291
+ },
2292
+ {
2293
+ "epoch": 6.16,
2294
+ "grad_norm": 5.089489936828613,
2295
+ "learning_rate": 7.690282855493555e-06,
2296
+ "loss": 0.4546,
2297
+ "step": 160000
2298
+ },
2299
+ {
2300
+ "epoch": 6.18,
2301
+ "grad_norm": 5.524288177490234,
2302
+ "learning_rate": 7.651799114873967e-06,
2303
+ "loss": 0.4444,
2304
+ "step": 160500
2305
+ },
2306
+ {
2307
+ "epoch": 6.2,
2308
+ "grad_norm": 5.413343906402588,
2309
+ "learning_rate": 7.613315374254379e-06,
2310
+ "loss": 0.4672,
2311
+ "step": 161000
2312
+ },
2313
+ {
2314
+ "epoch": 6.22,
2315
+ "grad_norm": 2.5175459384918213,
2316
+ "learning_rate": 7.5749086011160285e-06,
2317
+ "loss": 0.4534,
2318
+ "step": 161500
2319
+ },
2320
+ {
2321
+ "epoch": 6.23,
2322
+ "grad_norm": 9.147507667541504,
2323
+ "learning_rate": 7.5364248604964404e-06,
2324
+ "loss": 0.4642,
2325
+ "step": 162000
2326
+ },
2327
+ {
2328
+ "epoch": 6.25,
2329
+ "grad_norm": 4.600451946258545,
2330
+ "learning_rate": 7.497941119876852e-06,
2331
+ "loss": 0.4448,
2332
+ "step": 162500
2333
+ },
2334
+ {
2335
+ "epoch": 6.27,
2336
+ "grad_norm": 5.7864556312561035,
2337
+ "learning_rate": 7.459457379257264e-06,
2338
+ "loss": 0.4463,
2339
+ "step": 163000
2340
+ },
2341
+ {
2342
+ "epoch": 6.29,
2343
+ "grad_norm": 6.185606002807617,
2344
+ "learning_rate": 7.420973638637676e-06,
2345
+ "loss": 0.4608,
2346
+ "step": 163500
2347
+ },
2348
+ {
2349
+ "epoch": 6.31,
2350
+ "grad_norm": 4.961802959442139,
2351
+ "learning_rate": 7.382489898018088e-06,
2352
+ "loss": 0.4609,
2353
+ "step": 164000
2354
+ },
2355
+ {
2356
+ "epoch": 6.33,
2357
+ "grad_norm": 8.758057594299316,
2358
+ "learning_rate": 7.3440061573985e-06,
2359
+ "loss": 0.459,
2360
+ "step": 164500
2361
+ },
2362
+ {
2363
+ "epoch": 6.35,
2364
+ "grad_norm": 13.342981338500977,
2365
+ "learning_rate": 7.305522416778912e-06,
2366
+ "loss": 0.4684,
2367
+ "step": 165000
2368
+ },
2369
+ {
2370
+ "epoch": 6.37,
2371
+ "grad_norm": 3.678190231323242,
2372
+ "learning_rate": 7.267038676159324e-06,
2373
+ "loss": 0.4482,
2374
+ "step": 165500
2375
+ },
2376
+ {
2377
+ "epoch": 6.39,
2378
+ "grad_norm": 5.763678550720215,
2379
+ "learning_rate": 7.2285549355397356e-06,
2380
+ "loss": 0.4828,
2381
+ "step": 166000
2382
+ },
2383
+ {
2384
+ "epoch": 6.41,
2385
+ "grad_norm": 2.4365394115448,
2386
+ "learning_rate": 7.1900711949201475e-06,
2387
+ "loss": 0.4519,
2388
+ "step": 166500
2389
+ },
2390
+ {
2391
+ "epoch": 6.43,
2392
+ "grad_norm": 7.325679779052734,
2393
+ "learning_rate": 7.1515874543005585e-06,
2394
+ "loss": 0.4725,
2395
+ "step": 167000
2396
+ },
2397
+ {
2398
+ "epoch": 6.45,
2399
+ "grad_norm": 7.1001811027526855,
2400
+ "learning_rate": 7.11310371368097e-06,
2401
+ "loss": 0.4552,
2402
+ "step": 167500
2403
+ },
2404
+ {
2405
+ "epoch": 6.47,
2406
+ "grad_norm": 4.369767189025879,
2407
+ "learning_rate": 7.0746199730613814e-06,
2408
+ "loss": 0.4615,
2409
+ "step": 168000
2410
+ },
2411
+ {
2412
+ "epoch": 6.48,
2413
+ "grad_norm": 5.81917667388916,
2414
+ "learning_rate": 7.036213199923034e-06,
2415
+ "loss": 0.4518,
2416
+ "step": 168500
2417
+ },
2418
+ {
2419
+ "epoch": 6.5,
2420
+ "grad_norm": 4.713387966156006,
2421
+ "learning_rate": 6.997729459303445e-06,
2422
+ "loss": 0.473,
2423
+ "step": 169000
2424
+ },
2425
+ {
2426
+ "epoch": 6.52,
2427
+ "grad_norm": 7.274464130401611,
2428
+ "learning_rate": 6.9592457186838565e-06,
2429
+ "loss": 0.4729,
2430
+ "step": 169500
2431
+ },
2432
+ {
2433
+ "epoch": 6.54,
2434
+ "grad_norm": 8.085463523864746,
2435
+ "learning_rate": 6.9207619780642684e-06,
2436
+ "loss": 0.4723,
2437
+ "step": 170000
2438
+ },
2439
+ {
2440
+ "epoch": 6.56,
2441
+ "grad_norm": 9.212237358093262,
2442
+ "learning_rate": 6.88227823744468e-06,
2443
+ "loss": 0.4571,
2444
+ "step": 170500
2445
+ },
2446
+ {
2447
+ "epoch": 6.58,
2448
+ "grad_norm": 3.6301112174987793,
2449
+ "learning_rate": 6.843871464306331e-06,
2450
+ "loss": 0.4726,
2451
+ "step": 171000
2452
+ },
2453
+ {
2454
+ "epoch": 6.6,
2455
+ "grad_norm": 5.698800563812256,
2456
+ "learning_rate": 6.805464691167982e-06,
2457
+ "loss": 0.4831,
2458
+ "step": 171500
2459
+ },
2460
+ {
2461
+ "epoch": 6.62,
2462
+ "grad_norm": 5.532044887542725,
2463
+ "learning_rate": 6.766980950548394e-06,
2464
+ "loss": 0.4708,
2465
+ "step": 172000
2466
+ },
2467
+ {
2468
+ "epoch": 6.64,
2469
+ "grad_norm": 7.939385414123535,
2470
+ "learning_rate": 6.728497209928806e-06,
2471
+ "loss": 0.468,
2472
+ "step": 172500
2473
+ },
2474
+ {
2475
+ "epoch": 6.66,
2476
+ "grad_norm": 3.2672829627990723,
2477
+ "learning_rate": 6.690013469309218e-06,
2478
+ "loss": 0.4935,
2479
+ "step": 173000
2480
+ },
2481
+ {
2482
+ "epoch": 6.68,
2483
+ "grad_norm": 11.197509765625,
2484
+ "learning_rate": 6.65152972868963e-06,
2485
+ "loss": 0.4777,
2486
+ "step": 173500
2487
+ },
2488
+ {
2489
+ "epoch": 6.7,
2490
+ "grad_norm": 2.946389675140381,
2491
+ "learning_rate": 6.6130459880700416e-06,
2492
+ "loss": 0.4786,
2493
+ "step": 174000
2494
+ },
2495
+ {
2496
+ "epoch": 6.72,
2497
+ "grad_norm": 1.725356936454773,
2498
+ "learning_rate": 6.574562247450453e-06,
2499
+ "loss": 0.4471,
2500
+ "step": 174500
2501
+ },
2502
+ {
2503
+ "epoch": 6.73,
2504
+ "grad_norm": 3.4278807640075684,
2505
+ "learning_rate": 6.5360785068308645e-06,
2506
+ "loss": 0.472,
2507
+ "step": 175000
2508
+ },
2509
+ {
2510
+ "epoch": 6.75,
2511
+ "grad_norm": 7.619284629821777,
2512
+ "learning_rate": 6.4975947662112756e-06,
2513
+ "loss": 0.4566,
2514
+ "step": 175500
2515
+ },
2516
+ {
2517
+ "epoch": 6.77,
2518
+ "grad_norm": 3.8815529346466064,
2519
+ "learning_rate": 6.459187993072928e-06,
2520
+ "loss": 0.4738,
2521
+ "step": 176000
2522
+ },
2523
+ {
2524
+ "epoch": 6.79,
2525
+ "grad_norm": 4.85827112197876,
2526
+ "learning_rate": 6.420704252453339e-06,
2527
+ "loss": 0.4637,
2528
+ "step": 176500
2529
+ },
2530
+ {
2531
+ "epoch": 6.81,
2532
+ "grad_norm": 5.672398090362549,
2533
+ "learning_rate": 6.382220511833751e-06,
2534
+ "loss": 0.4646,
2535
+ "step": 177000
2536
+ },
2537
+ {
2538
+ "epoch": 6.83,
2539
+ "grad_norm": 3.5175821781158447,
2540
+ "learning_rate": 6.3437367712141626e-06,
2541
+ "loss": 0.4949,
2542
+ "step": 177500
2543
+ },
2544
+ {
2545
+ "epoch": 6.85,
2546
+ "grad_norm": 12.251681327819824,
2547
+ "learning_rate": 6.3052530305945744e-06,
2548
+ "loss": 0.4764,
2549
+ "step": 178000
2550
+ },
2551
+ {
2552
+ "epoch": 6.87,
2553
+ "grad_norm": 7.721076488494873,
2554
+ "learning_rate": 6.266846257456225e-06,
2555
+ "loss": 0.4982,
2556
+ "step": 178500
2557
+ },
2558
+ {
2559
+ "epoch": 6.89,
2560
+ "grad_norm": 3.9712600708007812,
2561
+ "learning_rate": 6.228362516836637e-06,
2562
+ "loss": 0.4859,
2563
+ "step": 179000
2564
+ },
2565
+ {
2566
+ "epoch": 6.91,
2567
+ "grad_norm": 5.2524213790893555,
2568
+ "learning_rate": 6.189878776217049e-06,
2569
+ "loss": 0.4697,
2570
+ "step": 179500
2571
+ },
2572
+ {
2573
+ "epoch": 6.93,
2574
+ "grad_norm": 4.3481340408325195,
2575
+ "learning_rate": 6.151395035597461e-06,
2576
+ "loss": 0.4916,
2577
+ "step": 180000
2578
+ },
2579
+ {
2580
+ "epoch": 6.95,
2581
+ "grad_norm": 5.635584354400635,
2582
+ "learning_rate": 6.1129112949778725e-06,
2583
+ "loss": 0.476,
2584
+ "step": 180500
2585
+ },
2586
+ {
2587
+ "epoch": 6.97,
2588
+ "grad_norm": 4.938673973083496,
2589
+ "learning_rate": 6.074504521839524e-06,
2590
+ "loss": 0.4842,
2591
+ "step": 181000
2592
+ },
2593
+ {
2594
+ "epoch": 6.98,
2595
+ "grad_norm": 3.648951530456543,
2596
+ "learning_rate": 6.036097748701174e-06,
2597
+ "loss": 0.4961,
2598
+ "step": 181500
2599
+ },
2600
+ {
2601
+ "epoch": 7.0,
2602
+ "eval_loss": 1.7367790937423706,
2603
+ "eval_runtime": 11.3431,
2604
+ "eval_samples_per_second": 350.961,
2605
+ "eval_steps_per_second": 87.806,
2606
+ "step": 181895
2607
+ },
2608
+ {
2609
+ "epoch": 7.0,
2610
+ "grad_norm": 6.272878646850586,
2611
+ "learning_rate": 5.997614008081586e-06,
2612
+ "loss": 0.4793,
2613
+ "step": 182000
2614
+ },
2615
+ {
2616
+ "epoch": 7.02,
2617
+ "grad_norm": 8.145177841186523,
2618
+ "learning_rate": 5.959130267461998e-06,
2619
+ "loss": 0.4003,
2620
+ "step": 182500
2621
+ },
2622
+ {
2623
+ "epoch": 7.04,
2624
+ "grad_norm": 3.661898374557495,
2625
+ "learning_rate": 5.92064652684241e-06,
2626
+ "loss": 0.4182,
2627
+ "step": 183000
2628
+ },
2629
+ {
2630
+ "epoch": 7.06,
2631
+ "grad_norm": 4.206645488739014,
2632
+ "learning_rate": 5.882162786222822e-06,
2633
+ "loss": 0.4278,
2634
+ "step": 183500
2635
+ },
2636
+ {
2637
+ "epoch": 7.08,
2638
+ "grad_norm": 2.098565101623535,
2639
+ "learning_rate": 5.843679045603233e-06,
2640
+ "loss": 0.4281,
2641
+ "step": 184000
2642
+ },
2643
+ {
2644
+ "epoch": 7.1,
2645
+ "grad_norm": 5.96292781829834,
2646
+ "learning_rate": 5.805195304983645e-06,
2647
+ "loss": 0.4374,
2648
+ "step": 184500
2649
+ },
2650
+ {
2651
+ "epoch": 7.12,
2652
+ "grad_norm": 2.972966194152832,
2653
+ "learning_rate": 5.766711564364057e-06,
2654
+ "loss": 0.4143,
2655
+ "step": 185000
2656
+ },
2657
+ {
2658
+ "epoch": 7.14,
2659
+ "grad_norm": 9.22352123260498,
2660
+ "learning_rate": 5.7282278237444686e-06,
2661
+ "loss": 0.401,
2662
+ "step": 185500
2663
+ },
2664
+ {
2665
+ "epoch": 7.16,
2666
+ "grad_norm": 8.188728332519531,
2667
+ "learning_rate": 5.689821050606119e-06,
2668
+ "loss": 0.4134,
2669
+ "step": 186000
2670
+ },
2671
+ {
2672
+ "epoch": 7.18,
2673
+ "grad_norm": 8.370118141174316,
2674
+ "learning_rate": 5.65141427746777e-06,
2675
+ "loss": 0.445,
2676
+ "step": 186500
2677
+ },
2678
+ {
2679
+ "epoch": 7.2,
2680
+ "grad_norm": 4.1989922523498535,
2681
+ "learning_rate": 5.612930536848182e-06,
2682
+ "loss": 0.4428,
2683
+ "step": 187000
2684
+ },
2685
+ {
2686
+ "epoch": 7.22,
2687
+ "grad_norm": 8.339373588562012,
2688
+ "learning_rate": 5.574446796228594e-06,
2689
+ "loss": 0.4091,
2690
+ "step": 187500
2691
+ },
2692
+ {
2693
+ "epoch": 7.23,
2694
+ "grad_norm": 5.580036640167236,
2695
+ "learning_rate": 5.535963055609006e-06,
2696
+ "loss": 0.4169,
2697
+ "step": 188000
2698
+ },
2699
+ {
2700
+ "epoch": 7.25,
2701
+ "grad_norm": 5.285128116607666,
2702
+ "learning_rate": 5.4975562824706565e-06,
2703
+ "loss": 0.4282,
2704
+ "step": 188500
2705
+ },
2706
+ {
2707
+ "epoch": 7.27,
2708
+ "grad_norm": 4.649808406829834,
2709
+ "learning_rate": 5.459072541851068e-06,
2710
+ "loss": 0.4199,
2711
+ "step": 189000
2712
+ },
2713
+ {
2714
+ "epoch": 7.29,
2715
+ "grad_norm": 7.711951732635498,
2716
+ "learning_rate": 5.42058880123148e-06,
2717
+ "loss": 0.4331,
2718
+ "step": 189500
2719
+ },
2720
+ {
2721
+ "epoch": 7.31,
2722
+ "grad_norm": 3.5178699493408203,
2723
+ "learning_rate": 5.382105060611892e-06,
2724
+ "loss": 0.4362,
2725
+ "step": 190000
2726
+ },
2727
+ {
2728
+ "epoch": 7.33,
2729
+ "grad_norm": 6.855963230133057,
2730
+ "learning_rate": 5.343621319992304e-06,
2731
+ "loss": 0.449,
2732
+ "step": 190500
2733
+ },
2734
+ {
2735
+ "epoch": 7.35,
2736
+ "grad_norm": 5.831274509429932,
2737
+ "learning_rate": 5.305137579372716e-06,
2738
+ "loss": 0.4286,
2739
+ "step": 191000
2740
+ },
2741
+ {
2742
+ "epoch": 7.37,
2743
+ "grad_norm": 5.696512699127197,
2744
+ "learning_rate": 5.266653838753127e-06,
2745
+ "loss": 0.4388,
2746
+ "step": 191500
2747
+ },
2748
+ {
2749
+ "epoch": 7.39,
2750
+ "grad_norm": 6.7039055824279785,
2751
+ "learning_rate": 5.228170098133539e-06,
2752
+ "loss": 0.4395,
2753
+ "step": 192000
2754
+ },
2755
+ {
2756
+ "epoch": 7.41,
2757
+ "grad_norm": 7.356657028198242,
2758
+ "learning_rate": 5.189686357513951e-06,
2759
+ "loss": 0.4488,
2760
+ "step": 192500
2761
+ },
2762
+ {
2763
+ "epoch": 7.43,
2764
+ "grad_norm": 4.289228439331055,
2765
+ "learning_rate": 5.151279584375601e-06,
2766
+ "loss": 0.4367,
2767
+ "step": 193000
2768
+ },
2769
+ {
2770
+ "epoch": 7.45,
2771
+ "grad_norm": 6.5351948738098145,
2772
+ "learning_rate": 5.112795843756013e-06,
2773
+ "loss": 0.438,
2774
+ "step": 193500
2775
+ },
2776
+ {
2777
+ "epoch": 7.47,
2778
+ "grad_norm": 6.228538990020752,
2779
+ "learning_rate": 5.0743890706176645e-06,
2780
+ "loss": 0.4419,
2781
+ "step": 194000
2782
+ },
2783
+ {
2784
+ "epoch": 7.49,
2785
+ "grad_norm": 8.515239715576172,
2786
+ "learning_rate": 5.035905329998076e-06,
2787
+ "loss": 0.4501,
2788
+ "step": 194500
2789
+ },
2790
+ {
2791
+ "epoch": 7.5,
2792
+ "grad_norm": 6.8445725440979,
2793
+ "learning_rate": 4.9974215893784874e-06,
2794
+ "loss": 0.4257,
2795
+ "step": 195000
2796
+ },
2797
+ {
2798
+ "epoch": 7.52,
2799
+ "grad_norm": 9.162801742553711,
2800
+ "learning_rate": 4.958937848758899e-06,
2801
+ "loss": 0.4349,
2802
+ "step": 195500
2803
+ },
2804
+ {
2805
+ "epoch": 7.54,
2806
+ "grad_norm": 10.447943687438965,
2807
+ "learning_rate": 4.920454108139311e-06,
2808
+ "loss": 0.4364,
2809
+ "step": 196000
2810
+ },
2811
+ {
2812
+ "epoch": 7.56,
2813
+ "grad_norm": 6.493594169616699,
2814
+ "learning_rate": 4.881970367519723e-06,
2815
+ "loss": 0.4436,
2816
+ "step": 196500
2817
+ },
2818
+ {
2819
+ "epoch": 7.58,
2820
+ "grad_norm": 6.790309429168701,
2821
+ "learning_rate": 4.843486626900135e-06,
2822
+ "loss": 0.4535,
2823
+ "step": 197000
2824
+ },
2825
+ {
2826
+ "epoch": 7.6,
2827
+ "grad_norm": 4.3611860275268555,
2828
+ "learning_rate": 4.805002886280547e-06,
2829
+ "loss": 0.443,
2830
+ "step": 197500
2831
+ },
2832
+ {
2833
+ "epoch": 7.62,
2834
+ "grad_norm": 7.639214992523193,
2835
+ "learning_rate": 4.766519145660959e-06,
2836
+ "loss": 0.4433,
2837
+ "step": 198000
2838
+ },
2839
+ {
2840
+ "epoch": 7.64,
2841
+ "grad_norm": 4.556090831756592,
2842
+ "learning_rate": 4.72811237252261e-06,
2843
+ "loss": 0.4433,
2844
+ "step": 198500
2845
+ },
2846
+ {
2847
+ "epoch": 7.66,
2848
+ "grad_norm": 6.17908239364624,
2849
+ "learning_rate": 4.689628631903021e-06,
2850
+ "loss": 0.4389,
2851
+ "step": 199000
2852
+ },
2853
+ {
2854
+ "epoch": 7.68,
2855
+ "grad_norm": 5.435972213745117,
2856
+ "learning_rate": 4.651144891283433e-06,
2857
+ "loss": 0.4413,
2858
+ "step": 199500
2859
+ },
2860
+ {
2861
+ "epoch": 7.7,
2862
+ "grad_norm": 8.422209739685059,
2863
+ "learning_rate": 4.612661150663845e-06,
2864
+ "loss": 0.4444,
2865
+ "step": 200000
2866
+ },
2867
+ {
2868
+ "epoch": 7.72,
2869
+ "grad_norm": 9.431715965270996,
2870
+ "learning_rate": 4.574254377525496e-06,
2871
+ "loss": 0.4495,
2872
+ "step": 200500
2873
+ },
2874
+ {
2875
+ "epoch": 7.74,
2876
+ "grad_norm": 2.9533908367156982,
2877
+ "learning_rate": 4.535847604387147e-06,
2878
+ "loss": 0.4417,
2879
+ "step": 201000
2880
+ },
2881
+ {
2882
+ "epoch": 7.75,
2883
+ "grad_norm": 4.222201824188232,
2884
+ "learning_rate": 4.497363863767559e-06,
2885
+ "loss": 0.4456,
2886
+ "step": 201500
2887
+ },
2888
+ {
2889
+ "epoch": 7.77,
2890
+ "grad_norm": 8.314460754394531,
2891
+ "learning_rate": 4.4588801231479705e-06,
2892
+ "loss": 0.4297,
2893
+ "step": 202000
2894
+ },
2895
+ {
2896
+ "epoch": 7.79,
2897
+ "grad_norm": 8.237698554992676,
2898
+ "learning_rate": 4.4203963825283815e-06,
2899
+ "loss": 0.4637,
2900
+ "step": 202500
2901
+ },
2902
+ {
2903
+ "epoch": 7.81,
2904
+ "grad_norm": 7.0175933837890625,
2905
+ "learning_rate": 4.3819126419087934e-06,
2906
+ "loss": 0.4253,
2907
+ "step": 203000
2908
+ },
2909
+ {
2910
+ "epoch": 7.83,
2911
+ "grad_norm": 9.906578063964844,
2912
+ "learning_rate": 4.343428901289205e-06,
2913
+ "loss": 0.4261,
2914
+ "step": 203500
2915
+ },
2916
+ {
2917
+ "epoch": 7.85,
2918
+ "grad_norm": 2.473310947418213,
2919
+ "learning_rate": 4.304945160669617e-06,
2920
+ "loss": 0.4576,
2921
+ "step": 204000
2922
+ },
2923
+ {
2924
+ "epoch": 7.87,
2925
+ "grad_norm": 7.461137771606445,
2926
+ "learning_rate": 4.266461420050029e-06,
2927
+ "loss": 0.458,
2928
+ "step": 204500
2929
+ },
2930
+ {
2931
+ "epoch": 7.89,
2932
+ "grad_norm": 4.642632007598877,
2933
+ "learning_rate": 4.227977679430441e-06,
2934
+ "loss": 0.4477,
2935
+ "step": 205000
2936
+ },
2937
+ {
2938
+ "epoch": 7.91,
2939
+ "grad_norm": 4.437539100646973,
2940
+ "learning_rate": 4.189570906292092e-06,
2941
+ "loss": 0.4441,
2942
+ "step": 205500
2943
+ },
2944
+ {
2945
+ "epoch": 7.93,
2946
+ "grad_norm": 6.345538139343262,
2947
+ "learning_rate": 4.151087165672504e-06,
2948
+ "loss": 0.4549,
2949
+ "step": 206000
2950
+ },
2951
+ {
2952
+ "epoch": 7.95,
2953
+ "grad_norm": 5.00540828704834,
2954
+ "learning_rate": 4.112603425052915e-06,
2955
+ "loss": 0.4563,
2956
+ "step": 206500
2957
+ },
2958
+ {
2959
+ "epoch": 7.97,
2960
+ "grad_norm": 2.3221421241760254,
2961
+ "learning_rate": 4.074119684433327e-06,
2962
+ "loss": 0.4462,
2963
+ "step": 207000
2964
+ },
2965
+ {
2966
+ "epoch": 7.99,
2967
+ "grad_norm": 4.091787815093994,
2968
+ "learning_rate": 4.0357129112949785e-06,
2969
+ "loss": 0.4588,
2970
+ "step": 207500
2971
+ },
2972
+ {
2973
+ "epoch": 8.0,
2974
+ "eval_loss": 1.752930998802185,
2975
+ "eval_runtime": 11.2624,
2976
+ "eval_samples_per_second": 353.477,
2977
+ "eval_steps_per_second": 88.436,
2978
+ "step": 207880
2979
+ },
2980
+ {
2981
+ "epoch": 8.0,
2982
+ "grad_norm": 10.46442985534668,
2983
+ "learning_rate": 3.99722917067539e-06,
2984
+ "loss": 0.4772,
2985
+ "step": 208000
2986
+ },
2987
+ {
2988
+ "epoch": 8.02,
2989
+ "grad_norm": 2.7908215522766113,
2990
+ "learning_rate": 3.958745430055801e-06,
2991
+ "loss": 0.4165,
2992
+ "step": 208500
2993
+ },
2994
+ {
2995
+ "epoch": 8.04,
2996
+ "grad_norm": 7.39349365234375,
2997
+ "learning_rate": 3.920261689436213e-06,
2998
+ "loss": 0.4127,
2999
+ "step": 209000
3000
+ },
3001
+ {
3002
+ "epoch": 8.06,
3003
+ "grad_norm": 7.740577697753906,
3004
+ "learning_rate": 3.881777948816625e-06,
3005
+ "loss": 0.4235,
3006
+ "step": 209500
3007
+ },
3008
+ {
3009
+ "epoch": 8.08,
3010
+ "grad_norm": 3.9219789505004883,
3011
+ "learning_rate": 3.843294208197037e-06,
3012
+ "loss": 0.4272,
3013
+ "step": 210000
3014
+ },
3015
+ {
3016
+ "epoch": 8.1,
3017
+ "grad_norm": 6.718163967132568,
3018
+ "learning_rate": 3.804887435058688e-06,
3019
+ "loss": 0.4062,
3020
+ "step": 210500
3021
+ },
3022
+ {
3023
+ "epoch": 8.12,
3024
+ "grad_norm": 4.927985191345215,
3025
+ "learning_rate": 3.7664036944391e-06,
3026
+ "loss": 0.4,
3027
+ "step": 211000
3028
+ },
3029
+ {
3030
+ "epoch": 8.14,
3031
+ "grad_norm": 6.605733871459961,
3032
+ "learning_rate": 3.7279199538195113e-06,
3033
+ "loss": 0.4224,
3034
+ "step": 211500
3035
+ },
3036
+ {
3037
+ "epoch": 8.16,
3038
+ "grad_norm": 5.900667190551758,
3039
+ "learning_rate": 3.6894362131999232e-06,
3040
+ "loss": 0.401,
3041
+ "step": 212000
3042
+ },
3043
+ {
3044
+ "epoch": 8.18,
3045
+ "grad_norm": 3.047395706176758,
3046
+ "learning_rate": 3.651029440061574e-06,
3047
+ "loss": 0.4099,
3048
+ "step": 212500
3049
+ },
3050
+ {
3051
+ "epoch": 8.2,
3052
+ "grad_norm": 10.504358291625977,
3053
+ "learning_rate": 3.612545699441986e-06,
3054
+ "loss": 0.4094,
3055
+ "step": 213000
3056
+ },
3057
+ {
3058
+ "epoch": 8.22,
3059
+ "grad_norm": 2.9013960361480713,
3060
+ "learning_rate": 3.574061958822398e-06,
3061
+ "loss": 0.4225,
3062
+ "step": 213500
3063
+ },
3064
+ {
3065
+ "epoch": 8.24,
3066
+ "grad_norm": 1.0684086084365845,
3067
+ "learning_rate": 3.53557821820281e-06,
3068
+ "loss": 0.4333,
3069
+ "step": 214000
3070
+ },
3071
+ {
3072
+ "epoch": 8.25,
3073
+ "grad_norm": 1.7722007036209106,
3074
+ "learning_rate": 3.4970944775832217e-06,
3075
+ "loss": 0.419,
3076
+ "step": 214500
3077
+ },
3078
+ {
3079
+ "epoch": 8.27,
3080
+ "grad_norm": 5.168225288391113,
3081
+ "learning_rate": 3.4586107369636327e-06,
3082
+ "loss": 0.429,
3083
+ "step": 215000
3084
+ },
3085
+ {
3086
+ "epoch": 8.29,
3087
+ "grad_norm": 9.540655136108398,
3088
+ "learning_rate": 3.4201269963440446e-06,
3089
+ "loss": 0.4024,
3090
+ "step": 215500
3091
+ },
3092
+ {
3093
+ "epoch": 8.31,
3094
+ "grad_norm": 7.926085472106934,
3095
+ "learning_rate": 3.3817202232056955e-06,
3096
+ "loss": 0.4081,
3097
+ "step": 216000
3098
+ },
3099
+ {
3100
+ "epoch": 8.33,
3101
+ "grad_norm": 3.9807188510894775,
3102
+ "learning_rate": 3.3432364825861074e-06,
3103
+ "loss": 0.4098,
3104
+ "step": 216500
3105
+ },
3106
+ {
3107
+ "epoch": 8.35,
3108
+ "grad_norm": 4.447396755218506,
3109
+ "learning_rate": 3.3047527419665193e-06,
3110
+ "loss": 0.4045,
3111
+ "step": 217000
3112
+ },
3113
+ {
3114
+ "epoch": 8.37,
3115
+ "grad_norm": 5.585531711578369,
3116
+ "learning_rate": 3.266269001346931e-06,
3117
+ "loss": 0.409,
3118
+ "step": 217500
3119
+ },
3120
+ {
3121
+ "epoch": 8.39,
3122
+ "grad_norm": 4.316734790802002,
3123
+ "learning_rate": 3.227862228208582e-06,
3124
+ "loss": 0.4239,
3125
+ "step": 218000
3126
+ },
3127
+ {
3128
+ "epoch": 8.41,
3129
+ "grad_norm": 7.370926380157471,
3130
+ "learning_rate": 3.189378487588994e-06,
3131
+ "loss": 0.4104,
3132
+ "step": 218500
3133
+ },
3134
+ {
3135
+ "epoch": 8.43,
3136
+ "grad_norm": 6.9890522956848145,
3137
+ "learning_rate": 3.1508947469694055e-06,
3138
+ "loss": 0.4024,
3139
+ "step": 219000
3140
+ },
3141
+ {
3142
+ "epoch": 8.45,
3143
+ "grad_norm": 5.685979843139648,
3144
+ "learning_rate": 3.1124110063498174e-06,
3145
+ "loss": 0.4167,
3146
+ "step": 219500
3147
+ },
3148
+ {
3149
+ "epoch": 8.47,
3150
+ "grad_norm": 5.407013416290283,
3151
+ "learning_rate": 3.0739272657302293e-06,
3152
+ "loss": 0.3967,
3153
+ "step": 220000
3154
+ },
3155
+ {
3156
+ "epoch": 8.49,
3157
+ "grad_norm": 3.3302626609802246,
3158
+ "learning_rate": 3.03552049259188e-06,
3159
+ "loss": 0.4303,
3160
+ "step": 220500
3161
+ },
3162
+ {
3163
+ "epoch": 8.5,
3164
+ "grad_norm": 9.727193832397461,
3165
+ "learning_rate": 2.997113719453531e-06,
3166
+ "loss": 0.4108,
3167
+ "step": 221000
3168
+ },
3169
+ {
3170
+ "epoch": 8.52,
3171
+ "grad_norm": 5.004329681396484,
3172
+ "learning_rate": 2.958629978833943e-06,
3173
+ "loss": 0.4135,
3174
+ "step": 221500
3175
+ },
3176
+ {
3177
+ "epoch": 8.54,
3178
+ "grad_norm": 5.899045467376709,
3179
+ "learning_rate": 2.920146238214355e-06,
3180
+ "loss": 0.4214,
3181
+ "step": 222000
3182
+ },
3183
+ {
3184
+ "epoch": 8.56,
3185
+ "grad_norm": 9.472413063049316,
3186
+ "learning_rate": 2.8816624975947667e-06,
3187
+ "loss": 0.4063,
3188
+ "step": 222500
3189
+ },
3190
+ {
3191
+ "epoch": 8.58,
3192
+ "grad_norm": 3.2998526096343994,
3193
+ "learning_rate": 2.8431787569751786e-06,
3194
+ "loss": 0.4152,
3195
+ "step": 223000
3196
+ },
3197
+ {
3198
+ "epoch": 8.6,
3199
+ "grad_norm": 9.45439338684082,
3200
+ "learning_rate": 2.8046950163555897e-06,
3201
+ "loss": 0.4156,
3202
+ "step": 223500
3203
+ },
3204
+ {
3205
+ "epoch": 8.62,
3206
+ "grad_norm": 10.916293144226074,
3207
+ "learning_rate": 2.7662112757360016e-06,
3208
+ "loss": 0.4177,
3209
+ "step": 224000
3210
+ },
3211
+ {
3212
+ "epoch": 8.64,
3213
+ "grad_norm": 9.067635536193848,
3214
+ "learning_rate": 2.7277275351164134e-06,
3215
+ "loss": 0.4107,
3216
+ "step": 224500
3217
+ },
3218
+ {
3219
+ "epoch": 8.66,
3220
+ "grad_norm": 8.393898010253906,
3221
+ "learning_rate": 2.6892437944968253e-06,
3222
+ "loss": 0.4167,
3223
+ "step": 225000
3224
+ },
3225
+ {
3226
+ "epoch": 8.68,
3227
+ "grad_norm": 4.151984691619873,
3228
+ "learning_rate": 2.6508370213584762e-06,
3229
+ "loss": 0.415,
3230
+ "step": 225500
3231
+ },
3232
+ {
3233
+ "epoch": 8.7,
3234
+ "grad_norm": 5.011842250823975,
3235
+ "learning_rate": 2.612353280738888e-06,
3236
+ "loss": 0.4358,
3237
+ "step": 226000
3238
+ },
3239
+ {
3240
+ "epoch": 8.72,
3241
+ "grad_norm": 2.5369720458984375,
3242
+ "learning_rate": 2.5738695401192996e-06,
3243
+ "loss": 0.4114,
3244
+ "step": 226500
3245
+ },
3246
+ {
3247
+ "epoch": 8.74,
3248
+ "grad_norm": 6.167263507843018,
3249
+ "learning_rate": 2.5353857994997115e-06,
3250
+ "loss": 0.4117,
3251
+ "step": 227000
3252
+ },
3253
+ {
3254
+ "epoch": 8.76,
3255
+ "grad_norm": 3.8160266876220703,
3256
+ "learning_rate": 2.4969020588801234e-06,
3257
+ "loss": 0.4195,
3258
+ "step": 227500
3259
+ },
3260
+ {
3261
+ "epoch": 8.77,
3262
+ "grad_norm": 6.264139175415039,
3263
+ "learning_rate": 2.4584952857417743e-06,
3264
+ "loss": 0.4182,
3265
+ "step": 228000
3266
+ },
3267
+ {
3268
+ "epoch": 8.79,
3269
+ "grad_norm": 2.6473159790039062,
3270
+ "learning_rate": 2.420011545122186e-06,
3271
+ "loss": 0.4109,
3272
+ "step": 228500
3273
+ },
3274
+ {
3275
+ "epoch": 8.81,
3276
+ "grad_norm": 5.942705154418945,
3277
+ "learning_rate": 2.3815278045025976e-06,
3278
+ "loss": 0.4385,
3279
+ "step": 229000
3280
+ },
3281
+ {
3282
+ "epoch": 8.83,
3283
+ "grad_norm": 6.846287250518799,
3284
+ "learning_rate": 2.3430440638830095e-06,
3285
+ "loss": 0.4406,
3286
+ "step": 229500
3287
+ },
3288
+ {
3289
+ "epoch": 8.85,
3290
+ "grad_norm": 3.9785611629486084,
3291
+ "learning_rate": 2.3045603232634214e-06,
3292
+ "loss": 0.4154,
3293
+ "step": 230000
3294
+ },
3295
+ {
3296
+ "epoch": 8.87,
3297
+ "grad_norm": 4.567147254943848,
3298
+ "learning_rate": 2.2661535501250723e-06,
3299
+ "loss": 0.4214,
3300
+ "step": 230500
3301
+ },
3302
+ {
3303
+ "epoch": 8.89,
3304
+ "grad_norm": 7.9429240226745605,
3305
+ "learning_rate": 2.227669809505484e-06,
3306
+ "loss": 0.446,
3307
+ "step": 231000
3308
+ },
3309
+ {
3310
+ "epoch": 8.91,
3311
+ "grad_norm": 6.396302223205566,
3312
+ "learning_rate": 2.189186068885896e-06,
3313
+ "loss": 0.4306,
3314
+ "step": 231500
3315
+ },
3316
+ {
3317
+ "epoch": 8.93,
3318
+ "grad_norm": 7.975338935852051,
3319
+ "learning_rate": 2.1507023282663076e-06,
3320
+ "loss": 0.419,
3321
+ "step": 232000
3322
+ },
3323
+ {
3324
+ "epoch": 8.95,
3325
+ "grad_norm": 6.531688690185547,
3326
+ "learning_rate": 2.1122955551279585e-06,
3327
+ "loss": 0.4144,
3328
+ "step": 232500
3329
+ },
3330
+ {
3331
+ "epoch": 8.97,
3332
+ "grad_norm": 4.969785213470459,
3333
+ "learning_rate": 2.0738118145083704e-06,
3334
+ "loss": 0.4283,
3335
+ "step": 233000
3336
+ },
3337
+ {
3338
+ "epoch": 8.99,
3339
+ "grad_norm": 7.7668538093566895,
3340
+ "learning_rate": 2.0353280738887822e-06,
3341
+ "loss": 0.4025,
3342
+ "step": 233500
3343
+ },
3344
+ {
3345
+ "epoch": 9.0,
3346
+ "eval_loss": 1.7583389282226562,
3347
+ "eval_runtime": 11.3764,
3348
+ "eval_samples_per_second": 349.935,
3349
+ "eval_steps_per_second": 87.55,
3350
+ "step": 233865
3351
+ },
3352
+ {
3353
+ "epoch": 9.01,
3354
+ "grad_norm": 6.1113409996032715,
3355
+ "learning_rate": 1.9968443332691937e-06,
3356
+ "loss": 0.4198,
3357
+ "step": 234000
3358
+ },
3359
+ {
3360
+ "epoch": 9.02,
3361
+ "grad_norm": 6.071191787719727,
3362
+ "learning_rate": 1.9583605926496056e-06,
3363
+ "loss": 0.3948,
3364
+ "step": 234500
3365
+ },
3366
+ {
3367
+ "epoch": 9.04,
3368
+ "grad_norm": 2.7634339332580566,
3369
+ "learning_rate": 1.9198768520300175e-06,
3370
+ "loss": 0.3819,
3371
+ "step": 235000
3372
+ },
3373
+ {
3374
+ "epoch": 9.06,
3375
+ "grad_norm": 5.456170082092285,
3376
+ "learning_rate": 1.8813931114104292e-06,
3377
+ "loss": 0.3905,
3378
+ "step": 235500
3379
+ },
3380
+ {
3381
+ "epoch": 9.08,
3382
+ "grad_norm": 6.395251750946045,
3383
+ "learning_rate": 1.842909370790841e-06,
3384
+ "loss": 0.3947,
3385
+ "step": 236000
3386
+ },
3387
+ {
3388
+ "epoch": 9.1,
3389
+ "grad_norm": 4.872887134552002,
3390
+ "learning_rate": 1.804425630171253e-06,
3391
+ "loss": 0.3775,
3392
+ "step": 236500
3393
+ },
3394
+ {
3395
+ "epoch": 9.12,
3396
+ "grad_norm": 7.3761162757873535,
3397
+ "learning_rate": 1.7660188570329036e-06,
3398
+ "loss": 0.3947,
3399
+ "step": 237000
3400
+ },
3401
+ {
3402
+ "epoch": 9.14,
3403
+ "grad_norm": 4.171831130981445,
3404
+ "learning_rate": 1.7275351164133155e-06,
3405
+ "loss": 0.4119,
3406
+ "step": 237500
3407
+ },
3408
+ {
3409
+ "epoch": 9.16,
3410
+ "grad_norm": 2.4952399730682373,
3411
+ "learning_rate": 1.6890513757937274e-06,
3412
+ "loss": 0.394,
3413
+ "step": 238000
3414
+ },
3415
+ {
3416
+ "epoch": 9.18,
3417
+ "grad_norm": 6.803173542022705,
3418
+ "learning_rate": 1.650567635174139e-06,
3419
+ "loss": 0.4007,
3420
+ "step": 238500
3421
+ },
3422
+ {
3423
+ "epoch": 9.2,
3424
+ "grad_norm": 3.5348174571990967,
3425
+ "learning_rate": 1.6120838945545508e-06,
3426
+ "loss": 0.4027,
3427
+ "step": 239000
3428
+ },
3429
+ {
3430
+ "epoch": 9.22,
3431
+ "grad_norm": 5.090949058532715,
3432
+ "learning_rate": 1.5736771214162017e-06,
3433
+ "loss": 0.3902,
3434
+ "step": 239500
3435
+ },
3436
+ {
3437
+ "epoch": 9.24,
3438
+ "grad_norm": 12.865615844726562,
3439
+ "learning_rate": 1.5351933807966136e-06,
3440
+ "loss": 0.4002,
3441
+ "step": 240000
3442
+ },
3443
+ {
3444
+ "epoch": 9.26,
3445
+ "grad_norm": 8.621195793151855,
3446
+ "learning_rate": 1.4967096401770253e-06,
3447
+ "loss": 0.4197,
3448
+ "step": 240500
3449
+ },
3450
+ {
3451
+ "epoch": 9.27,
3452
+ "grad_norm": 6.878337383270264,
3453
+ "learning_rate": 1.4582258995574371e-06,
3454
+ "loss": 0.4024,
3455
+ "step": 241000
3456
+ },
3457
+ {
3458
+ "epoch": 9.29,
3459
+ "grad_norm": 6.962910175323486,
3460
+ "learning_rate": 1.4197421589378488e-06,
3461
+ "loss": 0.4079,
3462
+ "step": 241500
3463
+ },
3464
+ {
3465
+ "epoch": 9.31,
3466
+ "grad_norm": 4.2871174812316895,
3467
+ "learning_rate": 1.3813353857995e-06,
3468
+ "loss": 0.3943,
3469
+ "step": 242000
3470
+ },
3471
+ {
3472
+ "epoch": 9.33,
3473
+ "grad_norm": 3.1228811740875244,
3474
+ "learning_rate": 1.3428516451799118e-06,
3475
+ "loss": 0.3982,
3476
+ "step": 242500
3477
+ },
3478
+ {
3479
+ "epoch": 9.35,
3480
+ "grad_norm": 4.895651340484619,
3481
+ "learning_rate": 1.3043679045603233e-06,
3482
+ "loss": 0.3939,
3483
+ "step": 243000
3484
+ },
3485
+ {
3486
+ "epoch": 9.37,
3487
+ "grad_norm": 3.8321521282196045,
3488
+ "learning_rate": 1.2658841639407352e-06,
3489
+ "loss": 0.4048,
3490
+ "step": 243500
3491
+ },
3492
+ {
3493
+ "epoch": 9.39,
3494
+ "grad_norm": 4.628911972045898,
3495
+ "learning_rate": 1.2274004233211469e-06,
3496
+ "loss": 0.4099,
3497
+ "step": 244000
3498
+ },
3499
+ {
3500
+ "epoch": 9.41,
3501
+ "grad_norm": 4.80955696105957,
3502
+ "learning_rate": 1.1889166827015588e-06,
3503
+ "loss": 0.3891,
3504
+ "step": 244500
3505
+ },
3506
+ {
3507
+ "epoch": 9.43,
3508
+ "grad_norm": 7.214938163757324,
3509
+ "learning_rate": 1.1504329420819704e-06,
3510
+ "loss": 0.4,
3511
+ "step": 245000
3512
+ },
3513
+ {
3514
+ "epoch": 9.45,
3515
+ "grad_norm": 5.537662982940674,
3516
+ "learning_rate": 1.1120261689436213e-06,
3517
+ "loss": 0.4052,
3518
+ "step": 245500
3519
+ },
3520
+ {
3521
+ "epoch": 9.47,
3522
+ "grad_norm": 4.672789573669434,
3523
+ "learning_rate": 1.0736193958052725e-06,
3524
+ "loss": 0.4121,
3525
+ "step": 246000
3526
+ },
3527
+ {
3528
+ "epoch": 9.49,
3529
+ "grad_norm": 7.284801959991455,
3530
+ "learning_rate": 1.0351356551856841e-06,
3531
+ "loss": 0.4168,
3532
+ "step": 246500
3533
+ },
3534
+ {
3535
+ "epoch": 9.51,
3536
+ "grad_norm": 8.112405776977539,
3537
+ "learning_rate": 9.96651914566096e-07,
3538
+ "loss": 0.3951,
3539
+ "step": 247000
3540
+ },
3541
+ {
3542
+ "epoch": 9.52,
3543
+ "grad_norm": 5.421095848083496,
3544
+ "learning_rate": 9.581681739465077e-07,
3545
+ "loss": 0.384,
3546
+ "step": 247500
3547
+ },
3548
+ {
3549
+ "epoch": 9.54,
3550
+ "grad_norm": 6.66250467300415,
3551
+ "learning_rate": 9.196844333269195e-07,
3552
+ "loss": 0.4037,
3553
+ "step": 248000
3554
+ },
3555
+ {
3556
+ "epoch": 9.56,
3557
+ "grad_norm": 7.84215784072876,
3558
+ "learning_rate": 8.812006927073312e-07,
3559
+ "loss": 0.4025,
3560
+ "step": 248500
3561
+ },
3562
+ {
3563
+ "epoch": 9.58,
3564
+ "grad_norm": 4.539148807525635,
3565
+ "learning_rate": 8.427169520877431e-07,
3566
+ "loss": 0.4072,
3567
+ "step": 249000
3568
+ },
3569
+ {
3570
+ "epoch": 9.6,
3571
+ "grad_norm": 4.91790246963501,
3572
+ "learning_rate": 8.042332114681547e-07,
3573
+ "loss": 0.3853,
3574
+ "step": 249500
3575
+ },
3576
+ {
3577
+ "epoch": 9.62,
3578
+ "grad_norm": 7.577627658843994,
3579
+ "learning_rate": 7.658264383298057e-07,
3580
+ "loss": 0.3997,
3581
+ "step": 250000
3582
+ },
3583
+ {
3584
+ "epoch": 9.64,
3585
+ "grad_norm": 14.857439994812012,
3586
+ "learning_rate": 7.273426977102174e-07,
3587
+ "loss": 0.3852,
3588
+ "step": 250500
3589
+ },
3590
+ {
3591
+ "epoch": 9.66,
3592
+ "grad_norm": 3.0629687309265137,
3593
+ "learning_rate": 6.888589570906293e-07,
3594
+ "loss": 0.3887,
3595
+ "step": 251000
3596
+ },
3597
+ {
3598
+ "epoch": 9.68,
3599
+ "grad_norm": 3.039710760116577,
3600
+ "learning_rate": 6.50375216471041e-07,
3601
+ "loss": 0.4124,
3602
+ "step": 251500
3603
+ },
3604
+ {
3605
+ "epoch": 9.7,
3606
+ "grad_norm": 7.001966953277588,
3607
+ "learning_rate": 6.11968443332692e-07,
3608
+ "loss": 0.4102,
3609
+ "step": 252000
3610
+ },
3611
+ {
3612
+ "epoch": 9.72,
3613
+ "grad_norm": 3.581230878829956,
3614
+ "learning_rate": 5.734847027131038e-07,
3615
+ "loss": 0.3916,
3616
+ "step": 252500
3617
+ },
3618
+ {
3619
+ "epoch": 9.74,
3620
+ "grad_norm": 3.673539638519287,
3621
+ "learning_rate": 5.350009620935155e-07,
3622
+ "loss": 0.3973,
3623
+ "step": 253000
3624
+ },
3625
+ {
3626
+ "epoch": 9.76,
3627
+ "grad_norm": 9.442609786987305,
3628
+ "learning_rate": 4.965172214739272e-07,
3629
+ "loss": 0.3913,
3630
+ "step": 253500
3631
+ },
3632
+ {
3633
+ "epoch": 9.77,
3634
+ "grad_norm": 4.602294921875,
3635
+ "learning_rate": 4.5811044833557826e-07,
3636
+ "loss": 0.3989,
3637
+ "step": 254000
3638
+ },
3639
+ {
3640
+ "epoch": 9.79,
3641
+ "grad_norm": 2.9956846237182617,
3642
+ "learning_rate": 4.1962670771599004e-07,
3643
+ "loss": 0.4018,
3644
+ "step": 254500
3645
+ },
3646
+ {
3647
+ "epoch": 9.81,
3648
+ "grad_norm": 5.884944438934326,
3649
+ "learning_rate": 3.811429670964018e-07,
3650
+ "loss": 0.3974,
3651
+ "step": 255000
3652
+ },
3653
+ {
3654
+ "epoch": 9.83,
3655
+ "grad_norm": 3.6451961994171143,
3656
+ "learning_rate": 3.426592264768136e-07,
3657
+ "loss": 0.3981,
3658
+ "step": 255500
3659
+ },
3660
+ {
3661
+ "epoch": 9.85,
3662
+ "grad_norm": 5.324589729309082,
3663
+ "learning_rate": 3.0417548585722534e-07,
3664
+ "loss": 0.3885,
3665
+ "step": 256000
3666
+ },
3667
+ {
3668
+ "epoch": 9.87,
3669
+ "grad_norm": 5.142848968505859,
3670
+ "learning_rate": 2.6569174523763713e-07,
3671
+ "loss": 0.4074,
3672
+ "step": 256500
3673
+ },
3674
+ {
3675
+ "epoch": 9.89,
3676
+ "grad_norm": 2.519761085510254,
3677
+ "learning_rate": 2.272080046180489e-07,
3678
+ "loss": 0.416,
3679
+ "step": 257000
3680
+ },
3681
+ {
3682
+ "epoch": 9.91,
3683
+ "grad_norm": 8.230846405029297,
3684
+ "learning_rate": 1.8880123147969984e-07,
3685
+ "loss": 0.4173,
3686
+ "step": 257500
3687
+ },
3688
+ {
3689
+ "epoch": 9.93,
3690
+ "grad_norm": 8.473825454711914,
3691
+ "learning_rate": 1.5031749086011162e-07,
3692
+ "loss": 0.3753,
3693
+ "step": 258000
3694
+ },
3695
+ {
3696
+ "epoch": 9.95,
3697
+ "grad_norm": 9.58931827545166,
3698
+ "learning_rate": 1.1183375024052338e-07,
3699
+ "loss": 0.4112,
3700
+ "step": 258500
3701
+ },
3702
+ {
3703
+ "epoch": 9.97,
3704
+ "grad_norm": 7.562945365905762,
3705
+ "learning_rate": 7.335000962093516e-08,
3706
+ "loss": 0.3932,
3707
+ "step": 259000
3708
+ },
3709
+ {
3710
+ "epoch": 9.99,
3711
+ "grad_norm": 3.519716501235962,
3712
+ "learning_rate": 3.4866269001346934e-08,
3713
+ "loss": 0.4157,
3714
+ "step": 259500
3715
+ },
3716
+ {
3717
+ "epoch": 10.0,
3718
+ "eval_loss": 1.769174575805664,
3719
+ "eval_runtime": 11.3755,
3720
+ "eval_samples_per_second": 349.964,
3721
+ "eval_steps_per_second": 87.557,
3722
+ "step": 259850
3723
+ }
3724
+ ],
3725
+ "logging_steps": 500,
3726
+ "max_steps": 259850,
3727
+ "num_input_tokens_seen": 0,
3728
+ "num_train_epochs": 10,
3729
+ "save_steps": 500,
3730
+ "total_flos": 1.0298513413767168e+16,
3731
+ "train_batch_size": 4,
3732
+ "trial_name": null,
3733
+ "trial_params": null
3734
+ }
checkpoint-259850/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f10481b97b85db779f30bdfe4858bc0eccd200747cd750186b7f3c88a54eeb56
3
+ size 4667
checkpoint-259850/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
config.json ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "opus-mt-zh-en-finetuned-zhen-checkpoints/checkpoint-51970",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "swish",
5
+ "add_bias_logits": false,
6
+ "add_final_layer_norm": false,
7
+ "architectures": [
8
+ "MarianMTModel"
9
+ ],
10
+ "attention_dropout": 0.0,
11
+ "bad_words_ids": [
12
+ [
13
+ 65000
14
+ ]
15
+ ],
16
+ "bos_token_id": 0,
17
+ "classif_dropout": 0.0,
18
+ "classifier_dropout": 0.0,
19
+ "d_model": 512,
20
+ "decoder_attention_heads": 8,
21
+ "decoder_ffn_dim": 2048,
22
+ "decoder_layerdrop": 0.0,
23
+ "decoder_layers": 6,
24
+ "decoder_start_token_id": 65000,
25
+ "decoder_vocab_size": 65001,
26
+ "dropout": 0.1,
27
+ "encoder_attention_heads": 8,
28
+ "encoder_ffn_dim": 2048,
29
+ "encoder_layerdrop": 0.0,
30
+ "encoder_layers": 6,
31
+ "eos_token_id": 0,
32
+ "extra_pos_embeddings": 65001,
33
+ "forced_eos_token_id": 0,
34
+ "id2label": {
35
+ "0": "LABEL_0",
36
+ "1": "LABEL_1",
37
+ "2": "LABEL_2"
38
+ },
39
+ "init_std": 0.02,
40
+ "is_encoder_decoder": true,
41
+ "label2id": {
42
+ "LABEL_0": 0,
43
+ "LABEL_1": 1,
44
+ "LABEL_2": 2
45
+ },
46
+ "max_length": 512,
47
+ "max_position_embeddings": 512,
48
+ "model_type": "marian",
49
+ "normalize_before": false,
50
+ "normalize_embedding": false,
51
+ "num_beams": 6,
52
+ "num_hidden_layers": 6,
53
+ "pad_token_id": 65000,
54
+ "scale_embedding": true,
55
+ "share_encoder_decoder_embeddings": true,
56
+ "static_position_embeddings": true,
57
+ "torch_dtype": "float32",
58
+ "transformers_version": "4.39.3",
59
+ "use_cache": true,
60
+ "vocab_size": 65001
61
+ }
generation_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bad_words_ids": [
3
+ [
4
+ 65000
5
+ ]
6
+ ],
7
+ "bos_token_id": 0,
8
+ "decoder_start_token_id": 65000,
9
+ "eos_token_id": 0,
10
+ "forced_eos_token_id": 0,
11
+ "max_length": 512,
12
+ "num_beams": 6,
13
+ "pad_token_id": 65000,
14
+ "renormalize_logits": true,
15
+ "transformers_version": "4.39.3"
16
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca3579c434713fc47d13e38d92ff2f0470237dc5bdcf686cdd8e18a6dc4ead80
3
+ size 309965092
runs/Apr16_07-58-43_129-213-63-124/events.out.tfevents.1713254333.129-213-63-124.2528.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4eab0cb3f0101eb4914f830d3bba207b8399c0c594fa386f413875758b201890
3
+ size 28561
runs/Apr16_12-36-45_129-213-89-192/events.out.tfevents.1713271010.129-213-89-192.2749.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7f3fa24ebd0e4b2c33ea176d17f53d0d090cb1bbc828751c1824468e52cfc70
3
+ size 6254
runs/Apr16_12-38-25_129-213-89-192/events.out.tfevents.1713271109.129-213-89-192.2749.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64c3ee391ebe8af7e05c05daeb4b4c25dee51192474a2b7a7b34623aa46971d0
3
+ size 120198
source.spm ADDED
Binary file (805 kB). View file
 
special_tokens_map.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "eos_token": "</s>",
3
+ "pad_token": "<pad>",
4
+ "unk_token": "<unk>"
5
+ }
target.spm ADDED
Binary file (807 kB). View file
 
tokenizer_config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "</s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<unk>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "65000": {
20
+ "content": "<pad>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ }
27
+ },
28
+ "clean_up_tokenization_spaces": true,
29
+ "eos_token": "</s>",
30
+ "model_max_length": 512,
31
+ "pad_token": "<pad>",
32
+ "separate_vocabs": false,
33
+ "source_lang": "zho",
34
+ "sp_model_kwargs": {},
35
+ "target_lang": "eng",
36
+ "tokenizer_class": "MarianTokenizer",
37
+ "unk_token": "<unk>"
38
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff