Hugging Face's logo

Leonardo6
/
tinyllava-collection

Safetensors
Model card Files
xet
Community
Leonardo6 commited on
Commit
a1a58ac
·
verified ·
1 Parent(s): 3d8e79c

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +6 -0
  2. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/config.json +108 -0
  3. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/generation_config.json +7 -0
  4. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/model.safetensors +3 -0
  5. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/preprocessor_config.json +27 -0
  6. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/trainer_state.json +0 -0
  7. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/training_args.bin +3 -0
  8. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/config.json +108 -0
  9. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/generation_config.json +7 -0
  10. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/model.safetensors +3 -0
  11. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/preprocessor_config.json +27 -0
  12. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/trainer_state.json +3534 -0
  13. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/training_args.bin +3 -0
  14. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/log.txt +4 -0
  15. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/added_tokens.json +24 -0
  16. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/chat_template.jinja +54 -0
  17. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/config.json +104 -0
  18. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/connector/pytorch_model.bin +3 -0
  19. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/language_model/config.json +54 -0
  20. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/language_model/pytorch_model.bin +3 -0
  21. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/log.txt +4 -0
  22. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/merges.txt +0 -0
  23. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/special_tokens_map.json +32 -0
  24. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/tokenizer.json +3 -0
  25. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/tokenizer_config.json +208 -0
  26. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/trainer_state.json +0 -0
  27. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/vision_tower/config.json +19 -0
  28. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/vision_tower/pytorch_model.bin +3 -0
  29. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/vocab.json +0 -0
  30. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/config.json +108 -0
  31. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/generation_config.json +7 -0
  32. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  33. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  34. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  35. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  36. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_model_states.pt +3 -0
  37. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_model_states.pt +3 -0
  38. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_model_states.pt +3 -0
  39. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_model_states.pt +3 -0
  40. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/latest +1 -0
  41. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/model.safetensors +3 -0
  42. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/preprocessor_config.json +27 -0
  43. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_0.pth +3 -0
  44. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_1.pth +3 -0
  45. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_2.pth +3 -0
  46. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_3.pth +3 -0
  47. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/scheduler.pt +3 -0
  48. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/trainer_state.json +0 -0
  49. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/training_args.bin +3 -0
  50. tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/zero_to_fp32.py +760 -0
.gitattributes CHANGED
@@ -33,3 +33,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/tokenizer.json filter=lfs diff=lfs merge=lfs -text
37
+ tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-pretrain/tokenizer.json filter=lfs diff=lfs merge=lfs -text
38
+ tiny-llava-Qwen2.5-1.5B-Instruct-clip-vit-base-patch16-qwen2.5-1.5b_instruct_patch-pretrain/tokenizer.json filter=lfs diff=lfs merge=lfs -text
39
+ tiny-llava-Qwen2.5-3B-Instruct-clip-vit-large-patch14-336-qwen2.5-3b_instruct-pretrain/tokenizer.json filter=lfs diff=lfs merge=lfs -text
40
+ tiny-llava-Qwen2.5-7B-Instruct-clip-vit-large-patch14-336-qwen2.5-7b_instruct-pretrain/tokenizer.json filter=lfs diff=lfs merge=lfs -text
41
+ tiny-llava-gemma-2b-it-clip-vit-large-patch14-336-gemma-pretrain/tokenizer.json filter=lfs diff=lfs merge=lfs -text
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "TinyLlavaForConditionalGeneration"
4
+ ],
5
+ "cache_dir": null,
6
+ "connector_type": "mlp2x_gelu",
7
+ "hidden_size": 896,
8
+ "ignore_index": -100,
9
+ "image_aspect_ratio": "square",
10
+ "image_token_index": -200,
11
+ "llm_model_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
12
+ "model_type": "tinyllava",
13
+ "num_queries": 128,
14
+ "num_resampler_layers": 3,
15
+ "pad_token": "<|endoftext|>",
16
+ "pad_token_id": 151643,
17
+ "resampler_hidden_size": 768,
18
+ "text_config": {
19
+ "_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
20
+ "architectures": [
21
+ "Qwen2ForCausalLM"
22
+ ],
23
+ "attention_dropout": 0.0,
24
+ "bos_token_id": 151643,
25
+ "eos_token_id": 151645,
26
+ "hidden_act": "silu",
27
+ "hidden_size": 896,
28
+ "initializer_range": 0.02,
29
+ "intermediate_size": 4864,
30
+ "layer_types": [
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention",
53
+ "full_attention",
54
+ "full_attention"
55
+ ],
56
+ "max_position_embeddings": 32768,
57
+ "max_window_layers": 21,
58
+ "model_type": "qwen2",
59
+ "num_attention_heads": 14,
60
+ "num_hidden_layers": 24,
61
+ "num_key_value_heads": 2,
62
+ "rms_norm_eps": 1e-06,
63
+ "rope_scaling": null,
64
+ "rope_theta": 1000000.0,
65
+ "sliding_window": null,
66
+ "tie_word_embeddings": true,
67
+ "torch_dtype": "bfloat16",
68
+ "use_cache": true,
69
+ "use_sliding_window": false,
70
+ "vocab_size": 151936
71
+ },
72
+ "tokenizer_model_max_length": 2048,
73
+ "tokenizer_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
74
+ "tokenizer_padding_side": "right",
75
+ "tokenizer_use_fast": true,
76
+ "torch_dtype": "bfloat16",
77
+ "transformers_version": "4.55.4",
78
+ "tune_type_connector": "full",
79
+ "tune_type_llm": "full",
80
+ "tune_type_vision_tower": "frozen",
81
+ "tune_vision_tower_from_layer": 0,
82
+ "use_cache": false,
83
+ "vision_config": {
84
+ "attention_dropout": 0.0,
85
+ "dropout": 0.0,
86
+ "hidden_act": "quick_gelu",
87
+ "hidden_size": 768,
88
+ "image_size": 224,
89
+ "initializer_factor": 1.0,
90
+ "initializer_range": 0.02,
91
+ "intermediate_size": 3072,
92
+ "layer_norm_eps": 1e-05,
93
+ "model_name_or_path": "openai/clip-vit-base-patch32",
94
+ "model_name_or_path2": "",
95
+ "model_type": "clip_vision_model",
96
+ "num_attention_heads": 12,
97
+ "num_channels": 3,
98
+ "num_hidden_layers": 12,
99
+ "patch_size": 32,
100
+ "projection_dim": 512
101
+ },
102
+ "vision_feature_layer": -2,
103
+ "vision_feature_select_strategy": "patch",
104
+ "vision_hidden_size": 768,
105
+ "vision_model_name_or_path": "openai/clip-vit-base-patch32",
106
+ "vision_model_name_or_path2": "",
107
+ "vocab_size": 151936
108
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 151643,
4
+ "eos_token_id": 151645,
5
+ "transformers_version": "4.55.4",
6
+ "use_cache": false
7
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2ed9b74f919fa40ec155aed8a6d915b865919cb452d8ea0cb7bf600a60807b6
3
+ size 1166029904
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/preprocessor_config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 224,
4
+ "width": 224
5
+ },
6
+ "do_center_crop": true,
7
+ "do_convert_rgb": true,
8
+ "do_normalize": true,
9
+ "do_rescale": true,
10
+ "do_resize": true,
11
+ "image_mean": [
12
+ 0.48145466,
13
+ 0.4578275,
14
+ 0.40821073
15
+ ],
16
+ "image_processor_type": "CLIPImageProcessor",
17
+ "image_std": [
18
+ 0.26862954,
19
+ 0.26130258,
20
+ 0.27577711
21
+ ],
22
+ "resample": 3,
23
+ "rescale_factor": 0.00392156862745098,
24
+ "size": {
25
+ "shortest_edge": 224
26
+ }
27
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc8bfffb4e2330aa971fbff9d356fe4402740b05604db488bc5593bd343b649f
3
+ size 8312
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "TinyLlavaForConditionalGeneration"
4
+ ],
5
+ "cache_dir": null,
6
+ "connector_type": "mlp2x_gelu",
7
+ "hidden_size": 896,
8
+ "ignore_index": -100,
9
+ "image_aspect_ratio": "square",
10
+ "image_token_index": -200,
11
+ "llm_model_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
12
+ "model_type": "tinyllava",
13
+ "num_queries": 128,
14
+ "num_resampler_layers": 3,
15
+ "pad_token": "<|endoftext|>",
16
+ "pad_token_id": 151643,
17
+ "resampler_hidden_size": 768,
18
+ "text_config": {
19
+ "_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
20
+ "architectures": [
21
+ "Qwen2ForCausalLM"
22
+ ],
23
+ "attention_dropout": 0.0,
24
+ "bos_token_id": 151643,
25
+ "eos_token_id": 151645,
26
+ "hidden_act": "silu",
27
+ "hidden_size": 896,
28
+ "initializer_range": 0.02,
29
+ "intermediate_size": 4864,
30
+ "layer_types": [
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention",
53
+ "full_attention",
54
+ "full_attention"
55
+ ],
56
+ "max_position_embeddings": 32768,
57
+ "max_window_layers": 21,
58
+ "model_type": "qwen2",
59
+ "num_attention_heads": 14,
60
+ "num_hidden_layers": 24,
61
+ "num_key_value_heads": 2,
62
+ "rms_norm_eps": 1e-06,
63
+ "rope_scaling": null,
64
+ "rope_theta": 1000000.0,
65
+ "sliding_window": null,
66
+ "tie_word_embeddings": true,
67
+ "torch_dtype": "bfloat16",
68
+ "use_cache": true,
69
+ "use_sliding_window": false,
70
+ "vocab_size": 151936
71
+ },
72
+ "tokenizer_model_max_length": 2048,
73
+ "tokenizer_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
74
+ "tokenizer_padding_side": "right",
75
+ "tokenizer_use_fast": true,
76
+ "torch_dtype": "bfloat16",
77
+ "transformers_version": "4.55.4",
78
+ "tune_type_connector": "full",
79
+ "tune_type_llm": "full",
80
+ "tune_type_vision_tower": "frozen",
81
+ "tune_vision_tower_from_layer": 0,
82
+ "use_cache": false,
83
+ "vision_config": {
84
+ "attention_dropout": 0.0,
85
+ "dropout": 0.0,
86
+ "hidden_act": "quick_gelu",
87
+ "hidden_size": 768,
88
+ "image_size": 224,
89
+ "initializer_factor": 1.0,
90
+ "initializer_range": 0.02,
91
+ "intermediate_size": 3072,
92
+ "layer_norm_eps": 1e-05,
93
+ "model_name_or_path": "openai/clip-vit-base-patch32",
94
+ "model_name_or_path2": "",
95
+ "model_type": "clip_vision_model",
96
+ "num_attention_heads": 12,
97
+ "num_channels": 3,
98
+ "num_hidden_layers": 12,
99
+ "patch_size": 32,
100
+ "projection_dim": 512
101
+ },
102
+ "vision_feature_layer": -2,
103
+ "vision_feature_select_strategy": "patch",
104
+ "vision_hidden_size": 768,
105
+ "vision_model_name_or_path": "openai/clip-vit-base-patch32",
106
+ "vision_model_name_or_path2": "",
107
+ "vocab_size": 151936
108
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 151643,
4
+ "eos_token_id": 151645,
5
+ "transformers_version": "4.55.4",
6
+ "use_cache": false
7
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:076acf72bb9a26f3244c08817eb5a9e97eae42f85d44712e4f1ac3315617c413
3
+ size 1166029904
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/preprocessor_config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 224,
4
+ "width": 224
5
+ },
6
+ "do_center_crop": true,
7
+ "do_convert_rgb": true,
8
+ "do_normalize": true,
9
+ "do_rescale": true,
10
+ "do_resize": true,
11
+ "image_mean": [
12
+ 0.48145466,
13
+ 0.4578275,
14
+ 0.40821073
15
+ ],
16
+ "image_processor_type": "CLIPImageProcessor",
17
+ "image_std": [
18
+ 0.26862954,
19
+ 0.26130258,
20
+ 0.27577711
21
+ ],
22
+ "resample": 3,
23
+ "rescale_factor": 0.00392156862745098,
24
+ "size": {
25
+ "shortest_edge": 224
26
+ }
27
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/trainer_state.json ADDED
@@ -0,0 +1,3534 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.096195469193401,
6
+ "eval_steps": 500,
7
+ "global_step": 500,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.00019239093838680199,
14
+ "grad_norm": 18.596346579815247,
15
+ "learning_rate": 0.0,
16
+ "loss": 2.0201,
17
+ "step": 1
18
+ },
19
+ {
20
+ "epoch": 0.00038478187677360397,
21
+ "grad_norm": 19.620062648993773,
22
+ "learning_rate": 1.282051282051282e-07,
23
+ "loss": 1.8757,
24
+ "step": 2
25
+ },
26
+ {
27
+ "epoch": 0.0005771728151604059,
28
+ "grad_norm": 18.15748603195869,
29
+ "learning_rate": 2.564102564102564e-07,
30
+ "loss": 1.7546,
31
+ "step": 3
32
+ },
33
+ {
34
+ "epoch": 0.0007695637535472079,
35
+ "grad_norm": 21.52078685000807,
36
+ "learning_rate": 3.846153846153847e-07,
37
+ "loss": 2.1009,
38
+ "step": 4
39
+ },
40
+ {
41
+ "epoch": 0.0009619546919340099,
42
+ "grad_norm": 17.547051938044735,
43
+ "learning_rate": 5.128205128205128e-07,
44
+ "loss": 1.9744,
45
+ "step": 5
46
+ },
47
+ {
48
+ "epoch": 0.0011543456303208118,
49
+ "grad_norm": 18.507275669342576,
50
+ "learning_rate": 6.41025641025641e-07,
51
+ "loss": 1.8365,
52
+ "step": 6
53
+ },
54
+ {
55
+ "epoch": 0.0013467365687076138,
56
+ "grad_norm": 18.842417883529933,
57
+ "learning_rate": 7.692307692307694e-07,
58
+ "loss": 1.9593,
59
+ "step": 7
60
+ },
61
+ {
62
+ "epoch": 0.0015391275070944159,
63
+ "grad_norm": 16.025960273667035,
64
+ "learning_rate": 8.974358974358975e-07,
65
+ "loss": 1.8733,
66
+ "step": 8
67
+ },
68
+ {
69
+ "epoch": 0.001731518445481218,
70
+ "grad_norm": 21.668397022061654,
71
+ "learning_rate": 1.0256410256410257e-06,
72
+ "loss": 1.9656,
73
+ "step": 9
74
+ },
75
+ {
76
+ "epoch": 0.0019239093838680197,
77
+ "grad_norm": 18.94192010349296,
78
+ "learning_rate": 1.153846153846154e-06,
79
+ "loss": 1.9052,
80
+ "step": 10
81
+ },
82
+ {
83
+ "epoch": 0.002116300322254822,
84
+ "grad_norm": 16.05155830510946,
85
+ "learning_rate": 1.282051282051282e-06,
86
+ "loss": 1.9139,
87
+ "step": 11
88
+ },
89
+ {
90
+ "epoch": 0.0023086912606416236,
91
+ "grad_norm": 12.758098756397695,
92
+ "learning_rate": 1.4102564102564104e-06,
93
+ "loss": 1.7169,
94
+ "step": 12
95
+ },
96
+ {
97
+ "epoch": 0.0025010821990284256,
98
+ "grad_norm": 11.980891906403086,
99
+ "learning_rate": 1.5384615384615387e-06,
100
+ "loss": 1.7637,
101
+ "step": 13
102
+ },
103
+ {
104
+ "epoch": 0.0026934731374152277,
105
+ "grad_norm": 17.001373294385477,
106
+ "learning_rate": 1.6666666666666667e-06,
107
+ "loss": 1.8849,
108
+ "step": 14
109
+ },
110
+ {
111
+ "epoch": 0.0028858640758020297,
112
+ "grad_norm": 15.147553545151323,
113
+ "learning_rate": 1.794871794871795e-06,
114
+ "loss": 1.7857,
115
+ "step": 15
116
+ },
117
+ {
118
+ "epoch": 0.0030782550141888318,
119
+ "grad_norm": 15.594132811638236,
120
+ "learning_rate": 1.9230769230769234e-06,
121
+ "loss": 1.8346,
122
+ "step": 16
123
+ },
124
+ {
125
+ "epoch": 0.003270645952575634,
126
+ "grad_norm": 9.769224673991353,
127
+ "learning_rate": 2.0512820512820513e-06,
128
+ "loss": 1.6662,
129
+ "step": 17
130
+ },
131
+ {
132
+ "epoch": 0.003463036890962436,
133
+ "grad_norm": 13.480771226608576,
134
+ "learning_rate": 2.1794871794871797e-06,
135
+ "loss": 1.6215,
136
+ "step": 18
137
+ },
138
+ {
139
+ "epoch": 0.0036554278293492374,
140
+ "grad_norm": 16.63076791213617,
141
+ "learning_rate": 2.307692307692308e-06,
142
+ "loss": 1.6979,
143
+ "step": 19
144
+ },
145
+ {
146
+ "epoch": 0.0038478187677360395,
147
+ "grad_norm": 10.739676752590206,
148
+ "learning_rate": 2.435897435897436e-06,
149
+ "loss": 1.5907,
150
+ "step": 20
151
+ },
152
+ {
153
+ "epoch": 0.004040209706122842,
154
+ "grad_norm": 9.330160169027454,
155
+ "learning_rate": 2.564102564102564e-06,
156
+ "loss": 1.6164,
157
+ "step": 21
158
+ },
159
+ {
160
+ "epoch": 0.004232600644509644,
161
+ "grad_norm": 6.069983499685054,
162
+ "learning_rate": 2.6923076923076923e-06,
163
+ "loss": 1.4696,
164
+ "step": 22
165
+ },
166
+ {
167
+ "epoch": 0.004424991582896445,
168
+ "grad_norm": 8.394618206978805,
169
+ "learning_rate": 2.8205128205128207e-06,
170
+ "loss": 1.6246,
171
+ "step": 23
172
+ },
173
+ {
174
+ "epoch": 0.004617382521283247,
175
+ "grad_norm": 9.813701596123554,
176
+ "learning_rate": 2.948717948717949e-06,
177
+ "loss": 1.6543,
178
+ "step": 24
179
+ },
180
+ {
181
+ "epoch": 0.004809773459670049,
182
+ "grad_norm": 7.1225776503344385,
183
+ "learning_rate": 3.0769230769230774e-06,
184
+ "loss": 1.5476,
185
+ "step": 25
186
+ },
187
+ {
188
+ "epoch": 0.005002164398056851,
189
+ "grad_norm": 7.025245990856855,
190
+ "learning_rate": 3.205128205128206e-06,
191
+ "loss": 1.5915,
192
+ "step": 26
193
+ },
194
+ {
195
+ "epoch": 0.005194555336443653,
196
+ "grad_norm": 4.911738916449327,
197
+ "learning_rate": 3.3333333333333333e-06,
198
+ "loss": 1.6135,
199
+ "step": 27
200
+ },
201
+ {
202
+ "epoch": 0.005386946274830455,
203
+ "grad_norm": 5.675687466411745,
204
+ "learning_rate": 3.4615384615384617e-06,
205
+ "loss": 1.5999,
206
+ "step": 28
207
+ },
208
+ {
209
+ "epoch": 0.005579337213217257,
210
+ "grad_norm": 5.130161167595587,
211
+ "learning_rate": 3.58974358974359e-06,
212
+ "loss": 1.5827,
213
+ "step": 29
214
+ },
215
+ {
216
+ "epoch": 0.0057717281516040594,
217
+ "grad_norm": 5.485827392758109,
218
+ "learning_rate": 3.7179487179487184e-06,
219
+ "loss": 1.4926,
220
+ "step": 30
221
+ },
222
+ {
223
+ "epoch": 0.0059641190899908615,
224
+ "grad_norm": 6.0312448528879035,
225
+ "learning_rate": 3.846153846153847e-06,
226
+ "loss": 1.6133,
227
+ "step": 31
228
+ },
229
+ {
230
+ "epoch": 0.0061565100283776635,
231
+ "grad_norm": 3.8723259090708018,
232
+ "learning_rate": 3.974358974358974e-06,
233
+ "loss": 1.5898,
234
+ "step": 32
235
+ },
236
+ {
237
+ "epoch": 0.0063489009667644656,
238
+ "grad_norm": 4.046198899481709,
239
+ "learning_rate": 4.102564102564103e-06,
240
+ "loss": 1.4708,
241
+ "step": 33
242
+ },
243
+ {
244
+ "epoch": 0.006541291905151268,
245
+ "grad_norm": 4.935190294150509,
246
+ "learning_rate": 4.230769230769231e-06,
247
+ "loss": 1.4294,
248
+ "step": 34
249
+ },
250
+ {
251
+ "epoch": 0.00673368284353807,
252
+ "grad_norm": 4.8124629959405905,
253
+ "learning_rate": 4.358974358974359e-06,
254
+ "loss": 1.5768,
255
+ "step": 35
256
+ },
257
+ {
258
+ "epoch": 0.006926073781924872,
259
+ "grad_norm": 4.612119540477251,
260
+ "learning_rate": 4.487179487179488e-06,
261
+ "loss": 1.4806,
262
+ "step": 36
263
+ },
264
+ {
265
+ "epoch": 0.007118464720311674,
266
+ "grad_norm": 3.539195202734351,
267
+ "learning_rate": 4.615384615384616e-06,
268
+ "loss": 1.3562,
269
+ "step": 37
270
+ },
271
+ {
272
+ "epoch": 0.007310855658698475,
273
+ "grad_norm": 4.280037095410956,
274
+ "learning_rate": 4.743589743589744e-06,
275
+ "loss": 1.4511,
276
+ "step": 38
277
+ },
278
+ {
279
+ "epoch": 0.007503246597085277,
280
+ "grad_norm": 4.045204919478714,
281
+ "learning_rate": 4.871794871794872e-06,
282
+ "loss": 1.4259,
283
+ "step": 39
284
+ },
285
+ {
286
+ "epoch": 0.007695637535472079,
287
+ "grad_norm": 3.5453992727499104,
288
+ "learning_rate": 5e-06,
289
+ "loss": 1.5242,
290
+ "step": 40
291
+ },
292
+ {
293
+ "epoch": 0.007888028473858881,
294
+ "grad_norm": 4.449788031385935,
295
+ "learning_rate": 5.128205128205128e-06,
296
+ "loss": 1.4537,
297
+ "step": 41
298
+ },
299
+ {
300
+ "epoch": 0.008080419412245684,
301
+ "grad_norm": 3.8637779520893356,
302
+ "learning_rate": 5.256410256410257e-06,
303
+ "loss": 1.4287,
304
+ "step": 42
305
+ },
306
+ {
307
+ "epoch": 0.008272810350632485,
308
+ "grad_norm": 3.9013870962052266,
309
+ "learning_rate": 5.384615384615385e-06,
310
+ "loss": 1.5265,
311
+ "step": 43
312
+ },
313
+ {
314
+ "epoch": 0.008465201289019288,
315
+ "grad_norm": 4.284934489842209,
316
+ "learning_rate": 5.512820512820514e-06,
317
+ "loss": 1.426,
318
+ "step": 44
319
+ },
320
+ {
321
+ "epoch": 0.00865759222740609,
322
+ "grad_norm": 3.348607031794847,
323
+ "learning_rate": 5.641025641025641e-06,
324
+ "loss": 1.268,
325
+ "step": 45
326
+ },
327
+ {
328
+ "epoch": 0.00884998316579289,
329
+ "grad_norm": 4.638115637341805,
330
+ "learning_rate": 5.769230769230769e-06,
331
+ "loss": 1.4783,
332
+ "step": 46
333
+ },
334
+ {
335
+ "epoch": 0.009042374104179693,
336
+ "grad_norm": 3.200270366772654,
337
+ "learning_rate": 5.897435897435898e-06,
338
+ "loss": 1.4425,
339
+ "step": 47
340
+ },
341
+ {
342
+ "epoch": 0.009234765042566494,
343
+ "grad_norm": 3.8953486663950874,
344
+ "learning_rate": 6.025641025641026e-06,
345
+ "loss": 1.4805,
346
+ "step": 48
347
+ },
348
+ {
349
+ "epoch": 0.009427155980953297,
350
+ "grad_norm": 4.091514206107613,
351
+ "learning_rate": 6.153846153846155e-06,
352
+ "loss": 1.3474,
353
+ "step": 49
354
+ },
355
+ {
356
+ "epoch": 0.009619546919340098,
357
+ "grad_norm": 3.7095611219591307,
358
+ "learning_rate": 6.282051282051282e-06,
359
+ "loss": 1.3823,
360
+ "step": 50
361
+ },
362
+ {
363
+ "epoch": 0.009811937857726901,
364
+ "grad_norm": 3.6239930257996775,
365
+ "learning_rate": 6.410256410256412e-06,
366
+ "loss": 1.3843,
367
+ "step": 51
368
+ },
369
+ {
370
+ "epoch": 0.010004328796113703,
371
+ "grad_norm": 4.46531244470849,
372
+ "learning_rate": 6.538461538461539e-06,
373
+ "loss": 1.4725,
374
+ "step": 52
375
+ },
376
+ {
377
+ "epoch": 0.010196719734500505,
378
+ "grad_norm": 3.226735262984168,
379
+ "learning_rate": 6.666666666666667e-06,
380
+ "loss": 1.4845,
381
+ "step": 53
382
+ },
383
+ {
384
+ "epoch": 0.010389110672887307,
385
+ "grad_norm": 3.6493935896982386,
386
+ "learning_rate": 6.794871794871796e-06,
387
+ "loss": 1.4604,
388
+ "step": 54
389
+ },
390
+ {
391
+ "epoch": 0.01058150161127411,
392
+ "grad_norm": 3.9066104095484047,
393
+ "learning_rate": 6.923076923076923e-06,
394
+ "loss": 1.3897,
395
+ "step": 55
396
+ },
397
+ {
398
+ "epoch": 0.01077389254966091,
399
+ "grad_norm": 3.78621978205094,
400
+ "learning_rate": 7.051282051282053e-06,
401
+ "loss": 1.5521,
402
+ "step": 56
403
+ },
404
+ {
405
+ "epoch": 0.010966283488047714,
406
+ "grad_norm": 3.7258672092661804,
407
+ "learning_rate": 7.17948717948718e-06,
408
+ "loss": 1.4454,
409
+ "step": 57
410
+ },
411
+ {
412
+ "epoch": 0.011158674426434515,
413
+ "grad_norm": 3.4641653455305557,
414
+ "learning_rate": 7.307692307692308e-06,
415
+ "loss": 1.4657,
416
+ "step": 58
417
+ },
418
+ {
419
+ "epoch": 0.011351065364821318,
420
+ "grad_norm": 2.80098702135382,
421
+ "learning_rate": 7.435897435897437e-06,
422
+ "loss": 1.2991,
423
+ "step": 59
424
+ },
425
+ {
426
+ "epoch": 0.011543456303208119,
427
+ "grad_norm": 2.975596898714371,
428
+ "learning_rate": 7.564102564102564e-06,
429
+ "loss": 1.3685,
430
+ "step": 60
431
+ },
432
+ {
433
+ "epoch": 0.01173584724159492,
434
+ "grad_norm": 3.531591616515045,
435
+ "learning_rate": 7.692307692307694e-06,
436
+ "loss": 1.4134,
437
+ "step": 61
438
+ },
439
+ {
440
+ "epoch": 0.011928238179981723,
441
+ "grad_norm": 2.914698932344494,
442
+ "learning_rate": 7.820512820512822e-06,
443
+ "loss": 1.3538,
444
+ "step": 62
445
+ },
446
+ {
447
+ "epoch": 0.012120629118368524,
448
+ "grad_norm": 3.252603432376639,
449
+ "learning_rate": 7.948717948717949e-06,
450
+ "loss": 1.4165,
451
+ "step": 63
452
+ },
453
+ {
454
+ "epoch": 0.012313020056755327,
455
+ "grad_norm": 2.8948487124390723,
456
+ "learning_rate": 8.076923076923077e-06,
457
+ "loss": 1.3985,
458
+ "step": 64
459
+ },
460
+ {
461
+ "epoch": 0.012505410995142128,
462
+ "grad_norm": 3.8806644911282984,
463
+ "learning_rate": 8.205128205128205e-06,
464
+ "loss": 1.451,
465
+ "step": 65
466
+ },
467
+ {
468
+ "epoch": 0.012697801933528931,
469
+ "grad_norm": 3.5361182720176827,
470
+ "learning_rate": 8.333333333333334e-06,
471
+ "loss": 1.2744,
472
+ "step": 66
473
+ },
474
+ {
475
+ "epoch": 0.012890192871915732,
476
+ "grad_norm": 3.840767961245834,
477
+ "learning_rate": 8.461538461538462e-06,
478
+ "loss": 1.508,
479
+ "step": 67
480
+ },
481
+ {
482
+ "epoch": 0.013082583810302535,
483
+ "grad_norm": 2.97293860255653,
484
+ "learning_rate": 8.58974358974359e-06,
485
+ "loss": 1.3571,
486
+ "step": 68
487
+ },
488
+ {
489
+ "epoch": 0.013274974748689336,
490
+ "grad_norm": 3.4243087862218466,
491
+ "learning_rate": 8.717948717948719e-06,
492
+ "loss": 1.448,
493
+ "step": 69
494
+ },
495
+ {
496
+ "epoch": 0.01346736568707614,
497
+ "grad_norm": 4.237210587283784,
498
+ "learning_rate": 8.846153846153847e-06,
499
+ "loss": 1.4946,
500
+ "step": 70
501
+ },
502
+ {
503
+ "epoch": 0.01365975662546294,
504
+ "grad_norm": 3.036912456620411,
505
+ "learning_rate": 8.974358974358976e-06,
506
+ "loss": 1.3244,
507
+ "step": 71
508
+ },
509
+ {
510
+ "epoch": 0.013852147563849743,
511
+ "grad_norm": 3.5282129576707697,
512
+ "learning_rate": 9.102564102564104e-06,
513
+ "loss": 1.5027,
514
+ "step": 72
515
+ },
516
+ {
517
+ "epoch": 0.014044538502236545,
518
+ "grad_norm": 3.4192316609333266,
519
+ "learning_rate": 9.230769230769232e-06,
520
+ "loss": 1.362,
521
+ "step": 73
522
+ },
523
+ {
524
+ "epoch": 0.014236929440623347,
525
+ "grad_norm": 2.9491736972856932,
526
+ "learning_rate": 9.358974358974359e-06,
527
+ "loss": 1.3689,
528
+ "step": 74
529
+ },
530
+ {
531
+ "epoch": 0.014429320379010149,
532
+ "grad_norm": 3.5500120300124665,
533
+ "learning_rate": 9.487179487179487e-06,
534
+ "loss": 1.4428,
535
+ "step": 75
536
+ },
537
+ {
538
+ "epoch": 0.01462171131739695,
539
+ "grad_norm": 3.2246422896239126,
540
+ "learning_rate": 9.615384615384616e-06,
541
+ "loss": 1.4442,
542
+ "step": 76
543
+ },
544
+ {
545
+ "epoch": 0.014814102255783753,
546
+ "grad_norm": 3.444051497962385,
547
+ "learning_rate": 9.743589743589744e-06,
548
+ "loss": 1.3738,
549
+ "step": 77
550
+ },
551
+ {
552
+ "epoch": 0.015006493194170554,
553
+ "grad_norm": 3.427461083364507,
554
+ "learning_rate": 9.871794871794872e-06,
555
+ "loss": 1.3402,
556
+ "step": 78
557
+ },
558
+ {
559
+ "epoch": 0.015198884132557357,
560
+ "grad_norm": 4.109517782791629,
561
+ "learning_rate": 1e-05,
562
+ "loss": 1.4267,
563
+ "step": 79
564
+ },
565
+ {
566
+ "epoch": 0.015391275070944158,
567
+ "grad_norm": 3.3202267566679096,
568
+ "learning_rate": 1.012820512820513e-05,
569
+ "loss": 1.4432,
570
+ "step": 80
571
+ },
572
+ {
573
+ "epoch": 0.01558366600933096,
574
+ "grad_norm": 2.971119422738144,
575
+ "learning_rate": 1.0256410256410256e-05,
576
+ "loss": 1.3762,
577
+ "step": 81
578
+ },
579
+ {
580
+ "epoch": 0.015776056947717762,
581
+ "grad_norm": 2.933293376961111,
582
+ "learning_rate": 1.0384615384615386e-05,
583
+ "loss": 1.2824,
584
+ "step": 82
585
+ },
586
+ {
587
+ "epoch": 0.015968447886104565,
588
+ "grad_norm": 3.9952279466954645,
589
+ "learning_rate": 1.0512820512820514e-05,
590
+ "loss": 1.4101,
591
+ "step": 83
592
+ },
593
+ {
594
+ "epoch": 0.016160838824491368,
595
+ "grad_norm": 3.0626448576219287,
596
+ "learning_rate": 1.0641025641025643e-05,
597
+ "loss": 1.3622,
598
+ "step": 84
599
+ },
600
+ {
601
+ "epoch": 0.016353229762878167,
602
+ "grad_norm": 3.8653871342308532,
603
+ "learning_rate": 1.076923076923077e-05,
604
+ "loss": 1.5116,
605
+ "step": 85
606
+ },
607
+ {
608
+ "epoch": 0.01654562070126497,
609
+ "grad_norm": 3.9049134003009294,
610
+ "learning_rate": 1.0897435897435898e-05,
611
+ "loss": 1.4936,
612
+ "step": 86
613
+ },
614
+ {
615
+ "epoch": 0.016738011639651773,
616
+ "grad_norm": 3.338095947230545,
617
+ "learning_rate": 1.1025641025641028e-05,
618
+ "loss": 1.4359,
619
+ "step": 87
620
+ },
621
+ {
622
+ "epoch": 0.016930402578038576,
623
+ "grad_norm": 3.3773351982809747,
624
+ "learning_rate": 1.1153846153846154e-05,
625
+ "loss": 1.4468,
626
+ "step": 88
627
+ },
628
+ {
629
+ "epoch": 0.017122793516425375,
630
+ "grad_norm": 3.2977728757562077,
631
+ "learning_rate": 1.1282051282051283e-05,
632
+ "loss": 1.3717,
633
+ "step": 89
634
+ },
635
+ {
636
+ "epoch": 0.01731518445481218,
637
+ "grad_norm": 4.912045804073467,
638
+ "learning_rate": 1.1410256410256411e-05,
639
+ "loss": 1.4495,
640
+ "step": 90
641
+ },
642
+ {
643
+ "epoch": 0.01750757539319898,
644
+ "grad_norm": 3.9376717357748827,
645
+ "learning_rate": 1.1538461538461538e-05,
646
+ "loss": 1.3626,
647
+ "step": 91
648
+ },
649
+ {
650
+ "epoch": 0.01769996633158578,
651
+ "grad_norm": 2.701339103157188,
652
+ "learning_rate": 1.1666666666666668e-05,
653
+ "loss": 1.2561,
654
+ "step": 92
655
+ },
656
+ {
657
+ "epoch": 0.017892357269972584,
658
+ "grad_norm": 3.5432827173371186,
659
+ "learning_rate": 1.1794871794871796e-05,
660
+ "loss": 1.5077,
661
+ "step": 93
662
+ },
663
+ {
664
+ "epoch": 0.018084748208359386,
665
+ "grad_norm": 3.263515633024985,
666
+ "learning_rate": 1.1923076923076925e-05,
667
+ "loss": 1.3667,
668
+ "step": 94
669
+ },
670
+ {
671
+ "epoch": 0.01827713914674619,
672
+ "grad_norm": 3.6583450365857635,
673
+ "learning_rate": 1.2051282051282051e-05,
674
+ "loss": 1.3833,
675
+ "step": 95
676
+ },
677
+ {
678
+ "epoch": 0.01846953008513299,
679
+ "grad_norm": 3.703710596859915,
680
+ "learning_rate": 1.217948717948718e-05,
681
+ "loss": 1.3969,
682
+ "step": 96
683
+ },
684
+ {
685
+ "epoch": 0.01866192102351979,
686
+ "grad_norm": 4.1236009457162,
687
+ "learning_rate": 1.230769230769231e-05,
688
+ "loss": 1.3239,
689
+ "step": 97
690
+ },
691
+ {
692
+ "epoch": 0.018854311961906595,
693
+ "grad_norm": 3.6166505934050295,
694
+ "learning_rate": 1.2435897435897436e-05,
695
+ "loss": 1.3506,
696
+ "step": 98
697
+ },
698
+ {
699
+ "epoch": 0.019046702900293398,
700
+ "grad_norm": 3.848654906277788,
701
+ "learning_rate": 1.2564102564102565e-05,
702
+ "loss": 1.4312,
703
+ "step": 99
704
+ },
705
+ {
706
+ "epoch": 0.019239093838680197,
707
+ "grad_norm": 3.4126019473397946,
708
+ "learning_rate": 1.2692307692307693e-05,
709
+ "loss": 1.4312,
710
+ "step": 100
711
+ },
712
+ {
713
+ "epoch": 0.019431484777067,
714
+ "grad_norm": 3.689305653605826,
715
+ "learning_rate": 1.2820512820512823e-05,
716
+ "loss": 1.3916,
717
+ "step": 101
718
+ },
719
+ {
720
+ "epoch": 0.019623875715453803,
721
+ "grad_norm": 3.6593592644955044,
722
+ "learning_rate": 1.294871794871795e-05,
723
+ "loss": 1.3092,
724
+ "step": 102
725
+ },
726
+ {
727
+ "epoch": 0.019816266653840606,
728
+ "grad_norm": 3.5942861063223015,
729
+ "learning_rate": 1.3076923076923078e-05,
730
+ "loss": 1.3691,
731
+ "step": 103
732
+ },
733
+ {
734
+ "epoch": 0.020008657592227405,
735
+ "grad_norm": 3.176733248751845,
736
+ "learning_rate": 1.3205128205128207e-05,
737
+ "loss": 1.3758,
738
+ "step": 104
739
+ },
740
+ {
741
+ "epoch": 0.020201048530614208,
742
+ "grad_norm": 3.2912285855746357,
743
+ "learning_rate": 1.3333333333333333e-05,
744
+ "loss": 1.3937,
745
+ "step": 105
746
+ },
747
+ {
748
+ "epoch": 0.02039343946900101,
749
+ "grad_norm": 4.193900652049377,
750
+ "learning_rate": 1.3461538461538463e-05,
751
+ "loss": 1.4287,
752
+ "step": 106
753
+ },
754
+ {
755
+ "epoch": 0.02058583040738781,
756
+ "grad_norm": 3.5896317618385427,
757
+ "learning_rate": 1.3589743589743592e-05,
758
+ "loss": 1.3873,
759
+ "step": 107
760
+ },
761
+ {
762
+ "epoch": 0.020778221345774613,
763
+ "grad_norm": 3.2491370771024313,
764
+ "learning_rate": 1.3717948717948718e-05,
765
+ "loss": 1.3245,
766
+ "step": 108
767
+ },
768
+ {
769
+ "epoch": 0.020970612284161416,
770
+ "grad_norm": 3.6307440256798866,
771
+ "learning_rate": 1.3846153846153847e-05,
772
+ "loss": 1.4377,
773
+ "step": 109
774
+ },
775
+ {
776
+ "epoch": 0.02116300322254822,
777
+ "grad_norm": 3.486354961473348,
778
+ "learning_rate": 1.3974358974358975e-05,
779
+ "loss": 1.5146,
780
+ "step": 110
781
+ },
782
+ {
783
+ "epoch": 0.02135539416093502,
784
+ "grad_norm": 3.577903589198358,
785
+ "learning_rate": 1.4102564102564105e-05,
786
+ "loss": 1.3277,
787
+ "step": 111
788
+ },
789
+ {
790
+ "epoch": 0.02154778509932182,
791
+ "grad_norm": 3.2973635772894605,
792
+ "learning_rate": 1.4230769230769232e-05,
793
+ "loss": 1.3273,
794
+ "step": 112
795
+ },
796
+ {
797
+ "epoch": 0.021740176037708624,
798
+ "grad_norm": 2.987453569450381,
799
+ "learning_rate": 1.435897435897436e-05,
800
+ "loss": 1.2799,
801
+ "step": 113
802
+ },
803
+ {
804
+ "epoch": 0.021932566976095427,
805
+ "grad_norm": 3.9880303827260364,
806
+ "learning_rate": 1.4487179487179489e-05,
807
+ "loss": 1.3746,
808
+ "step": 114
809
+ },
810
+ {
811
+ "epoch": 0.022124957914482227,
812
+ "grad_norm": 3.5241529416950836,
813
+ "learning_rate": 1.4615384615384615e-05,
814
+ "loss": 1.3982,
815
+ "step": 115
816
+ },
817
+ {
818
+ "epoch": 0.02231734885286903,
819
+ "grad_norm": 3.7922909812471173,
820
+ "learning_rate": 1.4743589743589745e-05,
821
+ "loss": 1.3225,
822
+ "step": 116
823
+ },
824
+ {
825
+ "epoch": 0.022509739791255833,
826
+ "grad_norm": 2.8536987417234867,
827
+ "learning_rate": 1.4871794871794874e-05,
828
+ "loss": 1.2426,
829
+ "step": 117
830
+ },
831
+ {
832
+ "epoch": 0.022702130729642635,
833
+ "grad_norm": 3.659818759988536,
834
+ "learning_rate": 1.5000000000000002e-05,
835
+ "loss": 1.4757,
836
+ "step": 118
837
+ },
838
+ {
839
+ "epoch": 0.022894521668029435,
840
+ "grad_norm": 3.9911745786242148,
841
+ "learning_rate": 1.5128205128205129e-05,
842
+ "loss": 1.5255,
843
+ "step": 119
844
+ },
845
+ {
846
+ "epoch": 0.023086912606416238,
847
+ "grad_norm": 3.8597136196034008,
848
+ "learning_rate": 1.5256410256410257e-05,
849
+ "loss": 1.4575,
850
+ "step": 120
851
+ },
852
+ {
853
+ "epoch": 0.02327930354480304,
854
+ "grad_norm": 3.0987796400020473,
855
+ "learning_rate": 1.5384615384615387e-05,
856
+ "loss": 1.3673,
857
+ "step": 121
858
+ },
859
+ {
860
+ "epoch": 0.02347169448318984,
861
+ "grad_norm": 3.168640337276763,
862
+ "learning_rate": 1.5512820512820516e-05,
863
+ "loss": 1.2768,
864
+ "step": 122
865
+ },
866
+ {
867
+ "epoch": 0.023664085421576643,
868
+ "grad_norm": 3.6247801431539894,
869
+ "learning_rate": 1.5641025641025644e-05,
870
+ "loss": 1.3968,
871
+ "step": 123
872
+ },
873
+ {
874
+ "epoch": 0.023856476359963446,
875
+ "grad_norm": 4.179633235800667,
876
+ "learning_rate": 1.576923076923077e-05,
877
+ "loss": 1.4356,
878
+ "step": 124
879
+ },
880
+ {
881
+ "epoch": 0.02404886729835025,
882
+ "grad_norm": 3.1995334883826247,
883
+ "learning_rate": 1.5897435897435897e-05,
884
+ "loss": 1.4334,
885
+ "step": 125
886
+ },
887
+ {
888
+ "epoch": 0.02424125823673705,
889
+ "grad_norm": 3.3358932949167257,
890
+ "learning_rate": 1.602564102564103e-05,
891
+ "loss": 1.3743,
892
+ "step": 126
893
+ },
894
+ {
895
+ "epoch": 0.02443364917512385,
896
+ "grad_norm": 3.6663669202992817,
897
+ "learning_rate": 1.6153846153846154e-05,
898
+ "loss": 1.3125,
899
+ "step": 127
900
+ },
901
+ {
902
+ "epoch": 0.024626040113510654,
903
+ "grad_norm": 3.54587312795278,
904
+ "learning_rate": 1.6282051282051282e-05,
905
+ "loss": 1.3034,
906
+ "step": 128
907
+ },
908
+ {
909
+ "epoch": 0.024818431051897457,
910
+ "grad_norm": 2.990408959617764,
911
+ "learning_rate": 1.641025641025641e-05,
912
+ "loss": 1.2316,
913
+ "step": 129
914
+ },
915
+ {
916
+ "epoch": 0.025010821990284256,
917
+ "grad_norm": 3.352866488505463,
918
+ "learning_rate": 1.653846153846154e-05,
919
+ "loss": 1.37,
920
+ "step": 130
921
+ },
922
+ {
923
+ "epoch": 0.02520321292867106,
924
+ "grad_norm": 3.5865676122517627,
925
+ "learning_rate": 1.6666666666666667e-05,
926
+ "loss": 1.4055,
927
+ "step": 131
928
+ },
929
+ {
930
+ "epoch": 0.025395603867057862,
931
+ "grad_norm": 3.543653306909296,
932
+ "learning_rate": 1.6794871794871796e-05,
933
+ "loss": 1.3231,
934
+ "step": 132
935
+ },
936
+ {
937
+ "epoch": 0.025587994805444665,
938
+ "grad_norm": 3.2104773882187208,
939
+ "learning_rate": 1.6923076923076924e-05,
940
+ "loss": 1.2883,
941
+ "step": 133
942
+ },
943
+ {
944
+ "epoch": 0.025780385743831465,
945
+ "grad_norm": 3.0894181765572895,
946
+ "learning_rate": 1.7051282051282053e-05,
947
+ "loss": 1.3078,
948
+ "step": 134
949
+ },
950
+ {
951
+ "epoch": 0.025972776682218267,
952
+ "grad_norm": 3.743120864308114,
953
+ "learning_rate": 1.717948717948718e-05,
954
+ "loss": 1.3553,
955
+ "step": 135
956
+ },
957
+ {
958
+ "epoch": 0.02616516762060507,
959
+ "grad_norm": 2.914188647541259,
960
+ "learning_rate": 1.730769230769231e-05,
961
+ "loss": 1.3056,
962
+ "step": 136
963
+ },
964
+ {
965
+ "epoch": 0.02635755855899187,
966
+ "grad_norm": 3.240363853506205,
967
+ "learning_rate": 1.7435897435897438e-05,
968
+ "loss": 1.4489,
969
+ "step": 137
970
+ },
971
+ {
972
+ "epoch": 0.026549949497378673,
973
+ "grad_norm": 2.910195053494495,
974
+ "learning_rate": 1.7564102564102566e-05,
975
+ "loss": 1.2466,
976
+ "step": 138
977
+ },
978
+ {
979
+ "epoch": 0.026742340435765476,
980
+ "grad_norm": 3.0600450548051925,
981
+ "learning_rate": 1.7692307692307694e-05,
982
+ "loss": 1.3814,
983
+ "step": 139
984
+ },
985
+ {
986
+ "epoch": 0.02693473137415228,
987
+ "grad_norm": 3.9214613674415753,
988
+ "learning_rate": 1.7820512820512823e-05,
989
+ "loss": 1.4135,
990
+ "step": 140
991
+ },
992
+ {
993
+ "epoch": 0.027127122312539078,
994
+ "grad_norm": 2.781818689754229,
995
+ "learning_rate": 1.794871794871795e-05,
996
+ "loss": 1.3794,
997
+ "step": 141
998
+ },
999
+ {
1000
+ "epoch": 0.02731951325092588,
1001
+ "grad_norm": 3.893947843546862,
1002
+ "learning_rate": 1.807692307692308e-05,
1003
+ "loss": 1.2674,
1004
+ "step": 142
1005
+ },
1006
+ {
1007
+ "epoch": 0.027511904189312684,
1008
+ "grad_norm": 3.5447646062600717,
1009
+ "learning_rate": 1.8205128205128208e-05,
1010
+ "loss": 1.3307,
1011
+ "step": 143
1012
+ },
1013
+ {
1014
+ "epoch": 0.027704295127699487,
1015
+ "grad_norm": 3.5638559218333277,
1016
+ "learning_rate": 1.8333333333333333e-05,
1017
+ "loss": 1.2972,
1018
+ "step": 144
1019
+ },
1020
+ {
1021
+ "epoch": 0.027896686066086286,
1022
+ "grad_norm": 3.708727272626663,
1023
+ "learning_rate": 1.8461538461538465e-05,
1024
+ "loss": 1.3865,
1025
+ "step": 145
1026
+ },
1027
+ {
1028
+ "epoch": 0.02808907700447309,
1029
+ "grad_norm": 4.886005630567603,
1030
+ "learning_rate": 1.8589743589743593e-05,
1031
+ "loss": 1.3649,
1032
+ "step": 146
1033
+ },
1034
+ {
1035
+ "epoch": 0.028281467942859892,
1036
+ "grad_norm": 6.426516945357641,
1037
+ "learning_rate": 1.8717948717948718e-05,
1038
+ "loss": 1.366,
1039
+ "step": 147
1040
+ },
1041
+ {
1042
+ "epoch": 0.028473858881246695,
1043
+ "grad_norm": 2.995740795784324,
1044
+ "learning_rate": 1.8846153846153846e-05,
1045
+ "loss": 1.3476,
1046
+ "step": 148
1047
+ },
1048
+ {
1049
+ "epoch": 0.028666249819633494,
1050
+ "grad_norm": 4.368829467600509,
1051
+ "learning_rate": 1.8974358974358975e-05,
1052
+ "loss": 1.5164,
1053
+ "step": 149
1054
+ },
1055
+ {
1056
+ "epoch": 0.028858640758020297,
1057
+ "grad_norm": 3.651762470904225,
1058
+ "learning_rate": 1.9102564102564106e-05,
1059
+ "loss": 1.3543,
1060
+ "step": 150
1061
+ },
1062
+ {
1063
+ "epoch": 0.0290510316964071,
1064
+ "grad_norm": 4.8055151737000115,
1065
+ "learning_rate": 1.923076923076923e-05,
1066
+ "loss": 1.4435,
1067
+ "step": 151
1068
+ },
1069
+ {
1070
+ "epoch": 0.0292434226347939,
1071
+ "grad_norm": 3.3394518957538946,
1072
+ "learning_rate": 1.935897435897436e-05,
1073
+ "loss": 1.3682,
1074
+ "step": 152
1075
+ },
1076
+ {
1077
+ "epoch": 0.029435813573180702,
1078
+ "grad_norm": 3.9341451123812763,
1079
+ "learning_rate": 1.9487179487179488e-05,
1080
+ "loss": 1.3937,
1081
+ "step": 153
1082
+ },
1083
+ {
1084
+ "epoch": 0.029628204511567505,
1085
+ "grad_norm": 4.478809321700407,
1086
+ "learning_rate": 1.9615384615384617e-05,
1087
+ "loss": 1.3579,
1088
+ "step": 154
1089
+ },
1090
+ {
1091
+ "epoch": 0.02982059544995431,
1092
+ "grad_norm": 3.820274705643633,
1093
+ "learning_rate": 1.9743589743589745e-05,
1094
+ "loss": 1.2489,
1095
+ "step": 155
1096
+ },
1097
+ {
1098
+ "epoch": 0.030012986388341108,
1099
+ "grad_norm": 4.683884284887553,
1100
+ "learning_rate": 1.9871794871794873e-05,
1101
+ "loss": 1.4288,
1102
+ "step": 156
1103
+ },
1104
+ {
1105
+ "epoch": 0.03020537732672791,
1106
+ "grad_norm": 3.380261066559227,
1107
+ "learning_rate": 2e-05,
1108
+ "loss": 1.3781,
1109
+ "step": 157
1110
+ },
1111
+ {
1112
+ "epoch": 0.030397768265114714,
1113
+ "grad_norm": 3.3679749105010424,
1114
+ "learning_rate": 1.9999998058827844e-05,
1115
+ "loss": 1.3917,
1116
+ "step": 158
1117
+ },
1118
+ {
1119
+ "epoch": 0.030590159203501516,
1120
+ "grad_norm": 3.630929540775464,
1121
+ "learning_rate": 1.9999992235312136e-05,
1122
+ "loss": 1.4554,
1123
+ "step": 159
1124
+ },
1125
+ {
1126
+ "epoch": 0.030782550141888316,
1127
+ "grad_norm": 3.4109224146650745,
1128
+ "learning_rate": 1.9999982529455127e-05,
1129
+ "loss": 1.3118,
1130
+ "step": 160
1131
+ },
1132
+ {
1133
+ "epoch": 0.03097494108027512,
1134
+ "grad_norm": 3.3613779344409433,
1135
+ "learning_rate": 1.9999968941260596e-05,
1136
+ "loss": 1.3947,
1137
+ "step": 161
1138
+ },
1139
+ {
1140
+ "epoch": 0.03116733201866192,
1141
+ "grad_norm": 3.2348663660117722,
1142
+ "learning_rate": 1.9999951470733808e-05,
1143
+ "loss": 1.3755,
1144
+ "step": 162
1145
+ },
1146
+ {
1147
+ "epoch": 0.031359722957048725,
1148
+ "grad_norm": 3.2335344322136086,
1149
+ "learning_rate": 1.9999930117881548e-05,
1150
+ "loss": 1.3816,
1151
+ "step": 163
1152
+ },
1153
+ {
1154
+ "epoch": 0.031552113895435524,
1155
+ "grad_norm": 3.512545902843456,
1156
+ "learning_rate": 1.9999904882712115e-05,
1157
+ "loss": 1.3295,
1158
+ "step": 164
1159
+ },
1160
+ {
1161
+ "epoch": 0.03174450483382233,
1162
+ "grad_norm": 3.345738598437256,
1163
+ "learning_rate": 1.99998757652353e-05,
1164
+ "loss": 1.3539,
1165
+ "step": 165
1166
+ },
1167
+ {
1168
+ "epoch": 0.03193689577220913,
1169
+ "grad_norm": 3.876098996594679,
1170
+ "learning_rate": 1.9999842765462403e-05,
1171
+ "loss": 1.3051,
1172
+ "step": 166
1173
+ },
1174
+ {
1175
+ "epoch": 0.03212928671059593,
1176
+ "grad_norm": 2.6409767257120693,
1177
+ "learning_rate": 1.999980588340624e-05,
1178
+ "loss": 1.335,
1179
+ "step": 167
1180
+ },
1181
+ {
1182
+ "epoch": 0.032321677648982736,
1183
+ "grad_norm": 3.1939631295159305,
1184
+ "learning_rate": 1.9999765119081132e-05,
1185
+ "loss": 1.3213,
1186
+ "step": 168
1187
+ },
1188
+ {
1189
+ "epoch": 0.032514068587369535,
1190
+ "grad_norm": 3.086653918671633,
1191
+ "learning_rate": 1.9999720472502902e-05,
1192
+ "loss": 1.2674,
1193
+ "step": 169
1194
+ },
1195
+ {
1196
+ "epoch": 0.032706459525756335,
1197
+ "grad_norm": 3.1439630365183446,
1198
+ "learning_rate": 1.9999671943688885e-05,
1199
+ "loss": 1.3513,
1200
+ "step": 170
1201
+ },
1202
+ {
1203
+ "epoch": 0.03289885046414314,
1204
+ "grad_norm": 3.841064679135219,
1205
+ "learning_rate": 1.9999619532657915e-05,
1206
+ "loss": 1.3333,
1207
+ "step": 171
1208
+ },
1209
+ {
1210
+ "epoch": 0.03309124140252994,
1211
+ "grad_norm": 2.8341030096813395,
1212
+ "learning_rate": 1.9999563239430352e-05,
1213
+ "loss": 1.2951,
1214
+ "step": 172
1215
+ },
1216
+ {
1217
+ "epoch": 0.03328363234091674,
1218
+ "grad_norm": 3.4228491056879333,
1219
+ "learning_rate": 1.9999503064028043e-05,
1220
+ "loss": 1.3976,
1221
+ "step": 173
1222
+ },
1223
+ {
1224
+ "epoch": 0.033476023279303546,
1225
+ "grad_norm": 3.89631117502807,
1226
+ "learning_rate": 1.999943900647435e-05,
1227
+ "loss": 1.4222,
1228
+ "step": 174
1229
+ },
1230
+ {
1231
+ "epoch": 0.033668414217690346,
1232
+ "grad_norm": 3.5036522220755395,
1233
+ "learning_rate": 1.9999371066794146e-05,
1234
+ "loss": 1.3788,
1235
+ "step": 175
1236
+ },
1237
+ {
1238
+ "epoch": 0.03386080515607715,
1239
+ "grad_norm": 3.192845525272364,
1240
+ "learning_rate": 1.9999299245013805e-05,
1241
+ "loss": 1.2369,
1242
+ "step": 176
1243
+ },
1244
+ {
1245
+ "epoch": 0.03405319609446395,
1246
+ "grad_norm": 3.164520170774654,
1247
+ "learning_rate": 1.999922354116121e-05,
1248
+ "loss": 1.3555,
1249
+ "step": 177
1250
+ },
1251
+ {
1252
+ "epoch": 0.03424558703285075,
1253
+ "grad_norm": 3.5440531277707112,
1254
+ "learning_rate": 1.999914395526575e-05,
1255
+ "loss": 1.3534,
1256
+ "step": 178
1257
+ },
1258
+ {
1259
+ "epoch": 0.03443797797123756,
1260
+ "grad_norm": 3.7482594869867496,
1261
+ "learning_rate": 1.9999060487358333e-05,
1262
+ "loss": 1.2831,
1263
+ "step": 179
1264
+ },
1265
+ {
1266
+ "epoch": 0.03463036890962436,
1267
+ "grad_norm": 3.3869594251480737,
1268
+ "learning_rate": 1.9998973137471352e-05,
1269
+ "loss": 1.3303,
1270
+ "step": 180
1271
+ },
1272
+ {
1273
+ "epoch": 0.034822759848011156,
1274
+ "grad_norm": 3.5041506688632262,
1275
+ "learning_rate": 1.9998881905638727e-05,
1276
+ "loss": 1.2953,
1277
+ "step": 181
1278
+ },
1279
+ {
1280
+ "epoch": 0.03501515078639796,
1281
+ "grad_norm": 3.7177571630470014,
1282
+ "learning_rate": 1.9998786791895874e-05,
1283
+ "loss": 1.3184,
1284
+ "step": 182
1285
+ },
1286
+ {
1287
+ "epoch": 0.03520754172478476,
1288
+ "grad_norm": 3.4631415940492034,
1289
+ "learning_rate": 1.999868779627972e-05,
1290
+ "loss": 1.347,
1291
+ "step": 183
1292
+ },
1293
+ {
1294
+ "epoch": 0.03539993266317156,
1295
+ "grad_norm": 3.1615782500888296,
1296
+ "learning_rate": 1.9998584918828695e-05,
1297
+ "loss": 1.398,
1298
+ "step": 184
1299
+ },
1300
+ {
1301
+ "epoch": 0.03559232360155837,
1302
+ "grad_norm": 3.3162312003012975,
1303
+ "learning_rate": 1.9998478159582747e-05,
1304
+ "loss": 1.2542,
1305
+ "step": 185
1306
+ },
1307
+ {
1308
+ "epoch": 0.03578471453994517,
1309
+ "grad_norm": 3.1127318177993795,
1310
+ "learning_rate": 1.999836751858332e-05,
1311
+ "loss": 1.2804,
1312
+ "step": 186
1313
+ },
1314
+ {
1315
+ "epoch": 0.035977105478331974,
1316
+ "grad_norm": 3.545026558242914,
1317
+ "learning_rate": 1.9998252995873367e-05,
1318
+ "loss": 1.3128,
1319
+ "step": 187
1320
+ },
1321
+ {
1322
+ "epoch": 0.03616949641671877,
1323
+ "grad_norm": 3.6906353871414015,
1324
+ "learning_rate": 1.999813459149735e-05,
1325
+ "loss": 1.3065,
1326
+ "step": 188
1327
+ },
1328
+ {
1329
+ "epoch": 0.03636188735510557,
1330
+ "grad_norm": 3.5288025073219247,
1331
+ "learning_rate": 1.9998012305501243e-05,
1332
+ "loss": 1.3684,
1333
+ "step": 189
1334
+ },
1335
+ {
1336
+ "epoch": 0.03655427829349238,
1337
+ "grad_norm": 3.9995489316086053,
1338
+ "learning_rate": 1.999788613793251e-05,
1339
+ "loss": 1.3675,
1340
+ "step": 190
1341
+ },
1342
+ {
1343
+ "epoch": 0.03674666923187918,
1344
+ "grad_norm": 3.6614932100545703,
1345
+ "learning_rate": 1.999775608884015e-05,
1346
+ "loss": 1.3286,
1347
+ "step": 191
1348
+ },
1349
+ {
1350
+ "epoch": 0.03693906017026598,
1351
+ "grad_norm": 3.0068432376094743,
1352
+ "learning_rate": 1.9997622158274635e-05,
1353
+ "loss": 1.2528,
1354
+ "step": 192
1355
+ },
1356
+ {
1357
+ "epoch": 0.037131451108652784,
1358
+ "grad_norm": 3.6268699518199408,
1359
+ "learning_rate": 1.9997484346287973e-05,
1360
+ "loss": 1.4303,
1361
+ "step": 193
1362
+ },
1363
+ {
1364
+ "epoch": 0.03732384204703958,
1365
+ "grad_norm": 3.7805948089256445,
1366
+ "learning_rate": 1.9997342652933668e-05,
1367
+ "loss": 1.3726,
1368
+ "step": 194
1369
+ },
1370
+ {
1371
+ "epoch": 0.03751623298542639,
1372
+ "grad_norm": 3.55714131239389,
1373
+ "learning_rate": 1.9997197078266723e-05,
1374
+ "loss": 1.4871,
1375
+ "step": 195
1376
+ },
1377
+ {
1378
+ "epoch": 0.03770862392381319,
1379
+ "grad_norm": 3.125036697040528,
1380
+ "learning_rate": 1.999704762234366e-05,
1381
+ "loss": 1.2855,
1382
+ "step": 196
1383
+ },
1384
+ {
1385
+ "epoch": 0.03790101486219999,
1386
+ "grad_norm": 3.504801529603517,
1387
+ "learning_rate": 1.99968942852225e-05,
1388
+ "loss": 1.5002,
1389
+ "step": 197
1390
+ },
1391
+ {
1392
+ "epoch": 0.038093405800586795,
1393
+ "grad_norm": 3.292027917569746,
1394
+ "learning_rate": 1.9996737066962778e-05,
1395
+ "loss": 1.2743,
1396
+ "step": 198
1397
+ },
1398
+ {
1399
+ "epoch": 0.038285796738973595,
1400
+ "grad_norm": 2.9110512352717026,
1401
+ "learning_rate": 1.9996575967625525e-05,
1402
+ "loss": 1.3161,
1403
+ "step": 199
1404
+ },
1405
+ {
1406
+ "epoch": 0.038478187677360394,
1407
+ "grad_norm": 3.2204145147058174,
1408
+ "learning_rate": 1.999641098727329e-05,
1409
+ "loss": 1.3671,
1410
+ "step": 200
1411
+ },
1412
+ {
1413
+ "epoch": 0.0386705786157472,
1414
+ "grad_norm": 3.9809360259542794,
1415
+ "learning_rate": 1.999624212597013e-05,
1416
+ "loss": 1.4259,
1417
+ "step": 201
1418
+ },
1419
+ {
1420
+ "epoch": 0.038862969554134,
1421
+ "grad_norm": 3.2095868744346707,
1422
+ "learning_rate": 1.9996069383781587e-05,
1423
+ "loss": 1.3867,
1424
+ "step": 202
1425
+ },
1426
+ {
1427
+ "epoch": 0.0390553604925208,
1428
+ "grad_norm": 3.1132405994687384,
1429
+ "learning_rate": 1.9995892760774738e-05,
1430
+ "loss": 1.2283,
1431
+ "step": 203
1432
+ },
1433
+ {
1434
+ "epoch": 0.039247751430907606,
1435
+ "grad_norm": 2.898787932977016,
1436
+ "learning_rate": 1.9995712257018153e-05,
1437
+ "loss": 1.2556,
1438
+ "step": 204
1439
+ },
1440
+ {
1441
+ "epoch": 0.039440142369294405,
1442
+ "grad_norm": 3.5540700380635233,
1443
+ "learning_rate": 1.9995527872581903e-05,
1444
+ "loss": 1.3077,
1445
+ "step": 205
1446
+ },
1447
+ {
1448
+ "epoch": 0.03963253330768121,
1449
+ "grad_norm": 3.119718056806577,
1450
+ "learning_rate": 1.9995339607537578e-05,
1451
+ "loss": 1.3327,
1452
+ "step": 206
1453
+ },
1454
+ {
1455
+ "epoch": 0.03982492424606801,
1456
+ "grad_norm": 3.1909478863501857,
1457
+ "learning_rate": 1.9995147461958267e-05,
1458
+ "loss": 1.2348,
1459
+ "step": 207
1460
+ },
1461
+ {
1462
+ "epoch": 0.04001731518445481,
1463
+ "grad_norm": 3.183977552390641,
1464
+ "learning_rate": 1.999495143591857e-05,
1465
+ "loss": 1.3447,
1466
+ "step": 208
1467
+ },
1468
+ {
1469
+ "epoch": 0.04020970612284162,
1470
+ "grad_norm": 3.5890218841815225,
1471
+ "learning_rate": 1.999475152949459e-05,
1472
+ "loss": 1.382,
1473
+ "step": 209
1474
+ },
1475
+ {
1476
+ "epoch": 0.040402097061228416,
1477
+ "grad_norm": 3.137601507427993,
1478
+ "learning_rate": 1.9994547742763935e-05,
1479
+ "loss": 1.3103,
1480
+ "step": 210
1481
+ },
1482
+ {
1483
+ "epoch": 0.040594487999615216,
1484
+ "grad_norm": 2.839808998235376,
1485
+ "learning_rate": 1.9994340075805724e-05,
1486
+ "loss": 1.2432,
1487
+ "step": 211
1488
+ },
1489
+ {
1490
+ "epoch": 0.04078687893800202,
1491
+ "grad_norm": 3.2985280894842597,
1492
+ "learning_rate": 1.9994128528700583e-05,
1493
+ "loss": 1.3139,
1494
+ "step": 212
1495
+ },
1496
+ {
1497
+ "epoch": 0.04097926987638882,
1498
+ "grad_norm": 3.384286549900991,
1499
+ "learning_rate": 1.9993913101530635e-05,
1500
+ "loss": 1.3828,
1501
+ "step": 213
1502
+ },
1503
+ {
1504
+ "epoch": 0.04117166081477562,
1505
+ "grad_norm": 4.0641821268551785,
1506
+ "learning_rate": 1.9993693794379525e-05,
1507
+ "loss": 1.2476,
1508
+ "step": 214
1509
+ },
1510
+ {
1511
+ "epoch": 0.04136405175316243,
1512
+ "grad_norm": 3.7449043216878812,
1513
+ "learning_rate": 1.9993470607332387e-05,
1514
+ "loss": 1.3685,
1515
+ "step": 215
1516
+ },
1517
+ {
1518
+ "epoch": 0.04155644269154923,
1519
+ "grad_norm": 3.2484051178977924,
1520
+ "learning_rate": 1.999324354047588e-05,
1521
+ "loss": 1.3172,
1522
+ "step": 216
1523
+ },
1524
+ {
1525
+ "epoch": 0.04174883362993603,
1526
+ "grad_norm": 3.8837763504253773,
1527
+ "learning_rate": 1.9993012593898146e-05,
1528
+ "loss": 1.4819,
1529
+ "step": 217
1530
+ },
1531
+ {
1532
+ "epoch": 0.04194122456832283,
1533
+ "grad_norm": 3.5461977708815353,
1534
+ "learning_rate": 1.9992777767688857e-05,
1535
+ "loss": 1.3774,
1536
+ "step": 218
1537
+ },
1538
+ {
1539
+ "epoch": 0.04213361550670963,
1540
+ "grad_norm": 3.3713883400154554,
1541
+ "learning_rate": 1.9992539061939175e-05,
1542
+ "loss": 1.4018,
1543
+ "step": 219
1544
+ },
1545
+ {
1546
+ "epoch": 0.04232600644509644,
1547
+ "grad_norm": 3.191224534750556,
1548
+ "learning_rate": 1.999229647674178e-05,
1549
+ "loss": 1.2658,
1550
+ "step": 220
1551
+ },
1552
+ {
1553
+ "epoch": 0.04251839738348324,
1554
+ "grad_norm": 3.4332335724116025,
1555
+ "learning_rate": 1.9992050012190845e-05,
1556
+ "loss": 1.2941,
1557
+ "step": 221
1558
+ },
1559
+ {
1560
+ "epoch": 0.04271078832187004,
1561
+ "grad_norm": 3.6417867926794854,
1562
+ "learning_rate": 1.9991799668382058e-05,
1563
+ "loss": 1.4573,
1564
+ "step": 222
1565
+ },
1566
+ {
1567
+ "epoch": 0.042903179260256843,
1568
+ "grad_norm": 3.9175631809574614,
1569
+ "learning_rate": 1.9991545445412614e-05,
1570
+ "loss": 1.388,
1571
+ "step": 223
1572
+ },
1573
+ {
1574
+ "epoch": 0.04309557019864364,
1575
+ "grad_norm": 3.3799414382304858,
1576
+ "learning_rate": 1.9991287343381208e-05,
1577
+ "loss": 1.3448,
1578
+ "step": 224
1579
+ },
1580
+ {
1581
+ "epoch": 0.04328796113703045,
1582
+ "grad_norm": 2.8695847443627924,
1583
+ "learning_rate": 1.9991025362388044e-05,
1584
+ "loss": 1.1804,
1585
+ "step": 225
1586
+ },
1587
+ {
1588
+ "epoch": 0.04348035207541725,
1589
+ "grad_norm": 2.986840229750879,
1590
+ "learning_rate": 1.9990759502534835e-05,
1591
+ "loss": 1.233,
1592
+ "step": 226
1593
+ },
1594
+ {
1595
+ "epoch": 0.04367274301380405,
1596
+ "grad_norm": 3.3994082422223806,
1597
+ "learning_rate": 1.9990489763924796e-05,
1598
+ "loss": 1.3798,
1599
+ "step": 227
1600
+ },
1601
+ {
1602
+ "epoch": 0.043865133952190855,
1603
+ "grad_norm": 3.5815691240989436,
1604
+ "learning_rate": 1.9990216146662648e-05,
1605
+ "loss": 1.2864,
1606
+ "step": 228
1607
+ },
1608
+ {
1609
+ "epoch": 0.044057524890577654,
1610
+ "grad_norm": 3.122714887319509,
1611
+ "learning_rate": 1.9989938650854618e-05,
1612
+ "loss": 1.2615,
1613
+ "step": 229
1614
+ },
1615
+ {
1616
+ "epoch": 0.04424991582896445,
1617
+ "grad_norm": 2.65578920741492,
1618
+ "learning_rate": 1.998965727660844e-05,
1619
+ "loss": 1.2789,
1620
+ "step": 230
1621
+ },
1622
+ {
1623
+ "epoch": 0.04444230676735126,
1624
+ "grad_norm": 2.931038012405391,
1625
+ "learning_rate": 1.9989372024033352e-05,
1626
+ "loss": 1.2192,
1627
+ "step": 231
1628
+ },
1629
+ {
1630
+ "epoch": 0.04463469770573806,
1631
+ "grad_norm": 3.459015830581156,
1632
+ "learning_rate": 1.99890828932401e-05,
1633
+ "loss": 1.3635,
1634
+ "step": 232
1635
+ },
1636
+ {
1637
+ "epoch": 0.04482708864412486,
1638
+ "grad_norm": 3.2385333524622224,
1639
+ "learning_rate": 1.9988789884340938e-05,
1640
+ "loss": 1.4044,
1641
+ "step": 233
1642
+ },
1643
+ {
1644
+ "epoch": 0.045019479582511665,
1645
+ "grad_norm": 3.4098986455401032,
1646
+ "learning_rate": 1.9988492997449615e-05,
1647
+ "loss": 1.3519,
1648
+ "step": 234
1649
+ },
1650
+ {
1651
+ "epoch": 0.045211870520898464,
1652
+ "grad_norm": 3.337408436829067,
1653
+ "learning_rate": 1.9988192232681398e-05,
1654
+ "loss": 1.326,
1655
+ "step": 235
1656
+ },
1657
+ {
1658
+ "epoch": 0.04540426145928527,
1659
+ "grad_norm": 3.419853742731603,
1660
+ "learning_rate": 1.9987887590153055e-05,
1661
+ "loss": 1.2586,
1662
+ "step": 236
1663
+ },
1664
+ {
1665
+ "epoch": 0.04559665239767207,
1666
+ "grad_norm": 4.177141396589547,
1667
+ "learning_rate": 1.9987579069982856e-05,
1668
+ "loss": 1.3743,
1669
+ "step": 237
1670
+ },
1671
+ {
1672
+ "epoch": 0.04578904333605887,
1673
+ "grad_norm": 3.1681630848098585,
1674
+ "learning_rate": 1.9987266672290577e-05,
1675
+ "loss": 1.3692,
1676
+ "step": 238
1677
+ },
1678
+ {
1679
+ "epoch": 0.045981434274445676,
1680
+ "grad_norm": 4.00896667291785,
1681
+ "learning_rate": 1.9986950397197503e-05,
1682
+ "loss": 1.4503,
1683
+ "step": 239
1684
+ },
1685
+ {
1686
+ "epoch": 0.046173825212832476,
1687
+ "grad_norm": 3.4645024170606002,
1688
+ "learning_rate": 1.9986630244826425e-05,
1689
+ "loss": 1.4493,
1690
+ "step": 240
1691
+ },
1692
+ {
1693
+ "epoch": 0.046366216151219275,
1694
+ "grad_norm": 3.276562318348876,
1695
+ "learning_rate": 1.998630621530164e-05,
1696
+ "loss": 1.3034,
1697
+ "step": 241
1698
+ },
1699
+ {
1700
+ "epoch": 0.04655860708960608,
1701
+ "grad_norm": 4.02524883018013,
1702
+ "learning_rate": 1.998597830874894e-05,
1703
+ "loss": 1.3799,
1704
+ "step": 242
1705
+ },
1706
+ {
1707
+ "epoch": 0.04675099802799288,
1708
+ "grad_norm": 3.434312855363554,
1709
+ "learning_rate": 1.9985646525295634e-05,
1710
+ "loss": 1.3007,
1711
+ "step": 243
1712
+ },
1713
+ {
1714
+ "epoch": 0.04694338896637968,
1715
+ "grad_norm": 3.418078308418391,
1716
+ "learning_rate": 1.998531086507053e-05,
1717
+ "loss": 1.3027,
1718
+ "step": 244
1719
+ },
1720
+ {
1721
+ "epoch": 0.04713577990476649,
1722
+ "grad_norm": 3.1023691176524366,
1723
+ "learning_rate": 1.9984971328203945e-05,
1724
+ "loss": 1.2706,
1725
+ "step": 245
1726
+ },
1727
+ {
1728
+ "epoch": 0.047328170843153286,
1729
+ "grad_norm": 3.2666280755621497,
1730
+ "learning_rate": 1.9984627914827698e-05,
1731
+ "loss": 1.2766,
1732
+ "step": 246
1733
+ },
1734
+ {
1735
+ "epoch": 0.04752056178154009,
1736
+ "grad_norm": 3.6108260878269434,
1737
+ "learning_rate": 1.9984280625075115e-05,
1738
+ "loss": 1.2738,
1739
+ "step": 247
1740
+ },
1741
+ {
1742
+ "epoch": 0.04771295271992689,
1743
+ "grad_norm": 3.229568114321586,
1744
+ "learning_rate": 1.9983929459081022e-05,
1745
+ "loss": 1.433,
1746
+ "step": 248
1747
+ },
1748
+ {
1749
+ "epoch": 0.04790534365831369,
1750
+ "grad_norm": 3.4463317549972077,
1751
+ "learning_rate": 1.998357441698176e-05,
1752
+ "loss": 1.339,
1753
+ "step": 249
1754
+ },
1755
+ {
1756
+ "epoch": 0.0480977345967005,
1757
+ "grad_norm": 4.049338389902545,
1758
+ "learning_rate": 1.998321549891516e-05,
1759
+ "loss": 1.3494,
1760
+ "step": 250
1761
+ },
1762
+ {
1763
+ "epoch": 0.0482901255350873,
1764
+ "grad_norm": 3.4203804192433376,
1765
+ "learning_rate": 1.9982852705020572e-05,
1766
+ "loss": 1.3678,
1767
+ "step": 251
1768
+ },
1769
+ {
1770
+ "epoch": 0.0484825164734741,
1771
+ "grad_norm": 3.0081776061094443,
1772
+ "learning_rate": 1.9982486035438848e-05,
1773
+ "loss": 1.2528,
1774
+ "step": 252
1775
+ },
1776
+ {
1777
+ "epoch": 0.0486749074118609,
1778
+ "grad_norm": 3.1568061419424205,
1779
+ "learning_rate": 1.9982115490312334e-05,
1780
+ "loss": 1.3451,
1781
+ "step": 253
1782
+ },
1783
+ {
1784
+ "epoch": 0.0488672983502477,
1785
+ "grad_norm": 3.159295243509227,
1786
+ "learning_rate": 1.9981741069784894e-05,
1787
+ "loss": 1.3433,
1788
+ "step": 254
1789
+ },
1790
+ {
1791
+ "epoch": 0.04905968928863451,
1792
+ "grad_norm": 2.932873904357268,
1793
+ "learning_rate": 1.9981362774001886e-05,
1794
+ "loss": 1.3459,
1795
+ "step": 255
1796
+ },
1797
+ {
1798
+ "epoch": 0.04925208022702131,
1799
+ "grad_norm": 3.327282410126083,
1800
+ "learning_rate": 1.9980980603110185e-05,
1801
+ "loss": 1.3692,
1802
+ "step": 256
1803
+ },
1804
+ {
1805
+ "epoch": 0.04944447116540811,
1806
+ "grad_norm": 3.40264879516427,
1807
+ "learning_rate": 1.9980594557258158e-05,
1808
+ "loss": 1.2934,
1809
+ "step": 257
1810
+ },
1811
+ {
1812
+ "epoch": 0.049636862103794914,
1813
+ "grad_norm": 3.5028564915897813,
1814
+ "learning_rate": 1.9980204636595682e-05,
1815
+ "loss": 1.3755,
1816
+ "step": 258
1817
+ },
1818
+ {
1819
+ "epoch": 0.04982925304218171,
1820
+ "grad_norm": 3.045820756242345,
1821
+ "learning_rate": 1.9979810841274135e-05,
1822
+ "loss": 1.3348,
1823
+ "step": 259
1824
+ },
1825
+ {
1826
+ "epoch": 0.05002164398056851,
1827
+ "grad_norm": 3.1060207575884746,
1828
+ "learning_rate": 1.9979413171446403e-05,
1829
+ "loss": 1.3514,
1830
+ "step": 260
1831
+ },
1832
+ {
1833
+ "epoch": 0.05021403491895532,
1834
+ "grad_norm": 3.228281564858791,
1835
+ "learning_rate": 1.9979011627266884e-05,
1836
+ "loss": 1.3146,
1837
+ "step": 261
1838
+ },
1839
+ {
1840
+ "epoch": 0.05040642585734212,
1841
+ "grad_norm": 2.8823930987390063,
1842
+ "learning_rate": 1.997860620889146e-05,
1843
+ "loss": 1.3184,
1844
+ "step": 262
1845
+ },
1846
+ {
1847
+ "epoch": 0.05059881679572892,
1848
+ "grad_norm": 3.161700641113056,
1849
+ "learning_rate": 1.997819691647753e-05,
1850
+ "loss": 1.3638,
1851
+ "step": 263
1852
+ },
1853
+ {
1854
+ "epoch": 0.050791207734115724,
1855
+ "grad_norm": 3.575497308874466,
1856
+ "learning_rate": 1.9977783750184e-05,
1857
+ "loss": 1.334,
1858
+ "step": 264
1859
+ },
1860
+ {
1861
+ "epoch": 0.050983598672502524,
1862
+ "grad_norm": 3.333371111323695,
1863
+ "learning_rate": 1.9977366710171274e-05,
1864
+ "loss": 1.3573,
1865
+ "step": 265
1866
+ },
1867
+ {
1868
+ "epoch": 0.05117598961088933,
1869
+ "grad_norm": 2.656362659722232,
1870
+ "learning_rate": 1.9976945796601258e-05,
1871
+ "loss": 1.1976,
1872
+ "step": 266
1873
+ },
1874
+ {
1875
+ "epoch": 0.05136838054927613,
1876
+ "grad_norm": 3.75578290416007,
1877
+ "learning_rate": 1.9976521009637366e-05,
1878
+ "loss": 1.396,
1879
+ "step": 267
1880
+ },
1881
+ {
1882
+ "epoch": 0.05156077148766293,
1883
+ "grad_norm": 3.0597216322239786,
1884
+ "learning_rate": 1.997609234944452e-05,
1885
+ "loss": 1.3497,
1886
+ "step": 268
1887
+ },
1888
+ {
1889
+ "epoch": 0.051753162426049736,
1890
+ "grad_norm": 3.2223394528301044,
1891
+ "learning_rate": 1.9975659816189137e-05,
1892
+ "loss": 1.313,
1893
+ "step": 269
1894
+ },
1895
+ {
1896
+ "epoch": 0.051945553364436535,
1897
+ "grad_norm": 2.827752375759985,
1898
+ "learning_rate": 1.997522341003914e-05,
1899
+ "loss": 1.2429,
1900
+ "step": 270
1901
+ },
1902
+ {
1903
+ "epoch": 0.052137944302823334,
1904
+ "grad_norm": 3.203533699054935,
1905
+ "learning_rate": 1.9974783131163957e-05,
1906
+ "loss": 1.358,
1907
+ "step": 271
1908
+ },
1909
+ {
1910
+ "epoch": 0.05233033524121014,
1911
+ "grad_norm": 3.448554963862977,
1912
+ "learning_rate": 1.9974338979734523e-05,
1913
+ "loss": 1.3713,
1914
+ "step": 272
1915
+ },
1916
+ {
1917
+ "epoch": 0.05252272617959694,
1918
+ "grad_norm": 3.0106084543566793,
1919
+ "learning_rate": 1.997389095592327e-05,
1920
+ "loss": 1.3622,
1921
+ "step": 273
1922
+ },
1923
+ {
1924
+ "epoch": 0.05271511711798374,
1925
+ "grad_norm": 3.2311132402524914,
1926
+ "learning_rate": 1.9973439059904133e-05,
1927
+ "loss": 1.4206,
1928
+ "step": 274
1929
+ },
1930
+ {
1931
+ "epoch": 0.052907508056370546,
1932
+ "grad_norm": 3.1311371497154488,
1933
+ "learning_rate": 1.9972983291852565e-05,
1934
+ "loss": 1.3187,
1935
+ "step": 275
1936
+ },
1937
+ {
1938
+ "epoch": 0.053099898994757345,
1939
+ "grad_norm": 3.3207243134592512,
1940
+ "learning_rate": 1.9972523651945496e-05,
1941
+ "loss": 1.4106,
1942
+ "step": 276
1943
+ },
1944
+ {
1945
+ "epoch": 0.05329228993314415,
1946
+ "grad_norm": 3.1171149258187456,
1947
+ "learning_rate": 1.9972060140361384e-05,
1948
+ "loss": 1.2392,
1949
+ "step": 277
1950
+ },
1951
+ {
1952
+ "epoch": 0.05348468087153095,
1953
+ "grad_norm": 2.9177528899257874,
1954
+ "learning_rate": 1.997159275728018e-05,
1955
+ "loss": 1.444,
1956
+ "step": 278
1957
+ },
1958
+ {
1959
+ "epoch": 0.05367707180991775,
1960
+ "grad_norm": 2.953116640632381,
1961
+ "learning_rate": 1.9971121502883332e-05,
1962
+ "loss": 1.2522,
1963
+ "step": 279
1964
+ },
1965
+ {
1966
+ "epoch": 0.05386946274830456,
1967
+ "grad_norm": 2.827803062016073,
1968
+ "learning_rate": 1.9970646377353802e-05,
1969
+ "loss": 1.2545,
1970
+ "step": 280
1971
+ },
1972
+ {
1973
+ "epoch": 0.05406185368669136,
1974
+ "grad_norm": 2.912366271545338,
1975
+ "learning_rate": 1.997016738087605e-05,
1976
+ "loss": 1.1966,
1977
+ "step": 281
1978
+ },
1979
+ {
1980
+ "epoch": 0.054254244625078156,
1981
+ "grad_norm": 3.4439124482125583,
1982
+ "learning_rate": 1.9969684513636035e-05,
1983
+ "loss": 1.3294,
1984
+ "step": 282
1985
+ },
1986
+ {
1987
+ "epoch": 0.05444663556346496,
1988
+ "grad_norm": 3.069794541278492,
1989
+ "learning_rate": 1.9969197775821227e-05,
1990
+ "loss": 1.289,
1991
+ "step": 283
1992
+ },
1993
+ {
1994
+ "epoch": 0.05463902650185176,
1995
+ "grad_norm": 3.009101858033586,
1996
+ "learning_rate": 1.9968707167620593e-05,
1997
+ "loss": 1.2529,
1998
+ "step": 284
1999
+ },
2000
+ {
2001
+ "epoch": 0.05483141744023857,
2002
+ "grad_norm": 3.279568609190259,
2003
+ "learning_rate": 1.9968212689224603e-05,
2004
+ "loss": 1.3773,
2005
+ "step": 285
2006
+ },
2007
+ {
2008
+ "epoch": 0.05502380837862537,
2009
+ "grad_norm": 3.2780224483676013,
2010
+ "learning_rate": 1.996771434082523e-05,
2011
+ "loss": 1.241,
2012
+ "step": 286
2013
+ },
2014
+ {
2015
+ "epoch": 0.05521619931701217,
2016
+ "grad_norm": 2.999995980325934,
2017
+ "learning_rate": 1.9967212122615958e-05,
2018
+ "loss": 1.3069,
2019
+ "step": 287
2020
+ },
2021
+ {
2022
+ "epoch": 0.05540859025539897,
2023
+ "grad_norm": 3.312729305799956,
2024
+ "learning_rate": 1.9966706034791752e-05,
2025
+ "loss": 1.3344,
2026
+ "step": 288
2027
+ },
2028
+ {
2029
+ "epoch": 0.05560098119378577,
2030
+ "grad_norm": 3.9630135771906057,
2031
+ "learning_rate": 1.9966196077549106e-05,
2032
+ "loss": 1.3276,
2033
+ "step": 289
2034
+ },
2035
+ {
2036
+ "epoch": 0.05579337213217257,
2037
+ "grad_norm": 3.353949651678405,
2038
+ "learning_rate": 1.996568225108599e-05,
2039
+ "loss": 1.3988,
2040
+ "step": 290
2041
+ },
2042
+ {
2043
+ "epoch": 0.05598576307055938,
2044
+ "grad_norm": 2.498748484170424,
2045
+ "learning_rate": 1.99651645556019e-05,
2046
+ "loss": 1.2112,
2047
+ "step": 291
2048
+ },
2049
+ {
2050
+ "epoch": 0.05617815400894618,
2051
+ "grad_norm": 3.3852729681658467,
2052
+ "learning_rate": 1.9964642991297817e-05,
2053
+ "loss": 1.353,
2054
+ "step": 292
2055
+ },
2056
+ {
2057
+ "epoch": 0.05637054494733298,
2058
+ "grad_norm": 2.6984438029648614,
2059
+ "learning_rate": 1.996411755837623e-05,
2060
+ "loss": 1.299,
2061
+ "step": 293
2062
+ },
2063
+ {
2064
+ "epoch": 0.056562935885719784,
2065
+ "grad_norm": 3.06342573072672,
2066
+ "learning_rate": 1.9963588257041137e-05,
2067
+ "loss": 1.3837,
2068
+ "step": 294
2069
+ },
2070
+ {
2071
+ "epoch": 0.05675532682410658,
2072
+ "grad_norm": 3.084728085361937,
2073
+ "learning_rate": 1.996305508749802e-05,
2074
+ "loss": 1.3168,
2075
+ "step": 295
2076
+ },
2077
+ {
2078
+ "epoch": 0.05694771776249339,
2079
+ "grad_norm": 3.2161188283431152,
2080
+ "learning_rate": 1.9962518049953887e-05,
2081
+ "loss": 1.3795,
2082
+ "step": 296
2083
+ },
2084
+ {
2085
+ "epoch": 0.05714010870088019,
2086
+ "grad_norm": 3.0334314681598924,
2087
+ "learning_rate": 1.9961977144617225e-05,
2088
+ "loss": 1.2531,
2089
+ "step": 297
2090
+ },
2091
+ {
2092
+ "epoch": 0.05733249963926699,
2093
+ "grad_norm": 3.107915568943348,
2094
+ "learning_rate": 1.996143237169803e-05,
2095
+ "loss": 1.2853,
2096
+ "step": 298
2097
+ },
2098
+ {
2099
+ "epoch": 0.057524890577653795,
2100
+ "grad_norm": 3.411054552233234,
2101
+ "learning_rate": 1.996088373140781e-05,
2102
+ "loss": 1.3917,
2103
+ "step": 299
2104
+ },
2105
+ {
2106
+ "epoch": 0.057717281516040594,
2107
+ "grad_norm": 3.021077832989322,
2108
+ "learning_rate": 1.9960331223959564e-05,
2109
+ "loss": 1.4026,
2110
+ "step": 300
2111
+ },
2112
+ {
2113
+ "epoch": 0.057909672454427394,
2114
+ "grad_norm": 2.6669934020451525,
2115
+ "learning_rate": 1.995977484956779e-05,
2116
+ "loss": 1.2223,
2117
+ "step": 301
2118
+ },
2119
+ {
2120
+ "epoch": 0.0581020633928142,
2121
+ "grad_norm": 3.003737061756357,
2122
+ "learning_rate": 1.9959214608448495e-05,
2123
+ "loss": 1.3617,
2124
+ "step": 302
2125
+ },
2126
+ {
2127
+ "epoch": 0.058294454331201,
2128
+ "grad_norm": 2.9416662154737265,
2129
+ "learning_rate": 1.9958650500819183e-05,
2130
+ "loss": 1.2639,
2131
+ "step": 303
2132
+ },
2133
+ {
2134
+ "epoch": 0.0584868452695878,
2135
+ "grad_norm": 3.5006891063041587,
2136
+ "learning_rate": 1.995808252689886e-05,
2137
+ "loss": 1.3415,
2138
+ "step": 304
2139
+ },
2140
+ {
2141
+ "epoch": 0.058679236207974605,
2142
+ "grad_norm": 3.645101562859597,
2143
+ "learning_rate": 1.9957510686908034e-05,
2144
+ "loss": 1.2652,
2145
+ "step": 305
2146
+ },
2147
+ {
2148
+ "epoch": 0.058871627146361405,
2149
+ "grad_norm": 3.0805525570605172,
2150
+ "learning_rate": 1.9956934981068713e-05,
2151
+ "loss": 1.2145,
2152
+ "step": 306
2153
+ },
2154
+ {
2155
+ "epoch": 0.05906401808474821,
2156
+ "grad_norm": 2.8694139260214384,
2157
+ "learning_rate": 1.9956355409604402e-05,
2158
+ "loss": 1.2865,
2159
+ "step": 307
2160
+ },
2161
+ {
2162
+ "epoch": 0.05925640902313501,
2163
+ "grad_norm": 2.984217296816447,
2164
+ "learning_rate": 1.9955771972740118e-05,
2165
+ "loss": 1.3294,
2166
+ "step": 308
2167
+ },
2168
+ {
2169
+ "epoch": 0.05944879996152181,
2170
+ "grad_norm": 2.904910765062724,
2171
+ "learning_rate": 1.9955184670702363e-05,
2172
+ "loss": 1.3932,
2173
+ "step": 309
2174
+ },
2175
+ {
2176
+ "epoch": 0.05964119089990862,
2177
+ "grad_norm": 3.341287444059739,
2178
+ "learning_rate": 1.995459350371915e-05,
2179
+ "loss": 1.3307,
2180
+ "step": 310
2181
+ },
2182
+ {
2183
+ "epoch": 0.059833581838295416,
2184
+ "grad_norm": 3.1997936104764038,
2185
+ "learning_rate": 1.9953998472019996e-05,
2186
+ "loss": 1.3315,
2187
+ "step": 311
2188
+ },
2189
+ {
2190
+ "epoch": 0.060025972776682215,
2191
+ "grad_norm": 2.8352595111672465,
2192
+ "learning_rate": 1.995339957583591e-05,
2193
+ "loss": 1.2747,
2194
+ "step": 312
2195
+ },
2196
+ {
2197
+ "epoch": 0.06021836371506902,
2198
+ "grad_norm": 3.568311140344332,
2199
+ "learning_rate": 1.9952796815399403e-05,
2200
+ "loss": 1.2788,
2201
+ "step": 313
2202
+ },
2203
+ {
2204
+ "epoch": 0.06041075465345582,
2205
+ "grad_norm": 2.9250295150154795,
2206
+ "learning_rate": 1.9952190190944484e-05,
2207
+ "loss": 1.3404,
2208
+ "step": 314
2209
+ },
2210
+ {
2211
+ "epoch": 0.06060314559184263,
2212
+ "grad_norm": 3.1263327294036833,
2213
+ "learning_rate": 1.9951579702706668e-05,
2214
+ "loss": 1.4053,
2215
+ "step": 315
2216
+ },
2217
+ {
2218
+ "epoch": 0.06079553653022943,
2219
+ "grad_norm": 2.5287565111688632,
2220
+ "learning_rate": 1.9950965350922975e-05,
2221
+ "loss": 1.2107,
2222
+ "step": 316
2223
+ },
2224
+ {
2225
+ "epoch": 0.060987927468616226,
2226
+ "grad_norm": 3.353749376034332,
2227
+ "learning_rate": 1.9950347135831907e-05,
2228
+ "loss": 1.3264,
2229
+ "step": 317
2230
+ },
2231
+ {
2232
+ "epoch": 0.06118031840700303,
2233
+ "grad_norm": 3.6321124730079406,
2234
+ "learning_rate": 1.994972505767348e-05,
2235
+ "loss": 1.3501,
2236
+ "step": 318
2237
+ },
2238
+ {
2239
+ "epoch": 0.06137270934538983,
2240
+ "grad_norm": 3.096254086295918,
2241
+ "learning_rate": 1.994909911668921e-05,
2242
+ "loss": 1.3059,
2243
+ "step": 319
2244
+ },
2245
+ {
2246
+ "epoch": 0.06156510028377663,
2247
+ "grad_norm": 2.5879351422051635,
2248
+ "learning_rate": 1.99484693131221e-05,
2249
+ "loss": 1.2356,
2250
+ "step": 320
2251
+ },
2252
+ {
2253
+ "epoch": 0.06175749122216344,
2254
+ "grad_norm": 3.165740678692725,
2255
+ "learning_rate": 1.994783564721667e-05,
2256
+ "loss": 1.2915,
2257
+ "step": 321
2258
+ },
2259
+ {
2260
+ "epoch": 0.06194988216055024,
2261
+ "grad_norm": 3.349102082685375,
2262
+ "learning_rate": 1.9947198119218924e-05,
2263
+ "loss": 1.4176,
2264
+ "step": 322
2265
+ },
2266
+ {
2267
+ "epoch": 0.06214227309893704,
2268
+ "grad_norm": 3.3976655578423562,
2269
+ "learning_rate": 1.994655672937638e-05,
2270
+ "loss": 1.3558,
2271
+ "step": 323
2272
+ },
2273
+ {
2274
+ "epoch": 0.06233466403732384,
2275
+ "grad_norm": 3.3271580735918094,
2276
+ "learning_rate": 1.9945911477938044e-05,
2277
+ "loss": 1.3463,
2278
+ "step": 324
2279
+ },
2280
+ {
2281
+ "epoch": 0.06252705497571065,
2282
+ "grad_norm": 2.5952891773272038,
2283
+ "learning_rate": 1.994526236515442e-05,
2284
+ "loss": 1.1785,
2285
+ "step": 325
2286
+ },
2287
+ {
2288
+ "epoch": 0.06271944591409745,
2289
+ "grad_norm": 3.544594701649707,
2290
+ "learning_rate": 1.994460939127753e-05,
2291
+ "loss": 1.2953,
2292
+ "step": 326
2293
+ },
2294
+ {
2295
+ "epoch": 0.06291183685248425,
2296
+ "grad_norm": 3.407515265644298,
2297
+ "learning_rate": 1.9943952556560863e-05,
2298
+ "loss": 1.2786,
2299
+ "step": 327
2300
+ },
2301
+ {
2302
+ "epoch": 0.06310422779087105,
2303
+ "grad_norm": 2.717214472551306,
2304
+ "learning_rate": 1.9943291861259433e-05,
2305
+ "loss": 1.2224,
2306
+ "step": 328
2307
+ },
2308
+ {
2309
+ "epoch": 0.06329661872925785,
2310
+ "grad_norm": 3.209674078047875,
2311
+ "learning_rate": 1.9942627305629747e-05,
2312
+ "loss": 1.3204,
2313
+ "step": 329
2314
+ },
2315
+ {
2316
+ "epoch": 0.06348900966764466,
2317
+ "grad_norm": 3.3551103826386988,
2318
+ "learning_rate": 1.9941958889929808e-05,
2319
+ "loss": 1.3348,
2320
+ "step": 330
2321
+ },
2322
+ {
2323
+ "epoch": 0.06368140060603146,
2324
+ "grad_norm": 2.732309891639565,
2325
+ "learning_rate": 1.9941286614419113e-05,
2326
+ "loss": 1.2209,
2327
+ "step": 331
2328
+ },
2329
+ {
2330
+ "epoch": 0.06387379154441826,
2331
+ "grad_norm": 3.0844892899055023,
2332
+ "learning_rate": 1.994061047935867e-05,
2333
+ "loss": 1.3371,
2334
+ "step": 332
2335
+ },
2336
+ {
2337
+ "epoch": 0.06406618248280506,
2338
+ "grad_norm": 3.052699248227704,
2339
+ "learning_rate": 1.9939930485010968e-05,
2340
+ "loss": 1.3391,
2341
+ "step": 333
2342
+ },
2343
+ {
2344
+ "epoch": 0.06425857342119186,
2345
+ "grad_norm": 3.006284824941497,
2346
+ "learning_rate": 1.9939246631640014e-05,
2347
+ "loss": 1.212,
2348
+ "step": 334
2349
+ },
2350
+ {
2351
+ "epoch": 0.06445096435957866,
2352
+ "grad_norm": 3.4018556229377697,
2353
+ "learning_rate": 1.99385589195113e-05,
2354
+ "loss": 1.3663,
2355
+ "step": 335
2356
+ },
2357
+ {
2358
+ "epoch": 0.06464335529796547,
2359
+ "grad_norm": 3.304200009640364,
2360
+ "learning_rate": 1.9937867348891815e-05,
2361
+ "loss": 1.3525,
2362
+ "step": 336
2363
+ },
2364
+ {
2365
+ "epoch": 0.06483574623635227,
2366
+ "grad_norm": 2.5928544668817244,
2367
+ "learning_rate": 1.9937171920050057e-05,
2368
+ "loss": 1.2591,
2369
+ "step": 337
2370
+ },
2371
+ {
2372
+ "epoch": 0.06502813717473907,
2373
+ "grad_norm": 2.76556082116995,
2374
+ "learning_rate": 1.9936472633256012e-05,
2375
+ "loss": 1.2593,
2376
+ "step": 338
2377
+ },
2378
+ {
2379
+ "epoch": 0.06522052811312587,
2380
+ "grad_norm": 3.3699396142634077,
2381
+ "learning_rate": 1.9935769488781167e-05,
2382
+ "loss": 1.3127,
2383
+ "step": 339
2384
+ },
2385
+ {
2386
+ "epoch": 0.06541291905151267,
2387
+ "grad_norm": 3.1849085896961316,
2388
+ "learning_rate": 1.993506248689851e-05,
2389
+ "loss": 1.1739,
2390
+ "step": 340
2391
+ },
2392
+ {
2393
+ "epoch": 0.06560530998989948,
2394
+ "grad_norm": 2.9446436123369737,
2395
+ "learning_rate": 1.993435162788252e-05,
2396
+ "loss": 1.2765,
2397
+ "step": 341
2398
+ },
2399
+ {
2400
+ "epoch": 0.06579770092828628,
2401
+ "grad_norm": 3.0250244847511687,
2402
+ "learning_rate": 1.993363691200918e-05,
2403
+ "loss": 1.3546,
2404
+ "step": 342
2405
+ },
2406
+ {
2407
+ "epoch": 0.06599009186667308,
2408
+ "grad_norm": 3.0134566329084818,
2409
+ "learning_rate": 1.9932918339555965e-05,
2410
+ "loss": 1.4037,
2411
+ "step": 343
2412
+ },
2413
+ {
2414
+ "epoch": 0.06618248280505988,
2415
+ "grad_norm": 3.566099513979839,
2416
+ "learning_rate": 1.9932195910801848e-05,
2417
+ "loss": 1.3606,
2418
+ "step": 344
2419
+ },
2420
+ {
2421
+ "epoch": 0.06637487374344668,
2422
+ "grad_norm": 3.089629096818802,
2423
+ "learning_rate": 1.9931469626027305e-05,
2424
+ "loss": 1.3351,
2425
+ "step": 345
2426
+ },
2427
+ {
2428
+ "epoch": 0.06656726468183348,
2429
+ "grad_norm": 3.075174135542356,
2430
+ "learning_rate": 1.9930739485514304e-05,
2431
+ "loss": 1.3332,
2432
+ "step": 346
2433
+ },
2434
+ {
2435
+ "epoch": 0.06675965562022029,
2436
+ "grad_norm": 2.9479620638129536,
2437
+ "learning_rate": 1.9930005489546308e-05,
2438
+ "loss": 1.3025,
2439
+ "step": 347
2440
+ },
2441
+ {
2442
+ "epoch": 0.06695204655860709,
2443
+ "grad_norm": 2.7246531110321337,
2444
+ "learning_rate": 1.9929267638408277e-05,
2445
+ "loss": 1.3417,
2446
+ "step": 348
2447
+ },
2448
+ {
2449
+ "epoch": 0.06714443749699389,
2450
+ "grad_norm": 2.8842020037105884,
2451
+ "learning_rate": 1.9928525932386678e-05,
2452
+ "loss": 1.2763,
2453
+ "step": 349
2454
+ },
2455
+ {
2456
+ "epoch": 0.06733682843538069,
2457
+ "grad_norm": 2.5013972955412,
2458
+ "learning_rate": 1.9927780371769463e-05,
2459
+ "loss": 1.2753,
2460
+ "step": 350
2461
+ },
2462
+ {
2463
+ "epoch": 0.06752921937376749,
2464
+ "grad_norm": 3.3629781382672492,
2465
+ "learning_rate": 1.9927030956846083e-05,
2466
+ "loss": 1.3662,
2467
+ "step": 351
2468
+ },
2469
+ {
2470
+ "epoch": 0.0677216103121543,
2471
+ "grad_norm": 3.2328395817663944,
2472
+ "learning_rate": 1.992627768790749e-05,
2473
+ "loss": 1.3414,
2474
+ "step": 352
2475
+ },
2476
+ {
2477
+ "epoch": 0.0679140012505411,
2478
+ "grad_norm": 3.1400605819086334,
2479
+ "learning_rate": 1.9925520565246125e-05,
2480
+ "loss": 1.3003,
2481
+ "step": 353
2482
+ },
2483
+ {
2484
+ "epoch": 0.0681063921889279,
2485
+ "grad_norm": 2.8203238470201955,
2486
+ "learning_rate": 1.9924759589155932e-05,
2487
+ "loss": 1.3375,
2488
+ "step": 354
2489
+ },
2490
+ {
2491
+ "epoch": 0.0682987831273147,
2492
+ "grad_norm": 2.6168351836486123,
2493
+ "learning_rate": 1.9923994759932344e-05,
2494
+ "loss": 1.3189,
2495
+ "step": 355
2496
+ },
2497
+ {
2498
+ "epoch": 0.0684911740657015,
2499
+ "grad_norm": 2.854079493104668,
2500
+ "learning_rate": 1.9923226077872296e-05,
2501
+ "loss": 1.271,
2502
+ "step": 356
2503
+ },
2504
+ {
2505
+ "epoch": 0.0686835650040883,
2506
+ "grad_norm": 3.329364068217029,
2507
+ "learning_rate": 1.9922453543274223e-05,
2508
+ "loss": 1.2837,
2509
+ "step": 357
2510
+ },
2511
+ {
2512
+ "epoch": 0.06887595594247511,
2513
+ "grad_norm": 2.1243476353135797,
2514
+ "learning_rate": 1.9921677156438044e-05,
2515
+ "loss": 1.1984,
2516
+ "step": 358
2517
+ },
2518
+ {
2519
+ "epoch": 0.06906834688086191,
2520
+ "grad_norm": 3.053154629492327,
2521
+ "learning_rate": 1.9920896917665178e-05,
2522
+ "loss": 1.3582,
2523
+ "step": 359
2524
+ },
2525
+ {
2526
+ "epoch": 0.06926073781924871,
2527
+ "grad_norm": 2.926534645895194,
2528
+ "learning_rate": 1.992011282725854e-05,
2529
+ "loss": 1.2684,
2530
+ "step": 360
2531
+ },
2532
+ {
2533
+ "epoch": 0.06945312875763551,
2534
+ "grad_norm": 3.1709911069343324,
2535
+ "learning_rate": 1.9919324885522548e-05,
2536
+ "loss": 1.2826,
2537
+ "step": 361
2538
+ },
2539
+ {
2540
+ "epoch": 0.06964551969602231,
2541
+ "grad_norm": 2.801010993498812,
2542
+ "learning_rate": 1.99185330927631e-05,
2543
+ "loss": 1.3309,
2544
+ "step": 362
2545
+ },
2546
+ {
2547
+ "epoch": 0.06983791063440913,
2548
+ "grad_norm": 2.8494696945364235,
2549
+ "learning_rate": 1.99177374492876e-05,
2550
+ "loss": 1.3195,
2551
+ "step": 363
2552
+ },
2553
+ {
2554
+ "epoch": 0.07003030157279592,
2555
+ "grad_norm": 2.920485356763821,
2556
+ "learning_rate": 1.991693795540494e-05,
2557
+ "loss": 1.2408,
2558
+ "step": 364
2559
+ },
2560
+ {
2561
+ "epoch": 0.07022269251118272,
2562
+ "grad_norm": 3.110090882812885,
2563
+ "learning_rate": 1.9916134611425522e-05,
2564
+ "loss": 1.3174,
2565
+ "step": 365
2566
+ },
2567
+ {
2568
+ "epoch": 0.07041508344956952,
2569
+ "grad_norm": 2.948566475573386,
2570
+ "learning_rate": 1.9915327417661226e-05,
2571
+ "loss": 1.3482,
2572
+ "step": 366
2573
+ },
2574
+ {
2575
+ "epoch": 0.07060747438795632,
2576
+ "grad_norm": 2.9617434830038496,
2577
+ "learning_rate": 1.991451637442543e-05,
2578
+ "loss": 1.3136,
2579
+ "step": 367
2580
+ },
2581
+ {
2582
+ "epoch": 0.07079986532634312,
2583
+ "grad_norm": 2.748824967842853,
2584
+ "learning_rate": 1.9913701482033008e-05,
2585
+ "loss": 1.2455,
2586
+ "step": 368
2587
+ },
2588
+ {
2589
+ "epoch": 0.07099225626472994,
2590
+ "grad_norm": 2.980655822963688,
2591
+ "learning_rate": 1.9912882740800336e-05,
2592
+ "loss": 1.2514,
2593
+ "step": 369
2594
+ },
2595
+ {
2596
+ "epoch": 0.07118464720311674,
2597
+ "grad_norm": 2.69702640294432,
2598
+ "learning_rate": 1.9912060151045273e-05,
2599
+ "loss": 1.3134,
2600
+ "step": 370
2601
+ },
2602
+ {
2603
+ "epoch": 0.07137703814150353,
2604
+ "grad_norm": 3.2067960918596423,
2605
+ "learning_rate": 1.9911233713087172e-05,
2606
+ "loss": 1.3189,
2607
+ "step": 371
2608
+ },
2609
+ {
2610
+ "epoch": 0.07156942907989033,
2611
+ "grad_norm": 3.292370215865587,
2612
+ "learning_rate": 1.9910403427246895e-05,
2613
+ "loss": 1.304,
2614
+ "step": 372
2615
+ },
2616
+ {
2617
+ "epoch": 0.07176182001827713,
2618
+ "grad_norm": 3.1295716881232774,
2619
+ "learning_rate": 1.990956929384678e-05,
2620
+ "loss": 1.2027,
2621
+ "step": 373
2622
+ },
2623
+ {
2624
+ "epoch": 0.07195421095666395,
2625
+ "grad_norm": 2.667394299241552,
2626
+ "learning_rate": 1.990873131321067e-05,
2627
+ "loss": 1.2186,
2628
+ "step": 374
2629
+ },
2630
+ {
2631
+ "epoch": 0.07214660189505075,
2632
+ "grad_norm": 3.1071516311875387,
2633
+ "learning_rate": 1.9907889485663897e-05,
2634
+ "loss": 1.2644,
2635
+ "step": 375
2636
+ },
2637
+ {
2638
+ "epoch": 0.07233899283343755,
2639
+ "grad_norm": 2.078747766698742,
2640
+ "learning_rate": 1.9907043811533283e-05,
2641
+ "loss": 1.1508,
2642
+ "step": 376
2643
+ },
2644
+ {
2645
+ "epoch": 0.07253138377182435,
2646
+ "grad_norm": 3.0596159516921073,
2647
+ "learning_rate": 1.9906194291147155e-05,
2648
+ "loss": 1.3451,
2649
+ "step": 377
2650
+ },
2651
+ {
2652
+ "epoch": 0.07272377471021114,
2653
+ "grad_norm": 2.6549562570625422,
2654
+ "learning_rate": 1.9905340924835322e-05,
2655
+ "loss": 1.3043,
2656
+ "step": 378
2657
+ },
2658
+ {
2659
+ "epoch": 0.07291616564859794,
2660
+ "grad_norm": 2.7464226896376274,
2661
+ "learning_rate": 1.9904483712929094e-05,
2662
+ "loss": 1.2999,
2663
+ "step": 379
2664
+ },
2665
+ {
2666
+ "epoch": 0.07310855658698476,
2667
+ "grad_norm": 2.962629947805071,
2668
+ "learning_rate": 1.9903622655761267e-05,
2669
+ "loss": 1.2264,
2670
+ "step": 380
2671
+ },
2672
+ {
2673
+ "epoch": 0.07330094752537156,
2674
+ "grad_norm": 2.8469757571782224,
2675
+ "learning_rate": 1.990275775366613e-05,
2676
+ "loss": 1.2846,
2677
+ "step": 381
2678
+ },
2679
+ {
2680
+ "epoch": 0.07349333846375836,
2681
+ "grad_norm": 2.71868754081009,
2682
+ "learning_rate": 1.9901889006979473e-05,
2683
+ "loss": 1.272,
2684
+ "step": 382
2685
+ },
2686
+ {
2687
+ "epoch": 0.07368572940214516,
2688
+ "grad_norm": 2.628122279355355,
2689
+ "learning_rate": 1.990101641603857e-05,
2690
+ "loss": 1.213,
2691
+ "step": 383
2692
+ },
2693
+ {
2694
+ "epoch": 0.07387812034053196,
2695
+ "grad_norm": 2.977898396465207,
2696
+ "learning_rate": 1.9900139981182193e-05,
2697
+ "loss": 1.2937,
2698
+ "step": 384
2699
+ },
2700
+ {
2701
+ "epoch": 0.07407051127891877,
2702
+ "grad_norm": 2.9968281760229973,
2703
+ "learning_rate": 1.9899259702750604e-05,
2704
+ "loss": 1.1936,
2705
+ "step": 385
2706
+ },
2707
+ {
2708
+ "epoch": 0.07426290221730557,
2709
+ "grad_norm": 3.0695820225444503,
2710
+ "learning_rate": 1.9898375581085555e-05,
2711
+ "loss": 1.1853,
2712
+ "step": 386
2713
+ },
2714
+ {
2715
+ "epoch": 0.07445529315569237,
2716
+ "grad_norm": 2.7907777595708527,
2717
+ "learning_rate": 1.9897487616530296e-05,
2718
+ "loss": 1.2413,
2719
+ "step": 387
2720
+ },
2721
+ {
2722
+ "epoch": 0.07464768409407917,
2723
+ "grad_norm": 3.1430693749997265,
2724
+ "learning_rate": 1.9896595809429565e-05,
2725
+ "loss": 1.3225,
2726
+ "step": 388
2727
+ },
2728
+ {
2729
+ "epoch": 0.07484007503246597,
2730
+ "grad_norm": 3.026907650840076,
2731
+ "learning_rate": 1.9895700160129593e-05,
2732
+ "loss": 1.319,
2733
+ "step": 389
2734
+ },
2735
+ {
2736
+ "epoch": 0.07503246597085278,
2737
+ "grad_norm": 3.006410608707178,
2738
+ "learning_rate": 1.9894800668978095e-05,
2739
+ "loss": 1.3512,
2740
+ "step": 390
2741
+ },
2742
+ {
2743
+ "epoch": 0.07522485690923958,
2744
+ "grad_norm": 2.929111268982832,
2745
+ "learning_rate": 1.9893897336324292e-05,
2746
+ "loss": 1.2775,
2747
+ "step": 391
2748
+ },
2749
+ {
2750
+ "epoch": 0.07541724784762638,
2751
+ "grad_norm": 2.9884932715351122,
2752
+ "learning_rate": 1.9892990162518884e-05,
2753
+ "loss": 1.295,
2754
+ "step": 392
2755
+ },
2756
+ {
2757
+ "epoch": 0.07560963878601318,
2758
+ "grad_norm": 2.604435825261819,
2759
+ "learning_rate": 1.9892079147914072e-05,
2760
+ "loss": 1.2506,
2761
+ "step": 393
2762
+ },
2763
+ {
2764
+ "epoch": 0.07580202972439998,
2765
+ "grad_norm": 2.960706185493458,
2766
+ "learning_rate": 1.9891164292863537e-05,
2767
+ "loss": 1.3212,
2768
+ "step": 394
2769
+ },
2770
+ {
2771
+ "epoch": 0.07599442066278678,
2772
+ "grad_norm": 2.8743122021995764,
2773
+ "learning_rate": 1.9890245597722465e-05,
2774
+ "loss": 1.3092,
2775
+ "step": 395
2776
+ },
2777
+ {
2778
+ "epoch": 0.07618681160117359,
2779
+ "grad_norm": 3.0021254612136805,
2780
+ "learning_rate": 1.9889323062847516e-05,
2781
+ "loss": 1.3408,
2782
+ "step": 396
2783
+ },
2784
+ {
2785
+ "epoch": 0.07637920253956039,
2786
+ "grad_norm": 4.583864444196446,
2787
+ "learning_rate": 1.988839668859686e-05,
2788
+ "loss": 1.2952,
2789
+ "step": 397
2790
+ },
2791
+ {
2792
+ "epoch": 0.07657159347794719,
2793
+ "grad_norm": 3.1298586971303,
2794
+ "learning_rate": 1.988746647533014e-05,
2795
+ "loss": 1.1942,
2796
+ "step": 398
2797
+ },
2798
+ {
2799
+ "epoch": 0.07676398441633399,
2800
+ "grad_norm": 2.937592590962088,
2801
+ "learning_rate": 1.9886532423408495e-05,
2802
+ "loss": 1.3736,
2803
+ "step": 399
2804
+ },
2805
+ {
2806
+ "epoch": 0.07695637535472079,
2807
+ "grad_norm": 2.637757364994724,
2808
+ "learning_rate": 1.9885594533194564e-05,
2809
+ "loss": 1.2337,
2810
+ "step": 400
2811
+ },
2812
+ {
2813
+ "epoch": 0.0771487662931076,
2814
+ "grad_norm": 2.7327513750783075,
2815
+ "learning_rate": 1.9884652805052465e-05,
2816
+ "loss": 1.2566,
2817
+ "step": 401
2818
+ },
2819
+ {
2820
+ "epoch": 0.0773411572314944,
2821
+ "grad_norm": 2.9979638637210275,
2822
+ "learning_rate": 1.9883707239347804e-05,
2823
+ "loss": 1.256,
2824
+ "step": 402
2825
+ },
2826
+ {
2827
+ "epoch": 0.0775335481698812,
2828
+ "grad_norm": 2.812698997715836,
2829
+ "learning_rate": 1.988275783644769e-05,
2830
+ "loss": 1.2901,
2831
+ "step": 403
2832
+ },
2833
+ {
2834
+ "epoch": 0.077725939108268,
2835
+ "grad_norm": 3.011186330190687,
2836
+ "learning_rate": 1.988180459672071e-05,
2837
+ "loss": 1.2334,
2838
+ "step": 404
2839
+ },
2840
+ {
2841
+ "epoch": 0.0779183300466548,
2842
+ "grad_norm": 3.3557654319792043,
2843
+ "learning_rate": 1.988084752053695e-05,
2844
+ "loss": 1.3222,
2845
+ "step": 405
2846
+ },
2847
+ {
2848
+ "epoch": 0.0781107209850416,
2849
+ "grad_norm": 2.5175432824352537,
2850
+ "learning_rate": 1.9879886608267967e-05,
2851
+ "loss": 1.2541,
2852
+ "step": 406
2853
+ },
2854
+ {
2855
+ "epoch": 0.07830311192342841,
2856
+ "grad_norm": 3.517377754673804,
2857
+ "learning_rate": 1.9878921860286832e-05,
2858
+ "loss": 1.3745,
2859
+ "step": 407
2860
+ },
2861
+ {
2862
+ "epoch": 0.07849550286181521,
2863
+ "grad_norm": 3.4750272624774006,
2864
+ "learning_rate": 1.9877953276968088e-05,
2865
+ "loss": 1.255,
2866
+ "step": 408
2867
+ },
2868
+ {
2869
+ "epoch": 0.07868789380020201,
2870
+ "grad_norm": 2.7337894951083817,
2871
+ "learning_rate": 1.9876980858687777e-05,
2872
+ "loss": 1.2899,
2873
+ "step": 409
2874
+ },
2875
+ {
2876
+ "epoch": 0.07888028473858881,
2877
+ "grad_norm": 3.7041115181490962,
2878
+ "learning_rate": 1.9876004605823417e-05,
2879
+ "loss": 1.3643,
2880
+ "step": 410
2881
+ },
2882
+ {
2883
+ "epoch": 0.07907267567697561,
2884
+ "grad_norm": 3.5224526874841198,
2885
+ "learning_rate": 1.987502451875403e-05,
2886
+ "loss": 1.2163,
2887
+ "step": 411
2888
+ },
2889
+ {
2890
+ "epoch": 0.07926506661536242,
2891
+ "grad_norm": 2.8124864304043617,
2892
+ "learning_rate": 1.987404059786012e-05,
2893
+ "loss": 1.3362,
2894
+ "step": 412
2895
+ },
2896
+ {
2897
+ "epoch": 0.07945745755374922,
2898
+ "grad_norm": 2.9070548449705913,
2899
+ "learning_rate": 1.9873052843523676e-05,
2900
+ "loss": 1.2785,
2901
+ "step": 413
2902
+ },
2903
+ {
2904
+ "epoch": 0.07964984849213602,
2905
+ "grad_norm": 2.5926438645105203,
2906
+ "learning_rate": 1.987206125612818e-05,
2907
+ "loss": 1.3217,
2908
+ "step": 414
2909
+ },
2910
+ {
2911
+ "epoch": 0.07984223943052282,
2912
+ "grad_norm": 2.9190006615720185,
2913
+ "learning_rate": 1.98710658360586e-05,
2914
+ "loss": 1.3586,
2915
+ "step": 415
2916
+ },
2917
+ {
2918
+ "epoch": 0.08003463036890962,
2919
+ "grad_norm": 3.140039316558201,
2920
+ "learning_rate": 1.987006658370139e-05,
2921
+ "loss": 1.3243,
2922
+ "step": 416
2923
+ },
2924
+ {
2925
+ "epoch": 0.08022702130729642,
2926
+ "grad_norm": 2.922936926419004,
2927
+ "learning_rate": 1.9869063499444495e-05,
2928
+ "loss": 1.2298,
2929
+ "step": 417
2930
+ },
2931
+ {
2932
+ "epoch": 0.08041941224568323,
2933
+ "grad_norm": 2.636852883215593,
2934
+ "learning_rate": 1.9868056583677346e-05,
2935
+ "loss": 1.1578,
2936
+ "step": 418
2937
+ },
2938
+ {
2939
+ "epoch": 0.08061180318407003,
2940
+ "grad_norm": 3.529561566024578,
2941
+ "learning_rate": 1.9867045836790867e-05,
2942
+ "loss": 1.414,
2943
+ "step": 419
2944
+ },
2945
+ {
2946
+ "epoch": 0.08080419412245683,
2947
+ "grad_norm": 2.466070761821265,
2948
+ "learning_rate": 1.9866031259177463e-05,
2949
+ "loss": 1.1781,
2950
+ "step": 420
2951
+ },
2952
+ {
2953
+ "epoch": 0.08099658506084363,
2954
+ "grad_norm": 2.826345144545681,
2955
+ "learning_rate": 1.9865012851231022e-05,
2956
+ "loss": 1.3059,
2957
+ "step": 421
2958
+ },
2959
+ {
2960
+ "epoch": 0.08118897599923043,
2961
+ "grad_norm": 3.7397425890168634,
2962
+ "learning_rate": 1.9863990613346936e-05,
2963
+ "loss": 1.351,
2964
+ "step": 422
2965
+ },
2966
+ {
2967
+ "epoch": 0.08138136693761724,
2968
+ "grad_norm": 2.868551515743501,
2969
+ "learning_rate": 1.986296454592206e-05,
2970
+ "loss": 1.2352,
2971
+ "step": 423
2972
+ },
2973
+ {
2974
+ "epoch": 0.08157375787600404,
2975
+ "grad_norm": 2.7925158191667845,
2976
+ "learning_rate": 1.9861934649354763e-05,
2977
+ "loss": 1.2381,
2978
+ "step": 424
2979
+ },
2980
+ {
2981
+ "epoch": 0.08176614881439084,
2982
+ "grad_norm": 2.8659646000033856,
2983
+ "learning_rate": 1.9860900924044873e-05,
2984
+ "loss": 1.2901,
2985
+ "step": 425
2986
+ },
2987
+ {
2988
+ "epoch": 0.08195853975277764,
2989
+ "grad_norm": 3.102953042824247,
2990
+ "learning_rate": 1.9859863370393726e-05,
2991
+ "loss": 1.3534,
2992
+ "step": 426
2993
+ },
2994
+ {
2995
+ "epoch": 0.08215093069116444,
2996
+ "grad_norm": 3.3228414381051765,
2997
+ "learning_rate": 1.9858821988804132e-05,
2998
+ "loss": 1.3376,
2999
+ "step": 427
3000
+ },
3001
+ {
3002
+ "epoch": 0.08234332162955124,
3003
+ "grad_norm": 3.0656209741787603,
3004
+ "learning_rate": 1.9857776779680393e-05,
3005
+ "loss": 1.2786,
3006
+ "step": 428
3007
+ },
3008
+ {
3009
+ "epoch": 0.08253571256793805,
3010
+ "grad_norm": 3.1307213970416465,
3011
+ "learning_rate": 1.98567277434283e-05,
3012
+ "loss": 1.1746,
3013
+ "step": 429
3014
+ },
3015
+ {
3016
+ "epoch": 0.08272810350632485,
3017
+ "grad_norm": 2.834516707054261,
3018
+ "learning_rate": 1.9855674880455115e-05,
3019
+ "loss": 1.2754,
3020
+ "step": 430
3021
+ },
3022
+ {
3023
+ "epoch": 0.08292049444471165,
3024
+ "grad_norm": 2.4489319542749444,
3025
+ "learning_rate": 1.98546181911696e-05,
3026
+ "loss": 1.2888,
3027
+ "step": 431
3028
+ },
3029
+ {
3030
+ "epoch": 0.08311288538309845,
3031
+ "grad_norm": 3.2727270671338853,
3032
+ "learning_rate": 1.9853557675982e-05,
3033
+ "loss": 1.3288,
3034
+ "step": 432
3035
+ },
3036
+ {
3037
+ "epoch": 0.08330527632148525,
3038
+ "grad_norm": 2.8765634504339714,
3039
+ "learning_rate": 1.985249333530404e-05,
3040
+ "loss": 1.24,
3041
+ "step": 433
3042
+ },
3043
+ {
3044
+ "epoch": 0.08349766725987207,
3045
+ "grad_norm": 3.201525324274916,
3046
+ "learning_rate": 1.9851425169548938e-05,
3047
+ "loss": 1.309,
3048
+ "step": 434
3049
+ },
3050
+ {
3051
+ "epoch": 0.08369005819825887,
3052
+ "grad_norm": 2.7376693431213237,
3053
+ "learning_rate": 1.9850353179131392e-05,
3054
+ "loss": 1.2404,
3055
+ "step": 435
3056
+ },
3057
+ {
3058
+ "epoch": 0.08388244913664566,
3059
+ "grad_norm": 3.114660092530913,
3060
+ "learning_rate": 1.9849277364467585e-05,
3061
+ "loss": 1.2663,
3062
+ "step": 436
3063
+ },
3064
+ {
3065
+ "epoch": 0.08407484007503246,
3066
+ "grad_norm": 2.798179483922201,
3067
+ "learning_rate": 1.984819772597518e-05,
3068
+ "loss": 1.2586,
3069
+ "step": 437
3070
+ },
3071
+ {
3072
+ "epoch": 0.08426723101341926,
3073
+ "grad_norm": 2.683770568591081,
3074
+ "learning_rate": 1.9847114264073336e-05,
3075
+ "loss": 1.2395,
3076
+ "step": 438
3077
+ },
3078
+ {
3079
+ "epoch": 0.08445962195180606,
3080
+ "grad_norm": 3.0139252488559323,
3081
+ "learning_rate": 1.984602697918269e-05,
3082
+ "loss": 1.3572,
3083
+ "step": 439
3084
+ },
3085
+ {
3086
+ "epoch": 0.08465201289019288,
3087
+ "grad_norm": 2.9234340379842885,
3088
+ "learning_rate": 1.9844935871725363e-05,
3089
+ "loss": 1.1886,
3090
+ "step": 440
3091
+ },
3092
+ {
3093
+ "epoch": 0.08484440382857968,
3094
+ "grad_norm": 2.4569443949640486,
3095
+ "learning_rate": 1.9843840942124956e-05,
3096
+ "loss": 1.1814,
3097
+ "step": 441
3098
+ },
3099
+ {
3100
+ "epoch": 0.08503679476696648,
3101
+ "grad_norm": 3.038838981081329,
3102
+ "learning_rate": 1.9842742190806566e-05,
3103
+ "loss": 1.2952,
3104
+ "step": 442
3105
+ },
3106
+ {
3107
+ "epoch": 0.08522918570535327,
3108
+ "grad_norm": 2.5890067632966134,
3109
+ "learning_rate": 1.984163961819676e-05,
3110
+ "loss": 1.2221,
3111
+ "step": 443
3112
+ },
3113
+ {
3114
+ "epoch": 0.08542157664374007,
3115
+ "grad_norm": 3.104433682300787,
3116
+ "learning_rate": 1.9840533224723595e-05,
3117
+ "loss": 1.2581,
3118
+ "step": 444
3119
+ },
3120
+ {
3121
+ "epoch": 0.08561396758212689,
3122
+ "grad_norm": 3.301079931102193,
3123
+ "learning_rate": 1.9839423010816616e-05,
3124
+ "loss": 1.4437,
3125
+ "step": 445
3126
+ },
3127
+ {
3128
+ "epoch": 0.08580635852051369,
3129
+ "grad_norm": 3.268076741858138,
3130
+ "learning_rate": 1.983830897690684e-05,
3131
+ "loss": 1.2524,
3132
+ "step": 446
3133
+ },
3134
+ {
3135
+ "epoch": 0.08599874945890049,
3136
+ "grad_norm": 2.9376826789582746,
3137
+ "learning_rate": 1.9837191123426777e-05,
3138
+ "loss": 1.2342,
3139
+ "step": 447
3140
+ },
3141
+ {
3142
+ "epoch": 0.08619114039728729,
3143
+ "grad_norm": 3.1592501316361092,
3144
+ "learning_rate": 1.983606945081042e-05,
3145
+ "loss": 1.292,
3146
+ "step": 448
3147
+ },
3148
+ {
3149
+ "epoch": 0.08638353133567409,
3150
+ "grad_norm": 3.4278246110105832,
3151
+ "learning_rate": 1.983494395949323e-05,
3152
+ "loss": 1.3506,
3153
+ "step": 449
3154
+ },
3155
+ {
3156
+ "epoch": 0.0865759222740609,
3157
+ "grad_norm": 3.301439056033598,
3158
+ "learning_rate": 1.983381464991217e-05,
3159
+ "loss": 1.268,
3160
+ "step": 450
3161
+ },
3162
+ {
3163
+ "epoch": 0.0867683132124477,
3164
+ "grad_norm": 3.134215277126008,
3165
+ "learning_rate": 1.9832681522505676e-05,
3166
+ "loss": 1.2718,
3167
+ "step": 451
3168
+ },
3169
+ {
3170
+ "epoch": 0.0869607041508345,
3171
+ "grad_norm": 2.7282195932308513,
3172
+ "learning_rate": 1.9831544577713663e-05,
3173
+ "loss": 1.232,
3174
+ "step": 452
3175
+ },
3176
+ {
3177
+ "epoch": 0.0871530950892213,
3178
+ "grad_norm": 3.566655208490068,
3179
+ "learning_rate": 1.983040381597754e-05,
3180
+ "loss": 1.2901,
3181
+ "step": 453
3182
+ },
3183
+ {
3184
+ "epoch": 0.0873454860276081,
3185
+ "grad_norm": 4.0827938512517505,
3186
+ "learning_rate": 1.982925923774018e-05,
3187
+ "loss": 1.3294,
3188
+ "step": 454
3189
+ },
3190
+ {
3191
+ "epoch": 0.0875378769659949,
3192
+ "grad_norm": 3.1833143467958136,
3193
+ "learning_rate": 1.9828110843445954e-05,
3194
+ "loss": 1.3849,
3195
+ "step": 455
3196
+ },
3197
+ {
3198
+ "epoch": 0.08773026790438171,
3199
+ "grad_norm": 2.886865781675371,
3200
+ "learning_rate": 1.982695863354071e-05,
3201
+ "loss": 1.3206,
3202
+ "step": 456
3203
+ },
3204
+ {
3205
+ "epoch": 0.08792265884276851,
3206
+ "grad_norm": 2.8882129478975482,
3207
+ "learning_rate": 1.9825802608471767e-05,
3208
+ "loss": 1.257,
3209
+ "step": 457
3210
+ },
3211
+ {
3212
+ "epoch": 0.08811504978115531,
3213
+ "grad_norm": 3.579409350474279,
3214
+ "learning_rate": 1.982464276868794e-05,
3215
+ "loss": 1.3016,
3216
+ "step": 458
3217
+ },
3218
+ {
3219
+ "epoch": 0.08830744071954211,
3220
+ "grad_norm": 2.5183420055818986,
3221
+ "learning_rate": 1.982347911463952e-05,
3222
+ "loss": 1.2641,
3223
+ "step": 459
3224
+ },
3225
+ {
3226
+ "epoch": 0.0884998316579289,
3227
+ "grad_norm": 2.9662335360087653,
3228
+ "learning_rate": 1.9822311646778277e-05,
3229
+ "loss": 1.3198,
3230
+ "step": 460
3231
+ },
3232
+ {
3233
+ "epoch": 0.08869222259631572,
3234
+ "grad_norm": 3.0551988321524495,
3235
+ "learning_rate": 1.982114036555746e-05,
3236
+ "loss": 1.3654,
3237
+ "step": 461
3238
+ },
3239
+ {
3240
+ "epoch": 0.08888461353470252,
3241
+ "grad_norm": 3.160519375747151,
3242
+ "learning_rate": 1.9819965271431797e-05,
3243
+ "loss": 1.2995,
3244
+ "step": 462
3245
+ },
3246
+ {
3247
+ "epoch": 0.08907700447308932,
3248
+ "grad_norm": 2.2728872121244885,
3249
+ "learning_rate": 1.9818786364857506e-05,
3250
+ "loss": 1.2769,
3251
+ "step": 463
3252
+ },
3253
+ {
3254
+ "epoch": 0.08926939541147612,
3255
+ "grad_norm": 2.9602257487305756,
3256
+ "learning_rate": 1.9817603646292278e-05,
3257
+ "loss": 1.2821,
3258
+ "step": 464
3259
+ },
3260
+ {
3261
+ "epoch": 0.08946178634986292,
3262
+ "grad_norm": 3.061196153944173,
3263
+ "learning_rate": 1.9816417116195287e-05,
3264
+ "loss": 1.2299,
3265
+ "step": 465
3266
+ },
3267
+ {
3268
+ "epoch": 0.08965417728824972,
3269
+ "grad_norm": 3.146094048921812,
3270
+ "learning_rate": 1.9815226775027182e-05,
3271
+ "loss": 1.3158,
3272
+ "step": 466
3273
+ },
3274
+ {
3275
+ "epoch": 0.08984656822663653,
3276
+ "grad_norm": 2.9911933319609156,
3277
+ "learning_rate": 1.9814032623250093e-05,
3278
+ "loss": 1.3071,
3279
+ "step": 467
3280
+ },
3281
+ {
3282
+ "epoch": 0.09003895916502333,
3283
+ "grad_norm": 2.7568590911994204,
3284
+ "learning_rate": 1.9812834661327632e-05,
3285
+ "loss": 1.2081,
3286
+ "step": 468
3287
+ },
3288
+ {
3289
+ "epoch": 0.09023135010341013,
3290
+ "grad_norm": 3.11968597455611,
3291
+ "learning_rate": 1.9811632889724888e-05,
3292
+ "loss": 1.4081,
3293
+ "step": 469
3294
+ },
3295
+ {
3296
+ "epoch": 0.09042374104179693,
3297
+ "grad_norm": 3.1208788661599334,
3298
+ "learning_rate": 1.9810427308908437e-05,
3299
+ "loss": 1.355,
3300
+ "step": 470
3301
+ },
3302
+ {
3303
+ "epoch": 0.09061613198018373,
3304
+ "grad_norm": 2.8947601049375504,
3305
+ "learning_rate": 1.9809217919346318e-05,
3306
+ "loss": 1.3009,
3307
+ "step": 471
3308
+ },
3309
+ {
3310
+ "epoch": 0.09080852291857054,
3311
+ "grad_norm": 3.0459909725817083,
3312
+ "learning_rate": 1.980800472150806e-05,
3313
+ "loss": 1.2803,
3314
+ "step": 472
3315
+ },
3316
+ {
3317
+ "epoch": 0.09100091385695734,
3318
+ "grad_norm": 3.1606351980966467,
3319
+ "learning_rate": 1.9806787715864674e-05,
3320
+ "loss": 1.3412,
3321
+ "step": 473
3322
+ },
3323
+ {
3324
+ "epoch": 0.09119330479534414,
3325
+ "grad_norm": 3.038519302465389,
3326
+ "learning_rate": 1.9805566902888637e-05,
3327
+ "loss": 1.3751,
3328
+ "step": 474
3329
+ },
3330
+ {
3331
+ "epoch": 0.09138569573373094,
3332
+ "grad_norm": 3.0475191423269847,
3333
+ "learning_rate": 1.9804342283053916e-05,
3334
+ "loss": 1.4062,
3335
+ "step": 475
3336
+ },
3337
+ {
3338
+ "epoch": 0.09157808667211774,
3339
+ "grad_norm": 3.2693609734978644,
3340
+ "learning_rate": 1.980311385683594e-05,
3341
+ "loss": 1.318,
3342
+ "step": 476
3343
+ },
3344
+ {
3345
+ "epoch": 0.09177047761050454,
3346
+ "grad_norm": 2.9296783952437533,
3347
+ "learning_rate": 1.980188162471164e-05,
3348
+ "loss": 1.2575,
3349
+ "step": 477
3350
+ },
3351
+ {
3352
+ "epoch": 0.09196286854889135,
3353
+ "grad_norm": 3.243830683463007,
3354
+ "learning_rate": 1.98006455871594e-05,
3355
+ "loss": 1.3257,
3356
+ "step": 478
3357
+ },
3358
+ {
3359
+ "epoch": 0.09215525948727815,
3360
+ "grad_norm": 3.085502780164087,
3361
+ "learning_rate": 1.97994057446591e-05,
3362
+ "loss": 1.3022,
3363
+ "step": 479
3364
+ },
3365
+ {
3366
+ "epoch": 0.09234765042566495,
3367
+ "grad_norm": 3.2009889511854044,
3368
+ "learning_rate": 1.979816209769209e-05,
3369
+ "loss": 1.2485,
3370
+ "step": 480
3371
+ },
3372
+ {
3373
+ "epoch": 0.09254004136405175,
3374
+ "grad_norm": 2.745353541498989,
3375
+ "learning_rate": 1.9796914646741187e-05,
3376
+ "loss": 1.1989,
3377
+ "step": 481
3378
+ },
3379
+ {
3380
+ "epoch": 0.09273243230243855,
3381
+ "grad_norm": 3.4276367013865405,
3382
+ "learning_rate": 1.9795663392290702e-05,
3383
+ "loss": 1.3483,
3384
+ "step": 482
3385
+ },
3386
+ {
3387
+ "epoch": 0.09292482324082536,
3388
+ "grad_norm": 3.2887970063358822,
3389
+ "learning_rate": 1.9794408334826415e-05,
3390
+ "loss": 1.2986,
3391
+ "step": 483
3392
+ },
3393
+ {
3394
+ "epoch": 0.09311721417921216,
3395
+ "grad_norm": 3.4282845145748584,
3396
+ "learning_rate": 1.979314947483558e-05,
3397
+ "loss": 1.2592,
3398
+ "step": 484
3399
+ },
3400
+ {
3401
+ "epoch": 0.09330960511759896,
3402
+ "grad_norm": 2.6452002105384826,
3403
+ "learning_rate": 1.9791886812806932e-05,
3404
+ "loss": 1.3495,
3405
+ "step": 485
3406
+ },
3407
+ {
3408
+ "epoch": 0.09350199605598576,
3409
+ "grad_norm": 2.9331858250183713,
3410
+ "learning_rate": 1.9790620349230676e-05,
3411
+ "loss": 1.2389,
3412
+ "step": 486
3413
+ },
3414
+ {
3415
+ "epoch": 0.09369438699437256,
3416
+ "grad_norm": 2.8760604611434255,
3417
+ "learning_rate": 1.9789350084598504e-05,
3418
+ "loss": 1.2327,
3419
+ "step": 487
3420
+ },
3421
+ {
3422
+ "epoch": 0.09388677793275936,
3423
+ "grad_norm": 2.4473574857627463,
3424
+ "learning_rate": 1.9788076019403565e-05,
3425
+ "loss": 1.3268,
3426
+ "step": 488
3427
+ },
3428
+ {
3429
+ "epoch": 0.09407916887114617,
3430
+ "grad_norm": 2.721934764874413,
3431
+ "learning_rate": 1.9786798154140507e-05,
3432
+ "loss": 1.2914,
3433
+ "step": 489
3434
+ },
3435
+ {
3436
+ "epoch": 0.09427155980953297,
3437
+ "grad_norm": 2.8933447595243846,
3438
+ "learning_rate": 1.9785516489305437e-05,
3439
+ "loss": 1.2922,
3440
+ "step": 490
3441
+ },
3442
+ {
3443
+ "epoch": 0.09446395074791977,
3444
+ "grad_norm": 2.6333870713891843,
3445
+ "learning_rate": 1.9784231025395936e-05,
3446
+ "loss": 1.3221,
3447
+ "step": 491
3448
+ },
3449
+ {
3450
+ "epoch": 0.09465634168630657,
3451
+ "grad_norm": 3.104443746796275,
3452
+ "learning_rate": 1.9782941762911075e-05,
3453
+ "loss": 1.2535,
3454
+ "step": 492
3455
+ },
3456
+ {
3457
+ "epoch": 0.09484873262469337,
3458
+ "grad_norm": 2.75686140395938,
3459
+ "learning_rate": 1.9781648702351383e-05,
3460
+ "loss": 1.163,
3461
+ "step": 493
3462
+ },
3463
+ {
3464
+ "epoch": 0.09504112356308018,
3465
+ "grad_norm": 3.027361038299022,
3466
+ "learning_rate": 1.9780351844218874e-05,
3467
+ "loss": 1.277,
3468
+ "step": 494
3469
+ },
3470
+ {
3471
+ "epoch": 0.09523351450146698,
3472
+ "grad_norm": 2.4759740186065295,
3473
+ "learning_rate": 1.977905118901703e-05,
3474
+ "loss": 1.2629,
3475
+ "step": 495
3476
+ },
3477
+ {
3478
+ "epoch": 0.09542590543985378,
3479
+ "grad_norm": 3.0864905231594135,
3480
+ "learning_rate": 1.977774673725081e-05,
3481
+ "loss": 1.3345,
3482
+ "step": 496
3483
+ },
3484
+ {
3485
+ "epoch": 0.09561829637824058,
3486
+ "grad_norm": 3.013401135514556,
3487
+ "learning_rate": 1.977643848942665e-05,
3488
+ "loss": 1.3752,
3489
+ "step": 497
3490
+ },
3491
+ {
3492
+ "epoch": 0.09581068731662738,
3493
+ "grad_norm": 2.4688355634557957,
3494
+ "learning_rate": 1.977512644605246e-05,
3495
+ "loss": 1.2978,
3496
+ "step": 498
3497
+ },
3498
+ {
3499
+ "epoch": 0.09600307825501418,
3500
+ "grad_norm": 2.6448816581709944,
3501
+ "learning_rate": 1.9773810607637612e-05,
3502
+ "loss": 1.2207,
3503
+ "step": 499
3504
+ },
3505
+ {
3506
+ "epoch": 0.096195469193401,
3507
+ "grad_norm": 3.422607714860232,
3508
+ "learning_rate": 1.9772490974692962e-05,
3509
+ "loss": 1.3535,
3510
+ "step": 500
3511
+ }
3512
+ ],
3513
+ "logging_steps": 1.0,
3514
+ "max_steps": 5198,
3515
+ "num_input_tokens_seen": 0,
3516
+ "num_train_epochs": 1,
3517
+ "save_steps": 500,
3518
+ "stateful_callbacks": {
3519
+ "TrainerControl": {
3520
+ "args": {
3521
+ "should_epoch_stop": false,
3522
+ "should_evaluate": false,
3523
+ "should_log": false,
3524
+ "should_save": true,
3525
+ "should_training_stop": false
3526
+ },
3527
+ "attributes": {}
3528
+ }
3529
+ },
3530
+ "total_flos": 141059953721344.0,
3531
+ "train_batch_size": 4,
3532
+ "trial_name": null,
3533
+ "trial_params": null
3534
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/checkpoint-500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc8bfffb4e2330aa971fbff9d356fe4402740b05604db488bc5593bd343b649f
3
+ size 8312
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-finetune/log.txt ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ 2025-09-03 17:00:14,005 | INFO: Total Parameters: 1492736, Total Trainable Parameters: 1492736
2
+ 2025-09-03 17:00:14,005 | INFO: Trainable Parameters:
3
+ 2025-09-03 17:00:14,016 | WARNING: Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
4
+ 2025-09-03 17:00:21,741 | WARNING: Gradient accumulation steps mismatch: GradientAccumulationPlugin has 1, DeepSpeed config has 4. Using DeepSpeed's value.
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/added_tokens.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</tool_call>": 151658,
3
+ "<tool_call>": 151657,
4
+ "<|box_end|>": 151649,
5
+ "<|box_start|>": 151648,
6
+ "<|endoftext|>": 151643,
7
+ "<|file_sep|>": 151664,
8
+ "<|fim_middle|>": 151660,
9
+ "<|fim_pad|>": 151662,
10
+ "<|fim_prefix|>": 151659,
11
+ "<|fim_suffix|>": 151661,
12
+ "<|im_end|>": 151645,
13
+ "<|im_start|>": 151644,
14
+ "<|image_pad|>": 151655,
15
+ "<|object_ref_end|>": 151647,
16
+ "<|object_ref_start|>": 151646,
17
+ "<|quad_end|>": 151651,
18
+ "<|quad_start|>": 151650,
19
+ "<|repo_name|>": 151663,
20
+ "<|video_pad|>": 151656,
21
+ "<|vision_end|>": 151653,
22
+ "<|vision_pad|>": 151654,
23
+ "<|vision_start|>": 151652
24
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/chat_template.jinja ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0]['role'] == 'system' %}
4
+ {{- messages[0]['content'] }}
5
+ {%- else %}
6
+ {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
7
+ {%- endif %}
8
+ {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
9
+ {%- for tool in tools %}
10
+ {{- "\n" }}
11
+ {{- tool | tojson }}
12
+ {%- endfor %}
13
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
14
+ {%- else %}
15
+ {%- if messages[0]['role'] == 'system' %}
16
+ {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
17
+ {%- else %}
18
+ {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
19
+ {%- endif %}
20
+ {%- endif %}
21
+ {%- for message in messages %}
22
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
23
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
24
+ {%- elif message.role == "assistant" %}
25
+ {{- '<|im_start|>' + message.role }}
26
+ {%- if message.content %}
27
+ {{- '\n' + message.content }}
28
+ {%- endif %}
29
+ {%- for tool_call in message.tool_calls %}
30
+ {%- if tool_call.function is defined %}
31
+ {%- set tool_call = tool_call.function %}
32
+ {%- endif %}
33
+ {{- '\n<tool_call>\n{"name": "' }}
34
+ {{- tool_call.name }}
35
+ {{- '", "arguments": ' }}
36
+ {{- tool_call.arguments | tojson }}
37
+ {{- '}\n</tool_call>' }}
38
+ {%- endfor %}
39
+ {{- '<|im_end|>\n' }}
40
+ {%- elif message.role == "tool" %}
41
+ {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
42
+ {{- '<|im_start|>user' }}
43
+ {%- endif %}
44
+ {{- '\n<tool_response>\n' }}
45
+ {{- message.content }}
46
+ {{- '\n</tool_response>' }}
47
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
48
+ {{- '<|im_end|>\n' }}
49
+ {%- endif %}
50
+ {%- endif %}
51
+ {%- endfor %}
52
+ {%- if add_generation_prompt %}
53
+ {{- '<|im_start|>assistant\n' }}
54
+ {%- endif %}
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/config.json ADDED
@@ -0,0 +1,104 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cache_dir": null,
3
+ "connector_type": "mlp2x_gelu",
4
+ "hidden_size": 896,
5
+ "ignore_index": -100,
6
+ "image_aspect_ratio": "square",
7
+ "image_token_index": -200,
8
+ "llm_model_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
9
+ "model_type": "tinyllava",
10
+ "num_queries": 128,
11
+ "num_resampler_layers": 3,
12
+ "pad_token": "<|endoftext|>",
13
+ "pad_token_id": 151643,
14
+ "resampler_hidden_size": 768,
15
+ "text_config": {
16
+ "_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
17
+ "architectures": [
18
+ "Qwen2ForCausalLM"
19
+ ],
20
+ "attention_dropout": 0.0,
21
+ "bos_token_id": 151643,
22
+ "eos_token_id": 151645,
23
+ "hidden_act": "silu",
24
+ "hidden_size": 896,
25
+ "initializer_range": 0.02,
26
+ "intermediate_size": 4864,
27
+ "layer_types": [
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention"
52
+ ],
53
+ "max_position_embeddings": 32768,
54
+ "max_window_layers": 21,
55
+ "model_type": "qwen2",
56
+ "num_attention_heads": 14,
57
+ "num_hidden_layers": 24,
58
+ "num_key_value_heads": 2,
59
+ "rms_norm_eps": 1e-06,
60
+ "rope_scaling": null,
61
+ "rope_theta": 1000000.0,
62
+ "sliding_window": null,
63
+ "tie_word_embeddings": true,
64
+ "torch_dtype": "bfloat16",
65
+ "use_cache": true,
66
+ "use_sliding_window": false,
67
+ "vocab_size": 151936
68
+ },
69
+ "tokenizer_model_max_length": 2048,
70
+ "tokenizer_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
71
+ "tokenizer_padding_side": "right",
72
+ "tokenizer_use_fast": true,
73
+ "transformers_version": "4.55.4",
74
+ "tune_type_connector": "full",
75
+ "tune_type_llm": "frozen",
76
+ "tune_type_vision_tower": "frozen",
77
+ "tune_vision_tower_from_layer": 0,
78
+ "use_cache": true,
79
+ "vision_config": {
80
+ "attention_dropout": 0.0,
81
+ "dropout": 0.0,
82
+ "hidden_act": "quick_gelu",
83
+ "hidden_size": 768,
84
+ "image_size": 224,
85
+ "initializer_factor": 1.0,
86
+ "initializer_range": 0.02,
87
+ "intermediate_size": 3072,
88
+ "layer_norm_eps": 1e-05,
89
+ "model_name_or_path": "openai/clip-vit-base-patch32",
90
+ "model_name_or_path2": "",
91
+ "model_type": "clip_vision_model",
92
+ "num_attention_heads": 12,
93
+ "num_channels": 3,
94
+ "num_hidden_layers": 12,
95
+ "patch_size": 32,
96
+ "projection_dim": 512
97
+ },
98
+ "vision_feature_layer": -2,
99
+ "vision_feature_select_strategy": "patch",
100
+ "vision_hidden_size": 768,
101
+ "vision_model_name_or_path": "openai/clip-vit-base-patch32",
102
+ "vision_model_name_or_path2": "",
103
+ "vocab_size": 151936
104
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/connector/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07a415b3dacaee1e25e71d3a5855cc924619599ba2303cc1323c8d5c90b87c85
3
+ size 2987520
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/language_model/config.json ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2ForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 151643,
7
+ "eos_token_id": 151645,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 896,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 4864,
12
+ "layer_types": [
13
+ "full_attention",
14
+ "full_attention",
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention"
37
+ ],
38
+ "max_position_embeddings": 32768,
39
+ "max_window_layers": 21,
40
+ "model_type": "qwen2",
41
+ "num_attention_heads": 14,
42
+ "num_hidden_layers": 24,
43
+ "num_key_value_heads": 2,
44
+ "rms_norm_eps": 1e-06,
45
+ "rope_scaling": null,
46
+ "rope_theta": 1000000.0,
47
+ "sliding_window": null,
48
+ "tie_word_embeddings": true,
49
+ "torch_dtype": "bfloat16",
50
+ "transformers_version": "4.55.4",
51
+ "use_cache": true,
52
+ "use_sliding_window": false,
53
+ "vocab_size": 151936
54
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/language_model/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40023e628d7300854aaba56b565c923b3742338696e31bb88cb794e50d8e936a
3
+ size 988158866
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/log.txt ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ 2025-09-03 16:40:50,559 | INFO: Total Parameters: 582981504, Total Trainable Parameters: 1492736
2
+ 2025-09-03 16:40:50,560 | INFO: Trainable Parameters:
3
+ 2025-09-03 16:40:50,568 | WARNING: Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
4
+ 2025-09-03 16:40:53,941 | WARNING: Gradient accumulation steps mismatch: GradientAccumulationPlugin has 1, DeepSpeed config has 4. Using DeepSpeed's value.
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/special_tokens_map.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ },
31
+ "unk_token": "<|endoftext|>"
32
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
3
+ size 11421896
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/tokenizer_config.json ADDED
@@ -0,0 +1,208 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ }
181
+ },
182
+ "additional_special_tokens": [
183
+ "<|im_start|>",
184
+ "<|im_end|>",
185
+ "<|object_ref_start|>",
186
+ "<|object_ref_end|>",
187
+ "<|box_start|>",
188
+ "<|box_end|>",
189
+ "<|quad_start|>",
190
+ "<|quad_end|>",
191
+ "<|vision_start|>",
192
+ "<|vision_end|>",
193
+ "<|vision_pad|>",
194
+ "<|image_pad|>",
195
+ "<|video_pad|>"
196
+ ],
197
+ "bos_token": null,
198
+ "clean_up_tokenization_spaces": false,
199
+ "eos_token": "<|im_end|>",
200
+ "errors": "replace",
201
+ "extra_special_tokens": {},
202
+ "model_max_length": 2048,
203
+ "pad_token": "<|endoftext|>",
204
+ "padding_side": "right",
205
+ "split_special_tokens": false,
206
+ "tokenizer_class": "Qwen2Tokenizer",
207
+ "unk_token": "<|endoftext|>"
208
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/vision_tower/config.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attention_dropout": 0.0,
3
+ "dropout": 0.0,
4
+ "hidden_act": "quick_gelu",
5
+ "hidden_size": 768,
6
+ "image_size": 224,
7
+ "initializer_factor": 1.0,
8
+ "initializer_range": 0.02,
9
+ "intermediate_size": 3072,
10
+ "layer_norm_eps": 1e-05,
11
+ "model_type": "clip_vision_model",
12
+ "num_attention_heads": 12,
13
+ "num_channels": 3,
14
+ "num_hidden_layers": 12,
15
+ "patch_size": 32,
16
+ "projection_dim": 512,
17
+ "torch_dtype": "float32",
18
+ "transformers_version": "4.55.4"
19
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/vision_tower/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca8da55626b1674ea28d8c681d4a5d50bf7a65bd0d2f047470b0f3a9325e92ef
3
+ size 174977958
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0.5b_instruct_patch-pretrain/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "TinyLlavaForConditionalGeneration"
4
+ ],
5
+ "cache_dir": null,
6
+ "connector_type": "mlp2x_gelu",
7
+ "hidden_size": 896,
8
+ "ignore_index": -100,
9
+ "image_aspect_ratio": "square",
10
+ "image_token_index": -200,
11
+ "llm_model_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
12
+ "model_type": "tinyllava",
13
+ "num_queries": 128,
14
+ "num_resampler_layers": 3,
15
+ "pad_token": "<|endoftext|>",
16
+ "pad_token_id": 151643,
17
+ "resampler_hidden_size": 768,
18
+ "text_config": {
19
+ "_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
20
+ "architectures": [
21
+ "Qwen2ForCausalLM"
22
+ ],
23
+ "attention_dropout": 0.0,
24
+ "bos_token_id": 151643,
25
+ "eos_token_id": 151645,
26
+ "hidden_act": "silu",
27
+ "hidden_size": 896,
28
+ "initializer_range": 0.02,
29
+ "intermediate_size": 4864,
30
+ "layer_types": [
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention",
53
+ "full_attention",
54
+ "full_attention"
55
+ ],
56
+ "max_position_embeddings": 32768,
57
+ "max_window_layers": 21,
58
+ "model_type": "qwen2",
59
+ "num_attention_heads": 14,
60
+ "num_hidden_layers": 24,
61
+ "num_key_value_heads": 2,
62
+ "rms_norm_eps": 1e-06,
63
+ "rope_scaling": null,
64
+ "rope_theta": 1000000.0,
65
+ "sliding_window": null,
66
+ "tie_word_embeddings": true,
67
+ "torch_dtype": "bfloat16",
68
+ "use_cache": true,
69
+ "use_sliding_window": false,
70
+ "vocab_size": 151936
71
+ },
72
+ "tokenizer_model_max_length": 2048,
73
+ "tokenizer_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
74
+ "tokenizer_padding_side": "right",
75
+ "tokenizer_use_fast": true,
76
+ "torch_dtype": "bfloat16",
77
+ "transformers_version": "4.55.4",
78
+ "tune_type_connector": "full",
79
+ "tune_type_llm": "full",
80
+ "tune_type_vision_tower": "frozen",
81
+ "tune_vision_tower_from_layer": 0,
82
+ "use_cache": false,
83
+ "vision_config": {
84
+ "attention_dropout": 0.0,
85
+ "dropout": 0.0,
86
+ "hidden_act": "quick_gelu",
87
+ "hidden_size": 768,
88
+ "image_size": 224,
89
+ "initializer_factor": 1.0,
90
+ "initializer_range": 0.02,
91
+ "intermediate_size": 3072,
92
+ "layer_norm_eps": 1e-05,
93
+ "model_name_or_path": "openai/clip-vit-base-patch32",
94
+ "model_name_or_path2": "",
95
+ "model_type": "clip_vision_model",
96
+ "num_attention_heads": 12,
97
+ "num_channels": 3,
98
+ "num_hidden_layers": 12,
99
+ "patch_size": 32,
100
+ "projection_dim": 512
101
+ },
102
+ "vision_feature_layer": -1,
103
+ "vision_feature_select_strategy": "cls",
104
+ "vision_hidden_size": 768,
105
+ "vision_model_name_or_path": "openai/clip-vit-base-patch32",
106
+ "vision_model_name_or_path2": "",
107
+ "vocab_size": 151936
108
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 151643,
4
+ "eos_token_id": 151645,
5
+ "transformers_version": "4.55.4",
6
+ "use_cache": false
7
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a36a9156e339c0f03716dabadffb92845782b8dd8d86f2a2f3ebc5a57e51319
3
+ size 1486581616
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3159fb748798cea6ebf3e0c49dcd8007ed9c050d7b8a00bfeb4a70b76a340fad
3
+ size 1486581616
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b34669a2658658f95184473aa57acbb57acf6a9f8fa30100f68e9e26127850d
3
+ size 1486581616
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:adb87ae9d2a68b0a83719a5430c13fe12ede9c8bf8f418835449260a9b06ebb1
3
+ size 1486581616
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b11a459401f0f153376d190a33b2dcdf77f5d4b632cd4c2b25e7df8cecbb90f
3
+ size 44061074
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38eed616b6001289f21382bbd2480638505bc732c6d34fc7855ecbfce450018c
3
+ size 44061010
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8928595ccee05a35b678c0c5d97eeee9d72486d8deaab9714c4e91b78508a70b
3
+ size 44061010
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f6e7c223bf123105893c2845e89176f69b0be8223bf6882d0b55f271629b8a3
3
+ size 44061010
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step1000
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84a9a6c6f66aaaca10e5e5ccb822c5f29615b24ff422ab24c689f2bbf7a010d6
3
+ size 1166029904
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/preprocessor_config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 224,
4
+ "width": 224
5
+ },
6
+ "do_center_crop": true,
7
+ "do_convert_rgb": true,
8
+ "do_normalize": true,
9
+ "do_rescale": true,
10
+ "do_resize": true,
11
+ "image_mean": [
12
+ 0.48145466,
13
+ 0.4578275,
14
+ 0.40821073
15
+ ],
16
+ "image_processor_type": "CLIPImageProcessor",
17
+ "image_std": [
18
+ 0.26862954,
19
+ 0.26130258,
20
+ 0.27577711
21
+ ],
22
+ "resample": 3,
23
+ "rescale_factor": 0.00392156862745098,
24
+ "size": {
25
+ "shortest_edge": 224
26
+ }
27
+ }
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:309699dce691459deaa920cf597347dea691f378ebfcd89fde4a27a1aa85bbbf
3
+ size 14960
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6683b9287e8233906f33968436bcef76fae1ed2ec06b8c096c89d5bf4b6eaf80
3
+ size 14960
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8b9e60cfc3b60697ed1837b751cb132107758b94099e04c334078a34b6a3f1a
3
+ size 14960
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23a02e61fee597cb0e3953cee15e98e4d189da8431d8c75b0a77dc9a6385e5ff
3
+ size 14960
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea9b4ee6de7612218d7ece8a54518c6cff9364a1e28116b5b19d0aac14661eaa
3
+ size 1064
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b3e04cfb3331714bc35d839f2cad603aae082954fa5f7daba7498a3ec43fcf3
3
+ size 8248
tiny-llava-Qwen2.5-0.5B-Instruct-clip-vit-base-patch32-qwen2.5-0_5b_instruct-finetune/checkpoint-1000/zero_to_fp32.py ADDED
@@ -0,0 +1,760 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example:
14
+ # python zero_to_fp32.py . output_dir/
15
+ # or
16
+ # python zero_to_fp32.py . output_dir/ --safe_serialization
17
+
18
+ import argparse
19
+ import torch
20
+ import glob
21
+ import math
22
+ import os
23
+ import re
24
+ import gc
25
+ import json
26
+ import numpy as np
27
+ from tqdm import tqdm
28
+ from collections import OrderedDict
29
+ from dataclasses import dataclass
30
+
31
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
32
+ # DeepSpeed data structures it has to be available in the current python environment.
33
+ from deepspeed.utils import logger
34
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
35
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
36
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
37
+
38
+
39
+ @dataclass
40
+ class zero_model_state:
41
+ buffers: dict()
42
+ param_shapes: dict()
43
+ shared_params: list
44
+ ds_version: int
45
+ frozen_param_shapes: dict()
46
+ frozen_param_fragments: dict()
47
+
48
+
49
+ debug = 0
50
+
51
+ # load to cpu
52
+ device = torch.device('cpu')
53
+
54
+
55
+ def atoi(text):
56
+ return int(text) if text.isdigit() else text
57
+
58
+
59
+ def natural_keys(text):
60
+ '''
61
+ alist.sort(key=natural_keys) sorts in human order
62
+ http://nedbatchelder.com/blog/200712/human_sorting.html
63
+ (See Toothy's implementation in the comments)
64
+ '''
65
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
66
+
67
+
68
+ def get_model_state_file(checkpoint_dir, zero_stage):
69
+ if not os.path.isdir(checkpoint_dir):
70
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
71
+
72
+ # there should be only one file
73
+ if zero_stage <= 2:
74
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
75
+ elif zero_stage == 3:
76
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
77
+
78
+ if not os.path.exists(file):
79
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
80
+
81
+ return file
82
+
83
+
84
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
85
+ # XXX: need to test that this simple glob rule works for multi-node setup too
86
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
87
+
88
+ if len(ckpt_files) == 0:
89
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
90
+
91
+ return ckpt_files
92
+
93
+
94
+ def get_optim_files(checkpoint_dir):
95
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
96
+
97
+
98
+ def get_model_state_files(checkpoint_dir):
99
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
100
+
101
+
102
+ def parse_model_states(files):
103
+ zero_model_states = []
104
+ for file in files:
105
+ state_dict = torch.load(file, map_location=device, weights_only=False)
106
+
107
+ if BUFFER_NAMES not in state_dict:
108
+ raise ValueError(f"{file} is not a model state checkpoint")
109
+ buffer_names = state_dict[BUFFER_NAMES]
110
+ if debug:
111
+ print("Found buffers:", buffer_names)
112
+
113
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
114
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
115
+ param_shapes = state_dict[PARAM_SHAPES]
116
+
117
+ # collect parameters that are included in param_shapes
118
+ param_names = []
119
+ for s in param_shapes:
120
+ for name in s.keys():
121
+ param_names.append(name)
122
+
123
+ # update with frozen parameters
124
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
125
+ if frozen_param_shapes is not None:
126
+ if debug:
127
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
128
+ param_names += list(frozen_param_shapes.keys())
129
+
130
+ # handle shared params
131
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
132
+
133
+ ds_version = state_dict.get(DS_VERSION, None)
134
+
135
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
136
+
137
+ z_model_state = zero_model_state(buffers=buffers,
138
+ param_shapes=param_shapes,
139
+ shared_params=shared_params,
140
+ ds_version=ds_version,
141
+ frozen_param_shapes=frozen_param_shapes,
142
+ frozen_param_fragments=frozen_param_fragments)
143
+ zero_model_states.append(z_model_state)
144
+
145
+ return zero_model_states
146
+
147
+
148
+ def parse_optim_states(files, ds_checkpoint_dir):
149
+ total_files = len(files)
150
+ state_dicts = []
151
+ for f in tqdm(files, desc='Loading checkpoint shards'):
152
+ state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
153
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
154
+ # and also handle the case where it was already removed by another helper script
155
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
156
+ state_dicts.append(state_dict)
157
+
158
+ if ZERO_STAGE not in state_dicts[0][OPTIMIZER_STATE_DICT]:
159
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
160
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
161
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
162
+
163
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
164
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
165
+ # use the max of the partition_count to get the dp world_size.
166
+
167
+ if type(world_size) is list:
168
+ world_size = max(world_size)
169
+
170
+ if world_size != total_files:
171
+ raise ValueError(
172
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
173
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
174
+ )
175
+
176
+ # the groups are named differently in each stage
177
+ if zero_stage <= 2:
178
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
179
+ elif zero_stage == 3:
180
+ fp32_groups_key = FP32_FLAT_GROUPS
181
+ else:
182
+ raise ValueError(f"unknown zero stage {zero_stage}")
183
+
184
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
185
+ return zero_stage, world_size, fp32_flat_groups
186
+
187
+
188
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
189
+ """
190
+ Returns fp32 state_dict reconstructed from ds checkpoint
191
+
192
+ Args:
193
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
194
+
195
+ """
196
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
197
+
198
+ optim_files = get_optim_files(ds_checkpoint_dir)
199
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
200
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
201
+
202
+ model_files = get_model_state_files(ds_checkpoint_dir)
203
+
204
+ zero_model_states = parse_model_states(model_files)
205
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
206
+
207
+ if zero_stage <= 2:
208
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
209
+ exclude_frozen_parameters)
210
+ elif zero_stage == 3:
211
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
212
+ exclude_frozen_parameters)
213
+
214
+
215
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
216
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
217
+ return
218
+
219
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
220
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
221
+
222
+ if debug:
223
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
224
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
225
+
226
+ wanted_params = len(frozen_param_shapes)
227
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
228
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
229
+ print(f'Frozen params: Have {avail_numel} numels to process.')
230
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
231
+
232
+ total_params = 0
233
+ total_numel = 0
234
+ for name, shape in frozen_param_shapes.items():
235
+ total_params += 1
236
+ unpartitioned_numel = shape.numel()
237
+ total_numel += unpartitioned_numel
238
+
239
+ state_dict[name] = frozen_param_fragments[name]
240
+
241
+ if debug:
242
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
243
+
244
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
245
+
246
+
247
+ def _has_callable(obj, fn):
248
+ attr = getattr(obj, fn, None)
249
+ return callable(attr)
250
+
251
+
252
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
253
+ param_shapes = zero_model_states[0].param_shapes
254
+
255
+ # Reconstruction protocol:
256
+ #
257
+ # XXX: document this
258
+
259
+ if debug:
260
+ for i in range(world_size):
261
+ for j in range(len(fp32_flat_groups[0])):
262
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
263
+
264
+ # XXX: memory usage doubles here (zero2)
265
+ num_param_groups = len(fp32_flat_groups[0])
266
+ merged_single_partition_of_fp32_groups = []
267
+ for i in range(num_param_groups):
268
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
269
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
270
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
271
+ avail_numel = sum(
272
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
273
+
274
+ if debug:
275
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
276
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
277
+ # not asserting if there is a mismatch due to possible padding
278
+ print(f"Have {avail_numel} numels to process.")
279
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
280
+
281
+ # params
282
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
283
+ # out-of-core computing solution
284
+ total_numel = 0
285
+ total_params = 0
286
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
287
+ offset = 0
288
+ avail_numel = full_single_fp32_vector.numel()
289
+ for name, shape in shapes.items():
290
+
291
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
292
+ total_numel += unpartitioned_numel
293
+ total_params += 1
294
+
295
+ if debug:
296
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
297
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
298
+ offset += unpartitioned_numel
299
+
300
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
301
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
302
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
303
+ # live optimizer object, so we are checking that the numbers are within the right range
304
+ align_to = 2 * world_size
305
+
306
+ def zero2_align(x):
307
+ return align_to * math.ceil(x / align_to)
308
+
309
+ if debug:
310
+ print(f"original offset={offset}, avail_numel={avail_numel}")
311
+
312
+ offset = zero2_align(offset)
313
+ avail_numel = zero2_align(avail_numel)
314
+
315
+ if debug:
316
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
317
+
318
+ # Sanity check
319
+ if offset != avail_numel:
320
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
321
+
322
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
323
+
324
+
325
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
326
+ exclude_frozen_parameters):
327
+ state_dict = OrderedDict()
328
+
329
+ # buffers
330
+ buffers = zero_model_states[0].buffers
331
+ state_dict.update(buffers)
332
+ if debug:
333
+ print(f"added {len(buffers)} buffers")
334
+
335
+ if not exclude_frozen_parameters:
336
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
337
+
338
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
339
+
340
+ # recover shared parameters
341
+ for pair in zero_model_states[0].shared_params:
342
+ if pair[1] in state_dict:
343
+ state_dict[pair[0]] = state_dict[pair[1]]
344
+
345
+ return state_dict
346
+
347
+
348
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
349
+ remainder = unpartitioned_numel % world_size
350
+ padding_numel = (world_size - remainder) if remainder else 0
351
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
352
+ return partitioned_numel, padding_numel
353
+
354
+
355
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
356
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
357
+ return
358
+
359
+ if debug:
360
+ for i in range(world_size):
361
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
362
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
363
+
364
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
365
+ wanted_params = len(frozen_param_shapes)
366
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
367
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
368
+ print(f'Frozen params: Have {avail_numel} numels to process.')
369
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
370
+
371
+ total_params = 0
372
+ total_numel = 0
373
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
374
+ total_params += 1
375
+ unpartitioned_numel = shape.numel()
376
+ total_numel += unpartitioned_numel
377
+
378
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
379
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
380
+
381
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
382
+
383
+ if debug:
384
+ print(
385
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
386
+ )
387
+
388
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
389
+
390
+
391
+ class GatheredTensor:
392
+ """
393
+ A pseudo tensor that collects partitioned weights.
394
+ It is more memory efficient when there are multiple groups.
395
+ """
396
+
397
+ def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
398
+ self.flat_groups = flat_groups
399
+ self.flat_groups_offset = flat_groups_offset
400
+ self.offset = offset
401
+ self.partitioned_numel = partitioned_numel
402
+ self.shape = shape
403
+ self.dtype = self.flat_groups[0][0].dtype
404
+
405
+ def contiguous(self):
406
+ """
407
+ Merge partitioned weights from flat_groups into a single tensor.
408
+ """
409
+ end_idx = self.offset + self.partitioned_numel
410
+ world_size = len(self.flat_groups)
411
+ pad_flat_param_chunks = []
412
+
413
+ for rank_i in range(world_size):
414
+ # for each rank, we need to collect weights from related group/groups
415
+ flat_groups_at_rank_i = self.flat_groups[rank_i]
416
+ start_group_id = None
417
+ end_group_id = None
418
+ for group_id in range(len(self.flat_groups_offset)):
419
+ if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
420
+ start_group_id = group_id
421
+ if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
422
+ end_group_id = group_id
423
+ break
424
+ # collect weights from related group/groups
425
+ for group_id in range(start_group_id, end_group_id + 1):
426
+ flat_tensor = flat_groups_at_rank_i[group_id]
427
+ start_offset = self.offset - self.flat_groups_offset[group_id]
428
+ end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
429
+ pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
430
+
431
+ # collect weights from all ranks
432
+ pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
433
+ param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
434
+ return param
435
+
436
+
437
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
438
+ param_shapes = zero_model_states[0].param_shapes
439
+ avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
440
+
441
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
442
+ # param, re-consolidating each param, while dealing with padding if any
443
+
444
+ # merge list of dicts, preserving order
445
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
446
+
447
+ if debug:
448
+ for i in range(world_size):
449
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
450
+
451
+ wanted_params = len(param_shapes)
452
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
453
+ # not asserting if there is a mismatch due to possible padding
454
+ avail_numel = fp32_flat_groups[0].numel() * world_size
455
+ print(f"Trainable params: Have {avail_numel} numels to process.")
456
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
457
+
458
+ # params
459
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
460
+ # out-of-core computing solution
461
+ offset = 0
462
+ total_numel = 0
463
+ total_params = 0
464
+ flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
465
+ for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
466
+ unpartitioned_numel = shape.numel()
467
+ total_numel += unpartitioned_numel
468
+ total_params += 1
469
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
470
+
471
+ if debug:
472
+ print(
473
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
474
+ )
475
+
476
+ # memory efficient tensor
477
+ tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
478
+ state_dict[name] = tensor
479
+ offset += partitioned_numel
480
+
481
+ offset *= world_size
482
+
483
+ # Sanity check
484
+ if offset != avail_numel:
485
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
486
+
487
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
488
+
489
+
490
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
491
+ exclude_frozen_parameters):
492
+ state_dict = OrderedDict()
493
+
494
+ # buffers
495
+ buffers = zero_model_states[0].buffers
496
+ state_dict.update(buffers)
497
+ if debug:
498
+ print(f"added {len(buffers)} buffers")
499
+
500
+ if not exclude_frozen_parameters:
501
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
502
+
503
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
504
+
505
+ # recover shared parameters
506
+ for pair in zero_model_states[0].shared_params:
507
+ if pair[1] in state_dict:
508
+ state_dict[pair[0]] = state_dict[pair[1]]
509
+
510
+ return state_dict
511
+
512
+
513
+ def to_torch_tensor(state_dict, return_empty_tensor=False):
514
+ """
515
+ Convert state_dict of GatheredTensor to torch tensor
516
+ """
517
+ torch_state_dict = {}
518
+ converted_tensors = {}
519
+ for name, tensor in state_dict.items():
520
+ tensor_id = id(tensor)
521
+ if tensor_id in converted_tensors: # shared tensors
522
+ shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
523
+ torch_state_dict[name] = shared_tensor
524
+ else:
525
+ converted_tensors[tensor_id] = name
526
+ if return_empty_tensor:
527
+ torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
528
+ else:
529
+ torch_state_dict[name] = tensor.contiguous()
530
+ return torch_state_dict
531
+
532
+
533
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
534
+ tag=None,
535
+ exclude_frozen_parameters=False,
536
+ lazy_mode=False):
537
+ """
538
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
539
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
540
+ via a model hub.
541
+
542
+ Args:
543
+ - ``checkpoint_dir``: path to the desired checkpoint folder
544
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
545
+ - ``exclude_frozen_parameters``: exclude frozen parameters
546
+ - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
547
+ Convert the pesduo tensor to torch tensor by ``.contiguous()``
548
+
549
+ Returns:
550
+ - pytorch ``state_dict``
551
+
552
+ A typical usage might be ::
553
+
554
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
555
+ # do the training and checkpoint saving
556
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
557
+ model = model.cpu() # move to cpu
558
+ model.load_state_dict(state_dict)
559
+ # submit to model hub or save the model to share with others
560
+
561
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
562
+ application. i.e. you will need to re-initialize the deepspeed engine, since
563
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
564
+
565
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
566
+
567
+ Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
568
+ You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
569
+ the checkpoint. Or you can load state_dict in lazy mode ::
570
+
571
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
572
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
573
+ for name, lazy_tensor in state_dict.item():
574
+ tensor = lazy_tensor.contiguous() # to cpu
575
+ print(name, tensor)
576
+ # del tensor to release memory if it no longer in use
577
+ """
578
+ if tag is None:
579
+ latest_path = os.path.join(checkpoint_dir, 'latest')
580
+ if os.path.isfile(latest_path):
581
+ with open(latest_path, 'r') as fd:
582
+ tag = fd.read().strip()
583
+ else:
584
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
585
+
586
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
587
+
588
+ if not os.path.isdir(ds_checkpoint_dir):
589
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
590
+
591
+ state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
592
+ if lazy_mode:
593
+ return state_dict
594
+ else:
595
+ return to_torch_tensor(state_dict)
596
+
597
+
598
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
599
+ output_dir,
600
+ max_shard_size="5GB",
601
+ safe_serialization=False,
602
+ tag=None,
603
+ exclude_frozen_parameters=False):
604
+ """
605
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
606
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
607
+
608
+ Args:
609
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
610
+ - ``output_dir``: directory to the pytorch fp32 state_dict output files
611
+ - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
612
+ - ``safe_serialization``: whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
613
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
614
+ - ``exclude_frozen_parameters``: exclude frozen parameters
615
+ """
616
+
617
+ # Dependency pre-check
618
+ if safe_serialization:
619
+ try:
620
+ from safetensors.torch import save_file
621
+ except ImportError:
622
+ print('If you want to use `safe_serialization`, please `pip install safetensors`')
623
+ raise
624
+ if max_shard_size is not None:
625
+ try:
626
+ from huggingface_hub import split_torch_state_dict_into_shards
627
+ except ImportError:
628
+ print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
629
+ raise
630
+
631
+ # Convert zero checkpoint to state_dict
632
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
633
+ tag,
634
+ exclude_frozen_parameters,
635
+ lazy_mode=True)
636
+
637
+ # Shard the model if it is too big.
638
+ weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
639
+ if max_shard_size is not None:
640
+ filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
641
+ # an memory-efficient approach for sharding
642
+ empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
643
+ state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
644
+ filename_pattern=filename_pattern,
645
+ max_shard_size=max_shard_size)
646
+ else:
647
+ from collections import namedtuple
648
+ StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
649
+ state_dict_split = StateDictSplit(is_sharded=False,
650
+ filename_to_tensors={weights_name: list(state_dict.keys())})
651
+
652
+ # Save the model by shard
653
+ os.makedirs(output_dir, exist_ok=True)
654
+ filename_to_tensors = state_dict_split.filename_to_tensors.items()
655
+ for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
656
+ shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
657
+ shard_state_dict = to_torch_tensor(shard_state_dict)
658
+ output_path = os.path.join(output_dir, shard_file)
659
+ if safe_serialization:
660
+ save_file(shard_state_dict, output_path, metadata={"format": "pt"})
661
+ else:
662
+ torch.save(shard_state_dict, output_path)
663
+ # release the memory of current shard
664
+ for tensor_name in list(shard_state_dict.keys()):
665
+ del state_dict[tensor_name]
666
+ del shard_state_dict[tensor_name]
667
+ del shard_state_dict
668
+ gc.collect()
669
+
670
+ # Save index if sharded
671
+ if state_dict_split.is_sharded:
672
+ index = {
673
+ "metadata": state_dict_split.metadata,
674
+ "weight_map": state_dict_split.tensor_to_filename,
675
+ }
676
+ save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
677
+ save_index_file = os.path.join(output_dir, save_index_file)
678
+ with open(save_index_file, "w", encoding="utf-8") as f:
679
+ content = json.dumps(index, indent=2, sort_keys=True) + "\n"
680
+ f.write(content)
681
+
682
+
683
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
684
+ """
685
+ 1. Put the provided model to cpu
686
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
687
+ 3. Load it into the provided model
688
+
689
+ Args:
690
+ - ``model``: the model object to update
691
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
692
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
693
+
694
+ Returns:
695
+ - ``model`: modified model
696
+
697
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
698
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
699
+ conveniently placed for you in the checkpoint folder.
700
+
701
+ A typical usage might be ::
702
+
703
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
704
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
705
+ # submit to model hub or save the model to share with others
706
+
707
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
708
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
709
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
710
+
711
+ """
712
+ logger.info("Extracting fp32 weights")
713
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
714
+
715
+ logger.info("Overwriting model with fp32 weights")
716
+ model = model.cpu()
717
+ model.load_state_dict(state_dict, strict=False)
718
+
719
+ return model
720
+
721
+
722
+ if __name__ == "__main__":
723
+ parser = argparse.ArgumentParser()
724
+ parser.add_argument("checkpoint_dir",
725
+ type=str,
726
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
727
+ parser.add_argument("output_dir",
728
+ type=str,
729
+ help="directory to the pytorch fp32 state_dict output files"
730
+ "(e.g. path/checkpoint-12-output/)")
731
+ parser.add_argument(
732
+ "--max_shard_size",
733
+ type=str,
734
+ default="5GB",
735
+ help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
736
+ "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
737
+ "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
738
+ "without CPU OOM issues.")
739
+ parser.add_argument(
740
+ "--safe_serialization",
741
+ default=False,
742
+ action='store_true',
743
+ help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
744
+ parser.add_argument("-t",
745
+ "--tag",
746
+ type=str,
747
+ default=None,
748
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
749
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
750
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
751
+ args = parser.parse_args()
752
+
753
+ debug = args.debug
754
+
755
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
756
+ args.output_dir,
757
+ max_shard_size=args.max_shard_size,
758
+ safe_serialization=args.safe_serialization,
759
+ tag=args.tag,
760
+ exclude_frozen_parameters=args.exclude_frozen_parameters)