Hugging Face's logo

ajaymin28
/
vl-1.5-tempcheckpoints

TensorBoard
llava
Model card Files Metrics Community
ajaymin28 commited on
Commit
610b750
•
1 Parent(s): 336d785

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. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/README.md +9 -0
  2. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/adapter_config.json +26 -0
  3. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/adapter_model.bin +3 -0
  4. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/config.json +41 -0
  5. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/non_lora_trainables.bin +3 -0
  6. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/runs/Sep22_22-59-22_evc44/events.out.tfevents.1727060508.evc44 +3 -0
  7. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/special_tokens_map.json +24 -0
  8. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/tokenizer.model +3 -0
  9. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/tokenizer_config.json +35 -0
  10. [lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/trainer_state.json +0 -0
  11. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/README.md +9 -0
  12. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/adapter_config.json +26 -0
  13. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/adapter_model.bin +3 -0
  14. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/config.json +41 -0
  15. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/non_lora_trainables.bin +3 -0
  16. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/runs/Sep22_23-52-18_evc48/events.out.tfevents.1727063690.evc48 +3 -0
  17. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/special_tokens_map.json +24 -0
  18. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/tokenizer.model +3 -0
  19. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/tokenizer_config.json +35 -0
  20. [lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/trainer_state.json +3415 -0
  21. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/README.md +9 -0
  22. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/adapter_config.json +26 -0
  23. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/adapter_model.bin +3 -0
  24. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/config.json +41 -0
  25. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/non_lora_trainables.bin +3 -0
  26. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/runs/Sep22_02-41-32_evc28/events.out.tfevents.1726987448.evc28 +3 -0
  27. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/special_tokens_map.json +24 -0
  28. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/tokenizer.model +3 -0
  29. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/tokenizer_config.json +35 -0
  30. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/trainer_state.json +0 -0
  31. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/README.md +10 -0
  32. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/adapter_config.json +26 -0
  33. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/adapter_model.bin +3 -0
  34. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/config.json +41 -0
  35. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/non_lora_trainables.bin +3 -0
  36. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/runs/Sep22_13-40-39_evc28/events.out.tfevents.1727027010.evc28 +0 -0
  37. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/runs/Sep22_13-46-32_evc28/events.out.tfevents.1727027315.evc28 +3 -0
  38. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/runs/Sep22_23-43-22_evc45/events.out.tfevents.1727063201.evc45 +3 -0
  39. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/special_tokens_map.json +24 -0
  40. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/tokenizer.model +3 -0
  41. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/tokenizer_config.json +35 -0
  42. [lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/trainer_state.json +3415 -0
  43. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/README.md +9 -0
  44. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/adapter_config.json +26 -0
  45. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/adapter_model.bin +3 -0
  46. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/config.json +41 -0
  47. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/non_lora_trainables.bin +3 -0
  48. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/runs/Sep24_08-23-17_evc43/events.out.tfevents.1727180744.evc43 +3 -0
  49. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/special_tokens_map.json +24 -0
  50. [lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/tokenizer.model +3 -0
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+
9
+ - PEFT 0.4.0
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "lmsys/vicuna-7b-v1.5",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 256,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 128,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "k_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "o_proj",
22
+ "down_proj",
23
+ "gate_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ba367345203ed44b33f83bc8efd7b797d570162d95fdbf8474fe5b80a46612d
3
+ size 639787082
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "freeze_mm_mlp_adapter": false,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "image_aspect_ratio": "pad",
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
14
+ "max_position_embeddings": 4096,
15
+ "mm_hidden_size": 1024,
16
+ "mm_image_tower": "LanguageBind/LanguageBind_Image",
17
+ "mm_projector_lr": 2e-05,
18
+ "mm_projector_type": "mlp2x_gelu",
19
+ "mm_use_im_patch_token": false,
20
+ "mm_use_im_start_end": false,
21
+ "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
22
+ "mm_vision_select_feature": "patch",
23
+ "mm_vision_select_layer": -2,
24
+ "model_type": "llava",
25
+ "num_attention_heads": 32,
26
+ "num_hidden_layers": 32,
27
+ "num_key_value_heads": 32,
28
+ "pad_token_id": 0,
29
+ "pretraining_tp": 1,
30
+ "rms_norm_eps": 1e-05,
31
+ "rope_scaling": null,
32
+ "tie_word_embeddings": false,
33
+ "tokenizer_model_max_length": 3072,
34
+ "tokenizer_padding_side": "right",
35
+ "torch_dtype": "float16",
36
+ "transformers_version": "4.31.0",
37
+ "tune_mm_mlp_adapter": false,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "vocab_size": 32000
41
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/non_lora_trainables.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a44c5f733e14800a2445426f649b94fa2e23e1ffdbed263f4c6364648f7e09e3
3
+ size 41961648
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/runs/Sep22_22-59-22_evc44/events.out.tfevents.1727060508.evc44 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f78fcfc227ab4b4071611e938b9195173db1ac9ee254a82f14767b8844fde458
3
+ size 271521
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/tokenizer_config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "<s>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": false,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "legacy": false,
22
+ "model_max_length": 2048,
23
+ "pad_token": null,
24
+ "padding_side": "right",
25
+ "sp_model_kwargs": {},
26
+ "tokenizer_class": "LlamaTokenizer",
27
+ "unk_token": {
28
+ "__type": "AddedToken",
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v2_1_nomax_p09_e03/videollava-7b-lora/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+
9
+ - PEFT 0.4.0
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "lmsys/vicuna-7b-v1.5",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 256,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 128,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "up_proj",
19
+ "k_proj",
20
+ "down_proj",
21
+ "v_proj",
22
+ "gate_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5153146f8495cb47c1bc282038af32dfe42cac61f24267ab56a0015914e30d10
3
+ size 639787082
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "freeze_mm_mlp_adapter": false,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "image_aspect_ratio": "pad",
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
14
+ "max_position_embeddings": 4096,
15
+ "mm_hidden_size": 1024,
16
+ "mm_image_tower": "LanguageBind/LanguageBind_Image",
17
+ "mm_projector_lr": 2e-05,
18
+ "mm_projector_type": "mlp2x_gelu",
19
+ "mm_use_im_patch_token": false,
20
+ "mm_use_im_start_end": false,
21
+ "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
22
+ "mm_vision_select_feature": "patch",
23
+ "mm_vision_select_layer": -2,
24
+ "model_type": "llava",
25
+ "num_attention_heads": 32,
26
+ "num_hidden_layers": 32,
27
+ "num_key_value_heads": 32,
28
+ "pad_token_id": 0,
29
+ "pretraining_tp": 1,
30
+ "rms_norm_eps": 1e-05,
31
+ "rope_scaling": null,
32
+ "tie_word_embeddings": false,
33
+ "tokenizer_model_max_length": 3072,
34
+ "tokenizer_padding_side": "right",
35
+ "torch_dtype": "float16",
36
+ "transformers_version": "4.31.0",
37
+ "tune_mm_mlp_adapter": false,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "vocab_size": 32000
41
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/non_lora_trainables.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69955a37c6a4e7ab5b955ecbd4edd73f91fcfc01a2db038cfcfd4927b0163898
3
+ size 41961648
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/runs/Sep22_23-52-18_evc48/events.out.tfevents.1727063690.evc48 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f75a3505a8900a7884988d4dea336314908371bf35a1eec1f7a60c6671529890
3
+ size 94093
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/tokenizer_config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "<s>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": false,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "legacy": false,
22
+ "model_max_length": 2048,
23
+ "pad_token": null,
24
+ "padding_side": "right",
25
+ "sp_model_kwargs": {},
26
+ "tokenizer_class": "LlamaTokenizer",
27
+ "unk_token": {
28
+ "__type": "AddedToken",
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v4_4_p09_e01/videollava-7b-lora/trainer_state.json ADDED
@@ -0,0 +1,3415 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "global_step": 565,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 1.1764705882352942e-05,
13
+ "loss": 0.3104,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 2.3529411764705884e-05,
19
+ "loss": 0.3195,
20
+ "step": 2
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 3.529411764705883e-05,
25
+ "loss": 0.309,
26
+ "step": 3
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 4.705882352941177e-05,
31
+ "loss": 0.2956,
32
+ "step": 4
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 5.882352941176471e-05,
37
+ "loss": 0.2491,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 7.058823529411765e-05,
43
+ "loss": 0.1945,
44
+ "step": 6
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 8.23529411764706e-05,
49
+ "loss": 0.1464,
50
+ "step": 7
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "learning_rate": 9.411764705882353e-05,
55
+ "loss": 0.1343,
56
+ "step": 8
57
+ },
58
+ {
59
+ "epoch": 0.02,
60
+ "learning_rate": 0.00010588235294117647,
61
+ "loss": 0.1051,
62
+ "step": 9
63
+ },
64
+ {
65
+ "epoch": 0.02,
66
+ "learning_rate": 0.00011764705882352942,
67
+ "loss": 0.1085,
68
+ "step": 10
69
+ },
70
+ {
71
+ "epoch": 0.02,
72
+ "learning_rate": 0.00012941176470588237,
73
+ "loss": 0.0719,
74
+ "step": 11
75
+ },
76
+ {
77
+ "epoch": 0.02,
78
+ "learning_rate": 0.0001411764705882353,
79
+ "loss": 0.0898,
80
+ "step": 12
81
+ },
82
+ {
83
+ "epoch": 0.02,
84
+ "learning_rate": 0.00015294117647058822,
85
+ "loss": 0.1091,
86
+ "step": 13
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "learning_rate": 0.0001647058823529412,
91
+ "loss": 0.0852,
92
+ "step": 14
93
+ },
94
+ {
95
+ "epoch": 0.03,
96
+ "learning_rate": 0.00017647058823529413,
97
+ "loss": 0.0603,
98
+ "step": 15
99
+ },
100
+ {
101
+ "epoch": 0.03,
102
+ "learning_rate": 0.00018823529411764707,
103
+ "loss": 0.0663,
104
+ "step": 16
105
+ },
106
+ {
107
+ "epoch": 0.03,
108
+ "learning_rate": 0.0002,
109
+ "loss": 0.0658,
110
+ "step": 17
111
+ },
112
+ {
113
+ "epoch": 0.03,
114
+ "learning_rate": 0.00019999835673561284,
115
+ "loss": 0.0571,
116
+ "step": 18
117
+ },
118
+ {
119
+ "epoch": 0.03,
120
+ "learning_rate": 0.00019999342699645773,
121
+ "loss": 0.0444,
122
+ "step": 19
123
+ },
124
+ {
125
+ "epoch": 0.04,
126
+ "learning_rate": 0.00019998521094455196,
127
+ "loss": 0.0491,
128
+ "step": 20
129
+ },
130
+ {
131
+ "epoch": 0.04,
132
+ "learning_rate": 0.00019997370884991842,
133
+ "loss": 0.048,
134
+ "step": 21
135
+ },
136
+ {
137
+ "epoch": 0.04,
138
+ "learning_rate": 0.00019995892109057675,
139
+ "loss": 0.0578,
140
+ "step": 22
141
+ },
142
+ {
143
+ "epoch": 0.04,
144
+ "learning_rate": 0.00019994084815253097,
145
+ "loss": 0.0568,
146
+ "step": 23
147
+ },
148
+ {
149
+ "epoch": 0.04,
150
+ "learning_rate": 0.00019991949062975336,
151
+ "loss": 0.0543,
152
+ "step": 24
153
+ },
154
+ {
155
+ "epoch": 0.04,
156
+ "learning_rate": 0.00019989484922416502,
157
+ "loss": 0.0424,
158
+ "step": 25
159
+ },
160
+ {
161
+ "epoch": 0.05,
162
+ "learning_rate": 0.00019986692474561292,
163
+ "loss": 0.0549,
164
+ "step": 26
165
+ },
166
+ {
167
+ "epoch": 0.05,
168
+ "learning_rate": 0.00019983571811184295,
169
+ "loss": 0.0425,
170
+ "step": 27
171
+ },
172
+ {
173
+ "epoch": 0.05,
174
+ "learning_rate": 0.00019980123034847024,
175
+ "loss": 0.0435,
176
+ "step": 28
177
+ },
178
+ {
179
+ "epoch": 0.05,
180
+ "learning_rate": 0.00019976346258894503,
181
+ "loss": 0.0367,
182
+ "step": 29
183
+ },
184
+ {
185
+ "epoch": 0.05,
186
+ "learning_rate": 0.00019972241607451552,
187
+ "loss": 0.0439,
188
+ "step": 30
189
+ },
190
+ {
191
+ "epoch": 0.05,
192
+ "learning_rate": 0.00019967809215418728,
193
+ "loss": 0.0418,
194
+ "step": 31
195
+ },
196
+ {
197
+ "epoch": 0.06,
198
+ "learning_rate": 0.00019963049228467874,
199
+ "loss": 0.0374,
200
+ "step": 32
201
+ },
202
+ {
203
+ "epoch": 0.06,
204
+ "learning_rate": 0.00019957961803037326,
205
+ "loss": 0.043,
206
+ "step": 33
207
+ },
208
+ {
209
+ "epoch": 0.06,
210
+ "learning_rate": 0.00019952547106326783,
211
+ "loss": 0.0419,
212
+ "step": 34
213
+ },
214
+ {
215
+ "epoch": 0.06,
216
+ "learning_rate": 0.00019946805316291814,
217
+ "loss": 0.0396,
218
+ "step": 35
219
+ },
220
+ {
221
+ "epoch": 0.06,
222
+ "learning_rate": 0.00019940736621638,
223
+ "loss": 0.049,
224
+ "step": 36
225
+ },
226
+ {
227
+ "epoch": 0.07,
228
+ "learning_rate": 0.00019934341221814739,
229
+ "loss": 0.0325,
230
+ "step": 37
231
+ },
232
+ {
233
+ "epoch": 0.07,
234
+ "learning_rate": 0.00019927619327008683,
235
+ "loss": 0.0398,
236
+ "step": 38
237
+ },
238
+ {
239
+ "epoch": 0.07,
240
+ "learning_rate": 0.0001992057115813684,
241
+ "loss": 0.0449,
242
+ "step": 39
243
+ },
244
+ {
245
+ "epoch": 0.07,
246
+ "learning_rate": 0.00019913196946839303,
247
+ "loss": 0.0301,
248
+ "step": 40
249
+ },
250
+ {
251
+ "epoch": 0.07,
252
+ "learning_rate": 0.00019905496935471658,
253
+ "loss": 0.0399,
254
+ "step": 41
255
+ },
256
+ {
257
+ "epoch": 0.07,
258
+ "learning_rate": 0.0001989747137709699,
259
+ "loss": 0.0561,
260
+ "step": 42
261
+ },
262
+ {
263
+ "epoch": 0.08,
264
+ "learning_rate": 0.00019889120535477583,
265
+ "loss": 0.0324,
266
+ "step": 43
267
+ },
268
+ {
269
+ "epoch": 0.08,
270
+ "learning_rate": 0.0001988044468506625,
271
+ "loss": 0.0225,
272
+ "step": 44
273
+ },
274
+ {
275
+ "epoch": 0.08,
276
+ "learning_rate": 0.0001987144411099731,
277
+ "loss": 0.037,
278
+ "step": 45
279
+ },
280
+ {
281
+ "epoch": 0.08,
282
+ "learning_rate": 0.00019862119109077223,
283
+ "loss": 0.0278,
284
+ "step": 46
285
+ },
286
+ {
287
+ "epoch": 0.08,
288
+ "learning_rate": 0.0001985246998577486,
289
+ "loss": 0.0244,
290
+ "step": 47
291
+ },
292
+ {
293
+ "epoch": 0.08,
294
+ "learning_rate": 0.00019842497058211432,
295
+ "loss": 0.0485,
296
+ "step": 48
297
+ },
298
+ {
299
+ "epoch": 0.09,
300
+ "learning_rate": 0.00019832200654150076,
301
+ "loss": 0.0393,
302
+ "step": 49
303
+ },
304
+ {
305
+ "epoch": 0.09,
306
+ "learning_rate": 0.00019821581111985071,
307
+ "loss": 0.0416,
308
+ "step": 50
309
+ },
310
+ {
311
+ "epoch": 0.09,
312
+ "learning_rate": 0.0001981063878073073,
313
+ "loss": 0.0406,
314
+ "step": 51
315
+ },
316
+ {
317
+ "epoch": 0.09,
318
+ "learning_rate": 0.0001979937402000991,
319
+ "loss": 0.0277,
320
+ "step": 52
321
+ },
322
+ {
323
+ "epoch": 0.09,
324
+ "learning_rate": 0.00019787787200042223,
325
+ "loss": 0.0493,
326
+ "step": 53
327
+ },
328
+ {
329
+ "epoch": 0.1,
330
+ "learning_rate": 0.00019775878701631838,
331
+ "loss": 0.0345,
332
+ "step": 54
333
+ },
334
+ {
335
+ "epoch": 0.1,
336
+ "learning_rate": 0.0001976364891615498,
337
+ "loss": 0.0346,
338
+ "step": 55
339
+ },
340
+ {
341
+ "epoch": 0.1,
342
+ "learning_rate": 0.0001975109824554707,
343
+ "loss": 0.0297,
344
+ "step": 56
345
+ },
346
+ {
347
+ "epoch": 0.1,
348
+ "learning_rate": 0.0001973822710228951,
349
+ "loss": 0.0326,
350
+ "step": 57
351
+ },
352
+ {
353
+ "epoch": 0.1,
354
+ "learning_rate": 0.00019725035909396121,
355
+ "loss": 0.039,
356
+ "step": 58
357
+ },
358
+ {
359
+ "epoch": 0.1,
360
+ "learning_rate": 0.0001971152510039926,
361
+ "loss": 0.0428,
362
+ "step": 59
363
+ },
364
+ {
365
+ "epoch": 0.11,
366
+ "learning_rate": 0.00019697695119335546,
367
+ "loss": 0.0278,
368
+ "step": 60
369
+ },
370
+ {
371
+ "epoch": 0.11,
372
+ "learning_rate": 0.0001968354642073129,
373
+ "loss": 0.0312,
374
+ "step": 61
375
+ },
376
+ {
377
+ "epoch": 0.11,
378
+ "learning_rate": 0.00019669079469587545,
379
+ "loss": 0.0224,
380
+ "step": 62
381
+ },
382
+ {
383
+ "epoch": 0.11,
384
+ "learning_rate": 0.0001965429474136482,
385
+ "loss": 0.0364,
386
+ "step": 63
387
+ },
388
+ {
389
+ "epoch": 0.11,
390
+ "learning_rate": 0.0001963919272196746,
391
+ "loss": 0.0326,
392
+ "step": 64
393
+ },
394
+ {
395
+ "epoch": 0.12,
396
+ "learning_rate": 0.00019623773907727682,
397
+ "loss": 0.0276,
398
+ "step": 65
399
+ },
400
+ {
401
+ "epoch": 0.12,
402
+ "learning_rate": 0.00019608038805389252,
403
+ "loss": 0.0426,
404
+ "step": 66
405
+ },
406
+ {
407
+ "epoch": 0.12,
408
+ "learning_rate": 0.00019591987932090834,
409
+ "loss": 0.0324,
410
+ "step": 67
411
+ },
412
+ {
413
+ "epoch": 0.12,
414
+ "learning_rate": 0.00019575621815348998,
415
+ "loss": 0.035,
416
+ "step": 68
417
+ },
418
+ {
419
+ "epoch": 0.12,
420
+ "learning_rate": 0.00019558940993040885,
421
+ "loss": 0.0244,
422
+ "step": 69
423
+ },
424
+ {
425
+ "epoch": 0.12,
426
+ "learning_rate": 0.0001954194601338651,
427
+ "loss": 0.0429,
428
+ "step": 70
429
+ },
430
+ {
431
+ "epoch": 0.13,
432
+ "learning_rate": 0.00019524637434930776,
433
+ "loss": 0.0275,
434
+ "step": 71
435
+ },
436
+ {
437
+ "epoch": 0.13,
438
+ "learning_rate": 0.00019507015826525094,
439
+ "loss": 0.0439,
440
+ "step": 72
441
+ },
442
+ {
443
+ "epoch": 0.13,
444
+ "learning_rate": 0.00019489081767308698,
445
+ "loss": 0.0268,
446
+ "step": 73
447
+ },
448
+ {
449
+ "epoch": 0.13,
450
+ "learning_rate": 0.00019470835846689596,
451
+ "loss": 0.038,
452
+ "step": 74
453
+ },
454
+ {
455
+ "epoch": 0.13,
456
+ "learning_rate": 0.00019452278664325228,
457
+ "loss": 0.0315,
458
+ "step": 75
459
+ },
460
+ {
461
+ "epoch": 0.13,
462
+ "learning_rate": 0.00019433410830102722,
463
+ "loss": 0.0366,
464
+ "step": 76
465
+ },
466
+ {
467
+ "epoch": 0.14,
468
+ "learning_rate": 0.00019414232964118892,
469
+ "loss": 0.0341,
470
+ "step": 77
471
+ },
472
+ {
473
+ "epoch": 0.14,
474
+ "learning_rate": 0.0001939474569665981,
475
+ "loss": 0.0225,
476
+ "step": 78
477
+ },
478
+ {
479
+ "epoch": 0.14,
480
+ "learning_rate": 0.00019374949668180138,
481
+ "loss": 0.0333,
482
+ "step": 79
483
+ },
484
+ {
485
+ "epoch": 0.14,
486
+ "learning_rate": 0.00019354845529282038,
487
+ "loss": 0.0268,
488
+ "step": 80
489
+ },
490
+ {
491
+ "epoch": 0.14,
492
+ "learning_rate": 0.0001933443394069383,
493
+ "loss": 0.0319,
494
+ "step": 81
495
+ },
496
+ {
497
+ "epoch": 0.15,
498
+ "learning_rate": 0.00019313715573248237,
499
+ "loss": 0.0254,
500
+ "step": 82
501
+ },
502
+ {
503
+ "epoch": 0.15,
504
+ "learning_rate": 0.00019292691107860376,
505
+ "loss": 0.0371,
506
+ "step": 83
507
+ },
508
+ {
509
+ "epoch": 0.15,
510
+ "learning_rate": 0.00019271361235505342,
511
+ "loss": 0.0235,
512
+ "step": 84
513
+ },
514
+ {
515
+ "epoch": 0.15,
516
+ "learning_rate": 0.00019249726657195532,
517
+ "loss": 0.0261,
518
+ "step": 85
519
+ },
520
+ {
521
+ "epoch": 0.15,
522
+ "learning_rate": 0.0001922778808395759,
523
+ "loss": 0.0314,
524
+ "step": 86
525
+ },
526
+ {
527
+ "epoch": 0.15,
528
+ "learning_rate": 0.00019205546236809035,
529
+ "loss": 0.0385,
530
+ "step": 87
531
+ },
532
+ {
533
+ "epoch": 0.16,
534
+ "learning_rate": 0.00019183001846734575,
535
+ "loss": 0.0433,
536
+ "step": 88
537
+ },
538
+ {
539
+ "epoch": 0.16,
540
+ "learning_rate": 0.00019160155654662076,
541
+ "loss": 0.0328,
542
+ "step": 89
543
+ },
544
+ {
545
+ "epoch": 0.16,
546
+ "learning_rate": 0.0001913700841143821,
547
+ "loss": 0.0352,
548
+ "step": 90
549
+ },
550
+ {
551
+ "epoch": 0.16,
552
+ "learning_rate": 0.00019113560877803796,
553
+ "loss": 0.0283,
554
+ "step": 91
555
+ },
556
+ {
557
+ "epoch": 0.16,
558
+ "learning_rate": 0.00019089813824368763,
559
+ "loss": 0.0472,
560
+ "step": 92
561
+ },
562
+ {
563
+ "epoch": 0.16,
564
+ "learning_rate": 0.0001906576803158686,
565
+ "loss": 0.0393,
566
+ "step": 93
567
+ },
568
+ {
569
+ "epoch": 0.17,
570
+ "learning_rate": 0.00019041424289729993,
571
+ "loss": 0.0286,
572
+ "step": 94
573
+ },
574
+ {
575
+ "epoch": 0.17,
576
+ "learning_rate": 0.00019016783398862229,
577
+ "loss": 0.0204,
578
+ "step": 95
579
+ },
580
+ {
581
+ "epoch": 0.17,
582
+ "learning_rate": 0.00018991846168813544,
583
+ "loss": 0.0259,
584
+ "step": 96
585
+ },
586
+ {
587
+ "epoch": 0.17,
588
+ "learning_rate": 0.0001896661341915318,
589
+ "loss": 0.0258,
590
+ "step": 97
591
+ },
592
+ {
593
+ "epoch": 0.17,
594
+ "learning_rate": 0.00018941085979162713,
595
+ "loss": 0.0284,
596
+ "step": 98
597
+ },
598
+ {
599
+ "epoch": 0.18,
600
+ "learning_rate": 0.00018915264687808806,
601
+ "loss": 0.0414,
602
+ "step": 99
603
+ },
604
+ {
605
+ "epoch": 0.18,
606
+ "learning_rate": 0.00018889150393715628,
607
+ "loss": 0.034,
608
+ "step": 100
609
+ },
610
+ {
611
+ "epoch": 0.18,
612
+ "learning_rate": 0.00018862743955136966,
613
+ "loss": 0.0336,
614
+ "step": 101
615
+ },
616
+ {
617
+ "epoch": 0.18,
618
+ "learning_rate": 0.00018836046239928023,
619
+ "loss": 0.0316,
620
+ "step": 102
621
+ },
622
+ {
623
+ "epoch": 0.18,
624
+ "learning_rate": 0.00018809058125516894,
625
+ "loss": 0.0321,
626
+ "step": 103
627
+ },
628
+ {
629
+ "epoch": 0.18,
630
+ "learning_rate": 0.00018781780498875725,
631
+ "loss": 0.0362,
632
+ "step": 104
633
+ },
634
+ {
635
+ "epoch": 0.19,
636
+ "learning_rate": 0.00018754214256491562,
637
+ "loss": 0.0292,
638
+ "step": 105
639
+ },
640
+ {
641
+ "epoch": 0.19,
642
+ "learning_rate": 0.00018726360304336894,
643
+ "loss": 0.0411,
644
+ "step": 106
645
+ },
646
+ {
647
+ "epoch": 0.19,
648
+ "learning_rate": 0.00018698219557839872,
649
+ "loss": 0.0275,
650
+ "step": 107
651
+ },
652
+ {
653
+ "epoch": 0.19,
654
+ "learning_rate": 0.00018669792941854228,
655
+ "loss": 0.0376,
656
+ "step": 108
657
+ },
658
+ {
659
+ "epoch": 0.19,
660
+ "learning_rate": 0.00018641081390628877,
661
+ "loss": 0.0279,
662
+ "step": 109
663
+ },
664
+ {
665
+ "epoch": 0.19,
666
+ "learning_rate": 0.00018612085847777215,
667
+ "loss": 0.0246,
668
+ "step": 110
669
+ },
670
+ {
671
+ "epoch": 0.2,
672
+ "learning_rate": 0.00018582807266246092,
673
+ "loss": 0.0249,
674
+ "step": 111
675
+ },
676
+ {
677
+ "epoch": 0.2,
678
+ "learning_rate": 0.0001855324660828452,
679
+ "loss": 0.0298,
680
+ "step": 112
681
+ },
682
+ {
683
+ "epoch": 0.2,
684
+ "learning_rate": 0.00018523404845412027,
685
+ "loss": 0.0294,
686
+ "step": 113
687
+ },
688
+ {
689
+ "epoch": 0.2,
690
+ "learning_rate": 0.0001849328295838674,
691
+ "loss": 0.0336,
692
+ "step": 114
693
+ },
694
+ {
695
+ "epoch": 0.2,
696
+ "learning_rate": 0.00018462881937173143,
697
+ "loss": 0.0292,
698
+ "step": 115
699
+ },
700
+ {
701
+ "epoch": 0.21,
702
+ "learning_rate": 0.0001843220278090954,
703
+ "loss": 0.0278,
704
+ "step": 116
705
+ },
706
+ {
707
+ "epoch": 0.21,
708
+ "learning_rate": 0.0001840124649787524,
709
+ "loss": 0.0441,
710
+ "step": 117
711
+ },
712
+ {
713
+ "epoch": 0.21,
714
+ "learning_rate": 0.00018370014105457378,
715
+ "loss": 0.0218,
716
+ "step": 118
717
+ },
718
+ {
719
+ "epoch": 0.21,
720
+ "learning_rate": 0.00018338506630117526,
721
+ "loss": 0.0243,
722
+ "step": 119
723
+ },
724
+ {
725
+ "epoch": 0.21,
726
+ "learning_rate": 0.00018306725107357933,
727
+ "loss": 0.0264,
728
+ "step": 120
729
+ },
730
+ {
731
+ "epoch": 0.21,
732
+ "learning_rate": 0.0001827467058168748,
733
+ "loss": 0.0251,
734
+ "step": 121
735
+ },
736
+ {
737
+ "epoch": 0.22,
738
+ "learning_rate": 0.0001824234410658738,
739
+ "loss": 0.0378,
740
+ "step": 122
741
+ },
742
+ {
743
+ "epoch": 0.22,
744
+ "learning_rate": 0.00018209746744476536,
745
+ "loss": 0.0294,
746
+ "step": 123
747
+ },
748
+ {
749
+ "epoch": 0.22,
750
+ "learning_rate": 0.0001817687956667664,
751
+ "loss": 0.0319,
752
+ "step": 124
753
+ },
754
+ {
755
+ "epoch": 0.22,
756
+ "learning_rate": 0.00018143743653376942,
757
+ "loss": 0.0293,
758
+ "step": 125
759
+ },
760
+ {
761
+ "epoch": 0.22,
762
+ "learning_rate": 0.0001811034009359877,
763
+ "loss": 0.0238,
764
+ "step": 126
765
+ },
766
+ {
767
+ "epoch": 0.22,
768
+ "learning_rate": 0.00018076669985159726,
769
+ "loss": 0.0292,
770
+ "step": 127
771
+ },
772
+ {
773
+ "epoch": 0.23,
774
+ "learning_rate": 0.00018042734434637613,
775
+ "loss": 0.0205,
776
+ "step": 128
777
+ },
778
+ {
779
+ "epoch": 0.23,
780
+ "learning_rate": 0.00018008534557334064,
781
+ "loss": 0.0271,
782
+ "step": 129
783
+ },
784
+ {
785
+ "epoch": 0.23,
786
+ "learning_rate": 0.00017974071477237885,
787
+ "loss": 0.0343,
788
+ "step": 130
789
+ },
790
+ {
791
+ "epoch": 0.23,
792
+ "learning_rate": 0.00017939346326988127,
793
+ "loss": 0.0264,
794
+ "step": 131
795
+ },
796
+ {
797
+ "epoch": 0.23,
798
+ "learning_rate": 0.00017904360247836834,
799
+ "loss": 0.0262,
800
+ "step": 132
801
+ },
802
+ {
803
+ "epoch": 0.24,
804
+ "learning_rate": 0.00017869114389611575,
805
+ "loss": 0.025,
806
+ "step": 133
807
+ },
808
+ {
809
+ "epoch": 0.24,
810
+ "learning_rate": 0.00017833609910677614,
811
+ "loss": 0.0296,
812
+ "step": 134
813
+ },
814
+ {
815
+ "epoch": 0.24,
816
+ "learning_rate": 0.00017797847977899873,
817
+ "loss": 0.0198,
818
+ "step": 135
819
+ },
820
+ {
821
+ "epoch": 0.24,
822
+ "learning_rate": 0.00017761829766604556,
823
+ "loss": 0.0205,
824
+ "step": 136
825
+ },
826
+ {
827
+ "epoch": 0.24,
828
+ "learning_rate": 0.0001772555646054055,
829
+ "loss": 0.0363,
830
+ "step": 137
831
+ },
832
+ {
833
+ "epoch": 0.24,
834
+ "learning_rate": 0.00017689029251840492,
835
+ "loss": 0.0398,
836
+ "step": 138
837
+ },
838
+ {
839
+ "epoch": 0.25,
840
+ "learning_rate": 0.00017652249340981607,
841
+ "loss": 0.037,
842
+ "step": 139
843
+ },
844
+ {
845
+ "epoch": 0.25,
846
+ "learning_rate": 0.00017615217936746246,
847
+ "loss": 0.0232,
848
+ "step": 140
849
+ },
850
+ {
851
+ "epoch": 0.25,
852
+ "learning_rate": 0.00017577936256182167,
853
+ "loss": 0.0235,
854
+ "step": 141
855
+ },
856
+ {
857
+ "epoch": 0.25,
858
+ "learning_rate": 0.0001754040552456253,
859
+ "loss": 0.0383,
860
+ "step": 142
861
+ },
862
+ {
863
+ "epoch": 0.25,
864
+ "learning_rate": 0.00017502626975345628,
865
+ "loss": 0.0287,
866
+ "step": 143
867
+ },
868
+ {
869
+ "epoch": 0.25,
870
+ "learning_rate": 0.00017464601850134353,
871
+ "loss": 0.0219,
872
+ "step": 144
873
+ },
874
+ {
875
+ "epoch": 0.26,
876
+ "learning_rate": 0.0001742633139863538,
877
+ "loss": 0.0247,
878
+ "step": 145
879
+ },
880
+ {
881
+ "epoch": 0.26,
882
+ "learning_rate": 0.0001738781687861812,
883
+ "loss": 0.0324,
884
+ "step": 146
885
+ },
886
+ {
887
+ "epoch": 0.26,
888
+ "learning_rate": 0.00017349059555873345,
889
+ "loss": 0.03,
890
+ "step": 147
891
+ },
892
+ {
893
+ "epoch": 0.26,
894
+ "learning_rate": 0.0001731006070417163,
895
+ "loss": 0.0293,
896
+ "step": 148
897
+ },
898
+ {
899
+ "epoch": 0.26,
900
+ "learning_rate": 0.0001727082160522145,
901
+ "loss": 0.0256,
902
+ "step": 149
903
+ },
904
+ {
905
+ "epoch": 0.27,
906
+ "learning_rate": 0.00017231343548627083,
907
+ "loss": 0.0275,
908
+ "step": 150
909
+ },
910
+ {
911
+ "epoch": 0.27,
912
+ "learning_rate": 0.00017191627831846226,
913
+ "loss": 0.0211,
914
+ "step": 151
915
+ },
916
+ {
917
+ "epoch": 0.27,
918
+ "learning_rate": 0.00017151675760147326,
919
+ "loss": 0.0306,
920
+ "step": 152
921
+ },
922
+ {
923
+ "epoch": 0.27,
924
+ "learning_rate": 0.00017111488646566727,
925
+ "loss": 0.0248,
926
+ "step": 153
927
+ },
928
+ {
929
+ "epoch": 0.27,
930
+ "learning_rate": 0.00017071067811865476,
931
+ "loss": 0.0363,
932
+ "step": 154
933
+ },
934
+ {
935
+ "epoch": 0.27,
936
+ "learning_rate": 0.00017030414584485936,
937
+ "loss": 0.0258,
938
+ "step": 155
939
+ },
940
+ {
941
+ "epoch": 0.28,
942
+ "learning_rate": 0.00016989530300508124,
943
+ "loss": 0.0175,
944
+ "step": 156
945
+ },
946
+ {
947
+ "epoch": 0.28,
948
+ "learning_rate": 0.00016948416303605795,
949
+ "loss": 0.0234,
950
+ "step": 157
951
+ },
952
+ {
953
+ "epoch": 0.28,
954
+ "learning_rate": 0.0001690707394500229,
955
+ "loss": 0.0263,
956
+ "step": 158
957
+ },
958
+ {
959
+ "epoch": 0.28,
960
+ "learning_rate": 0.00016865504583426117,
961
+ "loss": 0.0308,
962
+ "step": 159
963
+ },
964
+ {
965
+ "epoch": 0.28,
966
+ "learning_rate": 0.00016823709585066308,
967
+ "loss": 0.0216,
968
+ "step": 160
969
+ },
970
+ {
971
+ "epoch": 0.28,
972
+ "learning_rate": 0.00016781690323527511,
973
+ "loss": 0.0252,
974
+ "step": 161
975
+ },
976
+ {
977
+ "epoch": 0.29,
978
+ "learning_rate": 0.00016739448179784846,
979
+ "loss": 0.0235,
980
+ "step": 162
981
+ },
982
+ {
983
+ "epoch": 0.29,
984
+ "learning_rate": 0.0001669698454213852,
985
+ "loss": 0.0262,
986
+ "step": 163
987
+ },
988
+ {
989
+ "epoch": 0.29,
990
+ "learning_rate": 0.00016654300806168208,
991
+ "loss": 0.0174,
992
+ "step": 164
993
+ },
994
+ {
995
+ "epoch": 0.29,
996
+ "learning_rate": 0.0001661139837468717,
997
+ "loss": 0.0315,
998
+ "step": 165
999
+ },
1000
+ {
1001
+ "epoch": 0.29,
1002
+ "learning_rate": 0.00016568278657696164,
1003
+ "loss": 0.0256,
1004
+ "step": 166
1005
+ },
1006
+ {
1007
+ "epoch": 0.3,
1008
+ "learning_rate": 0.00016524943072337094,
1009
+ "loss": 0.032,
1010
+ "step": 167
1011
+ },
1012
+ {
1013
+ "epoch": 0.3,
1014
+ "learning_rate": 0.00016481393042846443,
1015
+ "loss": 0.0258,
1016
+ "step": 168
1017
+ },
1018
+ {
1019
+ "epoch": 0.3,
1020
+ "learning_rate": 0.00016437630000508464,
1021
+ "loss": 0.0217,
1022
+ "step": 169
1023
+ },
1024
+ {
1025
+ "epoch": 0.3,
1026
+ "learning_rate": 0.00016393655383608132,
1027
+ "loss": 0.0322,
1028
+ "step": 170
1029
+ },
1030
+ {
1031
+ "epoch": 0.3,
1032
+ "learning_rate": 0.0001634947063738389,
1033
+ "loss": 0.0278,
1034
+ "step": 171
1035
+ },
1036
+ {
1037
+ "epoch": 0.3,
1038
+ "learning_rate": 0.00016305077213980132,
1039
+ "loss": 0.0319,
1040
+ "step": 172
1041
+ },
1042
+ {
1043
+ "epoch": 0.31,
1044
+ "learning_rate": 0.00016260476572399496,
1045
+ "loss": 0.0219,
1046
+ "step": 173
1047
+ },
1048
+ {
1049
+ "epoch": 0.31,
1050
+ "learning_rate": 0.00016215670178454893,
1051
+ "loss": 0.0237,
1052
+ "step": 174
1053
+ },
1054
+ {
1055
+ "epoch": 0.31,
1056
+ "learning_rate": 0.00016170659504721364,
1057
+ "loss": 0.017,
1058
+ "step": 175
1059
+ },
1060
+ {
1061
+ "epoch": 0.31,
1062
+ "learning_rate": 0.00016125446030487643,
1063
+ "loss": 0.0264,
1064
+ "step": 176
1065
+ },
1066
+ {
1067
+ "epoch": 0.31,
1068
+ "learning_rate": 0.00016080031241707578,
1069
+ "loss": 0.0255,
1070
+ "step": 177
1071
+ },
1072
+ {
1073
+ "epoch": 0.32,
1074
+ "learning_rate": 0.00016034416630951267,
1075
+ "loss": 0.0283,
1076
+ "step": 178
1077
+ },
1078
+ {
1079
+ "epoch": 0.32,
1080
+ "learning_rate": 0.0001598860369735601,
1081
+ "loss": 0.0274,
1082
+ "step": 179
1083
+ },
1084
+ {
1085
+ "epoch": 0.32,
1086
+ "learning_rate": 0.00015942593946577068,
1087
+ "loss": 0.0179,
1088
+ "step": 180
1089
+ },
1090
+ {
1091
+ "epoch": 0.32,
1092
+ "learning_rate": 0.00015896388890738127,
1093
+ "loss": 0.0244,
1094
+ "step": 181
1095
+ },
1096
+ {
1097
+ "epoch": 0.32,
1098
+ "learning_rate": 0.0001584999004838165,
1099
+ "loss": 0.0352,
1100
+ "step": 182
1101
+ },
1102
+ {
1103
+ "epoch": 0.32,
1104
+ "learning_rate": 0.00015803398944418934,
1105
+ "loss": 0.0348,
1106
+ "step": 183
1107
+ },
1108
+ {
1109
+ "epoch": 0.33,
1110
+ "learning_rate": 0.00015756617110080024,
1111
+ "loss": 0.032,
1112
+ "step": 184
1113
+ },
1114
+ {
1115
+ "epoch": 0.33,
1116
+ "learning_rate": 0.0001570964608286336,
1117
+ "loss": 0.0296,
1118
+ "step": 185
1119
+ },
1120
+ {
1121
+ "epoch": 0.33,
1122
+ "learning_rate": 0.00015662487406485273,
1123
+ "loss": 0.0231,
1124
+ "step": 186
1125
+ },
1126
+ {
1127
+ "epoch": 0.33,
1128
+ "learning_rate": 0.00015615142630829229,
1129
+ "loss": 0.0203,
1130
+ "step": 187
1131
+ },
1132
+ {
1133
+ "epoch": 0.33,
1134
+ "learning_rate": 0.0001556761331189491,
1135
+ "loss": 0.0291,
1136
+ "step": 188
1137
+ },
1138
+ {
1139
+ "epoch": 0.33,
1140
+ "learning_rate": 0.00015519901011747044,
1141
+ "loss": 0.0337,
1142
+ "step": 189
1143
+ },
1144
+ {
1145
+ "epoch": 0.34,
1146
+ "learning_rate": 0.00015472007298464118,
1147
+ "loss": 0.0252,
1148
+ "step": 190
1149
+ },
1150
+ {
1151
+ "epoch": 0.34,
1152
+ "learning_rate": 0.00015423933746086795,
1153
+ "loss": 0.0229,
1154
+ "step": 191
1155
+ },
1156
+ {
1157
+ "epoch": 0.34,
1158
+ "learning_rate": 0.00015375681934566205,
1159
+ "loss": 0.0272,
1160
+ "step": 192
1161
+ },
1162
+ {
1163
+ "epoch": 0.34,
1164
+ "learning_rate": 0.0001532725344971202,
1165
+ "loss": 0.0291,
1166
+ "step": 193
1167
+ },
1168
+ {
1169
+ "epoch": 0.34,
1170
+ "learning_rate": 0.0001527864988314033,
1171
+ "loss": 0.0246,
1172
+ "step": 194
1173
+ },
1174
+ {
1175
+ "epoch": 0.35,
1176
+ "learning_rate": 0.00015229872832221334,
1177
+ "loss": 0.0299,
1178
+ "step": 195
1179
+ },
1180
+ {
1181
+ "epoch": 0.35,
1182
+ "learning_rate": 0.00015180923900026848,
1183
+ "loss": 0.0278,
1184
+ "step": 196
1185
+ },
1186
+ {
1187
+ "epoch": 0.35,
1188
+ "learning_rate": 0.00015131804695277612,
1189
+ "loss": 0.028,
1190
+ "step": 197
1191
+ },
1192
+ {
1193
+ "epoch": 0.35,
1194
+ "learning_rate": 0.00015082516832290422,
1195
+ "loss": 0.0315,
1196
+ "step": 198
1197
+ },
1198
+ {
1199
+ "epoch": 0.35,
1200
+ "learning_rate": 0.0001503306193092508,
1201
+ "loss": 0.0249,
1202
+ "step": 199
1203
+ },
1204
+ {
1205
+ "epoch": 0.35,
1206
+ "learning_rate": 0.0001498344161653115,
1207
+ "loss": 0.0224,
1208
+ "step": 200
1209
+ },
1210
+ {
1211
+ "epoch": 0.36,
1212
+ "learning_rate": 0.0001493365751989454,
1213
+ "loss": 0.0307,
1214
+ "step": 201
1215
+ },
1216
+ {
1217
+ "epoch": 0.36,
1218
+ "learning_rate": 0.00014883711277183915,
1219
+ "loss": 0.0201,
1220
+ "step": 202
1221
+ },
1222
+ {
1223
+ "epoch": 0.36,
1224
+ "learning_rate": 0.0001483360452989691,
1225
+ "loss": 0.0325,
1226
+ "step": 203
1227
+ },
1228
+ {
1229
+ "epoch": 0.36,
1230
+ "learning_rate": 0.00014783338924806192,
1231
+ "loss": 0.0226,
1232
+ "step": 204
1233
+ },
1234
+ {
1235
+ "epoch": 0.36,
1236
+ "learning_rate": 0.00014732916113905335,
1237
+ "loss": 0.0213,
1238
+ "step": 205
1239
+ },
1240
+ {
1241
+ "epoch": 0.36,
1242
+ "learning_rate": 0.00014682337754354534,
1243
+ "loss": 0.0203,
1244
+ "step": 206
1245
+ },
1246
+ {
1247
+ "epoch": 0.37,
1248
+ "learning_rate": 0.00014631605508426122,
1249
+ "loss": 0.0253,
1250
+ "step": 207
1251
+ },
1252
+ {
1253
+ "epoch": 0.37,
1254
+ "learning_rate": 0.00014580721043449968,
1255
+ "loss": 0.022,
1256
+ "step": 208
1257
+ },
1258
+ {
1259
+ "epoch": 0.37,
1260
+ "learning_rate": 0.00014529686031758643,
1261
+ "loss": 0.0274,
1262
+ "step": 209
1263
+ },
1264
+ {
1265
+ "epoch": 0.37,
1266
+ "learning_rate": 0.00014478502150632502,
1267
+ "loss": 0.0214,
1268
+ "step": 210
1269
+ },
1270
+ {
1271
+ "epoch": 0.37,
1272
+ "learning_rate": 0.00014427171082244522,
1273
+ "loss": 0.018,
1274
+ "step": 211
1275
+ },
1276
+ {
1277
+ "epoch": 0.38,
1278
+ "learning_rate": 0.00014375694513605035,
1279
+ "loss": 0.0242,
1280
+ "step": 212
1281
+ },
1282
+ {
1283
+ "epoch": 0.38,
1284
+ "learning_rate": 0.00014324074136506284,
1285
+ "loss": 0.0195,
1286
+ "step": 213
1287
+ },
1288
+ {
1289
+ "epoch": 0.38,
1290
+ "learning_rate": 0.0001427231164746681,
1291
+ "loss": 0.0247,
1292
+ "step": 214
1293
+ },
1294
+ {
1295
+ "epoch": 0.38,
1296
+ "learning_rate": 0.00014220408747675714,
1297
+ "loss": 0.0247,
1298
+ "step": 215
1299
+ },
1300
+ {
1301
+ "epoch": 0.38,
1302
+ "learning_rate": 0.00014168367142936735,
1303
+ "loss": 0.0229,
1304
+ "step": 216
1305
+ },
1306
+ {
1307
+ "epoch": 0.38,
1308
+ "learning_rate": 0.0001411618854361218,
1309
+ "loss": 0.0233,
1310
+ "step": 217
1311
+ },
1312
+ {
1313
+ "epoch": 0.39,
1314
+ "learning_rate": 0.00014063874664566734,
1315
+ "loss": 0.0291,
1316
+ "step": 218
1317
+ },
1318
+ {
1319
+ "epoch": 0.39,
1320
+ "learning_rate": 0.00014011427225111092,
1321
+ "loss": 0.0274,
1322
+ "step": 219
1323
+ },
1324
+ {
1325
+ "epoch": 0.39,
1326
+ "learning_rate": 0.00013958847948945428,
1327
+ "loss": 0.018,
1328
+ "step": 220
1329
+ },
1330
+ {
1331
+ "epoch": 0.39,
1332
+ "learning_rate": 0.00013906138564102793,
1333
+ "loss": 0.0267,
1334
+ "step": 221
1335
+ },
1336
+ {
1337
+ "epoch": 0.39,
1338
+ "learning_rate": 0.00013853300802892283,
1339
+ "loss": 0.0261,
1340
+ "step": 222
1341
+ },
1342
+ {
1343
+ "epoch": 0.39,
1344
+ "learning_rate": 0.0001380033640184213,
1345
+ "loss": 0.0232,
1346
+ "step": 223
1347
+ },
1348
+ {
1349
+ "epoch": 0.4,
1350
+ "learning_rate": 0.00013747247101642603,
1351
+ "loss": 0.0272,
1352
+ "step": 224
1353
+ },
1354
+ {
1355
+ "epoch": 0.4,
1356
+ "learning_rate": 0.0001369403464708884,
1357
+ "loss": 0.0313,
1358
+ "step": 225
1359
+ },
1360
+ {
1361
+ "epoch": 0.4,
1362
+ "learning_rate": 0.00013640700787023464,
1363
+ "loss": 0.0264,
1364
+ "step": 226
1365
+ },
1366
+ {
1367
+ "epoch": 0.4,
1368
+ "learning_rate": 0.0001358724727427914,
1369
+ "loss": 0.0204,
1370
+ "step": 227
1371
+ },
1372
+ {
1373
+ "epoch": 0.4,
1374
+ "learning_rate": 0.00013533675865620937,
1375
+ "loss": 0.0304,
1376
+ "step": 228
1377
+ },
1378
+ {
1379
+ "epoch": 0.41,
1380
+ "learning_rate": 0.0001347998832168862,
1381
+ "loss": 0.0234,
1382
+ "step": 229
1383
+ },
1384
+ {
1385
+ "epoch": 0.41,
1386
+ "learning_rate": 0.00013426186406938768,
1387
+ "loss": 0.0206,
1388
+ "step": 230
1389
+ },
1390
+ {
1391
+ "epoch": 0.41,
1392
+ "learning_rate": 0.0001337227188958679,
1393
+ "loss": 0.0229,
1394
+ "step": 231
1395
+ },
1396
+ {
1397
+ "epoch": 0.41,
1398
+ "learning_rate": 0.0001331824654154881,
1399
+ "loss": 0.0282,
1400
+ "step": 232
1401
+ },
1402
+ {
1403
+ "epoch": 0.41,
1404
+ "learning_rate": 0.00013264112138383445,
1405
+ "loss": 0.0218,
1406
+ "step": 233
1407
+ },
1408
+ {
1409
+ "epoch": 0.41,
1410
+ "learning_rate": 0.00013209870459233422,
1411
+ "loss": 0.0278,
1412
+ "step": 234
1413
+ },
1414
+ {
1415
+ "epoch": 0.42,
1416
+ "learning_rate": 0.0001315552328676714,
1417
+ "loss": 0.0208,
1418
+ "step": 235
1419
+ },
1420
+ {
1421
+ "epoch": 0.42,
1422
+ "learning_rate": 0.00013101072407120057,
1423
+ "loss": 0.0147,
1424
+ "step": 236
1425
+ },
1426
+ {
1427
+ "epoch": 0.42,
1428
+ "learning_rate": 0.00013046519609836,
1429
+ "loss": 0.0172,
1430
+ "step": 237
1431
+ },
1432
+ {
1433
+ "epoch": 0.42,
1434
+ "learning_rate": 0.00012991866687808354,
1435
+ "loss": 0.0162,
1436
+ "step": 238
1437
+ },
1438
+ {
1439
+ "epoch": 0.42,
1440
+ "learning_rate": 0.0001293711543722112,
1441
+ "loss": 0.0153,
1442
+ "step": 239
1443
+ },
1444
+ {
1445
+ "epoch": 0.42,
1446
+ "learning_rate": 0.00012882267657489909,
1447
+ "loss": 0.0164,
1448
+ "step": 240
1449
+ },
1450
+ {
1451
+ "epoch": 0.43,
1452
+ "learning_rate": 0.00012827325151202782,
1453
+ "loss": 0.0239,
1454
+ "step": 241
1455
+ },
1456
+ {
1457
+ "epoch": 0.43,
1458
+ "learning_rate": 0.00012772289724061016,
1459
+ "loss": 0.0183,
1460
+ "step": 242
1461
+ },
1462
+ {
1463
+ "epoch": 0.43,
1464
+ "learning_rate": 0.0001271716318481976,
1465
+ "loss": 0.0115,
1466
+ "step": 243
1467
+ },
1468
+ {
1469
+ "epoch": 0.43,
1470
+ "learning_rate": 0.00012661947345228594,
1471
+ "loss": 0.0223,
1472
+ "step": 244
1473
+ },
1474
+ {
1475
+ "epoch": 0.43,
1476
+ "learning_rate": 0.00012606644019971968,
1477
+ "loss": 0.0275,
1478
+ "step": 245
1479
+ },
1480
+ {
1481
+ "epoch": 0.44,
1482
+ "learning_rate": 0.0001255125502660958,
1483
+ "loss": 0.0172,
1484
+ "step": 246
1485
+ },
1486
+ {
1487
+ "epoch": 0.44,
1488
+ "learning_rate": 0.0001249578218551664,
1489
+ "loss": 0.0345,
1490
+ "step": 247
1491
+ },
1492
+ {
1493
+ "epoch": 0.44,
1494
+ "learning_rate": 0.00012440227319824023,
1495
+ "loss": 0.0267,
1496
+ "step": 248
1497
+ },
1498
+ {
1499
+ "epoch": 0.44,
1500
+ "learning_rate": 0.00012384592255358385,
1501
+ "loss": 0.0201,
1502
+ "step": 249
1503
+ },
1504
+ {
1505
+ "epoch": 0.44,
1506
+ "learning_rate": 0.0001232887882058212,
1507
+ "loss": 0.0197,
1508
+ "step": 250
1509
+ },
1510
+ {
1511
+ "epoch": 0.44,
1512
+ "learning_rate": 0.00012273088846533303,
1513
+ "loss": 0.0289,
1514
+ "step": 251
1515
+ },
1516
+ {
1517
+ "epoch": 0.45,
1518
+ "learning_rate": 0.00012217224166765478,
1519
+ "loss": 0.0243,
1520
+ "step": 252
1521
+ },
1522
+ {
1523
+ "epoch": 0.45,
1524
+ "learning_rate": 0.00012161286617287419,
1525
+ "loss": 0.02,
1526
+ "step": 253
1527
+ },
1528
+ {
1529
+ "epoch": 0.45,
1530
+ "learning_rate": 0.00012105278036502788,
1531
+ "loss": 0.0237,
1532
+ "step": 254
1533
+ },
1534
+ {
1535
+ "epoch": 0.45,
1536
+ "learning_rate": 0.00012049200265149709,
1537
+ "loss": 0.0191,
1538
+ "step": 255
1539
+ },
1540
+ {
1541
+ "epoch": 0.45,
1542
+ "learning_rate": 0.00011993055146240273,
1543
+ "loss": 0.0248,
1544
+ "step": 256
1545
+ },
1546
+ {
1547
+ "epoch": 0.45,
1548
+ "learning_rate": 0.00011936844524999966,
1549
+ "loss": 0.0131,
1550
+ "step": 257
1551
+ },
1552
+ {
1553
+ "epoch": 0.46,
1554
+ "learning_rate": 0.00011880570248807033,
1555
+ "loss": 0.0244,
1556
+ "step": 258
1557
+ },
1558
+ {
1559
+ "epoch": 0.46,
1560
+ "learning_rate": 0.00011824234167131747,
1561
+ "loss": 0.0204,
1562
+ "step": 259
1563
+ },
1564
+ {
1565
+ "epoch": 0.46,
1566
+ "learning_rate": 0.00011767838131475655,
1567
+ "loss": 0.016,
1568
+ "step": 260
1569
+ },
1570
+ {
1571
+ "epoch": 0.46,
1572
+ "learning_rate": 0.00011711383995310681,
1573
+ "loss": 0.0265,
1574
+ "step": 261
1575
+ },
1576
+ {
1577
+ "epoch": 0.46,
1578
+ "learning_rate": 0.00011654873614018267,
1579
+ "loss": 0.0193,
1580
+ "step": 262
1581
+ },
1582
+ {
1583
+ "epoch": 0.47,
1584
+ "learning_rate": 0.00011598308844828348,
1585
+ "loss": 0.0266,
1586
+ "step": 263
1587
+ },
1588
+ {
1589
+ "epoch": 0.47,
1590
+ "learning_rate": 0.00011541691546758342,
1591
+ "loss": 0.0182,
1592
+ "step": 264
1593
+ },
1594
+ {
1595
+ "epoch": 0.47,
1596
+ "learning_rate": 0.00011485023580552039,
1597
+ "loss": 0.0337,
1598
+ "step": 265
1599
+ },
1600
+ {
1601
+ "epoch": 0.47,
1602
+ "learning_rate": 0.00011428306808618456,
1603
+ "loss": 0.0302,
1604
+ "step": 266
1605
+ },
1606
+ {
1607
+ "epoch": 0.47,
1608
+ "learning_rate": 0.00011371543094970622,
1609
+ "loss": 0.0203,
1610
+ "step": 267
1611
+ },
1612
+ {
1613
+ "epoch": 0.47,
1614
+ "learning_rate": 0.00011314734305164321,
1615
+ "loss": 0.0345,
1616
+ "step": 268
1617
+ },
1618
+ {
1619
+ "epoch": 0.48,
1620
+ "learning_rate": 0.00011257882306236775,
1621
+ "loss": 0.0219,
1622
+ "step": 269
1623
+ },
1624
+ {
1625
+ "epoch": 0.48,
1626
+ "learning_rate": 0.00011200988966645284,
1627
+ "loss": 0.0225,
1628
+ "step": 270
1629
+ },
1630
+ {
1631
+ "epoch": 0.48,
1632
+ "learning_rate": 0.00011144056156205833,
1633
+ "loss": 0.0149,
1634
+ "step": 271
1635
+ },
1636
+ {
1637
+ "epoch": 0.48,
1638
+ "learning_rate": 0.00011087085746031612,
1639
+ "loss": 0.0182,
1640
+ "step": 272
1641
+ },
1642
+ {
1643
+ "epoch": 0.48,
1644
+ "learning_rate": 0.00011030079608471544,
1645
+ "loss": 0.0268,
1646
+ "step": 273
1647
+ },
1648
+ {
1649
+ "epoch": 0.48,
1650
+ "learning_rate": 0.00010973039617048748,
1651
+ "loss": 0.0188,
1652
+ "step": 274
1653
+ },
1654
+ {
1655
+ "epoch": 0.49,
1656
+ "learning_rate": 0.0001091596764639895,
1657
+ "loss": 0.0207,
1658
+ "step": 275
1659
+ },
1660
+ {
1661
+ "epoch": 0.49,
1662
+ "learning_rate": 0.00010858865572208892,
1663
+ "loss": 0.0271,
1664
+ "step": 276
1665
+ },
1666
+ {
1667
+ "epoch": 0.49,
1668
+ "learning_rate": 0.00010801735271154669,
1669
+ "loss": 0.0153,
1670
+ "step": 277
1671
+ },
1672
+ {
1673
+ "epoch": 0.49,
1674
+ "learning_rate": 0.00010744578620840064,
1675
+ "loss": 0.0189,
1676
+ "step": 278
1677
+ },
1678
+ {
1679
+ "epoch": 0.49,
1680
+ "learning_rate": 0.00010687397499734842,
1681
+ "loss": 0.017,
1682
+ "step": 279
1683
+ },
1684
+ {
1685
+ "epoch": 0.5,
1686
+ "learning_rate": 0.00010630193787112994,
1687
+ "loss": 0.0306,
1688
+ "step": 280
1689
+ },
1690
+ {
1691
+ "epoch": 0.5,
1692
+ "learning_rate": 0.00010572969362990998,
1693
+ "loss": 0.0217,
1694
+ "step": 281
1695
+ },
1696
+ {
1697
+ "epoch": 0.5,
1698
+ "learning_rate": 0.00010515726108066025,
1699
+ "loss": 0.0222,
1700
+ "step": 282
1701
+ },
1702
+ {
1703
+ "epoch": 0.5,
1704
+ "learning_rate": 0.00010458465903654107,
1705
+ "loss": 0.0241,
1706
+ "step": 283
1707
+ },
1708
+ {
1709
+ "epoch": 0.5,
1710
+ "learning_rate": 0.00010401190631628349,
1711
+ "loss": 0.0206,
1712
+ "step": 284
1713
+ },
1714
+ {
1715
+ "epoch": 0.5,
1716
+ "learning_rate": 0.00010343902174357039,
1717
+ "loss": 0.0218,
1718
+ "step": 285
1719
+ },
1720
+ {
1721
+ "epoch": 0.51,
1722
+ "learning_rate": 0.00010286602414641817,
1723
+ "loss": 0.0181,
1724
+ "step": 286
1725
+ },
1726
+ {
1727
+ "epoch": 0.51,
1728
+ "learning_rate": 0.00010229293235655768,
1729
+ "loss": 0.0228,
1730
+ "step": 287
1731
+ },
1732
+ {
1733
+ "epoch": 0.51,
1734
+ "learning_rate": 0.00010171976520881552,
1735
+ "loss": 0.0146,
1736
+ "step": 288
1737
+ },
1738
+ {
1739
+ "epoch": 0.51,
1740
+ "learning_rate": 0.0001011465415404949,
1741
+ "loss": 0.021,
1742
+ "step": 289
1743
+ },
1744
+ {
1745
+ "epoch": 0.51,
1746
+ "learning_rate": 0.00010057328019075669,
1747
+ "loss": 0.0193,
1748
+ "step": 290
1749
+ },
1750
+ {
1751
+ "epoch": 0.52,
1752
+ "learning_rate": 0.0001,
1753
+ "loss": 0.0172,
1754
+ "step": 291
1755
+ },
1756
+ {
1757
+ "epoch": 0.52,
1758
+ "learning_rate": 9.942671980924336e-05,
1759
+ "loss": 0.0204,
1760
+ "step": 292
1761
+ },
1762
+ {
1763
+ "epoch": 0.52,
1764
+ "learning_rate": 9.88534584595051e-05,
1765
+ "loss": 0.0203,
1766
+ "step": 293
1767
+ },
1768
+ {
1769
+ "epoch": 0.52,
1770
+ "learning_rate": 9.82802347911845e-05,
1771
+ "loss": 0.0224,
1772
+ "step": 294
1773
+ },
1774
+ {
1775
+ "epoch": 0.52,
1776
+ "learning_rate": 9.770706764344234e-05,
1777
+ "loss": 0.0165,
1778
+ "step": 295
1779
+ },
1780
+ {
1781
+ "epoch": 0.52,
1782
+ "learning_rate": 9.713397585358188e-05,
1783
+ "loss": 0.0226,
1784
+ "step": 296
1785
+ },
1786
+ {
1787
+ "epoch": 0.53,
1788
+ "learning_rate": 9.656097825642961e-05,
1789
+ "loss": 0.0199,
1790
+ "step": 297
1791
+ },
1792
+ {
1793
+ "epoch": 0.53,
1794
+ "learning_rate": 9.598809368371655e-05,
1795
+ "loss": 0.0165,
1796
+ "step": 298
1797
+ },
1798
+ {
1799
+ "epoch": 0.53,
1800
+ "learning_rate": 9.541534096345897e-05,
1801
+ "loss": 0.016,
1802
+ "step": 299
1803
+ },
1804
+ {
1805
+ "epoch": 0.53,
1806
+ "learning_rate": 9.484273891933982e-05,
1807
+ "loss": 0.0167,
1808
+ "step": 300
1809
+ },
1810
+ {
1811
+ "epoch": 0.53,
1812
+ "learning_rate": 9.427030637009003e-05,
1813
+ "loss": 0.0227,
1814
+ "step": 301
1815
+ },
1816
+ {
1817
+ "epoch": 0.53,
1818
+ "learning_rate": 9.369806212887008e-05,
1819
+ "loss": 0.0171,
1820
+ "step": 302
1821
+ },
1822
+ {
1823
+ "epoch": 0.54,
1824
+ "learning_rate": 9.31260250026516e-05,
1825
+ "loss": 0.0253,
1826
+ "step": 303
1827
+ },
1828
+ {
1829
+ "epoch": 0.54,
1830
+ "learning_rate": 9.255421379159934e-05,
1831
+ "loss": 0.0191,
1832
+ "step": 304
1833
+ },
1834
+ {
1835
+ "epoch": 0.54,
1836
+ "learning_rate": 9.198264728845332e-05,
1837
+ "loss": 0.0157,
1838
+ "step": 305
1839
+ },
1840
+ {
1841
+ "epoch": 0.54,
1842
+ "learning_rate": 9.14113442779111e-05,
1843
+ "loss": 0.0141,
1844
+ "step": 306
1845
+ },
1846
+ {
1847
+ "epoch": 0.54,
1848
+ "learning_rate": 9.084032353601053e-05,
1849
+ "loss": 0.0166,
1850
+ "step": 307
1851
+ },
1852
+ {
1853
+ "epoch": 0.55,
1854
+ "learning_rate": 9.026960382951253e-05,
1855
+ "loss": 0.0195,
1856
+ "step": 308
1857
+ },
1858
+ {
1859
+ "epoch": 0.55,
1860
+ "learning_rate": 8.969920391528458e-05,
1861
+ "loss": 0.0181,
1862
+ "step": 309
1863
+ },
1864
+ {
1865
+ "epoch": 0.55,
1866
+ "learning_rate": 8.912914253968392e-05,
1867
+ "loss": 0.024,
1868
+ "step": 310
1869
+ },
1870
+ {
1871
+ "epoch": 0.55,
1872
+ "learning_rate": 8.85594384379417e-05,
1873
+ "loss": 0.0191,
1874
+ "step": 311
1875
+ },
1876
+ {
1877
+ "epoch": 0.55,
1878
+ "learning_rate": 8.799011033354717e-05,
1879
+ "loss": 0.0156,
1880
+ "step": 312
1881
+ },
1882
+ {
1883
+ "epoch": 0.55,
1884
+ "learning_rate": 8.742117693763227e-05,
1885
+ "loss": 0.0116,
1886
+ "step": 313
1887
+ },
1888
+ {
1889
+ "epoch": 0.56,
1890
+ "learning_rate": 8.685265694835681e-05,
1891
+ "loss": 0.0147,
1892
+ "step": 314
1893
+ },
1894
+ {
1895
+ "epoch": 0.56,
1896
+ "learning_rate": 8.628456905029383e-05,
1897
+ "loss": 0.015,
1898
+ "step": 315
1899
+ },
1900
+ {
1901
+ "epoch": 0.56,
1902
+ "learning_rate": 8.571693191381545e-05,
1903
+ "loss": 0.0141,
1904
+ "step": 316
1905
+ },
1906
+ {
1907
+ "epoch": 0.56,
1908
+ "learning_rate": 8.514976419447964e-05,
1909
+ "loss": 0.0212,
1910
+ "step": 317
1911
+ },
1912
+ {
1913
+ "epoch": 0.56,
1914
+ "learning_rate": 8.458308453241663e-05,
1915
+ "loss": 0.0317,
1916
+ "step": 318
1917
+ },
1918
+ {
1919
+ "epoch": 0.56,
1920
+ "learning_rate": 8.401691155171653e-05,
1921
+ "loss": 0.0264,
1922
+ "step": 319
1923
+ },
1924
+ {
1925
+ "epoch": 0.57,
1926
+ "learning_rate": 8.345126385981736e-05,
1927
+ "loss": 0.0149,
1928
+ "step": 320
1929
+ },
1930
+ {
1931
+ "epoch": 0.57,
1932
+ "learning_rate": 8.28861600468932e-05,
1933
+ "loss": 0.0145,
1934
+ "step": 321
1935
+ },
1936
+ {
1937
+ "epoch": 0.57,
1938
+ "learning_rate": 8.232161868524351e-05,
1939
+ "loss": 0.0171,
1940
+ "step": 322
1941
+ },
1942
+ {
1943
+ "epoch": 0.57,
1944
+ "learning_rate": 8.175765832868251e-05,
1945
+ "loss": 0.02,
1946
+ "step": 323
1947
+ },
1948
+ {
1949
+ "epoch": 0.57,
1950
+ "learning_rate": 8.11942975119297e-05,
1951
+ "loss": 0.0239,
1952
+ "step": 324
1953
+ },
1954
+ {
1955
+ "epoch": 0.58,
1956
+ "learning_rate": 8.063155475000037e-05,
1957
+ "loss": 0.0164,
1958
+ "step": 325
1959
+ },
1960
+ {
1961
+ "epoch": 0.58,
1962
+ "learning_rate": 8.006944853759732e-05,
1963
+ "loss": 0.0172,
1964
+ "step": 326
1965
+ },
1966
+ {
1967
+ "epoch": 0.58,
1968
+ "learning_rate": 7.950799734850292e-05,
1969
+ "loss": 0.0213,
1970
+ "step": 327
1971
+ },
1972
+ {
1973
+ "epoch": 0.58,
1974
+ "learning_rate": 7.894721963497213e-05,
1975
+ "loss": 0.022,
1976
+ "step": 328
1977
+ },
1978
+ {
1979
+ "epoch": 0.58,
1980
+ "learning_rate": 7.838713382712583e-05,
1981
+ "loss": 0.0187,
1982
+ "step": 329
1983
+ },
1984
+ {
1985
+ "epoch": 0.58,
1986
+ "learning_rate": 7.782775833234521e-05,
1987
+ "loss": 0.0189,
1988
+ "step": 330
1989
+ },
1990
+ {
1991
+ "epoch": 0.59,
1992
+ "learning_rate": 7.726911153466698e-05,
1993
+ "loss": 0.0252,
1994
+ "step": 331
1995
+ },
1996
+ {
1997
+ "epoch": 0.59,
1998
+ "learning_rate": 7.67112117941788e-05,
1999
+ "loss": 0.0176,
2000
+ "step": 332
2001
+ },
2002
+ {
2003
+ "epoch": 0.59,
2004
+ "learning_rate": 7.615407744641619e-05,
2005
+ "loss": 0.0206,
2006
+ "step": 333
2007
+ },
2008
+ {
2009
+ "epoch": 0.59,
2010
+ "learning_rate": 7.559772680175978e-05,
2011
+ "loss": 0.0179,
2012
+ "step": 334
2013
+ },
2014
+ {
2015
+ "epoch": 0.59,
2016
+ "learning_rate": 7.504217814483365e-05,
2017
+ "loss": 0.0179,
2018
+ "step": 335
2019
+ },
2020
+ {
2021
+ "epoch": 0.59,
2022
+ "learning_rate": 7.448744973390422e-05,
2023
+ "loss": 0.0303,
2024
+ "step": 336
2025
+ },
2026
+ {
2027
+ "epoch": 0.6,
2028
+ "learning_rate": 7.393355980028039e-05,
2029
+ "loss": 0.0192,
2030
+ "step": 337
2031
+ },
2032
+ {
2033
+ "epoch": 0.6,
2034
+ "learning_rate": 7.338052654771407e-05,
2035
+ "loss": 0.013,
2036
+ "step": 338
2037
+ },
2038
+ {
2039
+ "epoch": 0.6,
2040
+ "learning_rate": 7.282836815180241e-05,
2041
+ "loss": 0.0199,
2042
+ "step": 339
2043
+ },
2044
+ {
2045
+ "epoch": 0.6,
2046
+ "learning_rate": 7.227710275938986e-05,
2047
+ "loss": 0.019,
2048
+ "step": 340
2049
+ },
2050
+ {
2051
+ "epoch": 0.6,
2052
+ "learning_rate": 7.172674848797219e-05,
2053
+ "loss": 0.0142,
2054
+ "step": 341
2055
+ },
2056
+ {
2057
+ "epoch": 0.61,
2058
+ "learning_rate": 7.117732342510092e-05,
2059
+ "loss": 0.0143,
2060
+ "step": 342
2061
+ },
2062
+ {
2063
+ "epoch": 0.61,
2064
+ "learning_rate": 7.062884562778882e-05,
2065
+ "loss": 0.0147,
2066
+ "step": 343
2067
+ },
2068
+ {
2069
+ "epoch": 0.61,
2070
+ "learning_rate": 7.00813331219165e-05,
2071
+ "loss": 0.0133,
2072
+ "step": 344
2073
+ },
2074
+ {
2075
+ "epoch": 0.61,
2076
+ "learning_rate": 6.953480390164e-05,
2077
+ "loss": 0.0177,
2078
+ "step": 345
2079
+ },
2080
+ {
2081
+ "epoch": 0.61,
2082
+ "learning_rate": 6.898927592879945e-05,
2083
+ "loss": 0.0202,
2084
+ "step": 346
2085
+ },
2086
+ {
2087
+ "epoch": 0.61,
2088
+ "learning_rate": 6.844476713232862e-05,
2089
+ "loss": 0.0193,
2090
+ "step": 347
2091
+ },
2092
+ {
2093
+ "epoch": 0.62,
2094
+ "learning_rate": 6.790129540766582e-05,
2095
+ "loss": 0.0107,
2096
+ "step": 348
2097
+ },
2098
+ {
2099
+ "epoch": 0.62,
2100
+ "learning_rate": 6.735887861616556e-05,
2101
+ "loss": 0.0236,
2102
+ "step": 349
2103
+ },
2104
+ {
2105
+ "epoch": 0.62,
2106
+ "learning_rate": 6.681753458451189e-05,
2107
+ "loss": 0.0133,
2108
+ "step": 350
2109
+ },
2110
+ {
2111
+ "epoch": 0.62,
2112
+ "learning_rate": 6.627728110413213e-05,
2113
+ "loss": 0.0197,
2114
+ "step": 351
2115
+ },
2116
+ {
2117
+ "epoch": 0.62,
2118
+ "learning_rate": 6.573813593061236e-05,
2119
+ "loss": 0.0192,
2120
+ "step": 352
2121
+ },
2122
+ {
2123
+ "epoch": 0.62,
2124
+ "learning_rate": 6.520011678311382e-05,
2125
+ "loss": 0.0263,
2126
+ "step": 353
2127
+ },
2128
+ {
2129
+ "epoch": 0.63,
2130
+ "learning_rate": 6.466324134379066e-05,
2131
+ "loss": 0.0191,
2132
+ "step": 354
2133
+ },
2134
+ {
2135
+ "epoch": 0.63,
2136
+ "learning_rate": 6.412752725720864e-05,
2137
+ "loss": 0.0126,
2138
+ "step": 355
2139
+ },
2140
+ {
2141
+ "epoch": 0.63,
2142
+ "learning_rate": 6.359299212976534e-05,
2143
+ "loss": 0.0189,
2144
+ "step": 356
2145
+ },
2146
+ {
2147
+ "epoch": 0.63,
2148
+ "learning_rate": 6.305965352911161e-05,
2149
+ "loss": 0.0149,
2150
+ "step": 357
2151
+ },
2152
+ {
2153
+ "epoch": 0.63,
2154
+ "learning_rate": 6.252752898357396e-05,
2155
+ "loss": 0.018,
2156
+ "step": 358
2157
+ },
2158
+ {
2159
+ "epoch": 0.64,
2160
+ "learning_rate": 6.199663598157876e-05,
2161
+ "loss": 0.0198,
2162
+ "step": 359
2163
+ },
2164
+ {
2165
+ "epoch": 0.64,
2166
+ "learning_rate": 6.146699197107715e-05,
2167
+ "loss": 0.0112,
2168
+ "step": 360
2169
+ },
2170
+ {
2171
+ "epoch": 0.64,
2172
+ "learning_rate": 6.093861435897208e-05,
2173
+ "loss": 0.0139,
2174
+ "step": 361
2175
+ },
2176
+ {
2177
+ "epoch": 0.64,
2178
+ "learning_rate": 6.041152051054575e-05,
2179
+ "loss": 0.0173,
2180
+ "step": 362
2181
+ },
2182
+ {
2183
+ "epoch": 0.64,
2184
+ "learning_rate": 5.988572774888913e-05,
2185
+ "loss": 0.0135,
2186
+ "step": 363
2187
+ },
2188
+ {
2189
+ "epoch": 0.64,
2190
+ "learning_rate": 5.936125335433265e-05,
2191
+ "loss": 0.026,
2192
+ "step": 364
2193
+ },
2194
+ {
2195
+ "epoch": 0.65,
2196
+ "learning_rate": 5.883811456387821e-05,
2197
+ "loss": 0.0111,
2198
+ "step": 365
2199
+ },
2200
+ {
2201
+ "epoch": 0.65,
2202
+ "learning_rate": 5.8316328570632706e-05,
2203
+ "loss": 0.02,
2204
+ "step": 366
2205
+ },
2206
+ {
2207
+ "epoch": 0.65,
2208
+ "learning_rate": 5.779591252324286e-05,
2209
+ "loss": 0.0128,
2210
+ "step": 367
2211
+ },
2212
+ {
2213
+ "epoch": 0.65,
2214
+ "learning_rate": 5.7276883525331915e-05,
2215
+ "loss": 0.0109,
2216
+ "step": 368
2217
+ },
2218
+ {
2219
+ "epoch": 0.65,
2220
+ "learning_rate": 5.675925863493721e-05,
2221
+ "loss": 0.018,
2222
+ "step": 369
2223
+ },
2224
+ {
2225
+ "epoch": 0.65,
2226
+ "learning_rate": 5.624305486394967e-05,
2227
+ "loss": 0.0136,
2228
+ "step": 370
2229
+ },
2230
+ {
2231
+ "epoch": 0.66,
2232
+ "learning_rate": 5.5728289177554805e-05,
2233
+ "loss": 0.0206,
2234
+ "step": 371
2235
+ },
2236
+ {
2237
+ "epoch": 0.66,
2238
+ "learning_rate": 5.5214978493675005e-05,
2239
+ "loss": 0.0148,
2240
+ "step": 372
2241
+ },
2242
+ {
2243
+ "epoch": 0.66,
2244
+ "learning_rate": 5.4703139682413586e-05,
2245
+ "loss": 0.0184,
2246
+ "step": 373
2247
+ },
2248
+ {
2249
+ "epoch": 0.66,
2250
+ "learning_rate": 5.419278956550037e-05,
2251
+ "loss": 0.0197,
2252
+ "step": 374
2253
+ },
2254
+ {
2255
+ "epoch": 0.66,
2256
+ "learning_rate": 5.368394491573876e-05,
2257
+ "loss": 0.0148,
2258
+ "step": 375
2259
+ },
2260
+ {
2261
+ "epoch": 0.67,
2262
+ "learning_rate": 5.3176622456454693e-05,
2263
+ "loss": 0.015,
2264
+ "step": 376
2265
+ },
2266
+ {
2267
+ "epoch": 0.67,
2268
+ "learning_rate": 5.267083886094668e-05,
2269
+ "loss": 0.012,
2270
+ "step": 377
2271
+ },
2272
+ {
2273
+ "epoch": 0.67,
2274
+ "learning_rate": 5.216661075193814e-05,
2275
+ "loss": 0.0126,
2276
+ "step": 378
2277
+ },
2278
+ {
2279
+ "epoch": 0.67,
2280
+ "learning_rate": 5.1663954701030914e-05,
2281
+ "loss": 0.0096,
2282
+ "step": 379
2283
+ },
2284
+ {
2285
+ "epoch": 0.67,
2286
+ "learning_rate": 5.116288722816086e-05,
2287
+ "loss": 0.0118,
2288
+ "step": 380
2289
+ },
2290
+ {
2291
+ "epoch": 0.67,
2292
+ "learning_rate": 5.0663424801054595e-05,
2293
+ "loss": 0.025,
2294
+ "step": 381
2295
+ },
2296
+ {
2297
+ "epoch": 0.68,
2298
+ "learning_rate": 5.016558383468851e-05,
2299
+ "loss": 0.0134,
2300
+ "step": 382
2301
+ },
2302
+ {
2303
+ "epoch": 0.68,
2304
+ "learning_rate": 4.966938069074921e-05,
2305
+ "loss": 0.0137,
2306
+ "step": 383
2307
+ },
2308
+ {
2309
+ "epoch": 0.68,
2310
+ "learning_rate": 4.9174831677095804e-05,
2311
+ "loss": 0.022,
2312
+ "step": 384
2313
+ },
2314
+ {
2315
+ "epoch": 0.68,
2316
+ "learning_rate": 4.8681953047223914e-05,
2317
+ "loss": 0.0162,
2318
+ "step": 385
2319
+ },
2320
+ {
2321
+ "epoch": 0.68,
2322
+ "learning_rate": 4.8190760999731524e-05,
2323
+ "loss": 0.0147,
2324
+ "step": 386
2325
+ },
2326
+ {
2327
+ "epoch": 0.68,
2328
+ "learning_rate": 4.7701271677786695e-05,
2329
+ "loss": 0.0227,
2330
+ "step": 387
2331
+ },
2332
+ {
2333
+ "epoch": 0.69,
2334
+ "learning_rate": 4.721350116859675e-05,
2335
+ "loss": 0.0137,
2336
+ "step": 388
2337
+ },
2338
+ {
2339
+ "epoch": 0.69,
2340
+ "learning_rate": 4.672746550287985e-05,
2341
+ "loss": 0.0162,
2342
+ "step": 389
2343
+ },
2344
+ {
2345
+ "epoch": 0.69,
2346
+ "learning_rate": 4.624318065433797e-05,
2347
+ "loss": 0.0165,
2348
+ "step": 390
2349
+ },
2350
+ {
2351
+ "epoch": 0.69,
2352
+ "learning_rate": 4.576066253913208e-05,
2353
+ "loss": 0.0196,
2354
+ "step": 391
2355
+ },
2356
+ {
2357
+ "epoch": 0.69,
2358
+ "learning_rate": 4.5279927015358836e-05,
2359
+ "loss": 0.019,
2360
+ "step": 392
2361
+ },
2362
+ {
2363
+ "epoch": 0.7,
2364
+ "learning_rate": 4.4800989882529574e-05,
2365
+ "loss": 0.0148,
2366
+ "step": 393
2367
+ },
2368
+ {
2369
+ "epoch": 0.7,
2370
+ "learning_rate": 4.432386688105095e-05,
2371
+ "loss": 0.0112,
2372
+ "step": 394
2373
+ },
2374
+ {
2375
+ "epoch": 0.7,
2376
+ "learning_rate": 4.3848573691707717e-05,
2377
+ "loss": 0.0253,
2378
+ "step": 395
2379
+ },
2380
+ {
2381
+ "epoch": 0.7,
2382
+ "learning_rate": 4.337512593514729e-05,
2383
+ "loss": 0.0214,
2384
+ "step": 396
2385
+ },
2386
+ {
2387
+ "epoch": 0.7,
2388
+ "learning_rate": 4.2903539171366393e-05,
2389
+ "loss": 0.0209,
2390
+ "step": 397
2391
+ },
2392
+ {
2393
+ "epoch": 0.7,
2394
+ "learning_rate": 4.243382889919981e-05,
2395
+ "loss": 0.0118,
2396
+ "step": 398
2397
+ },
2398
+ {
2399
+ "epoch": 0.71,
2400
+ "learning_rate": 4.19660105558107e-05,
2401
+ "loss": 0.0192,
2402
+ "step": 399
2403
+ },
2404
+ {
2405
+ "epoch": 0.71,
2406
+ "learning_rate": 4.1500099516183555e-05,
2407
+ "loss": 0.014,
2408
+ "step": 400
2409
+ },
2410
+ {
2411
+ "epoch": 0.71,
2412
+ "learning_rate": 4.1036111092618725e-05,
2413
+ "loss": 0.0183,
2414
+ "step": 401
2415
+ },
2416
+ {
2417
+ "epoch": 0.71,
2418
+ "learning_rate": 4.057406053422933e-05,
2419
+ "loss": 0.0157,
2420
+ "step": 402
2421
+ },
2422
+ {
2423
+ "epoch": 0.71,
2424
+ "learning_rate": 4.0113963026439884e-05,
2425
+ "loss": 0.021,
2426
+ "step": 403
2427
+ },
2428
+ {
2429
+ "epoch": 0.72,
2430
+ "learning_rate": 3.965583369048738e-05,
2431
+ "loss": 0.0131,
2432
+ "step": 404
2433
+ },
2434
+ {
2435
+ "epoch": 0.72,
2436
+ "learning_rate": 3.919968758292425e-05,
2437
+ "loss": 0.0126,
2438
+ "step": 405
2439
+ },
2440
+ {
2441
+ "epoch": 0.72,
2442
+ "learning_rate": 3.8745539695123575e-05,
2443
+ "loss": 0.0087,
2444
+ "step": 406
2445
+ },
2446
+ {
2447
+ "epoch": 0.72,
2448
+ "learning_rate": 3.829340495278639e-05,
2449
+ "loss": 0.0103,
2450
+ "step": 407
2451
+ },
2452
+ {
2453
+ "epoch": 0.72,
2454
+ "learning_rate": 3.784329821545105e-05,
2455
+ "loss": 0.0132,
2456
+ "step": 408
2457
+ },
2458
+ {
2459
+ "epoch": 0.72,
2460
+ "learning_rate": 3.7395234276005087e-05,
2461
+ "loss": 0.0202,
2462
+ "step": 409
2463
+ },
2464
+ {
2465
+ "epoch": 0.73,
2466
+ "learning_rate": 3.694922786019871e-05,
2467
+ "loss": 0.013,
2468
+ "step": 410
2469
+ },
2470
+ {
2471
+ "epoch": 0.73,
2472
+ "learning_rate": 3.650529362616113e-05,
2473
+ "loss": 0.0117,
2474
+ "step": 411
2475
+ },
2476
+ {
2477
+ "epoch": 0.73,
2478
+ "learning_rate": 3.6063446163918667e-05,
2479
+ "loss": 0.0117,
2480
+ "step": 412
2481
+ },
2482
+ {
2483
+ "epoch": 0.73,
2484
+ "learning_rate": 3.562369999491536e-05,
2485
+ "loss": 0.0182,
2486
+ "step": 413
2487
+ },
2488
+ {
2489
+ "epoch": 0.73,
2490
+ "learning_rate": 3.5186069571535576e-05,
2491
+ "loss": 0.0148,
2492
+ "step": 414
2493
+ },
2494
+ {
2495
+ "epoch": 0.73,
2496
+ "learning_rate": 3.4750569276629117e-05,
2497
+ "loss": 0.0183,
2498
+ "step": 415
2499
+ },
2500
+ {
2501
+ "epoch": 0.74,
2502
+ "learning_rate": 3.431721342303839e-05,
2503
+ "loss": 0.0112,
2504
+ "step": 416
2505
+ },
2506
+ {
2507
+ "epoch": 0.74,
2508
+ "learning_rate": 3.3886016253128326e-05,
2509
+ "loss": 0.0117,
2510
+ "step": 417
2511
+ },
2512
+ {
2513
+ "epoch": 0.74,
2514
+ "learning_rate": 3.345699193831795e-05,
2515
+ "loss": 0.0168,
2516
+ "step": 418
2517
+ },
2518
+ {
2519
+ "epoch": 0.74,
2520
+ "learning_rate": 3.303015457861478e-05,
2521
+ "loss": 0.0221,
2522
+ "step": 419
2523
+ },
2524
+ {
2525
+ "epoch": 0.74,
2526
+ "learning_rate": 3.2605518202151576e-05,
2527
+ "loss": 0.0162,
2528
+ "step": 420
2529
+ },
2530
+ {
2531
+ "epoch": 0.75,
2532
+ "learning_rate": 3.2183096764724915e-05,
2533
+ "loss": 0.0133,
2534
+ "step": 421
2535
+ },
2536
+ {
2537
+ "epoch": 0.75,
2538
+ "learning_rate": 3.176290414933695e-05,
2539
+ "loss": 0.0149,
2540
+ "step": 422
2541
+ },
2542
+ {
2543
+ "epoch": 0.75,
2544
+ "learning_rate": 3.1344954165738835e-05,
2545
+ "loss": 0.0143,
2546
+ "step": 423
2547
+ },
2548
+ {
2549
+ "epoch": 0.75,
2550
+ "learning_rate": 3.092926054997711e-05,
2551
+ "loss": 0.0125,
2552
+ "step": 424
2553
+ },
2554
+ {
2555
+ "epoch": 0.75,
2556
+ "learning_rate": 3.0515836963942056e-05,
2557
+ "loss": 0.0139,
2558
+ "step": 425
2559
+ },
2560
+ {
2561
+ "epoch": 0.75,
2562
+ "learning_rate": 3.01046969949188e-05,
2563
+ "loss": 0.0216,
2564
+ "step": 426
2565
+ },
2566
+ {
2567
+ "epoch": 0.76,
2568
+ "learning_rate": 2.9695854155140646e-05,
2569
+ "loss": 0.0156,
2570
+ "step": 427
2571
+ },
2572
+ {
2573
+ "epoch": 0.76,
2574
+ "learning_rate": 2.9289321881345254e-05,
2575
+ "loss": 0.0206,
2576
+ "step": 428
2577
+ },
2578
+ {
2579
+ "epoch": 0.76,
2580
+ "learning_rate": 2.888511353433274e-05,
2581
+ "loss": 0.0227,
2582
+ "step": 429
2583
+ },
2584
+ {
2585
+ "epoch": 0.76,
2586
+ "learning_rate": 2.848324239852672e-05,
2587
+ "loss": 0.0111,
2588
+ "step": 430
2589
+ },
2590
+ {
2591
+ "epoch": 0.76,
2592
+ "learning_rate": 2.80837216815378e-05,
2593
+ "loss": 0.009,
2594
+ "step": 431
2595
+ },
2596
+ {
2597
+ "epoch": 0.76,
2598
+ "learning_rate": 2.7686564513729198e-05,
2599
+ "loss": 0.0145,
2600
+ "step": 432
2601
+ },
2602
+ {
2603
+ "epoch": 0.77,
2604
+ "learning_rate": 2.7291783947785543e-05,
2605
+ "loss": 0.0155,
2606
+ "step": 433
2607
+ },
2608
+ {
2609
+ "epoch": 0.77,
2610
+ "learning_rate": 2.6899392958283708e-05,
2611
+ "loss": 0.0145,
2612
+ "step": 434
2613
+ },
2614
+ {
2615
+ "epoch": 0.77,
2616
+ "learning_rate": 2.650940444126654e-05,
2617
+ "loss": 0.0204,
2618
+ "step": 435
2619
+ },
2620
+ {
2621
+ "epoch": 0.77,
2622
+ "learning_rate": 2.6121831213818827e-05,
2623
+ "loss": 0.0104,
2624
+ "step": 436
2625
+ },
2626
+ {
2627
+ "epoch": 0.77,
2628
+ "learning_rate": 2.5736686013646228e-05,
2629
+ "loss": 0.0186,
2630
+ "step": 437
2631
+ },
2632
+ {
2633
+ "epoch": 0.78,
2634
+ "learning_rate": 2.535398149865651e-05,
2635
+ "loss": 0.0201,
2636
+ "step": 438
2637
+ },
2638
+ {
2639
+ "epoch": 0.78,
2640
+ "learning_rate": 2.4973730246543735e-05,
2641
+ "loss": 0.0217,
2642
+ "step": 439
2643
+ },
2644
+ {
2645
+ "epoch": 0.78,
2646
+ "learning_rate": 2.4595944754374722e-05,
2647
+ "loss": 0.0227,
2648
+ "step": 440
2649
+ },
2650
+ {
2651
+ "epoch": 0.78,
2652
+ "learning_rate": 2.4220637438178317e-05,
2653
+ "loss": 0.0155,
2654
+ "step": 441
2655
+ },
2656
+ {
2657
+ "epoch": 0.78,
2658
+ "learning_rate": 2.3847820632537564e-05,
2659
+ "loss": 0.0139,
2660
+ "step": 442
2661
+ },
2662
+ {
2663
+ "epoch": 0.78,
2664
+ "learning_rate": 2.3477506590183972e-05,
2665
+ "loss": 0.0124,
2666
+ "step": 443
2667
+ },
2668
+ {
2669
+ "epoch": 0.79,
2670
+ "learning_rate": 2.310970748159511e-05,
2671
+ "loss": 0.0159,
2672
+ "step": 444
2673
+ },
2674
+ {
2675
+ "epoch": 0.79,
2676
+ "learning_rate": 2.2744435394594497e-05,
2677
+ "loss": 0.0132,
2678
+ "step": 445
2679
+ },
2680
+ {
2681
+ "epoch": 0.79,
2682
+ "learning_rate": 2.2381702333954434e-05,
2683
+ "loss": 0.0073,
2684
+ "step": 446
2685
+ },
2686
+ {
2687
+ "epoch": 0.79,
2688
+ "learning_rate": 2.20215202210013e-05,
2689
+ "loss": 0.0177,
2690
+ "step": 447
2691
+ },
2692
+ {
2693
+ "epoch": 0.79,
2694
+ "learning_rate": 2.1663900893223898e-05,
2695
+ "loss": 0.0151,
2696
+ "step": 448
2697
+ },
2698
+ {
2699
+ "epoch": 0.79,
2700
+ "learning_rate": 2.130885610388428e-05,
2701
+ "loss": 0.0082,
2702
+ "step": 449
2703
+ },
2704
+ {
2705
+ "epoch": 0.8,
2706
+ "learning_rate": 2.0956397521631664e-05,
2707
+ "loss": 0.0129,
2708
+ "step": 450
2709
+ },
2710
+ {
2711
+ "epoch": 0.8,
2712
+ "learning_rate": 2.0606536730118763e-05,
2713
+ "loss": 0.0124,
2714
+ "step": 451
2715
+ },
2716
+ {
2717
+ "epoch": 0.8,
2718
+ "learning_rate": 2.025928522762115e-05,
2719
+ "loss": 0.0109,
2720
+ "step": 452
2721
+ },
2722
+ {
2723
+ "epoch": 0.8,
2724
+ "learning_rate": 1.9914654426659374e-05,
2725
+ "loss": 0.0138,
2726
+ "step": 453
2727
+ },
2728
+ {
2729
+ "epoch": 0.8,
2730
+ "learning_rate": 1.9572655653623884e-05,
2731
+ "loss": 0.0096,
2732
+ "step": 454
2733
+ },
2734
+ {
2735
+ "epoch": 0.81,
2736
+ "learning_rate": 1.9233300148402765e-05,
2737
+ "loss": 0.0108,
2738
+ "step": 455
2739
+ },
2740
+ {
2741
+ "epoch": 0.81,
2742
+ "learning_rate": 1.8896599064012298e-05,
2743
+ "loss": 0.0099,
2744
+ "step": 456
2745
+ },
2746
+ {
2747
+ "epoch": 0.81,
2748
+ "learning_rate": 1.8562563466230576e-05,
2749
+ "loss": 0.0103,
2750
+ "step": 457
2751
+ },
2752
+ {
2753
+ "epoch": 0.81,
2754
+ "learning_rate": 1.823120433323361e-05,
2755
+ "loss": 0.0119,
2756
+ "step": 458
2757
+ },
2758
+ {
2759
+ "epoch": 0.81,
2760
+ "learning_rate": 1.790253255523465e-05,
2761
+ "loss": 0.0117,
2762
+ "step": 459
2763
+ },
2764
+ {
2765
+ "epoch": 0.81,
2766
+ "learning_rate": 1.757655893412622e-05,
2767
+ "loss": 0.0161,
2768
+ "step": 460
2769
+ },
2770
+ {
2771
+ "epoch": 0.82,
2772
+ "learning_rate": 1.7253294183125223e-05,
2773
+ "loss": 0.0183,
2774
+ "step": 461
2775
+ },
2776
+ {
2777
+ "epoch": 0.82,
2778
+ "learning_rate": 1.6932748926420693e-05,
2779
+ "loss": 0.0084,
2780
+ "step": 462
2781
+ },
2782
+ {
2783
+ "epoch": 0.82,
2784
+ "learning_rate": 1.6614933698824732e-05,
2785
+ "loss": 0.0153,
2786
+ "step": 463
2787
+ },
2788
+ {
2789
+ "epoch": 0.82,
2790
+ "learning_rate": 1.629985894542625e-05,
2791
+ "loss": 0.0179,
2792
+ "step": 464
2793
+ },
2794
+ {
2795
+ "epoch": 0.82,
2796
+ "learning_rate": 1.5987535021247667e-05,
2797
+ "loss": 0.0144,
2798
+ "step": 465
2799
+ },
2800
+ {
2801
+ "epoch": 0.82,
2802
+ "learning_rate": 1.5677972190904622e-05,
2803
+ "loss": 0.0178,
2804
+ "step": 466
2805
+ },
2806
+ {
2807
+ "epoch": 0.83,
2808
+ "learning_rate": 1.5371180628268587e-05,
2809
+ "loss": 0.0155,
2810
+ "step": 467
2811
+ },
2812
+ {
2813
+ "epoch": 0.83,
2814
+ "learning_rate": 1.5067170416132603e-05,
2815
+ "loss": 0.0163,
2816
+ "step": 468
2817
+ },
2818
+ {
2819
+ "epoch": 0.83,
2820
+ "learning_rate": 1.476595154587973e-05,
2821
+ "loss": 0.0159,
2822
+ "step": 469
2823
+ },
2824
+ {
2825
+ "epoch": 0.83,
2826
+ "learning_rate": 1.4467533917154841e-05,
2827
+ "loss": 0.0124,
2828
+ "step": 470
2829
+ },
2830
+ {
2831
+ "epoch": 0.83,
2832
+ "learning_rate": 1.4171927337539104e-05,
2833
+ "loss": 0.0171,
2834
+ "step": 471
2835
+ },
2836
+ {
2837
+ "epoch": 0.84,
2838
+ "learning_rate": 1.3879141522227878e-05,
2839
+ "loss": 0.0154,
2840
+ "step": 472
2841
+ },
2842
+ {
2843
+ "epoch": 0.84,
2844
+ "learning_rate": 1.3589186093711226e-05,
2845
+ "loss": 0.0088,
2846
+ "step": 473
2847
+ },
2848
+ {
2849
+ "epoch": 0.84,
2850
+ "learning_rate": 1.3302070581457715e-05,
2851
+ "loss": 0.0131,
2852
+ "step": 474
2853
+ },
2854
+ {
2855
+ "epoch": 0.84,
2856
+ "learning_rate": 1.3017804421601298e-05,
2857
+ "loss": 0.0114,
2858
+ "step": 475
2859
+ },
2860
+ {
2861
+ "epoch": 0.84,
2862
+ "learning_rate": 1.273639695663108e-05,
2863
+ "loss": 0.0197,
2864
+ "step": 476
2865
+ },
2866
+ {
2867
+ "epoch": 0.84,
2868
+ "learning_rate": 1.2457857435084408e-05,
2869
+ "loss": 0.0105,
2870
+ "step": 477
2871
+ },
2872
+ {
2873
+ "epoch": 0.85,
2874
+ "learning_rate": 1.2182195011242748e-05,
2875
+ "loss": 0.0066,
2876
+ "step": 478
2877
+ },
2878
+ {
2879
+ "epoch": 0.85,
2880
+ "learning_rate": 1.1909418744831047e-05,
2881
+ "loss": 0.0141,
2882
+ "step": 479
2883
+ },
2884
+ {
2885
+ "epoch": 0.85,
2886
+ "learning_rate": 1.1639537600719763e-05,
2887
+ "loss": 0.0146,
2888
+ "step": 480
2889
+ },
2890
+ {
2891
+ "epoch": 0.85,
2892
+ "learning_rate": 1.1372560448630376e-05,
2893
+ "loss": 0.0113,
2894
+ "step": 481
2895
+ },
2896
+ {
2897
+ "epoch": 0.85,
2898
+ "learning_rate": 1.1108496062843742e-05,
2899
+ "loss": 0.0124,
2900
+ "step": 482
2901
+ },
2902
+ {
2903
+ "epoch": 0.85,
2904
+ "learning_rate": 1.0847353121911951e-05,
2905
+ "loss": 0.0126,
2906
+ "step": 483
2907
+ },
2908
+ {
2909
+ "epoch": 0.86,
2910
+ "learning_rate": 1.0589140208372873e-05,
2911
+ "loss": 0.0082,
2912
+ "step": 484
2913
+ },
2914
+ {
2915
+ "epoch": 0.86,
2916
+ "learning_rate": 1.0333865808468202e-05,
2917
+ "loss": 0.0182,
2918
+ "step": 485
2919
+ },
2920
+ {
2921
+ "epoch": 0.86,
2922
+ "learning_rate": 1.0081538311864569e-05,
2923
+ "loss": 0.0144,
2924
+ "step": 486
2925
+ },
2926
+ {
2927
+ "epoch": 0.86,
2928
+ "learning_rate": 9.83216601137773e-06,
2929
+ "loss": 0.0135,
2930
+ "step": 487
2931
+ },
2932
+ {
2933
+ "epoch": 0.86,
2934
+ "learning_rate": 9.58575710270011e-06,
2935
+ "loss": 0.0116,
2936
+ "step": 488
2937
+ },
2938
+ {
2939
+ "epoch": 0.87,
2940
+ "learning_rate": 9.342319684131395e-06,
2941
+ "loss": 0.0135,
2942
+ "step": 489
2943
+ },
2944
+ {
2945
+ "epoch": 0.87,
2946
+ "learning_rate": 9.10186175631237e-06,
2947
+ "loss": 0.0115,
2948
+ "step": 490
2949
+ },
2950
+ {
2951
+ "epoch": 0.87,
2952
+ "learning_rate": 8.864391221962065e-06,
2953
+ "loss": 0.0139,
2954
+ "step": 491
2955
+ },
2956
+ {
2957
+ "epoch": 0.87,
2958
+ "learning_rate": 8.629915885617911e-06,
2959
+ "loss": 0.0155,
2960
+ "step": 492
2961
+ },
2962
+ {
2963
+ "epoch": 0.87,
2964
+ "learning_rate": 8.398443453379267e-06,
2965
+ "loss": 0.0107,
2966
+ "step": 493
2967
+ },
2968
+ {
2969
+ "epoch": 0.87,
2970
+ "learning_rate": 8.16998153265427e-06,
2971
+ "loss": 0.0113,
2972
+ "step": 494
2973
+ },
2974
+ {
2975
+ "epoch": 0.88,
2976
+ "learning_rate": 7.944537631909665e-06,
2977
+ "loss": 0.0077,
2978
+ "step": 495
2979
+ },
2980
+ {
2981
+ "epoch": 0.88,
2982
+ "learning_rate": 7.722119160424112e-06,
2983
+ "loss": 0.011,
2984
+ "step": 496
2985
+ },
2986
+ {
2987
+ "epoch": 0.88,
2988
+ "learning_rate": 7.502733428044683e-06,
2989
+ "loss": 0.01,
2990
+ "step": 497
2991
+ },
2992
+ {
2993
+ "epoch": 0.88,
2994
+ "learning_rate": 7.2863876449466016e-06,
2995
+ "loss": 0.0094,
2996
+ "step": 498
2997
+ },
2998
+ {
2999
+ "epoch": 0.88,
3000
+ "learning_rate": 7.073088921396287e-06,
3001
+ "loss": 0.0086,
3002
+ "step": 499
3003
+ },
3004
+ {
3005
+ "epoch": 0.88,
3006
+ "learning_rate": 6.862844267517643e-06,
3007
+ "loss": 0.0123,
3008
+ "step": 500
3009
+ },
3010
+ {
3011
+ "epoch": 0.89,
3012
+ "learning_rate": 6.655660593061719e-06,
3013
+ "loss": 0.015,
3014
+ "step": 501
3015
+ },
3016
+ {
3017
+ "epoch": 0.89,
3018
+ "learning_rate": 6.451544707179635e-06,
3019
+ "loss": 0.0076,
3020
+ "step": 502
3021
+ },
3022
+ {
3023
+ "epoch": 0.89,
3024
+ "learning_rate": 6.250503318198664e-06,
3025
+ "loss": 0.0065,
3026
+ "step": 503
3027
+ },
3028
+ {
3029
+ "epoch": 0.89,
3030
+ "learning_rate": 6.0525430334018915e-06,
3031
+ "loss": 0.0161,
3032
+ "step": 504
3033
+ },
3034
+ {
3035
+ "epoch": 0.89,
3036
+ "learning_rate": 5.857670358811096e-06,
3037
+ "loss": 0.0113,
3038
+ "step": 505
3039
+ },
3040
+ {
3041
+ "epoch": 0.9,
3042
+ "learning_rate": 5.6658916989727695e-06,
3043
+ "loss": 0.0117,
3044
+ "step": 506
3045
+ },
3046
+ {
3047
+ "epoch": 0.9,
3048
+ "learning_rate": 5.477213356747746e-06,
3049
+ "loss": 0.0119,
3050
+ "step": 507
3051
+ },
3052
+ {
3053
+ "epoch": 0.9,
3054
+ "learning_rate": 5.291641533104052e-06,
3055
+ "loss": 0.0127,
3056
+ "step": 508
3057
+ },
3058
+ {
3059
+ "epoch": 0.9,
3060
+ "learning_rate": 5.109182326913054e-06,
3061
+ "loss": 0.0143,
3062
+ "step": 509
3063
+ },
3064
+ {
3065
+ "epoch": 0.9,
3066
+ "learning_rate": 4.929841734749063e-06,
3067
+ "loss": 0.0154,
3068
+ "step": 510
3069
+ },
3070
+ {
3071
+ "epoch": 0.9,
3072
+ "learning_rate": 4.7536256506922506e-06,
3073
+ "loss": 0.0125,
3074
+ "step": 511
3075
+ },
3076
+ {
3077
+ "epoch": 0.91,
3078
+ "learning_rate": 4.5805398661349146e-06,
3079
+ "loss": 0.0134,
3080
+ "step": 512
3081
+ },
3082
+ {
3083
+ "epoch": 0.91,
3084
+ "learning_rate": 4.410590069591192e-06,
3085
+ "loss": 0.0139,
3086
+ "step": 513
3087
+ },
3088
+ {
3089
+ "epoch": 0.91,
3090
+ "learning_rate": 4.243781846510031e-06,
3091
+ "loss": 0.0097,
3092
+ "step": 514
3093
+ },
3094
+ {
3095
+ "epoch": 0.91,
3096
+ "learning_rate": 4.0801206790916814e-06,
3097
+ "loss": 0.0128,
3098
+ "step": 515
3099
+ },
3100
+ {
3101
+ "epoch": 0.91,
3102
+ "learning_rate": 3.919611946107493e-06,
3103
+ "loss": 0.02,
3104
+ "step": 516
3105
+ },
3106
+ {
3107
+ "epoch": 0.92,
3108
+ "learning_rate": 3.7622609227231818e-06,
3109
+ "loss": 0.0146,
3110
+ "step": 517
3111
+ },
3112
+ {
3113
+ "epoch": 0.92,
3114
+ "learning_rate": 3.6080727803254e-06,
3115
+ "loss": 0.0156,
3116
+ "step": 518
3117
+ },
3118
+ {
3119
+ "epoch": 0.92,
3120
+ "learning_rate": 3.4570525863518165e-06,
3121
+ "loss": 0.014,
3122
+ "step": 519
3123
+ },
3124
+ {
3125
+ "epoch": 0.92,
3126
+ "learning_rate": 3.3092053041245517e-06,
3127
+ "loss": 0.0143,
3128
+ "step": 520
3129
+ },
3130
+ {
3131
+ "epoch": 0.92,
3132
+ "learning_rate": 3.1645357926870955e-06,
3133
+ "loss": 0.0118,
3134
+ "step": 521
3135
+ },
3136
+ {
3137
+ "epoch": 0.92,
3138
+ "learning_rate": 3.0230488066445462e-06,
3139
+ "loss": 0.017,
3140
+ "step": 522
3141
+ },
3142
+ {
3143
+ "epoch": 0.93,
3144
+ "learning_rate": 2.8847489960074136e-06,
3145
+ "loss": 0.018,
3146
+ "step": 523
3147
+ },
3148
+ {
3149
+ "epoch": 0.93,
3150
+ "learning_rate": 2.7496409060387973e-06,
3151
+ "loss": 0.0227,
3152
+ "step": 524
3153
+ },
3154
+ {
3155
+ "epoch": 0.93,
3156
+ "learning_rate": 2.6177289771049274e-06,
3157
+ "loss": 0.0089,
3158
+ "step": 525
3159
+ },
3160
+ {
3161
+ "epoch": 0.93,
3162
+ "learning_rate": 2.489017544529315e-06,
3163
+ "loss": 0.0095,
3164
+ "step": 526
3165
+ },
3166
+ {
3167
+ "epoch": 0.93,
3168
+ "learning_rate": 2.3635108384502e-06,
3169
+ "loss": 0.0181,
3170
+ "step": 527
3171
+ },
3172
+ {
3173
+ "epoch": 0.93,
3174
+ "learning_rate": 2.2412129836816286e-06,
3175
+ "loss": 0.0139,
3176
+ "step": 528
3177
+ },
3178
+ {
3179
+ "epoch": 0.94,
3180
+ "learning_rate": 2.122127999577783e-06,
3181
+ "loss": 0.0115,
3182
+ "step": 529
3183
+ },
3184
+ {
3185
+ "epoch": 0.94,
3186
+ "learning_rate": 2.0062597999009116e-06,
3187
+ "loss": 0.0233,
3188
+ "step": 530
3189
+ },
3190
+ {
3191
+ "epoch": 0.94,
3192
+ "learning_rate": 1.893612192692751e-06,
3193
+ "loss": 0.015,
3194
+ "step": 531
3195
+ },
3196
+ {
3197
+ "epoch": 0.94,
3198
+ "learning_rate": 1.7841888801493178e-06,
3199
+ "loss": 0.011,
3200
+ "step": 532
3201
+ },
3202
+ {
3203
+ "epoch": 0.94,
3204
+ "learning_rate": 1.6779934584992718e-06,
3205
+ "loss": 0.0087,
3206
+ "step": 533
3207
+ },
3208
+ {
3209
+ "epoch": 0.95,
3210
+ "learning_rate": 1.5750294178856873e-06,
3211
+ "loss": 0.0078,
3212
+ "step": 534
3213
+ },
3214
+ {
3215
+ "epoch": 0.95,
3216
+ "learning_rate": 1.4753001422514123e-06,
3217
+ "loss": 0.0113,
3218
+ "step": 535
3219
+ },
3220
+ {
3221
+ "epoch": 0.95,
3222
+ "learning_rate": 1.378808909227769e-06,
3223
+ "loss": 0.006,
3224
+ "step": 536
3225
+ },
3226
+ {
3227
+ "epoch": 0.95,
3228
+ "learning_rate": 1.2855588900269056e-06,
3229
+ "loss": 0.008,
3230
+ "step": 537
3231
+ },
3232
+ {
3233
+ "epoch": 0.95,
3234
+ "learning_rate": 1.1955531493375139e-06,
3235
+ "loss": 0.0093,
3236
+ "step": 538
3237
+ },
3238
+ {
3239
+ "epoch": 0.95,
3240
+ "learning_rate": 1.108794645224187e-06,
3241
+ "loss": 0.0124,
3242
+ "step": 539
3243
+ },
3244
+ {
3245
+ "epoch": 0.96,
3246
+ "learning_rate": 1.0252862290301091e-06,
3247
+ "loss": 0.0084,
3248
+ "step": 540
3249
+ },
3250
+ {
3251
+ "epoch": 0.96,
3252
+ "learning_rate": 9.450306452834179e-07,
3253
+ "loss": 0.0051,
3254
+ "step": 541
3255
+ },
3256
+ {
3257
+ "epoch": 0.96,
3258
+ "learning_rate": 8.68030531606967e-07,
3259
+ "loss": 0.0135,
3260
+ "step": 542
3261
+ },
3262
+ {
3263
+ "epoch": 0.96,
3264
+ "learning_rate": 7.94288418631639e-07,
3265
+ "loss": 0.0117,
3266
+ "step": 543
3267
+ },
3268
+ {
3269
+ "epoch": 0.96,
3270
+ "learning_rate": 7.2380672991319e-07,
3271
+ "loss": 0.0146,
3272
+ "step": 544
3273
+ },
3274
+ {
3275
+ "epoch": 0.96,
3276
+ "learning_rate": 6.565877818526245e-07,
3277
+ "loss": 0.0152,
3278
+ "step": 545
3279
+ },
3280
+ {
3281
+ "epoch": 0.97,
3282
+ "learning_rate": 5.926337836199891e-07,
3283
+ "loss": 0.0105,
3284
+ "step": 546
3285
+ },
3286
+ {
3287
+ "epoch": 0.97,
3288
+ "learning_rate": 5.319468370818537e-07,
3289
+ "loss": 0.013,
3290
+ "step": 547
3291
+ },
3292
+ {
3293
+ "epoch": 0.97,
3294
+ "learning_rate": 4.745289367321659e-07,
3295
+ "loss": 0.0097,
3296
+ "step": 548
3297
+ },
3298
+ {
3299
+ "epoch": 0.97,
3300
+ "learning_rate": 4.203819696267486e-07,
3301
+ "loss": 0.0101,
3302
+ "step": 549
3303
+ },
3304
+ {
3305
+ "epoch": 0.97,
3306
+ "learning_rate": 3.6950771532126006e-07,
3307
+ "loss": 0.013,
3308
+ "step": 550
3309
+ },
3310
+ {
3311
+ "epoch": 0.98,
3312
+ "learning_rate": 3.2190784581270786e-07,
3313
+ "loss": 0.017,
3314
+ "step": 551
3315
+ },
3316
+ {
3317
+ "epoch": 0.98,
3318
+ "learning_rate": 2.7758392548449255e-07,
3319
+ "loss": 0.0133,
3320
+ "step": 552
3321
+ },
3322
+ {
3323
+ "epoch": 0.98,
3324
+ "learning_rate": 2.3653741105499338e-07,
3325
+ "loss": 0.012,
3326
+ "step": 553
3327
+ },
3328
+ {
3329
+ "epoch": 0.98,
3330
+ "learning_rate": 1.9876965152975103e-07,
3331
+ "loss": 0.0198,
3332
+ "step": 554
3333
+ },
3334
+ {
3335
+ "epoch": 0.98,
3336
+ "learning_rate": 1.6428188815703626e-07,
3337
+ "loss": 0.0119,
3338
+ "step": 555
3339
+ },
3340
+ {
3341
+ "epoch": 0.98,
3342
+ "learning_rate": 1.330752543871161e-07,
3343
+ "loss": 0.0206,
3344
+ "step": 556
3345
+ },
3346
+ {
3347
+ "epoch": 0.99,
3348
+ "learning_rate": 1.0515077583498344e-07,
3349
+ "loss": 0.0107,
3350
+ "step": 557
3351
+ },
3352
+ {
3353
+ "epoch": 0.99,
3354
+ "learning_rate": 8.050937024666194e-08,
3355
+ "loss": 0.0089,
3356
+ "step": 558
3357
+ },
3358
+ {
3359
+ "epoch": 0.99,
3360
+ "learning_rate": 5.915184746904112e-08,
3361
+ "loss": 0.0109,
3362
+ "step": 559
3363
+ },
3364
+ {
3365
+ "epoch": 0.99,
3366
+ "learning_rate": 4.107890942325332e-08,
3367
+ "loss": 0.0123,
3368
+ "step": 560
3369
+ },
3370
+ {
3371
+ "epoch": 0.99,
3372
+ "learning_rate": 2.6291150081603212e-08,
3373
+ "loss": 0.0166,
3374
+ "step": 561
3375
+ },
3376
+ {
3377
+ "epoch": 0.99,
3378
+ "learning_rate": 1.4789055448061196e-08,
3379
+ "loss": 0.0119,
3380
+ "step": 562
3381
+ },
3382
+ {
3383
+ "epoch": 1.0,
3384
+ "learning_rate": 6.57300354227619e-09,
3385
+ "loss": 0.0115,
3386
+ "step": 563
3387
+ },
3388
+ {
3389
+ "epoch": 1.0,
3390
+ "learning_rate": 1.6432643871633347e-09,
3391
+ "loss": 0.0176,
3392
+ "step": 564
3393
+ },
3394
+ {
3395
+ "epoch": 1.0,
3396
+ "learning_rate": 0.0,
3397
+ "loss": 0.0149,
3398
+ "step": 565
3399
+ },
3400
+ {
3401
+ "epoch": 1.0,
3402
+ "step": 565,
3403
+ "total_flos": 8.494954981915361e+17,
3404
+ "train_loss": 0.02618465211934748,
3405
+ "train_runtime": 7315.8291,
3406
+ "train_samples_per_second": 1.234,
3407
+ "train_steps_per_second": 0.077
3408
+ }
3409
+ ],
3410
+ "max_steps": 565,
3411
+ "num_train_epochs": 1,
3412
+ "total_flos": 8.494954981915361e+17,
3413
+ "trial_name": null,
3414
+ "trial_params": null
3415
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+
9
+ - PEFT 0.4.0
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "lmsys/vicuna-7b-v1.5",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 256,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 128,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "up_proj",
18
+ "v_proj",
19
+ "q_proj",
20
+ "gate_proj",
21
+ "down_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8d9b1ba29ec50e1325e6b1b294e0d7ba1477736271484242141d07ae603f392
3
+ size 639787082
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "freeze_mm_mlp_adapter": true,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "image_aspect_ratio": "pad",
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
14
+ "max_position_embeddings": 4096,
15
+ "mm_hidden_size": 1024,
16
+ "mm_image_tower": "LanguageBind/LanguageBind_Image",
17
+ "mm_projector_lr": 2e-05,
18
+ "mm_projector_type": "mlp2x_gelu",
19
+ "mm_use_im_patch_token": false,
20
+ "mm_use_im_start_end": false,
21
+ "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
22
+ "mm_vision_select_feature": "patch",
23
+ "mm_vision_select_layer": -2,
24
+ "model_type": "llava",
25
+ "num_attention_heads": 32,
26
+ "num_hidden_layers": 32,
27
+ "num_key_value_heads": 32,
28
+ "pad_token_id": 0,
29
+ "pretraining_tp": 1,
30
+ "rms_norm_eps": 1e-05,
31
+ "rope_scaling": null,
32
+ "tie_word_embeddings": false,
33
+ "tokenizer_model_max_length": 4000,
34
+ "tokenizer_padding_side": "right",
35
+ "torch_dtype": "float16",
36
+ "transformers_version": "4.31.0",
37
+ "tune_mm_mlp_adapter": false,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "vocab_size": 32000
41
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/non_lora_trainables.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d55f63a1513cd2df27c7811d07f7468f2d1ed3bf88fa85b602b0aea1a63f015
3
+ size 248950952
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/runs/Sep22_02-41-32_evc28/events.out.tfevents.1726987448.evc28 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d0e1a36d160581215b9452d6f51b6eebc9363c83457d08afa86e2a162db9905
3
+ size 299825
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/tokenizer_config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "<s>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": false,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "legacy": false,
22
+ "model_max_length": 4000,
23
+ "pad_token": null,
24
+ "padding_side": "right",
25
+ "sp_model_kwargs": {},
26
+ "tokenizer_class": "LlamaTokenizer",
27
+ "unk_token": {
28
+ "__type": "AddedToken",
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p05_e05/videollava-7b-lora/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/README.md ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+ - PEFT 0.4.0
9
+
10
+ - PEFT 0.4.0
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "lmsys/vicuna-7b-v1.5",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 256,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 128,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "down_proj",
18
+ "o_proj",
19
+ "v_proj",
20
+ "k_proj",
21
+ "up_proj",
22
+ "q_proj",
23
+ "gate_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55903822d8f61b66fcf18bb23d760430f031a199c45a6c72708158a8738c4877
3
+ size 639787082
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "freeze_mm_mlp_adapter": false,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "image_aspect_ratio": "pad",
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
14
+ "max_position_embeddings": 4096,
15
+ "mm_hidden_size": 1024,
16
+ "mm_image_tower": "LanguageBind/LanguageBind_Image",
17
+ "mm_projector_lr": 2e-05,
18
+ "mm_projector_type": "mlp2x_gelu",
19
+ "mm_use_im_patch_token": false,
20
+ "mm_use_im_start_end": false,
21
+ "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
22
+ "mm_vision_select_feature": "patch",
23
+ "mm_vision_select_layer": -2,
24
+ "model_type": "llava",
25
+ "num_attention_heads": 32,
26
+ "num_hidden_layers": 32,
27
+ "num_key_value_heads": 32,
28
+ "pad_token_id": 0,
29
+ "pretraining_tp": 1,
30
+ "rms_norm_eps": 1e-05,
31
+ "rope_scaling": null,
32
+ "tie_word_embeddings": false,
33
+ "tokenizer_model_max_length": 3072,
34
+ "tokenizer_padding_side": "right",
35
+ "torch_dtype": "float16",
36
+ "transformers_version": "4.31.0",
37
+ "tune_mm_mlp_adapter": false,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "vocab_size": 32000
41
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/non_lora_trainables.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c08cda1027c5f8cdc3046496469a70c52c95ac9fbb0a21a58b9befcaa4dac184
3
+ size 41961648
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/runs/Sep22_13-40-39_evc28/events.out.tfevents.1727027010.evc28 ADDED
File without changes
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/runs/Sep22_13-46-32_evc28/events.out.tfevents.1727027315.evc28 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb9cfcfc3aff898285f32a124277164fc8ed2a97b19e4c26dce3759e848c437a
3
+ size 94155
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/runs/Sep22_23-43-22_evc45/events.out.tfevents.1727063201.evc45 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11149d9fe528004e54321d3f3e620f387946e841b504549d02e533d103ada176
3
+ size 94126
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/tokenizer_config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "<s>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": false,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "legacy": false,
22
+ "model_max_length": 2048,
23
+ "pad_token": null,
24
+ "padding_side": "right",
25
+ "sp_model_kwargs": {},
26
+ "tokenizer_class": "LlamaTokenizer",
27
+ "unk_token": {
28
+ "__type": "AddedToken",
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_1_withsysmsg_p09_e01/videollava-7b-lora/trainer_state.json ADDED
@@ -0,0 +1,3415 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "global_step": 565,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 1.1764705882352942e-05,
13
+ "loss": 0.6361,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 2.3529411764705884e-05,
19
+ "loss": 0.7074,
20
+ "step": 2
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 3.529411764705883e-05,
25
+ "loss": 0.6856,
26
+ "step": 3
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 4.705882352941177e-05,
31
+ "loss": 0.5011,
32
+ "step": 4
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 5.882352941176471e-05,
37
+ "loss": 0.424,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 7.058823529411765e-05,
43
+ "loss": 0.3363,
44
+ "step": 6
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 8.23529411764706e-05,
49
+ "loss": 0.2835,
50
+ "step": 7
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "learning_rate": 9.411764705882353e-05,
55
+ "loss": 0.2229,
56
+ "step": 8
57
+ },
58
+ {
59
+ "epoch": 0.02,
60
+ "learning_rate": 0.00010588235294117647,
61
+ "loss": 0.212,
62
+ "step": 9
63
+ },
64
+ {
65
+ "epoch": 0.02,
66
+ "learning_rate": 0.00011764705882352942,
67
+ "loss": 0.1655,
68
+ "step": 10
69
+ },
70
+ {
71
+ "epoch": 0.02,
72
+ "learning_rate": 0.00012941176470588237,
73
+ "loss": 0.147,
74
+ "step": 11
75
+ },
76
+ {
77
+ "epoch": 0.02,
78
+ "learning_rate": 0.0001411764705882353,
79
+ "loss": 0.16,
80
+ "step": 12
81
+ },
82
+ {
83
+ "epoch": 0.02,
84
+ "learning_rate": 0.00015294117647058822,
85
+ "loss": 0.1546,
86
+ "step": 13
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "learning_rate": 0.0001647058823529412,
91
+ "loss": 0.1501,
92
+ "step": 14
93
+ },
94
+ {
95
+ "epoch": 0.03,
96
+ "learning_rate": 0.00017647058823529413,
97
+ "loss": 0.0948,
98
+ "step": 15
99
+ },
100
+ {
101
+ "epoch": 0.03,
102
+ "learning_rate": 0.00018823529411764707,
103
+ "loss": 0.1389,
104
+ "step": 16
105
+ },
106
+ {
107
+ "epoch": 0.03,
108
+ "learning_rate": 0.0002,
109
+ "loss": 0.1365,
110
+ "step": 17
111
+ },
112
+ {
113
+ "epoch": 0.03,
114
+ "learning_rate": 0.00019999835673561284,
115
+ "loss": 0.1414,
116
+ "step": 18
117
+ },
118
+ {
119
+ "epoch": 0.03,
120
+ "learning_rate": 0.00019999342699645773,
121
+ "loss": 0.0967,
122
+ "step": 19
123
+ },
124
+ {
125
+ "epoch": 0.04,
126
+ "learning_rate": 0.00019998521094455196,
127
+ "loss": 0.1107,
128
+ "step": 20
129
+ },
130
+ {
131
+ "epoch": 0.04,
132
+ "learning_rate": 0.00019997370884991842,
133
+ "loss": 0.1022,
134
+ "step": 21
135
+ },
136
+ {
137
+ "epoch": 0.04,
138
+ "learning_rate": 0.00019995892109057675,
139
+ "loss": 0.1147,
140
+ "step": 22
141
+ },
142
+ {
143
+ "epoch": 0.04,
144
+ "learning_rate": 0.00019994084815253097,
145
+ "loss": 0.1245,
146
+ "step": 23
147
+ },
148
+ {
149
+ "epoch": 0.04,
150
+ "learning_rate": 0.00019991949062975336,
151
+ "loss": 0.106,
152
+ "step": 24
153
+ },
154
+ {
155
+ "epoch": 0.04,
156
+ "learning_rate": 0.00019989484922416502,
157
+ "loss": 0.1095,
158
+ "step": 25
159
+ },
160
+ {
161
+ "epoch": 0.05,
162
+ "learning_rate": 0.00019986692474561292,
163
+ "loss": 0.1205,
164
+ "step": 26
165
+ },
166
+ {
167
+ "epoch": 0.05,
168
+ "learning_rate": 0.00019983571811184295,
169
+ "loss": 0.0934,
170
+ "step": 27
171
+ },
172
+ {
173
+ "epoch": 0.05,
174
+ "learning_rate": 0.00019980123034847024,
175
+ "loss": 0.1068,
176
+ "step": 28
177
+ },
178
+ {
179
+ "epoch": 0.05,
180
+ "learning_rate": 0.00019976346258894503,
181
+ "loss": 0.0973,
182
+ "step": 29
183
+ },
184
+ {
185
+ "epoch": 0.05,
186
+ "learning_rate": 0.00019972241607451552,
187
+ "loss": 0.0757,
188
+ "step": 30
189
+ },
190
+ {
191
+ "epoch": 0.05,
192
+ "learning_rate": 0.00019967809215418728,
193
+ "loss": 0.1034,
194
+ "step": 31
195
+ },
196
+ {
197
+ "epoch": 0.06,
198
+ "learning_rate": 0.00019963049228467874,
199
+ "loss": 0.0755,
200
+ "step": 32
201
+ },
202
+ {
203
+ "epoch": 0.06,
204
+ "learning_rate": 0.00019957961803037326,
205
+ "loss": 0.0938,
206
+ "step": 33
207
+ },
208
+ {
209
+ "epoch": 0.06,
210
+ "learning_rate": 0.00019952547106326783,
211
+ "loss": 0.1022,
212
+ "step": 34
213
+ },
214
+ {
215
+ "epoch": 0.06,
216
+ "learning_rate": 0.00019946805316291814,
217
+ "loss": 0.1098,
218
+ "step": 35
219
+ },
220
+ {
221
+ "epoch": 0.06,
222
+ "learning_rate": 0.00019940736621638,
223
+ "loss": 0.0943,
224
+ "step": 36
225
+ },
226
+ {
227
+ "epoch": 0.07,
228
+ "learning_rate": 0.00019934341221814739,
229
+ "loss": 0.1018,
230
+ "step": 37
231
+ },
232
+ {
233
+ "epoch": 0.07,
234
+ "learning_rate": 0.00019927619327008683,
235
+ "loss": 0.0749,
236
+ "step": 38
237
+ },
238
+ {
239
+ "epoch": 0.07,
240
+ "learning_rate": 0.0001992057115813684,
241
+ "loss": 0.0929,
242
+ "step": 39
243
+ },
244
+ {
245
+ "epoch": 0.07,
246
+ "learning_rate": 0.00019913196946839303,
247
+ "loss": 0.0755,
248
+ "step": 40
249
+ },
250
+ {
251
+ "epoch": 0.07,
252
+ "learning_rate": 0.00019905496935471658,
253
+ "loss": 0.1068,
254
+ "step": 41
255
+ },
256
+ {
257
+ "epoch": 0.07,
258
+ "learning_rate": 0.0001989747137709699,
259
+ "loss": 0.1349,
260
+ "step": 42
261
+ },
262
+ {
263
+ "epoch": 0.08,
264
+ "learning_rate": 0.00019889120535477583,
265
+ "loss": 0.0815,
266
+ "step": 43
267
+ },
268
+ {
269
+ "epoch": 0.08,
270
+ "learning_rate": 0.0001988044468506625,
271
+ "loss": 0.0469,
272
+ "step": 44
273
+ },
274
+ {
275
+ "epoch": 0.08,
276
+ "learning_rate": 0.0001987144411099731,
277
+ "loss": 0.0916,
278
+ "step": 45
279
+ },
280
+ {
281
+ "epoch": 0.08,
282
+ "learning_rate": 0.00019862119109077223,
283
+ "loss": 0.065,
284
+ "step": 46
285
+ },
286
+ {
287
+ "epoch": 0.08,
288
+ "learning_rate": 0.0001985246998577486,
289
+ "loss": 0.0508,
290
+ "step": 47
291
+ },
292
+ {
293
+ "epoch": 0.08,
294
+ "learning_rate": 0.00019842497058211432,
295
+ "loss": 0.1073,
296
+ "step": 48
297
+ },
298
+ {
299
+ "epoch": 0.09,
300
+ "learning_rate": 0.00019832200654150076,
301
+ "loss": 0.0775,
302
+ "step": 49
303
+ },
304
+ {
305
+ "epoch": 0.09,
306
+ "learning_rate": 0.00019821581111985071,
307
+ "loss": 0.0927,
308
+ "step": 50
309
+ },
310
+ {
311
+ "epoch": 0.09,
312
+ "learning_rate": 0.0001981063878073073,
313
+ "loss": 0.0893,
314
+ "step": 51
315
+ },
316
+ {
317
+ "epoch": 0.09,
318
+ "learning_rate": 0.0001979937402000991,
319
+ "loss": 0.0893,
320
+ "step": 52
321
+ },
322
+ {
323
+ "epoch": 0.09,
324
+ "learning_rate": 0.00019787787200042223,
325
+ "loss": 0.0939,
326
+ "step": 53
327
+ },
328
+ {
329
+ "epoch": 0.1,
330
+ "learning_rate": 0.00019775878701631838,
331
+ "loss": 0.0918,
332
+ "step": 54
333
+ },
334
+ {
335
+ "epoch": 0.1,
336
+ "learning_rate": 0.0001976364891615498,
337
+ "loss": 0.074,
338
+ "step": 55
339
+ },
340
+ {
341
+ "epoch": 0.1,
342
+ "learning_rate": 0.0001975109824554707,
343
+ "loss": 0.0573,
344
+ "step": 56
345
+ },
346
+ {
347
+ "epoch": 0.1,
348
+ "learning_rate": 0.0001973822710228951,
349
+ "loss": 0.076,
350
+ "step": 57
351
+ },
352
+ {
353
+ "epoch": 0.1,
354
+ "learning_rate": 0.00019725035909396121,
355
+ "loss": 0.0978,
356
+ "step": 58
357
+ },
358
+ {
359
+ "epoch": 0.1,
360
+ "learning_rate": 0.0001971152510039926,
361
+ "loss": 0.0989,
362
+ "step": 59
363
+ },
364
+ {
365
+ "epoch": 0.11,
366
+ "learning_rate": 0.00019697695119335546,
367
+ "loss": 0.0782,
368
+ "step": 60
369
+ },
370
+ {
371
+ "epoch": 0.11,
372
+ "learning_rate": 0.0001968354642073129,
373
+ "loss": 0.0606,
374
+ "step": 61
375
+ },
376
+ {
377
+ "epoch": 0.11,
378
+ "learning_rate": 0.00019669079469587545,
379
+ "loss": 0.0644,
380
+ "step": 62
381
+ },
382
+ {
383
+ "epoch": 0.11,
384
+ "learning_rate": 0.0001965429474136482,
385
+ "loss": 0.074,
386
+ "step": 63
387
+ },
388
+ {
389
+ "epoch": 0.11,
390
+ "learning_rate": 0.0001963919272196746,
391
+ "loss": 0.0751,
392
+ "step": 64
393
+ },
394
+ {
395
+ "epoch": 0.12,
396
+ "learning_rate": 0.00019623773907727682,
397
+ "loss": 0.0788,
398
+ "step": 65
399
+ },
400
+ {
401
+ "epoch": 0.12,
402
+ "learning_rate": 0.00019608038805389252,
403
+ "loss": 0.1,
404
+ "step": 66
405
+ },
406
+ {
407
+ "epoch": 0.12,
408
+ "learning_rate": 0.00019591987932090834,
409
+ "loss": 0.0859,
410
+ "step": 67
411
+ },
412
+ {
413
+ "epoch": 0.12,
414
+ "learning_rate": 0.00019575621815348998,
415
+ "loss": 0.072,
416
+ "step": 68
417
+ },
418
+ {
419
+ "epoch": 0.12,
420
+ "learning_rate": 0.00019558940993040885,
421
+ "loss": 0.0651,
422
+ "step": 69
423
+ },
424
+ {
425
+ "epoch": 0.12,
426
+ "learning_rate": 0.0001954194601338651,
427
+ "loss": 0.0939,
428
+ "step": 70
429
+ },
430
+ {
431
+ "epoch": 0.13,
432
+ "learning_rate": 0.00019524637434930776,
433
+ "loss": 0.0676,
434
+ "step": 71
435
+ },
436
+ {
437
+ "epoch": 0.13,
438
+ "learning_rate": 0.00019507015826525094,
439
+ "loss": 0.0646,
440
+ "step": 72
441
+ },
442
+ {
443
+ "epoch": 0.13,
444
+ "learning_rate": 0.00019489081767308698,
445
+ "loss": 0.0773,
446
+ "step": 73
447
+ },
448
+ {
449
+ "epoch": 0.13,
450
+ "learning_rate": 0.00019470835846689596,
451
+ "loss": 0.0822,
452
+ "step": 74
453
+ },
454
+ {
455
+ "epoch": 0.13,
456
+ "learning_rate": 0.00019452278664325228,
457
+ "loss": 0.0767,
458
+ "step": 75
459
+ },
460
+ {
461
+ "epoch": 0.13,
462
+ "learning_rate": 0.00019433410830102722,
463
+ "loss": 0.0873,
464
+ "step": 76
465
+ },
466
+ {
467
+ "epoch": 0.14,
468
+ "learning_rate": 0.00019414232964118892,
469
+ "loss": 0.0839,
470
+ "step": 77
471
+ },
472
+ {
473
+ "epoch": 0.14,
474
+ "learning_rate": 0.0001939474569665981,
475
+ "loss": 0.0519,
476
+ "step": 78
477
+ },
478
+ {
479
+ "epoch": 0.14,
480
+ "learning_rate": 0.00019374949668180138,
481
+ "loss": 0.0717,
482
+ "step": 79
483
+ },
484
+ {
485
+ "epoch": 0.14,
486
+ "learning_rate": 0.00019354845529282038,
487
+ "loss": 0.0648,
488
+ "step": 80
489
+ },
490
+ {
491
+ "epoch": 0.14,
492
+ "learning_rate": 0.0001933443394069383,
493
+ "loss": 0.0742,
494
+ "step": 81
495
+ },
496
+ {
497
+ "epoch": 0.15,
498
+ "learning_rate": 0.00019313715573248237,
499
+ "loss": 0.0614,
500
+ "step": 82
501
+ },
502
+ {
503
+ "epoch": 0.15,
504
+ "learning_rate": 0.00019292691107860376,
505
+ "loss": 0.048,
506
+ "step": 83
507
+ },
508
+ {
509
+ "epoch": 0.15,
510
+ "learning_rate": 0.00019271361235505342,
511
+ "loss": 0.0657,
512
+ "step": 84
513
+ },
514
+ {
515
+ "epoch": 0.15,
516
+ "learning_rate": 0.00019249726657195532,
517
+ "loss": 0.0511,
518
+ "step": 85
519
+ },
520
+ {
521
+ "epoch": 0.15,
522
+ "learning_rate": 0.0001922778808395759,
523
+ "loss": 0.0834,
524
+ "step": 86
525
+ },
526
+ {
527
+ "epoch": 0.15,
528
+ "learning_rate": 0.00019205546236809035,
529
+ "loss": 0.0778,
530
+ "step": 87
531
+ },
532
+ {
533
+ "epoch": 0.16,
534
+ "learning_rate": 0.00019183001846734575,
535
+ "loss": 0.0742,
536
+ "step": 88
537
+ },
538
+ {
539
+ "epoch": 0.16,
540
+ "learning_rate": 0.00019160155654662076,
541
+ "loss": 0.0814,
542
+ "step": 89
543
+ },
544
+ {
545
+ "epoch": 0.16,
546
+ "learning_rate": 0.0001913700841143821,
547
+ "loss": 0.0783,
548
+ "step": 90
549
+ },
550
+ {
551
+ "epoch": 0.16,
552
+ "learning_rate": 0.00019113560877803796,
553
+ "loss": 0.0784,
554
+ "step": 91
555
+ },
556
+ {
557
+ "epoch": 0.16,
558
+ "learning_rate": 0.00019089813824368763,
559
+ "loss": 0.0676,
560
+ "step": 92
561
+ },
562
+ {
563
+ "epoch": 0.16,
564
+ "learning_rate": 0.0001906576803158686,
565
+ "loss": 0.1092,
566
+ "step": 93
567
+ },
568
+ {
569
+ "epoch": 0.17,
570
+ "learning_rate": 0.00019041424289729993,
571
+ "loss": 0.0776,
572
+ "step": 94
573
+ },
574
+ {
575
+ "epoch": 0.17,
576
+ "learning_rate": 0.00019016783398862229,
577
+ "loss": 0.0454,
578
+ "step": 95
579
+ },
580
+ {
581
+ "epoch": 0.17,
582
+ "learning_rate": 0.00018991846168813544,
583
+ "loss": 0.0607,
584
+ "step": 96
585
+ },
586
+ {
587
+ "epoch": 0.17,
588
+ "learning_rate": 0.0001896661341915318,
589
+ "loss": 0.0722,
590
+ "step": 97
591
+ },
592
+ {
593
+ "epoch": 0.17,
594
+ "learning_rate": 0.00018941085979162713,
595
+ "loss": 0.0596,
596
+ "step": 98
597
+ },
598
+ {
599
+ "epoch": 0.18,
600
+ "learning_rate": 0.00018915264687808806,
601
+ "loss": 0.0924,
602
+ "step": 99
603
+ },
604
+ {
605
+ "epoch": 0.18,
606
+ "learning_rate": 0.00018889150393715628,
607
+ "loss": 0.0891,
608
+ "step": 100
609
+ },
610
+ {
611
+ "epoch": 0.18,
612
+ "learning_rate": 0.00018862743955136966,
613
+ "loss": 0.079,
614
+ "step": 101
615
+ },
616
+ {
617
+ "epoch": 0.18,
618
+ "learning_rate": 0.00018836046239928023,
619
+ "loss": 0.0661,
620
+ "step": 102
621
+ },
622
+ {
623
+ "epoch": 0.18,
624
+ "learning_rate": 0.00018809058125516894,
625
+ "loss": 0.0706,
626
+ "step": 103
627
+ },
628
+ {
629
+ "epoch": 0.18,
630
+ "learning_rate": 0.00018781780498875725,
631
+ "loss": 0.0876,
632
+ "step": 104
633
+ },
634
+ {
635
+ "epoch": 0.19,
636
+ "learning_rate": 0.00018754214256491562,
637
+ "loss": 0.0616,
638
+ "step": 105
639
+ },
640
+ {
641
+ "epoch": 0.19,
642
+ "learning_rate": 0.00018726360304336894,
643
+ "loss": 0.0652,
644
+ "step": 106
645
+ },
646
+ {
647
+ "epoch": 0.19,
648
+ "learning_rate": 0.00018698219557839872,
649
+ "loss": 0.0656,
650
+ "step": 107
651
+ },
652
+ {
653
+ "epoch": 0.19,
654
+ "learning_rate": 0.00018669792941854228,
655
+ "loss": 0.0915,
656
+ "step": 108
657
+ },
658
+ {
659
+ "epoch": 0.19,
660
+ "learning_rate": 0.00018641081390628877,
661
+ "loss": 0.0845,
662
+ "step": 109
663
+ },
664
+ {
665
+ "epoch": 0.19,
666
+ "learning_rate": 0.00018612085847777215,
667
+ "loss": 0.0654,
668
+ "step": 110
669
+ },
670
+ {
671
+ "epoch": 0.2,
672
+ "learning_rate": 0.00018582807266246092,
673
+ "loss": 0.071,
674
+ "step": 111
675
+ },
676
+ {
677
+ "epoch": 0.2,
678
+ "learning_rate": 0.0001855324660828452,
679
+ "loss": 0.061,
680
+ "step": 112
681
+ },
682
+ {
683
+ "epoch": 0.2,
684
+ "learning_rate": 0.00018523404845412027,
685
+ "loss": 0.0695,
686
+ "step": 113
687
+ },
688
+ {
689
+ "epoch": 0.2,
690
+ "learning_rate": 0.0001849328295838674,
691
+ "loss": 0.083,
692
+ "step": 114
693
+ },
694
+ {
695
+ "epoch": 0.2,
696
+ "learning_rate": 0.00018462881937173143,
697
+ "loss": 0.0492,
698
+ "step": 115
699
+ },
700
+ {
701
+ "epoch": 0.21,
702
+ "learning_rate": 0.0001843220278090954,
703
+ "loss": 0.0643,
704
+ "step": 116
705
+ },
706
+ {
707
+ "epoch": 0.21,
708
+ "learning_rate": 0.0001840124649787524,
709
+ "loss": 0.0863,
710
+ "step": 117
711
+ },
712
+ {
713
+ "epoch": 0.21,
714
+ "learning_rate": 0.00018370014105457378,
715
+ "loss": 0.0576,
716
+ "step": 118
717
+ },
718
+ {
719
+ "epoch": 0.21,
720
+ "learning_rate": 0.00018338506630117526,
721
+ "loss": 0.0594,
722
+ "step": 119
723
+ },
724
+ {
725
+ "epoch": 0.21,
726
+ "learning_rate": 0.00018306725107357933,
727
+ "loss": 0.0707,
728
+ "step": 120
729
+ },
730
+ {
731
+ "epoch": 0.21,
732
+ "learning_rate": 0.0001827467058168748,
733
+ "loss": 0.0665,
734
+ "step": 121
735
+ },
736
+ {
737
+ "epoch": 0.22,
738
+ "learning_rate": 0.0001824234410658738,
739
+ "loss": 0.0693,
740
+ "step": 122
741
+ },
742
+ {
743
+ "epoch": 0.22,
744
+ "learning_rate": 0.00018209746744476536,
745
+ "loss": 0.0628,
746
+ "step": 123
747
+ },
748
+ {
749
+ "epoch": 0.22,
750
+ "learning_rate": 0.0001817687956667664,
751
+ "loss": 0.0751,
752
+ "step": 124
753
+ },
754
+ {
755
+ "epoch": 0.22,
756
+ "learning_rate": 0.00018143743653376942,
757
+ "loss": 0.0643,
758
+ "step": 125
759
+ },
760
+ {
761
+ "epoch": 0.22,
762
+ "learning_rate": 0.0001811034009359877,
763
+ "loss": 0.0686,
764
+ "step": 126
765
+ },
766
+ {
767
+ "epoch": 0.22,
768
+ "learning_rate": 0.00018076669985159726,
769
+ "loss": 0.0639,
770
+ "step": 127
771
+ },
772
+ {
773
+ "epoch": 0.23,
774
+ "learning_rate": 0.00018042734434637613,
775
+ "loss": 0.056,
776
+ "step": 128
777
+ },
778
+ {
779
+ "epoch": 0.23,
780
+ "learning_rate": 0.00018008534557334064,
781
+ "loss": 0.0674,
782
+ "step": 129
783
+ },
784
+ {
785
+ "epoch": 0.23,
786
+ "learning_rate": 0.00017974071477237885,
787
+ "loss": 0.0799,
788
+ "step": 130
789
+ },
790
+ {
791
+ "epoch": 0.23,
792
+ "learning_rate": 0.00017939346326988127,
793
+ "loss": 0.0567,
794
+ "step": 131
795
+ },
796
+ {
797
+ "epoch": 0.23,
798
+ "learning_rate": 0.00017904360247836834,
799
+ "loss": 0.0746,
800
+ "step": 132
801
+ },
802
+ {
803
+ "epoch": 0.24,
804
+ "learning_rate": 0.00017869114389611575,
805
+ "loss": 0.055,
806
+ "step": 133
807
+ },
808
+ {
809
+ "epoch": 0.24,
810
+ "learning_rate": 0.00017833609910677614,
811
+ "loss": 0.0692,
812
+ "step": 134
813
+ },
814
+ {
815
+ "epoch": 0.24,
816
+ "learning_rate": 0.00017797847977899873,
817
+ "loss": 0.0365,
818
+ "step": 135
819
+ },
820
+ {
821
+ "epoch": 0.24,
822
+ "learning_rate": 0.00017761829766604556,
823
+ "loss": 0.0466,
824
+ "step": 136
825
+ },
826
+ {
827
+ "epoch": 0.24,
828
+ "learning_rate": 0.0001772555646054055,
829
+ "loss": 0.0744,
830
+ "step": 137
831
+ },
832
+ {
833
+ "epoch": 0.24,
834
+ "learning_rate": 0.00017689029251840492,
835
+ "loss": 0.1026,
836
+ "step": 138
837
+ },
838
+ {
839
+ "epoch": 0.25,
840
+ "learning_rate": 0.00017652249340981607,
841
+ "loss": 0.0872,
842
+ "step": 139
843
+ },
844
+ {
845
+ "epoch": 0.25,
846
+ "learning_rate": 0.00017615217936746246,
847
+ "loss": 0.056,
848
+ "step": 140
849
+ },
850
+ {
851
+ "epoch": 0.25,
852
+ "learning_rate": 0.00017577936256182167,
853
+ "loss": 0.0714,
854
+ "step": 141
855
+ },
856
+ {
857
+ "epoch": 0.25,
858
+ "learning_rate": 0.0001754040552456253,
859
+ "loss": 0.0908,
860
+ "step": 142
861
+ },
862
+ {
863
+ "epoch": 0.25,
864
+ "learning_rate": 0.00017502626975345628,
865
+ "loss": 0.0587,
866
+ "step": 143
867
+ },
868
+ {
869
+ "epoch": 0.25,
870
+ "learning_rate": 0.00017464601850134353,
871
+ "loss": 0.0534,
872
+ "step": 144
873
+ },
874
+ {
875
+ "epoch": 0.26,
876
+ "learning_rate": 0.0001742633139863538,
877
+ "loss": 0.062,
878
+ "step": 145
879
+ },
880
+ {
881
+ "epoch": 0.26,
882
+ "learning_rate": 0.0001738781687861812,
883
+ "loss": 0.0742,
884
+ "step": 146
885
+ },
886
+ {
887
+ "epoch": 0.26,
888
+ "learning_rate": 0.00017349059555873345,
889
+ "loss": 0.0553,
890
+ "step": 147
891
+ },
892
+ {
893
+ "epoch": 0.26,
894
+ "learning_rate": 0.0001731006070417163,
895
+ "loss": 0.077,
896
+ "step": 148
897
+ },
898
+ {
899
+ "epoch": 0.26,
900
+ "learning_rate": 0.0001727082160522145,
901
+ "loss": 0.0705,
902
+ "step": 149
903
+ },
904
+ {
905
+ "epoch": 0.27,
906
+ "learning_rate": 0.00017231343548627083,
907
+ "loss": 0.0728,
908
+ "step": 150
909
+ },
910
+ {
911
+ "epoch": 0.27,
912
+ "learning_rate": 0.00017191627831846226,
913
+ "loss": 0.0515,
914
+ "step": 151
915
+ },
916
+ {
917
+ "epoch": 0.27,
918
+ "learning_rate": 0.00017151675760147326,
919
+ "loss": 0.0616,
920
+ "step": 152
921
+ },
922
+ {
923
+ "epoch": 0.27,
924
+ "learning_rate": 0.00017111488646566727,
925
+ "loss": 0.0633,
926
+ "step": 153
927
+ },
928
+ {
929
+ "epoch": 0.27,
930
+ "learning_rate": 0.00017071067811865476,
931
+ "loss": 0.0739,
932
+ "step": 154
933
+ },
934
+ {
935
+ "epoch": 0.27,
936
+ "learning_rate": 0.00017030414584485936,
937
+ "loss": 0.0668,
938
+ "step": 155
939
+ },
940
+ {
941
+ "epoch": 0.28,
942
+ "learning_rate": 0.00016989530300508124,
943
+ "loss": 0.0448,
944
+ "step": 156
945
+ },
946
+ {
947
+ "epoch": 0.28,
948
+ "learning_rate": 0.00016948416303605795,
949
+ "loss": 0.0488,
950
+ "step": 157
951
+ },
952
+ {
953
+ "epoch": 0.28,
954
+ "learning_rate": 0.0001690707394500229,
955
+ "loss": 0.0574,
956
+ "step": 158
957
+ },
958
+ {
959
+ "epoch": 0.28,
960
+ "learning_rate": 0.00016865504583426117,
961
+ "loss": 0.0656,
962
+ "step": 159
963
+ },
964
+ {
965
+ "epoch": 0.28,
966
+ "learning_rate": 0.00016823709585066308,
967
+ "loss": 0.0432,
968
+ "step": 160
969
+ },
970
+ {
971
+ "epoch": 0.28,
972
+ "learning_rate": 0.00016781690323527511,
973
+ "loss": 0.056,
974
+ "step": 161
975
+ },
976
+ {
977
+ "epoch": 0.29,
978
+ "learning_rate": 0.00016739448179784846,
979
+ "loss": 0.0645,
980
+ "step": 162
981
+ },
982
+ {
983
+ "epoch": 0.29,
984
+ "learning_rate": 0.0001669698454213852,
985
+ "loss": 0.0615,
986
+ "step": 163
987
+ },
988
+ {
989
+ "epoch": 0.29,
990
+ "learning_rate": 0.00016654300806168208,
991
+ "loss": 0.0377,
992
+ "step": 164
993
+ },
994
+ {
995
+ "epoch": 0.29,
996
+ "learning_rate": 0.0001661139837468717,
997
+ "loss": 0.074,
998
+ "step": 165
999
+ },
1000
+ {
1001
+ "epoch": 0.29,
1002
+ "learning_rate": 0.00016568278657696164,
1003
+ "loss": 0.066,
1004
+ "step": 166
1005
+ },
1006
+ {
1007
+ "epoch": 0.3,
1008
+ "learning_rate": 0.00016524943072337094,
1009
+ "loss": 0.074,
1010
+ "step": 167
1011
+ },
1012
+ {
1013
+ "epoch": 0.3,
1014
+ "learning_rate": 0.00016481393042846443,
1015
+ "loss": 0.0569,
1016
+ "step": 168
1017
+ },
1018
+ {
1019
+ "epoch": 0.3,
1020
+ "learning_rate": 0.00016437630000508464,
1021
+ "loss": 0.0436,
1022
+ "step": 169
1023
+ },
1024
+ {
1025
+ "epoch": 0.3,
1026
+ "learning_rate": 0.00016393655383608132,
1027
+ "loss": 0.0725,
1028
+ "step": 170
1029
+ },
1030
+ {
1031
+ "epoch": 0.3,
1032
+ "learning_rate": 0.0001634947063738389,
1033
+ "loss": 0.0785,
1034
+ "step": 171
1035
+ },
1036
+ {
1037
+ "epoch": 0.3,
1038
+ "learning_rate": 0.00016305077213980132,
1039
+ "loss": 0.0538,
1040
+ "step": 172
1041
+ },
1042
+ {
1043
+ "epoch": 0.31,
1044
+ "learning_rate": 0.00016260476572399496,
1045
+ "loss": 0.0583,
1046
+ "step": 173
1047
+ },
1048
+ {
1049
+ "epoch": 0.31,
1050
+ "learning_rate": 0.00016215670178454893,
1051
+ "loss": 0.0526,
1052
+ "step": 174
1053
+ },
1054
+ {
1055
+ "epoch": 0.31,
1056
+ "learning_rate": 0.00016170659504721364,
1057
+ "loss": 0.0499,
1058
+ "step": 175
1059
+ },
1060
+ {
1061
+ "epoch": 0.31,
1062
+ "learning_rate": 0.00016125446030487643,
1063
+ "loss": 0.0649,
1064
+ "step": 176
1065
+ },
1066
+ {
1067
+ "epoch": 0.31,
1068
+ "learning_rate": 0.00016080031241707578,
1069
+ "loss": 0.06,
1070
+ "step": 177
1071
+ },
1072
+ {
1073
+ "epoch": 0.32,
1074
+ "learning_rate": 0.00016034416630951267,
1075
+ "loss": 0.0803,
1076
+ "step": 178
1077
+ },
1078
+ {
1079
+ "epoch": 0.32,
1080
+ "learning_rate": 0.0001598860369735601,
1081
+ "loss": 0.0734,
1082
+ "step": 179
1083
+ },
1084
+ {
1085
+ "epoch": 0.32,
1086
+ "learning_rate": 0.00015942593946577068,
1087
+ "loss": 0.0446,
1088
+ "step": 180
1089
+ },
1090
+ {
1091
+ "epoch": 0.32,
1092
+ "learning_rate": 0.00015896388890738127,
1093
+ "loss": 0.057,
1094
+ "step": 181
1095
+ },
1096
+ {
1097
+ "epoch": 0.32,
1098
+ "learning_rate": 0.0001584999004838165,
1099
+ "loss": 0.0647,
1100
+ "step": 182
1101
+ },
1102
+ {
1103
+ "epoch": 0.32,
1104
+ "learning_rate": 0.00015803398944418934,
1105
+ "loss": 0.0606,
1106
+ "step": 183
1107
+ },
1108
+ {
1109
+ "epoch": 0.33,
1110
+ "learning_rate": 0.00015756617110080024,
1111
+ "loss": 0.0696,
1112
+ "step": 184
1113
+ },
1114
+ {
1115
+ "epoch": 0.33,
1116
+ "learning_rate": 0.0001570964608286336,
1117
+ "loss": 0.059,
1118
+ "step": 185
1119
+ },
1120
+ {
1121
+ "epoch": 0.33,
1122
+ "learning_rate": 0.00015662487406485273,
1123
+ "loss": 0.0535,
1124
+ "step": 186
1125
+ },
1126
+ {
1127
+ "epoch": 0.33,
1128
+ "learning_rate": 0.00015615142630829229,
1129
+ "loss": 0.0503,
1130
+ "step": 187
1131
+ },
1132
+ {
1133
+ "epoch": 0.33,
1134
+ "learning_rate": 0.0001556761331189491,
1135
+ "loss": 0.0708,
1136
+ "step": 188
1137
+ },
1138
+ {
1139
+ "epoch": 0.33,
1140
+ "learning_rate": 0.00015519901011747044,
1141
+ "loss": 0.0816,
1142
+ "step": 189
1143
+ },
1144
+ {
1145
+ "epoch": 0.34,
1146
+ "learning_rate": 0.00015472007298464118,
1147
+ "loss": 0.06,
1148
+ "step": 190
1149
+ },
1150
+ {
1151
+ "epoch": 0.34,
1152
+ "learning_rate": 0.00015423933746086795,
1153
+ "loss": 0.0529,
1154
+ "step": 191
1155
+ },
1156
+ {
1157
+ "epoch": 0.34,
1158
+ "learning_rate": 0.00015375681934566205,
1159
+ "loss": 0.0474,
1160
+ "step": 192
1161
+ },
1162
+ {
1163
+ "epoch": 0.34,
1164
+ "learning_rate": 0.0001532725344971202,
1165
+ "loss": 0.0684,
1166
+ "step": 193
1167
+ },
1168
+ {
1169
+ "epoch": 0.34,
1170
+ "learning_rate": 0.0001527864988314033,
1171
+ "loss": 0.0656,
1172
+ "step": 194
1173
+ },
1174
+ {
1175
+ "epoch": 0.35,
1176
+ "learning_rate": 0.00015229872832221334,
1177
+ "loss": 0.0764,
1178
+ "step": 195
1179
+ },
1180
+ {
1181
+ "epoch": 0.35,
1182
+ "learning_rate": 0.00015180923900026848,
1183
+ "loss": 0.0662,
1184
+ "step": 196
1185
+ },
1186
+ {
1187
+ "epoch": 0.35,
1188
+ "learning_rate": 0.00015131804695277612,
1189
+ "loss": 0.0766,
1190
+ "step": 197
1191
+ },
1192
+ {
1193
+ "epoch": 0.35,
1194
+ "learning_rate": 0.00015082516832290422,
1195
+ "loss": 0.0589,
1196
+ "step": 198
1197
+ },
1198
+ {
1199
+ "epoch": 0.35,
1200
+ "learning_rate": 0.0001503306193092508,
1201
+ "loss": 0.0623,
1202
+ "step": 199
1203
+ },
1204
+ {
1205
+ "epoch": 0.35,
1206
+ "learning_rate": 0.0001498344161653115,
1207
+ "loss": 0.0548,
1208
+ "step": 200
1209
+ },
1210
+ {
1211
+ "epoch": 0.36,
1212
+ "learning_rate": 0.0001493365751989454,
1213
+ "loss": 0.0604,
1214
+ "step": 201
1215
+ },
1216
+ {
1217
+ "epoch": 0.36,
1218
+ "learning_rate": 0.00014883711277183915,
1219
+ "loss": 0.049,
1220
+ "step": 202
1221
+ },
1222
+ {
1223
+ "epoch": 0.36,
1224
+ "learning_rate": 0.0001483360452989691,
1225
+ "loss": 0.0599,
1226
+ "step": 203
1227
+ },
1228
+ {
1229
+ "epoch": 0.36,
1230
+ "learning_rate": 0.00014783338924806192,
1231
+ "loss": 0.058,
1232
+ "step": 204
1233
+ },
1234
+ {
1235
+ "epoch": 0.36,
1236
+ "learning_rate": 0.00014732916113905335,
1237
+ "loss": 0.0612,
1238
+ "step": 205
1239
+ },
1240
+ {
1241
+ "epoch": 0.36,
1242
+ "learning_rate": 0.00014682337754354534,
1243
+ "loss": 0.0664,
1244
+ "step": 206
1245
+ },
1246
+ {
1247
+ "epoch": 0.37,
1248
+ "learning_rate": 0.00014631605508426122,
1249
+ "loss": 0.0634,
1250
+ "step": 207
1251
+ },
1252
+ {
1253
+ "epoch": 0.37,
1254
+ "learning_rate": 0.00014580721043449968,
1255
+ "loss": 0.049,
1256
+ "step": 208
1257
+ },
1258
+ {
1259
+ "epoch": 0.37,
1260
+ "learning_rate": 0.00014529686031758643,
1261
+ "loss": 0.0717,
1262
+ "step": 209
1263
+ },
1264
+ {
1265
+ "epoch": 0.37,
1266
+ "learning_rate": 0.00014478502150632502,
1267
+ "loss": 0.0487,
1268
+ "step": 210
1269
+ },
1270
+ {
1271
+ "epoch": 0.37,
1272
+ "learning_rate": 0.00014427171082244522,
1273
+ "loss": 0.0394,
1274
+ "step": 211
1275
+ },
1276
+ {
1277
+ "epoch": 0.38,
1278
+ "learning_rate": 0.00014375694513605035,
1279
+ "loss": 0.0548,
1280
+ "step": 212
1281
+ },
1282
+ {
1283
+ "epoch": 0.38,
1284
+ "learning_rate": 0.00014324074136506284,
1285
+ "loss": 0.0441,
1286
+ "step": 213
1287
+ },
1288
+ {
1289
+ "epoch": 0.38,
1290
+ "learning_rate": 0.0001427231164746681,
1291
+ "loss": 0.0565,
1292
+ "step": 214
1293
+ },
1294
+ {
1295
+ "epoch": 0.38,
1296
+ "learning_rate": 0.00014220408747675714,
1297
+ "loss": 0.0697,
1298
+ "step": 215
1299
+ },
1300
+ {
1301
+ "epoch": 0.38,
1302
+ "learning_rate": 0.00014168367142936735,
1303
+ "loss": 0.0657,
1304
+ "step": 216
1305
+ },
1306
+ {
1307
+ "epoch": 0.38,
1308
+ "learning_rate": 0.0001411618854361218,
1309
+ "loss": 0.0635,
1310
+ "step": 217
1311
+ },
1312
+ {
1313
+ "epoch": 0.39,
1314
+ "learning_rate": 0.00014063874664566734,
1315
+ "loss": 0.0689,
1316
+ "step": 218
1317
+ },
1318
+ {
1319
+ "epoch": 0.39,
1320
+ "learning_rate": 0.00014011427225111092,
1321
+ "loss": 0.0689,
1322
+ "step": 219
1323
+ },
1324
+ {
1325
+ "epoch": 0.39,
1326
+ "learning_rate": 0.00013958847948945428,
1327
+ "loss": 0.05,
1328
+ "step": 220
1329
+ },
1330
+ {
1331
+ "epoch": 0.39,
1332
+ "learning_rate": 0.00013906138564102793,
1333
+ "loss": 0.0598,
1334
+ "step": 221
1335
+ },
1336
+ {
1337
+ "epoch": 0.39,
1338
+ "learning_rate": 0.00013853300802892283,
1339
+ "loss": 0.0757,
1340
+ "step": 222
1341
+ },
1342
+ {
1343
+ "epoch": 0.39,
1344
+ "learning_rate": 0.0001380033640184213,
1345
+ "loss": 0.0477,
1346
+ "step": 223
1347
+ },
1348
+ {
1349
+ "epoch": 0.4,
1350
+ "learning_rate": 0.00013747247101642603,
1351
+ "loss": 0.056,
1352
+ "step": 224
1353
+ },
1354
+ {
1355
+ "epoch": 0.4,
1356
+ "learning_rate": 0.0001369403464708884,
1357
+ "loss": 0.0559,
1358
+ "step": 225
1359
+ },
1360
+ {
1361
+ "epoch": 0.4,
1362
+ "learning_rate": 0.00013640700787023464,
1363
+ "loss": 0.0725,
1364
+ "step": 226
1365
+ },
1366
+ {
1367
+ "epoch": 0.4,
1368
+ "learning_rate": 0.0001358724727427914,
1369
+ "loss": 0.0568,
1370
+ "step": 227
1371
+ },
1372
+ {
1373
+ "epoch": 0.4,
1374
+ "learning_rate": 0.00013533675865620937,
1375
+ "loss": 0.0522,
1376
+ "step": 228
1377
+ },
1378
+ {
1379
+ "epoch": 0.41,
1380
+ "learning_rate": 0.0001347998832168862,
1381
+ "loss": 0.0607,
1382
+ "step": 229
1383
+ },
1384
+ {
1385
+ "epoch": 0.41,
1386
+ "learning_rate": 0.00013426186406938768,
1387
+ "loss": 0.0494,
1388
+ "step": 230
1389
+ },
1390
+ {
1391
+ "epoch": 0.41,
1392
+ "learning_rate": 0.0001337227188958679,
1393
+ "loss": 0.0588,
1394
+ "step": 231
1395
+ },
1396
+ {
1397
+ "epoch": 0.41,
1398
+ "learning_rate": 0.0001331824654154881,
1399
+ "loss": 0.0671,
1400
+ "step": 232
1401
+ },
1402
+ {
1403
+ "epoch": 0.41,
1404
+ "learning_rate": 0.00013264112138383445,
1405
+ "loss": 0.0613,
1406
+ "step": 233
1407
+ },
1408
+ {
1409
+ "epoch": 0.41,
1410
+ "learning_rate": 0.00013209870459233422,
1411
+ "loss": 0.0609,
1412
+ "step": 234
1413
+ },
1414
+ {
1415
+ "epoch": 0.42,
1416
+ "learning_rate": 0.0001315552328676714,
1417
+ "loss": 0.0512,
1418
+ "step": 235
1419
+ },
1420
+ {
1421
+ "epoch": 0.42,
1422
+ "learning_rate": 0.00013101072407120057,
1423
+ "loss": 0.04,
1424
+ "step": 236
1425
+ },
1426
+ {
1427
+ "epoch": 0.42,
1428
+ "learning_rate": 0.00013046519609836,
1429
+ "loss": 0.0516,
1430
+ "step": 237
1431
+ },
1432
+ {
1433
+ "epoch": 0.42,
1434
+ "learning_rate": 0.00012991866687808354,
1435
+ "loss": 0.0399,
1436
+ "step": 238
1437
+ },
1438
+ {
1439
+ "epoch": 0.42,
1440
+ "learning_rate": 0.0001293711543722112,
1441
+ "loss": 0.0467,
1442
+ "step": 239
1443
+ },
1444
+ {
1445
+ "epoch": 0.42,
1446
+ "learning_rate": 0.00012882267657489909,
1447
+ "loss": 0.039,
1448
+ "step": 240
1449
+ },
1450
+ {
1451
+ "epoch": 0.43,
1452
+ "learning_rate": 0.00012827325151202782,
1453
+ "loss": 0.0534,
1454
+ "step": 241
1455
+ },
1456
+ {
1457
+ "epoch": 0.43,
1458
+ "learning_rate": 0.00012772289724061016,
1459
+ "loss": 0.0451,
1460
+ "step": 242
1461
+ },
1462
+ {
1463
+ "epoch": 0.43,
1464
+ "learning_rate": 0.0001271716318481976,
1465
+ "loss": 0.0341,
1466
+ "step": 243
1467
+ },
1468
+ {
1469
+ "epoch": 0.43,
1470
+ "learning_rate": 0.00012661947345228594,
1471
+ "loss": 0.0352,
1472
+ "step": 244
1473
+ },
1474
+ {
1475
+ "epoch": 0.43,
1476
+ "learning_rate": 0.00012606644019971968,
1477
+ "loss": 0.0601,
1478
+ "step": 245
1479
+ },
1480
+ {
1481
+ "epoch": 0.44,
1482
+ "learning_rate": 0.0001255125502660958,
1483
+ "loss": 0.0539,
1484
+ "step": 246
1485
+ },
1486
+ {
1487
+ "epoch": 0.44,
1488
+ "learning_rate": 0.0001249578218551664,
1489
+ "loss": 0.0695,
1490
+ "step": 247
1491
+ },
1492
+ {
1493
+ "epoch": 0.44,
1494
+ "learning_rate": 0.00012440227319824023,
1495
+ "loss": 0.0467,
1496
+ "step": 248
1497
+ },
1498
+ {
1499
+ "epoch": 0.44,
1500
+ "learning_rate": 0.00012384592255358385,
1501
+ "loss": 0.039,
1502
+ "step": 249
1503
+ },
1504
+ {
1505
+ "epoch": 0.44,
1506
+ "learning_rate": 0.0001232887882058212,
1507
+ "loss": 0.0493,
1508
+ "step": 250
1509
+ },
1510
+ {
1511
+ "epoch": 0.44,
1512
+ "learning_rate": 0.00012273088846533303,
1513
+ "loss": 0.0593,
1514
+ "step": 251
1515
+ },
1516
+ {
1517
+ "epoch": 0.45,
1518
+ "learning_rate": 0.00012217224166765478,
1519
+ "loss": 0.0593,
1520
+ "step": 252
1521
+ },
1522
+ {
1523
+ "epoch": 0.45,
1524
+ "learning_rate": 0.00012161286617287419,
1525
+ "loss": 0.0454,
1526
+ "step": 253
1527
+ },
1528
+ {
1529
+ "epoch": 0.45,
1530
+ "learning_rate": 0.00012105278036502788,
1531
+ "loss": 0.054,
1532
+ "step": 254
1533
+ },
1534
+ {
1535
+ "epoch": 0.45,
1536
+ "learning_rate": 0.00012049200265149709,
1537
+ "loss": 0.0349,
1538
+ "step": 255
1539
+ },
1540
+ {
1541
+ "epoch": 0.45,
1542
+ "learning_rate": 0.00011993055146240273,
1543
+ "loss": 0.0487,
1544
+ "step": 256
1545
+ },
1546
+ {
1547
+ "epoch": 0.45,
1548
+ "learning_rate": 0.00011936844524999966,
1549
+ "loss": 0.0371,
1550
+ "step": 257
1551
+ },
1552
+ {
1553
+ "epoch": 0.46,
1554
+ "learning_rate": 0.00011880570248807033,
1555
+ "loss": 0.0482,
1556
+ "step": 258
1557
+ },
1558
+ {
1559
+ "epoch": 0.46,
1560
+ "learning_rate": 0.00011824234167131747,
1561
+ "loss": 0.0478,
1562
+ "step": 259
1563
+ },
1564
+ {
1565
+ "epoch": 0.46,
1566
+ "learning_rate": 0.00011767838131475655,
1567
+ "loss": 0.0508,
1568
+ "step": 260
1569
+ },
1570
+ {
1571
+ "epoch": 0.46,
1572
+ "learning_rate": 0.00011711383995310681,
1573
+ "loss": 0.0563,
1574
+ "step": 261
1575
+ },
1576
+ {
1577
+ "epoch": 0.46,
1578
+ "learning_rate": 0.00011654873614018267,
1579
+ "loss": 0.0506,
1580
+ "step": 262
1581
+ },
1582
+ {
1583
+ "epoch": 0.47,
1584
+ "learning_rate": 0.00011598308844828348,
1585
+ "loss": 0.066,
1586
+ "step": 263
1587
+ },
1588
+ {
1589
+ "epoch": 0.47,
1590
+ "learning_rate": 0.00011541691546758342,
1591
+ "loss": 0.0437,
1592
+ "step": 264
1593
+ },
1594
+ {
1595
+ "epoch": 0.47,
1596
+ "learning_rate": 0.00011485023580552039,
1597
+ "loss": 0.0768,
1598
+ "step": 265
1599
+ },
1600
+ {
1601
+ "epoch": 0.47,
1602
+ "learning_rate": 0.00011428306808618456,
1603
+ "loss": 0.0658,
1604
+ "step": 266
1605
+ },
1606
+ {
1607
+ "epoch": 0.47,
1608
+ "learning_rate": 0.00011371543094970622,
1609
+ "loss": 0.0465,
1610
+ "step": 267
1611
+ },
1612
+ {
1613
+ "epoch": 0.47,
1614
+ "learning_rate": 0.00011314734305164321,
1615
+ "loss": 0.0794,
1616
+ "step": 268
1617
+ },
1618
+ {
1619
+ "epoch": 0.48,
1620
+ "learning_rate": 0.00011257882306236775,
1621
+ "loss": 0.0607,
1622
+ "step": 269
1623
+ },
1624
+ {
1625
+ "epoch": 0.48,
1626
+ "learning_rate": 0.00011200988966645284,
1627
+ "loss": 0.0547,
1628
+ "step": 270
1629
+ },
1630
+ {
1631
+ "epoch": 0.48,
1632
+ "learning_rate": 0.00011144056156205833,
1633
+ "loss": 0.0315,
1634
+ "step": 271
1635
+ },
1636
+ {
1637
+ "epoch": 0.48,
1638
+ "learning_rate": 0.00011087085746031612,
1639
+ "loss": 0.0585,
1640
+ "step": 272
1641
+ },
1642
+ {
1643
+ "epoch": 0.48,
1644
+ "learning_rate": 0.00011030079608471544,
1645
+ "loss": 0.0529,
1646
+ "step": 273
1647
+ },
1648
+ {
1649
+ "epoch": 0.48,
1650
+ "learning_rate": 0.00010973039617048748,
1651
+ "loss": 0.0452,
1652
+ "step": 274
1653
+ },
1654
+ {
1655
+ "epoch": 0.49,
1656
+ "learning_rate": 0.0001091596764639895,
1657
+ "loss": 0.0548,
1658
+ "step": 275
1659
+ },
1660
+ {
1661
+ "epoch": 0.49,
1662
+ "learning_rate": 0.00010858865572208892,
1663
+ "loss": 0.0614,
1664
+ "step": 276
1665
+ },
1666
+ {
1667
+ "epoch": 0.49,
1668
+ "learning_rate": 0.00010801735271154669,
1669
+ "loss": 0.0342,
1670
+ "step": 277
1671
+ },
1672
+ {
1673
+ "epoch": 0.49,
1674
+ "learning_rate": 0.00010744578620840064,
1675
+ "loss": 0.0408,
1676
+ "step": 278
1677
+ },
1678
+ {
1679
+ "epoch": 0.49,
1680
+ "learning_rate": 0.00010687397499734842,
1681
+ "loss": 0.0387,
1682
+ "step": 279
1683
+ },
1684
+ {
1685
+ "epoch": 0.5,
1686
+ "learning_rate": 0.00010630193787112994,
1687
+ "loss": 0.0572,
1688
+ "step": 280
1689
+ },
1690
+ {
1691
+ "epoch": 0.5,
1692
+ "learning_rate": 0.00010572969362990998,
1693
+ "loss": 0.0608,
1694
+ "step": 281
1695
+ },
1696
+ {
1697
+ "epoch": 0.5,
1698
+ "learning_rate": 0.00010515726108066025,
1699
+ "loss": 0.0458,
1700
+ "step": 282
1701
+ },
1702
+ {
1703
+ "epoch": 0.5,
1704
+ "learning_rate": 0.00010458465903654107,
1705
+ "loss": 0.0486,
1706
+ "step": 283
1707
+ },
1708
+ {
1709
+ "epoch": 0.5,
1710
+ "learning_rate": 0.00010401190631628349,
1711
+ "loss": 0.0459,
1712
+ "step": 284
1713
+ },
1714
+ {
1715
+ "epoch": 0.5,
1716
+ "learning_rate": 0.00010343902174357039,
1717
+ "loss": 0.0437,
1718
+ "step": 285
1719
+ },
1720
+ {
1721
+ "epoch": 0.51,
1722
+ "learning_rate": 0.00010286602414641817,
1723
+ "loss": 0.0473,
1724
+ "step": 286
1725
+ },
1726
+ {
1727
+ "epoch": 0.51,
1728
+ "learning_rate": 0.00010229293235655768,
1729
+ "loss": 0.0482,
1730
+ "step": 287
1731
+ },
1732
+ {
1733
+ "epoch": 0.51,
1734
+ "learning_rate": 0.00010171976520881552,
1735
+ "loss": 0.0444,
1736
+ "step": 288
1737
+ },
1738
+ {
1739
+ "epoch": 0.51,
1740
+ "learning_rate": 0.0001011465415404949,
1741
+ "loss": 0.0452,
1742
+ "step": 289
1743
+ },
1744
+ {
1745
+ "epoch": 0.51,
1746
+ "learning_rate": 0.00010057328019075669,
1747
+ "loss": 0.043,
1748
+ "step": 290
1749
+ },
1750
+ {
1751
+ "epoch": 0.52,
1752
+ "learning_rate": 0.0001,
1753
+ "loss": 0.0305,
1754
+ "step": 291
1755
+ },
1756
+ {
1757
+ "epoch": 0.52,
1758
+ "learning_rate": 9.942671980924336e-05,
1759
+ "loss": 0.0615,
1760
+ "step": 292
1761
+ },
1762
+ {
1763
+ "epoch": 0.52,
1764
+ "learning_rate": 9.88534584595051e-05,
1765
+ "loss": 0.0558,
1766
+ "step": 293
1767
+ },
1768
+ {
1769
+ "epoch": 0.52,
1770
+ "learning_rate": 9.82802347911845e-05,
1771
+ "loss": 0.0412,
1772
+ "step": 294
1773
+ },
1774
+ {
1775
+ "epoch": 0.52,
1776
+ "learning_rate": 9.770706764344234e-05,
1777
+ "loss": 0.0436,
1778
+ "step": 295
1779
+ },
1780
+ {
1781
+ "epoch": 0.52,
1782
+ "learning_rate": 9.713397585358188e-05,
1783
+ "loss": 0.0622,
1784
+ "step": 296
1785
+ },
1786
+ {
1787
+ "epoch": 0.53,
1788
+ "learning_rate": 9.656097825642961e-05,
1789
+ "loss": 0.0441,
1790
+ "step": 297
1791
+ },
1792
+ {
1793
+ "epoch": 0.53,
1794
+ "learning_rate": 9.598809368371655e-05,
1795
+ "loss": 0.0313,
1796
+ "step": 298
1797
+ },
1798
+ {
1799
+ "epoch": 0.53,
1800
+ "learning_rate": 9.541534096345897e-05,
1801
+ "loss": 0.0385,
1802
+ "step": 299
1803
+ },
1804
+ {
1805
+ "epoch": 0.53,
1806
+ "learning_rate": 9.484273891933982e-05,
1807
+ "loss": 0.0304,
1808
+ "step": 300
1809
+ },
1810
+ {
1811
+ "epoch": 0.53,
1812
+ "learning_rate": 9.427030637009003e-05,
1813
+ "loss": 0.0484,
1814
+ "step": 301
1815
+ },
1816
+ {
1817
+ "epoch": 0.53,
1818
+ "learning_rate": 9.369806212887008e-05,
1819
+ "loss": 0.0407,
1820
+ "step": 302
1821
+ },
1822
+ {
1823
+ "epoch": 0.54,
1824
+ "learning_rate": 9.31260250026516e-05,
1825
+ "loss": 0.0474,
1826
+ "step": 303
1827
+ },
1828
+ {
1829
+ "epoch": 0.54,
1830
+ "learning_rate": 9.255421379159934e-05,
1831
+ "loss": 0.053,
1832
+ "step": 304
1833
+ },
1834
+ {
1835
+ "epoch": 0.54,
1836
+ "learning_rate": 9.198264728845332e-05,
1837
+ "loss": 0.0351,
1838
+ "step": 305
1839
+ },
1840
+ {
1841
+ "epoch": 0.54,
1842
+ "learning_rate": 9.14113442779111e-05,
1843
+ "loss": 0.0296,
1844
+ "step": 306
1845
+ },
1846
+ {
1847
+ "epoch": 0.54,
1848
+ "learning_rate": 9.084032353601053e-05,
1849
+ "loss": 0.0373,
1850
+ "step": 307
1851
+ },
1852
+ {
1853
+ "epoch": 0.55,
1854
+ "learning_rate": 9.026960382951253e-05,
1855
+ "loss": 0.0503,
1856
+ "step": 308
1857
+ },
1858
+ {
1859
+ "epoch": 0.55,
1860
+ "learning_rate": 8.969920391528458e-05,
1861
+ "loss": 0.0369,
1862
+ "step": 309
1863
+ },
1864
+ {
1865
+ "epoch": 0.55,
1866
+ "learning_rate": 8.912914253968392e-05,
1867
+ "loss": 0.0516,
1868
+ "step": 310
1869
+ },
1870
+ {
1871
+ "epoch": 0.55,
1872
+ "learning_rate": 8.85594384379417e-05,
1873
+ "loss": 0.0465,
1874
+ "step": 311
1875
+ },
1876
+ {
1877
+ "epoch": 0.55,
1878
+ "learning_rate": 8.799011033354717e-05,
1879
+ "loss": 0.0421,
1880
+ "step": 312
1881
+ },
1882
+ {
1883
+ "epoch": 0.55,
1884
+ "learning_rate": 8.742117693763227e-05,
1885
+ "loss": 0.0312,
1886
+ "step": 313
1887
+ },
1888
+ {
1889
+ "epoch": 0.56,
1890
+ "learning_rate": 8.685265694835681e-05,
1891
+ "loss": 0.0424,
1892
+ "step": 314
1893
+ },
1894
+ {
1895
+ "epoch": 0.56,
1896
+ "learning_rate": 8.628456905029383e-05,
1897
+ "loss": 0.029,
1898
+ "step": 315
1899
+ },
1900
+ {
1901
+ "epoch": 0.56,
1902
+ "learning_rate": 8.571693191381545e-05,
1903
+ "loss": 0.0361,
1904
+ "step": 316
1905
+ },
1906
+ {
1907
+ "epoch": 0.56,
1908
+ "learning_rate": 8.514976419447964e-05,
1909
+ "loss": 0.0519,
1910
+ "step": 317
1911
+ },
1912
+ {
1913
+ "epoch": 0.56,
1914
+ "learning_rate": 8.458308453241663e-05,
1915
+ "loss": 0.0911,
1916
+ "step": 318
1917
+ },
1918
+ {
1919
+ "epoch": 0.56,
1920
+ "learning_rate": 8.401691155171653e-05,
1921
+ "loss": 0.0532,
1922
+ "step": 319
1923
+ },
1924
+ {
1925
+ "epoch": 0.57,
1926
+ "learning_rate": 8.345126385981736e-05,
1927
+ "loss": 0.0399,
1928
+ "step": 320
1929
+ },
1930
+ {
1931
+ "epoch": 0.57,
1932
+ "learning_rate": 8.28861600468932e-05,
1933
+ "loss": 0.0337,
1934
+ "step": 321
1935
+ },
1936
+ {
1937
+ "epoch": 0.57,
1938
+ "learning_rate": 8.232161868524351e-05,
1939
+ "loss": 0.0343,
1940
+ "step": 322
1941
+ },
1942
+ {
1943
+ "epoch": 0.57,
1944
+ "learning_rate": 8.175765832868251e-05,
1945
+ "loss": 0.0413,
1946
+ "step": 323
1947
+ },
1948
+ {
1949
+ "epoch": 0.57,
1950
+ "learning_rate": 8.11942975119297e-05,
1951
+ "loss": 0.0669,
1952
+ "step": 324
1953
+ },
1954
+ {
1955
+ "epoch": 0.58,
1956
+ "learning_rate": 8.063155475000037e-05,
1957
+ "loss": 0.0424,
1958
+ "step": 325
1959
+ },
1960
+ {
1961
+ "epoch": 0.58,
1962
+ "learning_rate": 8.006944853759732e-05,
1963
+ "loss": 0.0319,
1964
+ "step": 326
1965
+ },
1966
+ {
1967
+ "epoch": 0.58,
1968
+ "learning_rate": 7.950799734850292e-05,
1969
+ "loss": 0.0459,
1970
+ "step": 327
1971
+ },
1972
+ {
1973
+ "epoch": 0.58,
1974
+ "learning_rate": 7.894721963497213e-05,
1975
+ "loss": 0.0605,
1976
+ "step": 328
1977
+ },
1978
+ {
1979
+ "epoch": 0.58,
1980
+ "learning_rate": 7.838713382712583e-05,
1981
+ "loss": 0.037,
1982
+ "step": 329
1983
+ },
1984
+ {
1985
+ "epoch": 0.58,
1986
+ "learning_rate": 7.782775833234521e-05,
1987
+ "loss": 0.0436,
1988
+ "step": 330
1989
+ },
1990
+ {
1991
+ "epoch": 0.59,
1992
+ "learning_rate": 7.726911153466698e-05,
1993
+ "loss": 0.0522,
1994
+ "step": 331
1995
+ },
1996
+ {
1997
+ "epoch": 0.59,
1998
+ "learning_rate": 7.67112117941788e-05,
1999
+ "loss": 0.0364,
2000
+ "step": 332
2001
+ },
2002
+ {
2003
+ "epoch": 0.59,
2004
+ "learning_rate": 7.615407744641619e-05,
2005
+ "loss": 0.0486,
2006
+ "step": 333
2007
+ },
2008
+ {
2009
+ "epoch": 0.59,
2010
+ "learning_rate": 7.559772680175978e-05,
2011
+ "loss": 0.0498,
2012
+ "step": 334
2013
+ },
2014
+ {
2015
+ "epoch": 0.59,
2016
+ "learning_rate": 7.504217814483365e-05,
2017
+ "loss": 0.0454,
2018
+ "step": 335
2019
+ },
2020
+ {
2021
+ "epoch": 0.59,
2022
+ "learning_rate": 7.448744973390422e-05,
2023
+ "loss": 0.0531,
2024
+ "step": 336
2025
+ },
2026
+ {
2027
+ "epoch": 0.6,
2028
+ "learning_rate": 7.393355980028039e-05,
2029
+ "loss": 0.0423,
2030
+ "step": 337
2031
+ },
2032
+ {
2033
+ "epoch": 0.6,
2034
+ "learning_rate": 7.338052654771407e-05,
2035
+ "loss": 0.0327,
2036
+ "step": 338
2037
+ },
2038
+ {
2039
+ "epoch": 0.6,
2040
+ "learning_rate": 7.282836815180241e-05,
2041
+ "loss": 0.0442,
2042
+ "step": 339
2043
+ },
2044
+ {
2045
+ "epoch": 0.6,
2046
+ "learning_rate": 7.227710275938986e-05,
2047
+ "loss": 0.0453,
2048
+ "step": 340
2049
+ },
2050
+ {
2051
+ "epoch": 0.6,
2052
+ "learning_rate": 7.172674848797219e-05,
2053
+ "loss": 0.0242,
2054
+ "step": 341
2055
+ },
2056
+ {
2057
+ "epoch": 0.61,
2058
+ "learning_rate": 7.117732342510092e-05,
2059
+ "loss": 0.0364,
2060
+ "step": 342
2061
+ },
2062
+ {
2063
+ "epoch": 0.61,
2064
+ "learning_rate": 7.062884562778882e-05,
2065
+ "loss": 0.0348,
2066
+ "step": 343
2067
+ },
2068
+ {
2069
+ "epoch": 0.61,
2070
+ "learning_rate": 7.00813331219165e-05,
2071
+ "loss": 0.033,
2072
+ "step": 344
2073
+ },
2074
+ {
2075
+ "epoch": 0.61,
2076
+ "learning_rate": 6.953480390164e-05,
2077
+ "loss": 0.0349,
2078
+ "step": 345
2079
+ },
2080
+ {
2081
+ "epoch": 0.61,
2082
+ "learning_rate": 6.898927592879945e-05,
2083
+ "loss": 0.0554,
2084
+ "step": 346
2085
+ },
2086
+ {
2087
+ "epoch": 0.61,
2088
+ "learning_rate": 6.844476713232862e-05,
2089
+ "loss": 0.0457,
2090
+ "step": 347
2091
+ },
2092
+ {
2093
+ "epoch": 0.62,
2094
+ "learning_rate": 6.790129540766582e-05,
2095
+ "loss": 0.0287,
2096
+ "step": 348
2097
+ },
2098
+ {
2099
+ "epoch": 0.62,
2100
+ "learning_rate": 6.735887861616556e-05,
2101
+ "loss": 0.0588,
2102
+ "step": 349
2103
+ },
2104
+ {
2105
+ "epoch": 0.62,
2106
+ "learning_rate": 6.681753458451189e-05,
2107
+ "loss": 0.0385,
2108
+ "step": 350
2109
+ },
2110
+ {
2111
+ "epoch": 0.62,
2112
+ "learning_rate": 6.627728110413213e-05,
2113
+ "loss": 0.0485,
2114
+ "step": 351
2115
+ },
2116
+ {
2117
+ "epoch": 0.62,
2118
+ "learning_rate": 6.573813593061236e-05,
2119
+ "loss": 0.0427,
2120
+ "step": 352
2121
+ },
2122
+ {
2123
+ "epoch": 0.62,
2124
+ "learning_rate": 6.520011678311382e-05,
2125
+ "loss": 0.0458,
2126
+ "step": 353
2127
+ },
2128
+ {
2129
+ "epoch": 0.63,
2130
+ "learning_rate": 6.466324134379066e-05,
2131
+ "loss": 0.0391,
2132
+ "step": 354
2133
+ },
2134
+ {
2135
+ "epoch": 0.63,
2136
+ "learning_rate": 6.412752725720864e-05,
2137
+ "loss": 0.0429,
2138
+ "step": 355
2139
+ },
2140
+ {
2141
+ "epoch": 0.63,
2142
+ "learning_rate": 6.359299212976534e-05,
2143
+ "loss": 0.0416,
2144
+ "step": 356
2145
+ },
2146
+ {
2147
+ "epoch": 0.63,
2148
+ "learning_rate": 6.305965352911161e-05,
2149
+ "loss": 0.0296,
2150
+ "step": 357
2151
+ },
2152
+ {
2153
+ "epoch": 0.63,
2154
+ "learning_rate": 6.252752898357396e-05,
2155
+ "loss": 0.0491,
2156
+ "step": 358
2157
+ },
2158
+ {
2159
+ "epoch": 0.64,
2160
+ "learning_rate": 6.199663598157876e-05,
2161
+ "loss": 0.053,
2162
+ "step": 359
2163
+ },
2164
+ {
2165
+ "epoch": 0.64,
2166
+ "learning_rate": 6.146699197107715e-05,
2167
+ "loss": 0.032,
2168
+ "step": 360
2169
+ },
2170
+ {
2171
+ "epoch": 0.64,
2172
+ "learning_rate": 6.093861435897208e-05,
2173
+ "loss": 0.0361,
2174
+ "step": 361
2175
+ },
2176
+ {
2177
+ "epoch": 0.64,
2178
+ "learning_rate": 6.041152051054575e-05,
2179
+ "loss": 0.0456,
2180
+ "step": 362
2181
+ },
2182
+ {
2183
+ "epoch": 0.64,
2184
+ "learning_rate": 5.988572774888913e-05,
2185
+ "loss": 0.0404,
2186
+ "step": 363
2187
+ },
2188
+ {
2189
+ "epoch": 0.64,
2190
+ "learning_rate": 5.936125335433265e-05,
2191
+ "loss": 0.0641,
2192
+ "step": 364
2193
+ },
2194
+ {
2195
+ "epoch": 0.65,
2196
+ "learning_rate": 5.883811456387821e-05,
2197
+ "loss": 0.0281,
2198
+ "step": 365
2199
+ },
2200
+ {
2201
+ "epoch": 0.65,
2202
+ "learning_rate": 5.8316328570632706e-05,
2203
+ "loss": 0.046,
2204
+ "step": 366
2205
+ },
2206
+ {
2207
+ "epoch": 0.65,
2208
+ "learning_rate": 5.779591252324286e-05,
2209
+ "loss": 0.03,
2210
+ "step": 367
2211
+ },
2212
+ {
2213
+ "epoch": 0.65,
2214
+ "learning_rate": 5.7276883525331915e-05,
2215
+ "loss": 0.0237,
2216
+ "step": 368
2217
+ },
2218
+ {
2219
+ "epoch": 0.65,
2220
+ "learning_rate": 5.675925863493721e-05,
2221
+ "loss": 0.038,
2222
+ "step": 369
2223
+ },
2224
+ {
2225
+ "epoch": 0.65,
2226
+ "learning_rate": 5.624305486394967e-05,
2227
+ "loss": 0.0279,
2228
+ "step": 370
2229
+ },
2230
+ {
2231
+ "epoch": 0.66,
2232
+ "learning_rate": 5.5728289177554805e-05,
2233
+ "loss": 0.0412,
2234
+ "step": 371
2235
+ },
2236
+ {
2237
+ "epoch": 0.66,
2238
+ "learning_rate": 5.5214978493675005e-05,
2239
+ "loss": 0.0357,
2240
+ "step": 372
2241
+ },
2242
+ {
2243
+ "epoch": 0.66,
2244
+ "learning_rate": 5.4703139682413586e-05,
2245
+ "loss": 0.0419,
2246
+ "step": 373
2247
+ },
2248
+ {
2249
+ "epoch": 0.66,
2250
+ "learning_rate": 5.419278956550037e-05,
2251
+ "loss": 0.0483,
2252
+ "step": 374
2253
+ },
2254
+ {
2255
+ "epoch": 0.66,
2256
+ "learning_rate": 5.368394491573876e-05,
2257
+ "loss": 0.0379,
2258
+ "step": 375
2259
+ },
2260
+ {
2261
+ "epoch": 0.67,
2262
+ "learning_rate": 5.3176622456454693e-05,
2263
+ "loss": 0.0452,
2264
+ "step": 376
2265
+ },
2266
+ {
2267
+ "epoch": 0.67,
2268
+ "learning_rate": 5.267083886094668e-05,
2269
+ "loss": 0.0276,
2270
+ "step": 377
2271
+ },
2272
+ {
2273
+ "epoch": 0.67,
2274
+ "learning_rate": 5.216661075193814e-05,
2275
+ "loss": 0.0352,
2276
+ "step": 378
2277
+ },
2278
+ {
2279
+ "epoch": 0.67,
2280
+ "learning_rate": 5.1663954701030914e-05,
2281
+ "loss": 0.0193,
2282
+ "step": 379
2283
+ },
2284
+ {
2285
+ "epoch": 0.67,
2286
+ "learning_rate": 5.116288722816086e-05,
2287
+ "loss": 0.0366,
2288
+ "step": 380
2289
+ },
2290
+ {
2291
+ "epoch": 0.67,
2292
+ "learning_rate": 5.0663424801054595e-05,
2293
+ "loss": 0.0458,
2294
+ "step": 381
2295
+ },
2296
+ {
2297
+ "epoch": 0.68,
2298
+ "learning_rate": 5.016558383468851e-05,
2299
+ "loss": 0.036,
2300
+ "step": 382
2301
+ },
2302
+ {
2303
+ "epoch": 0.68,
2304
+ "learning_rate": 4.966938069074921e-05,
2305
+ "loss": 0.0297,
2306
+ "step": 383
2307
+ },
2308
+ {
2309
+ "epoch": 0.68,
2310
+ "learning_rate": 4.9174831677095804e-05,
2311
+ "loss": 0.0642,
2312
+ "step": 384
2313
+ },
2314
+ {
2315
+ "epoch": 0.68,
2316
+ "learning_rate": 4.8681953047223914e-05,
2317
+ "loss": 0.0419,
2318
+ "step": 385
2319
+ },
2320
+ {
2321
+ "epoch": 0.68,
2322
+ "learning_rate": 4.8190760999731524e-05,
2323
+ "loss": 0.0356,
2324
+ "step": 386
2325
+ },
2326
+ {
2327
+ "epoch": 0.68,
2328
+ "learning_rate": 4.7701271677786695e-05,
2329
+ "loss": 0.0522,
2330
+ "step": 387
2331
+ },
2332
+ {
2333
+ "epoch": 0.69,
2334
+ "learning_rate": 4.721350116859675e-05,
2335
+ "loss": 0.0339,
2336
+ "step": 388
2337
+ },
2338
+ {
2339
+ "epoch": 0.69,
2340
+ "learning_rate": 4.672746550287985e-05,
2341
+ "loss": 0.0447,
2342
+ "step": 389
2343
+ },
2344
+ {
2345
+ "epoch": 0.69,
2346
+ "learning_rate": 4.624318065433797e-05,
2347
+ "loss": 0.026,
2348
+ "step": 390
2349
+ },
2350
+ {
2351
+ "epoch": 0.69,
2352
+ "learning_rate": 4.576066253913208e-05,
2353
+ "loss": 0.0383,
2354
+ "step": 391
2355
+ },
2356
+ {
2357
+ "epoch": 0.69,
2358
+ "learning_rate": 4.5279927015358836e-05,
2359
+ "loss": 0.0454,
2360
+ "step": 392
2361
+ },
2362
+ {
2363
+ "epoch": 0.7,
2364
+ "learning_rate": 4.4800989882529574e-05,
2365
+ "loss": 0.0302,
2366
+ "step": 393
2367
+ },
2368
+ {
2369
+ "epoch": 0.7,
2370
+ "learning_rate": 4.432386688105095e-05,
2371
+ "loss": 0.0293,
2372
+ "step": 394
2373
+ },
2374
+ {
2375
+ "epoch": 0.7,
2376
+ "learning_rate": 4.3848573691707717e-05,
2377
+ "loss": 0.0544,
2378
+ "step": 395
2379
+ },
2380
+ {
2381
+ "epoch": 0.7,
2382
+ "learning_rate": 4.337512593514729e-05,
2383
+ "loss": 0.0411,
2384
+ "step": 396
2385
+ },
2386
+ {
2387
+ "epoch": 0.7,
2388
+ "learning_rate": 4.2903539171366393e-05,
2389
+ "loss": 0.0356,
2390
+ "step": 397
2391
+ },
2392
+ {
2393
+ "epoch": 0.7,
2394
+ "learning_rate": 4.243382889919981e-05,
2395
+ "loss": 0.0313,
2396
+ "step": 398
2397
+ },
2398
+ {
2399
+ "epoch": 0.71,
2400
+ "learning_rate": 4.19660105558107e-05,
2401
+ "loss": 0.0424,
2402
+ "step": 399
2403
+ },
2404
+ {
2405
+ "epoch": 0.71,
2406
+ "learning_rate": 4.1500099516183555e-05,
2407
+ "loss": 0.0315,
2408
+ "step": 400
2409
+ },
2410
+ {
2411
+ "epoch": 0.71,
2412
+ "learning_rate": 4.1036111092618725e-05,
2413
+ "loss": 0.0462,
2414
+ "step": 401
2415
+ },
2416
+ {
2417
+ "epoch": 0.71,
2418
+ "learning_rate": 4.057406053422933e-05,
2419
+ "loss": 0.0423,
2420
+ "step": 402
2421
+ },
2422
+ {
2423
+ "epoch": 0.71,
2424
+ "learning_rate": 4.0113963026439884e-05,
2425
+ "loss": 0.0368,
2426
+ "step": 403
2427
+ },
2428
+ {
2429
+ "epoch": 0.72,
2430
+ "learning_rate": 3.965583369048738e-05,
2431
+ "loss": 0.036,
2432
+ "step": 404
2433
+ },
2434
+ {
2435
+ "epoch": 0.72,
2436
+ "learning_rate": 3.919968758292425e-05,
2437
+ "loss": 0.0277,
2438
+ "step": 405
2439
+ },
2440
+ {
2441
+ "epoch": 0.72,
2442
+ "learning_rate": 3.8745539695123575e-05,
2443
+ "loss": 0.0232,
2444
+ "step": 406
2445
+ },
2446
+ {
2447
+ "epoch": 0.72,
2448
+ "learning_rate": 3.829340495278639e-05,
2449
+ "loss": 0.03,
2450
+ "step": 407
2451
+ },
2452
+ {
2453
+ "epoch": 0.72,
2454
+ "learning_rate": 3.784329821545105e-05,
2455
+ "loss": 0.0285,
2456
+ "step": 408
2457
+ },
2458
+ {
2459
+ "epoch": 0.72,
2460
+ "learning_rate": 3.7395234276005087e-05,
2461
+ "loss": 0.0298,
2462
+ "step": 409
2463
+ },
2464
+ {
2465
+ "epoch": 0.73,
2466
+ "learning_rate": 3.694922786019871e-05,
2467
+ "loss": 0.027,
2468
+ "step": 410
2469
+ },
2470
+ {
2471
+ "epoch": 0.73,
2472
+ "learning_rate": 3.650529362616113e-05,
2473
+ "loss": 0.03,
2474
+ "step": 411
2475
+ },
2476
+ {
2477
+ "epoch": 0.73,
2478
+ "learning_rate": 3.6063446163918667e-05,
2479
+ "loss": 0.0294,
2480
+ "step": 412
2481
+ },
2482
+ {
2483
+ "epoch": 0.73,
2484
+ "learning_rate": 3.562369999491536e-05,
2485
+ "loss": 0.0428,
2486
+ "step": 413
2487
+ },
2488
+ {
2489
+ "epoch": 0.73,
2490
+ "learning_rate": 3.5186069571535576e-05,
2491
+ "loss": 0.0218,
2492
+ "step": 414
2493
+ },
2494
+ {
2495
+ "epoch": 0.73,
2496
+ "learning_rate": 3.4750569276629117e-05,
2497
+ "loss": 0.0461,
2498
+ "step": 415
2499
+ },
2500
+ {
2501
+ "epoch": 0.74,
2502
+ "learning_rate": 3.431721342303839e-05,
2503
+ "loss": 0.0333,
2504
+ "step": 416
2505
+ },
2506
+ {
2507
+ "epoch": 0.74,
2508
+ "learning_rate": 3.3886016253128326e-05,
2509
+ "loss": 0.0329,
2510
+ "step": 417
2511
+ },
2512
+ {
2513
+ "epoch": 0.74,
2514
+ "learning_rate": 3.345699193831795e-05,
2515
+ "loss": 0.0427,
2516
+ "step": 418
2517
+ },
2518
+ {
2519
+ "epoch": 0.74,
2520
+ "learning_rate": 3.303015457861478e-05,
2521
+ "loss": 0.055,
2522
+ "step": 419
2523
+ },
2524
+ {
2525
+ "epoch": 0.74,
2526
+ "learning_rate": 3.2605518202151576e-05,
2527
+ "loss": 0.0337,
2528
+ "step": 420
2529
+ },
2530
+ {
2531
+ "epoch": 0.75,
2532
+ "learning_rate": 3.2183096764724915e-05,
2533
+ "loss": 0.0387,
2534
+ "step": 421
2535
+ },
2536
+ {
2537
+ "epoch": 0.75,
2538
+ "learning_rate": 3.176290414933695e-05,
2539
+ "loss": 0.0409,
2540
+ "step": 422
2541
+ },
2542
+ {
2543
+ "epoch": 0.75,
2544
+ "learning_rate": 3.1344954165738835e-05,
2545
+ "loss": 0.0362,
2546
+ "step": 423
2547
+ },
2548
+ {
2549
+ "epoch": 0.75,
2550
+ "learning_rate": 3.092926054997711e-05,
2551
+ "loss": 0.0401,
2552
+ "step": 424
2553
+ },
2554
+ {
2555
+ "epoch": 0.75,
2556
+ "learning_rate": 3.0515836963942056e-05,
2557
+ "loss": 0.027,
2558
+ "step": 425
2559
+ },
2560
+ {
2561
+ "epoch": 0.75,
2562
+ "learning_rate": 3.01046969949188e-05,
2563
+ "loss": 0.0515,
2564
+ "step": 426
2565
+ },
2566
+ {
2567
+ "epoch": 0.76,
2568
+ "learning_rate": 2.9695854155140646e-05,
2569
+ "loss": 0.0368,
2570
+ "step": 427
2571
+ },
2572
+ {
2573
+ "epoch": 0.76,
2574
+ "learning_rate": 2.9289321881345254e-05,
2575
+ "loss": 0.0355,
2576
+ "step": 428
2577
+ },
2578
+ {
2579
+ "epoch": 0.76,
2580
+ "learning_rate": 2.888511353433274e-05,
2581
+ "loss": 0.0436,
2582
+ "step": 429
2583
+ },
2584
+ {
2585
+ "epoch": 0.76,
2586
+ "learning_rate": 2.848324239852672e-05,
2587
+ "loss": 0.027,
2588
+ "step": 430
2589
+ },
2590
+ {
2591
+ "epoch": 0.76,
2592
+ "learning_rate": 2.80837216815378e-05,
2593
+ "loss": 0.0222,
2594
+ "step": 431
2595
+ },
2596
+ {
2597
+ "epoch": 0.76,
2598
+ "learning_rate": 2.7686564513729198e-05,
2599
+ "loss": 0.0415,
2600
+ "step": 432
2601
+ },
2602
+ {
2603
+ "epoch": 0.77,
2604
+ "learning_rate": 2.7291783947785543e-05,
2605
+ "loss": 0.0361,
2606
+ "step": 433
2607
+ },
2608
+ {
2609
+ "epoch": 0.77,
2610
+ "learning_rate": 2.6899392958283708e-05,
2611
+ "loss": 0.0359,
2612
+ "step": 434
2613
+ },
2614
+ {
2615
+ "epoch": 0.77,
2616
+ "learning_rate": 2.650940444126654e-05,
2617
+ "loss": 0.0383,
2618
+ "step": 435
2619
+ },
2620
+ {
2621
+ "epoch": 0.77,
2622
+ "learning_rate": 2.6121831213818827e-05,
2623
+ "loss": 0.0238,
2624
+ "step": 436
2625
+ },
2626
+ {
2627
+ "epoch": 0.77,
2628
+ "learning_rate": 2.5736686013646228e-05,
2629
+ "loss": 0.0423,
2630
+ "step": 437
2631
+ },
2632
+ {
2633
+ "epoch": 0.78,
2634
+ "learning_rate": 2.535398149865651e-05,
2635
+ "loss": 0.037,
2636
+ "step": 438
2637
+ },
2638
+ {
2639
+ "epoch": 0.78,
2640
+ "learning_rate": 2.4973730246543735e-05,
2641
+ "loss": 0.0451,
2642
+ "step": 439
2643
+ },
2644
+ {
2645
+ "epoch": 0.78,
2646
+ "learning_rate": 2.4595944754374722e-05,
2647
+ "loss": 0.0475,
2648
+ "step": 440
2649
+ },
2650
+ {
2651
+ "epoch": 0.78,
2652
+ "learning_rate": 2.4220637438178317e-05,
2653
+ "loss": 0.0317,
2654
+ "step": 441
2655
+ },
2656
+ {
2657
+ "epoch": 0.78,
2658
+ "learning_rate": 2.3847820632537564e-05,
2659
+ "loss": 0.0387,
2660
+ "step": 442
2661
+ },
2662
+ {
2663
+ "epoch": 0.78,
2664
+ "learning_rate": 2.3477506590183972e-05,
2665
+ "loss": 0.0262,
2666
+ "step": 443
2667
+ },
2668
+ {
2669
+ "epoch": 0.79,
2670
+ "learning_rate": 2.310970748159511e-05,
2671
+ "loss": 0.043,
2672
+ "step": 444
2673
+ },
2674
+ {
2675
+ "epoch": 0.79,
2676
+ "learning_rate": 2.2744435394594497e-05,
2677
+ "loss": 0.0383,
2678
+ "step": 445
2679
+ },
2680
+ {
2681
+ "epoch": 0.79,
2682
+ "learning_rate": 2.2381702333954434e-05,
2683
+ "loss": 0.0223,
2684
+ "step": 446
2685
+ },
2686
+ {
2687
+ "epoch": 0.79,
2688
+ "learning_rate": 2.20215202210013e-05,
2689
+ "loss": 0.0421,
2690
+ "step": 447
2691
+ },
2692
+ {
2693
+ "epoch": 0.79,
2694
+ "learning_rate": 2.1663900893223898e-05,
2695
+ "loss": 0.0339,
2696
+ "step": 448
2697
+ },
2698
+ {
2699
+ "epoch": 0.79,
2700
+ "learning_rate": 2.130885610388428e-05,
2701
+ "loss": 0.0203,
2702
+ "step": 449
2703
+ },
2704
+ {
2705
+ "epoch": 0.8,
2706
+ "learning_rate": 2.0956397521631664e-05,
2707
+ "loss": 0.0435,
2708
+ "step": 450
2709
+ },
2710
+ {
2711
+ "epoch": 0.8,
2712
+ "learning_rate": 2.0606536730118763e-05,
2713
+ "loss": 0.0309,
2714
+ "step": 451
2715
+ },
2716
+ {
2717
+ "epoch": 0.8,
2718
+ "learning_rate": 2.025928522762115e-05,
2719
+ "loss": 0.0286,
2720
+ "step": 452
2721
+ },
2722
+ {
2723
+ "epoch": 0.8,
2724
+ "learning_rate": 1.9914654426659374e-05,
2725
+ "loss": 0.0382,
2726
+ "step": 453
2727
+ },
2728
+ {
2729
+ "epoch": 0.8,
2730
+ "learning_rate": 1.9572655653623884e-05,
2731
+ "loss": 0.025,
2732
+ "step": 454
2733
+ },
2734
+ {
2735
+ "epoch": 0.81,
2736
+ "learning_rate": 1.9233300148402765e-05,
2737
+ "loss": 0.0275,
2738
+ "step": 455
2739
+ },
2740
+ {
2741
+ "epoch": 0.81,
2742
+ "learning_rate": 1.8896599064012298e-05,
2743
+ "loss": 0.0276,
2744
+ "step": 456
2745
+ },
2746
+ {
2747
+ "epoch": 0.81,
2748
+ "learning_rate": 1.8562563466230576e-05,
2749
+ "loss": 0.0247,
2750
+ "step": 457
2751
+ },
2752
+ {
2753
+ "epoch": 0.81,
2754
+ "learning_rate": 1.823120433323361e-05,
2755
+ "loss": 0.0309,
2756
+ "step": 458
2757
+ },
2758
+ {
2759
+ "epoch": 0.81,
2760
+ "learning_rate": 1.790253255523465e-05,
2761
+ "loss": 0.0301,
2762
+ "step": 459
2763
+ },
2764
+ {
2765
+ "epoch": 0.81,
2766
+ "learning_rate": 1.757655893412622e-05,
2767
+ "loss": 0.0288,
2768
+ "step": 460
2769
+ },
2770
+ {
2771
+ "epoch": 0.82,
2772
+ "learning_rate": 1.7253294183125223e-05,
2773
+ "loss": 0.0345,
2774
+ "step": 461
2775
+ },
2776
+ {
2777
+ "epoch": 0.82,
2778
+ "learning_rate": 1.6932748926420693e-05,
2779
+ "loss": 0.0227,
2780
+ "step": 462
2781
+ },
2782
+ {
2783
+ "epoch": 0.82,
2784
+ "learning_rate": 1.6614933698824732e-05,
2785
+ "loss": 0.0436,
2786
+ "step": 463
2787
+ },
2788
+ {
2789
+ "epoch": 0.82,
2790
+ "learning_rate": 1.629985894542625e-05,
2791
+ "loss": 0.0432,
2792
+ "step": 464
2793
+ },
2794
+ {
2795
+ "epoch": 0.82,
2796
+ "learning_rate": 1.5987535021247667e-05,
2797
+ "loss": 0.0312,
2798
+ "step": 465
2799
+ },
2800
+ {
2801
+ "epoch": 0.82,
2802
+ "learning_rate": 1.5677972190904622e-05,
2803
+ "loss": 0.0473,
2804
+ "step": 466
2805
+ },
2806
+ {
2807
+ "epoch": 0.83,
2808
+ "learning_rate": 1.5371180628268587e-05,
2809
+ "loss": 0.042,
2810
+ "step": 467
2811
+ },
2812
+ {
2813
+ "epoch": 0.83,
2814
+ "learning_rate": 1.5067170416132603e-05,
2815
+ "loss": 0.0335,
2816
+ "step": 468
2817
+ },
2818
+ {
2819
+ "epoch": 0.83,
2820
+ "learning_rate": 1.476595154587973e-05,
2821
+ "loss": 0.0515,
2822
+ "step": 469
2823
+ },
2824
+ {
2825
+ "epoch": 0.83,
2826
+ "learning_rate": 1.4467533917154841e-05,
2827
+ "loss": 0.0204,
2828
+ "step": 470
2829
+ },
2830
+ {
2831
+ "epoch": 0.83,
2832
+ "learning_rate": 1.4171927337539104e-05,
2833
+ "loss": 0.0417,
2834
+ "step": 471
2835
+ },
2836
+ {
2837
+ "epoch": 0.84,
2838
+ "learning_rate": 1.3879141522227878e-05,
2839
+ "loss": 0.0357,
2840
+ "step": 472
2841
+ },
2842
+ {
2843
+ "epoch": 0.84,
2844
+ "learning_rate": 1.3589186093711226e-05,
2845
+ "loss": 0.0219,
2846
+ "step": 473
2847
+ },
2848
+ {
2849
+ "epoch": 0.84,
2850
+ "learning_rate": 1.3302070581457715e-05,
2851
+ "loss": 0.0349,
2852
+ "step": 474
2853
+ },
2854
+ {
2855
+ "epoch": 0.84,
2856
+ "learning_rate": 1.3017804421601298e-05,
2857
+ "loss": 0.0346,
2858
+ "step": 475
2859
+ },
2860
+ {
2861
+ "epoch": 0.84,
2862
+ "learning_rate": 1.273639695663108e-05,
2863
+ "loss": 0.0392,
2864
+ "step": 476
2865
+ },
2866
+ {
2867
+ "epoch": 0.84,
2868
+ "learning_rate": 1.2457857435084408e-05,
2869
+ "loss": 0.0237,
2870
+ "step": 477
2871
+ },
2872
+ {
2873
+ "epoch": 0.85,
2874
+ "learning_rate": 1.2182195011242748e-05,
2875
+ "loss": 0.0156,
2876
+ "step": 478
2877
+ },
2878
+ {
2879
+ "epoch": 0.85,
2880
+ "learning_rate": 1.1909418744831047e-05,
2881
+ "loss": 0.0271,
2882
+ "step": 479
2883
+ },
2884
+ {
2885
+ "epoch": 0.85,
2886
+ "learning_rate": 1.1639537600719763e-05,
2887
+ "loss": 0.0395,
2888
+ "step": 480
2889
+ },
2890
+ {
2891
+ "epoch": 0.85,
2892
+ "learning_rate": 1.1372560448630376e-05,
2893
+ "loss": 0.0345,
2894
+ "step": 481
2895
+ },
2896
+ {
2897
+ "epoch": 0.85,
2898
+ "learning_rate": 1.1108496062843742e-05,
2899
+ "loss": 0.034,
2900
+ "step": 482
2901
+ },
2902
+ {
2903
+ "epoch": 0.85,
2904
+ "learning_rate": 1.0847353121911951e-05,
2905
+ "loss": 0.032,
2906
+ "step": 483
2907
+ },
2908
+ {
2909
+ "epoch": 0.86,
2910
+ "learning_rate": 1.0589140208372873e-05,
2911
+ "loss": 0.0218,
2912
+ "step": 484
2913
+ },
2914
+ {
2915
+ "epoch": 0.86,
2916
+ "learning_rate": 1.0333865808468202e-05,
2917
+ "loss": 0.0342,
2918
+ "step": 485
2919
+ },
2920
+ {
2921
+ "epoch": 0.86,
2922
+ "learning_rate": 1.0081538311864569e-05,
2923
+ "loss": 0.0405,
2924
+ "step": 486
2925
+ },
2926
+ {
2927
+ "epoch": 0.86,
2928
+ "learning_rate": 9.83216601137773e-06,
2929
+ "loss": 0.0347,
2930
+ "step": 487
2931
+ },
2932
+ {
2933
+ "epoch": 0.86,
2934
+ "learning_rate": 9.58575710270011e-06,
2935
+ "loss": 0.027,
2936
+ "step": 488
2937
+ },
2938
+ {
2939
+ "epoch": 0.87,
2940
+ "learning_rate": 9.342319684131395e-06,
2941
+ "loss": 0.0285,
2942
+ "step": 489
2943
+ },
2944
+ {
2945
+ "epoch": 0.87,
2946
+ "learning_rate": 9.10186175631237e-06,
2947
+ "loss": 0.038,
2948
+ "step": 490
2949
+ },
2950
+ {
2951
+ "epoch": 0.87,
2952
+ "learning_rate": 8.864391221962065e-06,
2953
+ "loss": 0.0257,
2954
+ "step": 491
2955
+ },
2956
+ {
2957
+ "epoch": 0.87,
2958
+ "learning_rate": 8.629915885617911e-06,
2959
+ "loss": 0.0314,
2960
+ "step": 492
2961
+ },
2962
+ {
2963
+ "epoch": 0.87,
2964
+ "learning_rate": 8.398443453379267e-06,
2965
+ "loss": 0.0314,
2966
+ "step": 493
2967
+ },
2968
+ {
2969
+ "epoch": 0.87,
2970
+ "learning_rate": 8.16998153265427e-06,
2971
+ "loss": 0.0275,
2972
+ "step": 494
2973
+ },
2974
+ {
2975
+ "epoch": 0.88,
2976
+ "learning_rate": 7.944537631909665e-06,
2977
+ "loss": 0.0206,
2978
+ "step": 495
2979
+ },
2980
+ {
2981
+ "epoch": 0.88,
2982
+ "learning_rate": 7.722119160424112e-06,
2983
+ "loss": 0.0287,
2984
+ "step": 496
2985
+ },
2986
+ {
2987
+ "epoch": 0.88,
2988
+ "learning_rate": 7.502733428044683e-06,
2989
+ "loss": 0.0223,
2990
+ "step": 497
2991
+ },
2992
+ {
2993
+ "epoch": 0.88,
2994
+ "learning_rate": 7.2863876449466016e-06,
2995
+ "loss": 0.0214,
2996
+ "step": 498
2997
+ },
2998
+ {
2999
+ "epoch": 0.88,
3000
+ "learning_rate": 7.073088921396287e-06,
3001
+ "loss": 0.0224,
3002
+ "step": 499
3003
+ },
3004
+ {
3005
+ "epoch": 0.88,
3006
+ "learning_rate": 6.862844267517643e-06,
3007
+ "loss": 0.0267,
3008
+ "step": 500
3009
+ },
3010
+ {
3011
+ "epoch": 0.89,
3012
+ "learning_rate": 6.655660593061719e-06,
3013
+ "loss": 0.0337,
3014
+ "step": 501
3015
+ },
3016
+ {
3017
+ "epoch": 0.89,
3018
+ "learning_rate": 6.451544707179635e-06,
3019
+ "loss": 0.0152,
3020
+ "step": 502
3021
+ },
3022
+ {
3023
+ "epoch": 0.89,
3024
+ "learning_rate": 6.250503318198664e-06,
3025
+ "loss": 0.0186,
3026
+ "step": 503
3027
+ },
3028
+ {
3029
+ "epoch": 0.89,
3030
+ "learning_rate": 6.0525430334018915e-06,
3031
+ "loss": 0.0352,
3032
+ "step": 504
3033
+ },
3034
+ {
3035
+ "epoch": 0.89,
3036
+ "learning_rate": 5.857670358811096e-06,
3037
+ "loss": 0.0311,
3038
+ "step": 505
3039
+ },
3040
+ {
3041
+ "epoch": 0.9,
3042
+ "learning_rate": 5.6658916989727695e-06,
3043
+ "loss": 0.0321,
3044
+ "step": 506
3045
+ },
3046
+ {
3047
+ "epoch": 0.9,
3048
+ "learning_rate": 5.477213356747746e-06,
3049
+ "loss": 0.0376,
3050
+ "step": 507
3051
+ },
3052
+ {
3053
+ "epoch": 0.9,
3054
+ "learning_rate": 5.291641533104052e-06,
3055
+ "loss": 0.0283,
3056
+ "step": 508
3057
+ },
3058
+ {
3059
+ "epoch": 0.9,
3060
+ "learning_rate": 5.109182326913054e-06,
3061
+ "loss": 0.0279,
3062
+ "step": 509
3063
+ },
3064
+ {
3065
+ "epoch": 0.9,
3066
+ "learning_rate": 4.929841734749063e-06,
3067
+ "loss": 0.0412,
3068
+ "step": 510
3069
+ },
3070
+ {
3071
+ "epoch": 0.9,
3072
+ "learning_rate": 4.7536256506922506e-06,
3073
+ "loss": 0.0311,
3074
+ "step": 511
3075
+ },
3076
+ {
3077
+ "epoch": 0.91,
3078
+ "learning_rate": 4.5805398661349146e-06,
3079
+ "loss": 0.0241,
3080
+ "step": 512
3081
+ },
3082
+ {
3083
+ "epoch": 0.91,
3084
+ "learning_rate": 4.410590069591192e-06,
3085
+ "loss": 0.0339,
3086
+ "step": 513
3087
+ },
3088
+ {
3089
+ "epoch": 0.91,
3090
+ "learning_rate": 4.243781846510031e-06,
3091
+ "loss": 0.0325,
3092
+ "step": 514
3093
+ },
3094
+ {
3095
+ "epoch": 0.91,
3096
+ "learning_rate": 4.0801206790916814e-06,
3097
+ "loss": 0.0246,
3098
+ "step": 515
3099
+ },
3100
+ {
3101
+ "epoch": 0.91,
3102
+ "learning_rate": 3.919611946107493e-06,
3103
+ "loss": 0.039,
3104
+ "step": 516
3105
+ },
3106
+ {
3107
+ "epoch": 0.92,
3108
+ "learning_rate": 3.7622609227231818e-06,
3109
+ "loss": 0.0281,
3110
+ "step": 517
3111
+ },
3112
+ {
3113
+ "epoch": 0.92,
3114
+ "learning_rate": 3.6080727803254e-06,
3115
+ "loss": 0.0374,
3116
+ "step": 518
3117
+ },
3118
+ {
3119
+ "epoch": 0.92,
3120
+ "learning_rate": 3.4570525863518165e-06,
3121
+ "loss": 0.0309,
3122
+ "step": 519
3123
+ },
3124
+ {
3125
+ "epoch": 0.92,
3126
+ "learning_rate": 3.3092053041245517e-06,
3127
+ "loss": 0.0277,
3128
+ "step": 520
3129
+ },
3130
+ {
3131
+ "epoch": 0.92,
3132
+ "learning_rate": 3.1645357926870955e-06,
3133
+ "loss": 0.0237,
3134
+ "step": 521
3135
+ },
3136
+ {
3137
+ "epoch": 0.92,
3138
+ "learning_rate": 3.0230488066445462e-06,
3139
+ "loss": 0.0404,
3140
+ "step": 522
3141
+ },
3142
+ {
3143
+ "epoch": 0.93,
3144
+ "learning_rate": 2.8847489960074136e-06,
3145
+ "loss": 0.0428,
3146
+ "step": 523
3147
+ },
3148
+ {
3149
+ "epoch": 0.93,
3150
+ "learning_rate": 2.7496409060387973e-06,
3151
+ "loss": 0.043,
3152
+ "step": 524
3153
+ },
3154
+ {
3155
+ "epoch": 0.93,
3156
+ "learning_rate": 2.6177289771049274e-06,
3157
+ "loss": 0.0181,
3158
+ "step": 525
3159
+ },
3160
+ {
3161
+ "epoch": 0.93,
3162
+ "learning_rate": 2.489017544529315e-06,
3163
+ "loss": 0.0233,
3164
+ "step": 526
3165
+ },
3166
+ {
3167
+ "epoch": 0.93,
3168
+ "learning_rate": 2.3635108384502e-06,
3169
+ "loss": 0.033,
3170
+ "step": 527
3171
+ },
3172
+ {
3173
+ "epoch": 0.93,
3174
+ "learning_rate": 2.2412129836816286e-06,
3175
+ "loss": 0.0335,
3176
+ "step": 528
3177
+ },
3178
+ {
3179
+ "epoch": 0.94,
3180
+ "learning_rate": 2.122127999577783e-06,
3181
+ "loss": 0.0182,
3182
+ "step": 529
3183
+ },
3184
+ {
3185
+ "epoch": 0.94,
3186
+ "learning_rate": 2.0062597999009116e-06,
3187
+ "loss": 0.0379,
3188
+ "step": 530
3189
+ },
3190
+ {
3191
+ "epoch": 0.94,
3192
+ "learning_rate": 1.893612192692751e-06,
3193
+ "loss": 0.0407,
3194
+ "step": 531
3195
+ },
3196
+ {
3197
+ "epoch": 0.94,
3198
+ "learning_rate": 1.7841888801493178e-06,
3199
+ "loss": 0.0207,
3200
+ "step": 532
3201
+ },
3202
+ {
3203
+ "epoch": 0.94,
3204
+ "learning_rate": 1.6779934584992718e-06,
3205
+ "loss": 0.0185,
3206
+ "step": 533
3207
+ },
3208
+ {
3209
+ "epoch": 0.95,
3210
+ "learning_rate": 1.5750294178856873e-06,
3211
+ "loss": 0.0182,
3212
+ "step": 534
3213
+ },
3214
+ {
3215
+ "epoch": 0.95,
3216
+ "learning_rate": 1.4753001422514123e-06,
3217
+ "loss": 0.0342,
3218
+ "step": 535
3219
+ },
3220
+ {
3221
+ "epoch": 0.95,
3222
+ "learning_rate": 1.378808909227769e-06,
3223
+ "loss": 0.0187,
3224
+ "step": 536
3225
+ },
3226
+ {
3227
+ "epoch": 0.95,
3228
+ "learning_rate": 1.2855588900269056e-06,
3229
+ "loss": 0.014,
3230
+ "step": 537
3231
+ },
3232
+ {
3233
+ "epoch": 0.95,
3234
+ "learning_rate": 1.1955531493375139e-06,
3235
+ "loss": 0.0164,
3236
+ "step": 538
3237
+ },
3238
+ {
3239
+ "epoch": 0.95,
3240
+ "learning_rate": 1.108794645224187e-06,
3241
+ "loss": 0.0207,
3242
+ "step": 539
3243
+ },
3244
+ {
3245
+ "epoch": 0.96,
3246
+ "learning_rate": 1.0252862290301091e-06,
3247
+ "loss": 0.0188,
3248
+ "step": 540
3249
+ },
3250
+ {
3251
+ "epoch": 0.96,
3252
+ "learning_rate": 9.450306452834179e-07,
3253
+ "loss": 0.0124,
3254
+ "step": 541
3255
+ },
3256
+ {
3257
+ "epoch": 0.96,
3258
+ "learning_rate": 8.68030531606967e-07,
3259
+ "loss": 0.0276,
3260
+ "step": 542
3261
+ },
3262
+ {
3263
+ "epoch": 0.96,
3264
+ "learning_rate": 7.94288418631639e-07,
3265
+ "loss": 0.0299,
3266
+ "step": 543
3267
+ },
3268
+ {
3269
+ "epoch": 0.96,
3270
+ "learning_rate": 7.2380672991319e-07,
3271
+ "loss": 0.0371,
3272
+ "step": 544
3273
+ },
3274
+ {
3275
+ "epoch": 0.96,
3276
+ "learning_rate": 6.565877818526245e-07,
3277
+ "loss": 0.0358,
3278
+ "step": 545
3279
+ },
3280
+ {
3281
+ "epoch": 0.97,
3282
+ "learning_rate": 5.926337836199891e-07,
3283
+ "loss": 0.0255,
3284
+ "step": 546
3285
+ },
3286
+ {
3287
+ "epoch": 0.97,
3288
+ "learning_rate": 5.319468370818537e-07,
3289
+ "loss": 0.0348,
3290
+ "step": 547
3291
+ },
3292
+ {
3293
+ "epoch": 0.97,
3294
+ "learning_rate": 4.745289367321659e-07,
3295
+ "loss": 0.025,
3296
+ "step": 548
3297
+ },
3298
+ {
3299
+ "epoch": 0.97,
3300
+ "learning_rate": 4.203819696267486e-07,
3301
+ "loss": 0.0289,
3302
+ "step": 549
3303
+ },
3304
+ {
3305
+ "epoch": 0.97,
3306
+ "learning_rate": 3.6950771532126006e-07,
3307
+ "loss": 0.0243,
3308
+ "step": 550
3309
+ },
3310
+ {
3311
+ "epoch": 0.98,
3312
+ "learning_rate": 3.2190784581270786e-07,
3313
+ "loss": 0.0353,
3314
+ "step": 551
3315
+ },
3316
+ {
3317
+ "epoch": 0.98,
3318
+ "learning_rate": 2.7758392548449255e-07,
3319
+ "loss": 0.025,
3320
+ "step": 552
3321
+ },
3322
+ {
3323
+ "epoch": 0.98,
3324
+ "learning_rate": 2.3653741105499338e-07,
3325
+ "loss": 0.0259,
3326
+ "step": 553
3327
+ },
3328
+ {
3329
+ "epoch": 0.98,
3330
+ "learning_rate": 1.9876965152975103e-07,
3331
+ "loss": 0.042,
3332
+ "step": 554
3333
+ },
3334
+ {
3335
+ "epoch": 0.98,
3336
+ "learning_rate": 1.6428188815703626e-07,
3337
+ "loss": 0.0247,
3338
+ "step": 555
3339
+ },
3340
+ {
3341
+ "epoch": 0.98,
3342
+ "learning_rate": 1.330752543871161e-07,
3343
+ "loss": 0.0477,
3344
+ "step": 556
3345
+ },
3346
+ {
3347
+ "epoch": 0.99,
3348
+ "learning_rate": 1.0515077583498344e-07,
3349
+ "loss": 0.0316,
3350
+ "step": 557
3351
+ },
3352
+ {
3353
+ "epoch": 0.99,
3354
+ "learning_rate": 8.050937024666194e-08,
3355
+ "loss": 0.0177,
3356
+ "step": 558
3357
+ },
3358
+ {
3359
+ "epoch": 0.99,
3360
+ "learning_rate": 5.915184746904112e-08,
3361
+ "loss": 0.0279,
3362
+ "step": 559
3363
+ },
3364
+ {
3365
+ "epoch": 0.99,
3366
+ "learning_rate": 4.107890942325332e-08,
3367
+ "loss": 0.0346,
3368
+ "step": 560
3369
+ },
3370
+ {
3371
+ "epoch": 0.99,
3372
+ "learning_rate": 2.6291150081603212e-08,
3373
+ "loss": 0.0375,
3374
+ "step": 561
3375
+ },
3376
+ {
3377
+ "epoch": 0.99,
3378
+ "learning_rate": 1.4789055448061196e-08,
3379
+ "loss": 0.0347,
3380
+ "step": 562
3381
+ },
3382
+ {
3383
+ "epoch": 1.0,
3384
+ "learning_rate": 6.57300354227619e-09,
3385
+ "loss": 0.0281,
3386
+ "step": 563
3387
+ },
3388
+ {
3389
+ "epoch": 1.0,
3390
+ "learning_rate": 1.6432643871633347e-09,
3391
+ "loss": 0.04,
3392
+ "step": 564
3393
+ },
3394
+ {
3395
+ "epoch": 1.0,
3396
+ "learning_rate": 0.0,
3397
+ "loss": 0.0323,
3398
+ "step": 565
3399
+ },
3400
+ {
3401
+ "epoch": 1.0,
3402
+ "step": 565,
3403
+ "total_flos": 8.478714177802732e+17,
3404
+ "train_loss": 0.05881430652256297,
3405
+ "train_runtime": 7484.6873,
3406
+ "train_samples_per_second": 1.207,
3407
+ "train_steps_per_second": 0.075
3408
+ }
3409
+ ],
3410
+ "max_steps": 565,
3411
+ "num_train_epochs": 1,
3412
+ "total_flos": 8.478714177802732e+17,
3413
+ "trial_name": null,
3414
+ "trial_params": null
3415
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/README.md ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+
9
+ - PEFT 0.4.0
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "lmsys/vicuna-7b-v1.5",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 256,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 128,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "up_proj",
18
+ "q_proj",
19
+ "gate_proj",
20
+ "k_proj",
21
+ "o_proj",
22
+ "v_proj",
23
+ "down_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9db3a4669383796428373a7d0f923c3af731b20998c809b964228dce3a46fe2
3
+ size 639787082
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "freeze_mm_mlp_adapter": false,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "image_aspect_ratio": "pad",
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
14
+ "max_position_embeddings": 4096,
15
+ "mm_hidden_size": 1024,
16
+ "mm_image_tower": "LanguageBind/LanguageBind_Image",
17
+ "mm_projector_lr": 2e-05,
18
+ "mm_projector_type": "mlp2x_gelu",
19
+ "mm_use_im_patch_token": false,
20
+ "mm_use_im_start_end": false,
21
+ "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
22
+ "mm_vision_select_feature": "patch",
23
+ "mm_vision_select_layer": -2,
24
+ "model_type": "llava",
25
+ "num_attention_heads": 32,
26
+ "num_hidden_layers": 32,
27
+ "num_key_value_heads": 32,
28
+ "pad_token_id": 0,
29
+ "pretraining_tp": 1,
30
+ "rms_norm_eps": 1e-05,
31
+ "rope_scaling": null,
32
+ "tie_word_embeddings": false,
33
+ "tokenizer_model_max_length": 3072,
34
+ "tokenizer_padding_side": "right",
35
+ "torch_dtype": "float16",
36
+ "transformers_version": "4.31.0",
37
+ "tune_mm_mlp_adapter": false,
38
+ "use_cache": true,
39
+ "use_mm_proj": true,
40
+ "vocab_size": 32000
41
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/non_lora_trainables.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aaeabfe58c15d76d9e7a38dc4928dd572a7994b2862d62d1b05be8814b614613
3
+ size 41961648
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/runs/Sep24_08-23-17_evc43/events.out.tfevents.1727180744.evc43 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:81a3740fc8feddcd7b4d954ea4f43a4f0e2c6df187ace97d305ba6b09de95a43
3
+ size 271458
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
[lora]alpha256_video_llava_vidvrd_annotations_v5_3_e03/videollava-7b-lora/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723