Hugging Face's logo

divyjx
/
VideoLLaVA

Text Generation
Transformers
PyTorch
TensorBoard
llava
Model card Files
xet
Metrics Community
divyjx commited on
Commit
d5fedec
·
1 Parent(s): 16230db

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. config.json +43 -0
  2. generation_config.json +9 -0
  3. pytorch_model-00001-of-00002.bin +3 -0
  4. pytorch_model-00002-of-00002.bin +3 -0
  5. pytorch_model.bin.index.json +0 -0
  6. runs/Dec15_07-04-15_fb4250aaf359/events.out.tfevents.1702624039.fb4250aaf359 +3 -0
  7. special_tokens_map.json +24 -0
  8. tokenizer.model +3 -0
  9. tokenizer_config.json +35 -0
  10. trainer_state.json +4711 -0
  11. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "X": [
3
+ "Video",
4
+ "Image"
5
+ ],
6
+ "_name_or_path": "lmsys/vicuna-7b-v1.5",
7
+ "architectures": [
8
+ "LlavaLlamaForCausalLM"
9
+ ],
10
+ "bos_token_id": 1,
11
+ "eos_token_id": 2,
12
+ "freeze_mm_mlp_adapter": false,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 4096,
15
+ "image_aspect_ratio": "pad",
16
+ "image_grid_pinpoints": null,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 11008,
19
+ "max_position_embeddings": 4096,
20
+ "mm_hidden_size": 1024,
21
+ "mm_image_tower": "LanguageBind/LanguageBind_Image",
22
+ "mm_projector_type": "mlp2x_gelu",
23
+ "mm_use_x_patch_token": false,
24
+ "mm_use_x_start_end": false,
25
+ "mm_video_tower": "LanguageBind/LanguageBind_Video_merge",
26
+ "mm_vision_select_feature": "patch",
27
+ "mm_vision_select_layer": -2,
28
+ "model_type": "llava",
29
+ "num_attention_heads": 32,
30
+ "num_hidden_layers": 32,
31
+ "num_key_value_heads": 32,
32
+ "pad_token_id": 0,
33
+ "pretraining_tp": 1,
34
+ "rms_norm_eps": 1e-05,
35
+ "rope_scaling": null,
36
+ "tie_word_embeddings": false,
37
+ "torch_dtype": "bfloat16",
38
+ "transformers_version": "4.31.0",
39
+ "tune_mm_mlp_adapter": false,
40
+ "use_cache": true,
41
+ "use_mm_proj": true,
42
+ "vocab_size": 32000
43
+ }
generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 1,
3
+ "eos_token_id": 2,
4
+ "max_length": 4096,
5
+ "pad_token_id": 0,
6
+ "temperature": 0.9,
7
+ "top_p": 0.6,
8
+ "transformers_version": "4.31.0"
9
+ }
pytorch_model-00001-of-00002.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28860a6bab2bc8624cfad26a8adaeaedc4afed32a3e205d563ec4bdffd2fc9d9
3
+ size 9976631486
pytorch_model-00002-of-00002.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08ab65d6079e27a881d988e7b8c2d4d795891d97a8cda8395cfd6c55f12fbec8
3
+ size 4957395839
pytorch_model.bin.index.json ADDED
The diff for this file is too large to render. See raw diff
 
runs/Dec15_07-04-15_fb4250aaf359/events.out.tfevents.1702624039.fb4250aaf359 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d6d7d490327aef9c1c579da424366db0a542c94597f600ba42b41f46d40ba7d
3
+ size 127499
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
+ }
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
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
+ }
trainer_state.json ADDED
@@ -0,0 +1,4711 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.999360204734485,
5
+ "global_step": 781,
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": 8.333333333333333e-07,
13
+ "loss": 3.0762,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 1.6666666666666667e-06,
19
+ "loss": 3.1973,
20
+ "step": 2
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 2.5e-06,
25
+ "loss": 3.2324,
26
+ "step": 3
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 3.3333333333333333e-06,
31
+ "loss": 3.21,
32
+ "step": 4
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 4.166666666666667e-06,
37
+ "loss": 3.0703,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 5e-06,
43
+ "loss": 2.9287,
44
+ "step": 6
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 5.833333333333334e-06,
49
+ "loss": 2.9326,
50
+ "step": 7
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "learning_rate": 6.666666666666667e-06,
55
+ "loss": 2.7061,
56
+ "step": 8
57
+ },
58
+ {
59
+ "epoch": 0.01,
60
+ "learning_rate": 7.500000000000001e-06,
61
+ "loss": 2.6777,
62
+ "step": 9
63
+ },
64
+ {
65
+ "epoch": 0.01,
66
+ "learning_rate": 8.333333333333334e-06,
67
+ "loss": 2.6816,
68
+ "step": 10
69
+ },
70
+ {
71
+ "epoch": 0.01,
72
+ "learning_rate": 9.166666666666666e-06,
73
+ "loss": 2.3823,
74
+ "step": 11
75
+ },
76
+ {
77
+ "epoch": 0.02,
78
+ "learning_rate": 1e-05,
79
+ "loss": 2.4517,
80
+ "step": 12
81
+ },
82
+ {
83
+ "epoch": 0.02,
84
+ "learning_rate": 1.0833333333333334e-05,
85
+ "loss": 2.5,
86
+ "step": 13
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "learning_rate": 1.1666666666666668e-05,
91
+ "loss": 2.4087,
92
+ "step": 14
93
+ },
94
+ {
95
+ "epoch": 0.02,
96
+ "learning_rate": 1.25e-05,
97
+ "loss": 2.4521,
98
+ "step": 15
99
+ },
100
+ {
101
+ "epoch": 0.02,
102
+ "learning_rate": 1.3333333333333333e-05,
103
+ "loss": 2.4209,
104
+ "step": 16
105
+ },
106
+ {
107
+ "epoch": 0.02,
108
+ "learning_rate": 1.416666666666667e-05,
109
+ "loss": 2.3804,
110
+ "step": 17
111
+ },
112
+ {
113
+ "epoch": 0.02,
114
+ "learning_rate": 1.5000000000000002e-05,
115
+ "loss": 2.3135,
116
+ "step": 18
117
+ },
118
+ {
119
+ "epoch": 0.02,
120
+ "learning_rate": 1.5833333333333333e-05,
121
+ "loss": 2.3203,
122
+ "step": 19
123
+ },
124
+ {
125
+ "epoch": 0.03,
126
+ "learning_rate": 1.6666666666666667e-05,
127
+ "loss": 2.3721,
128
+ "step": 20
129
+ },
130
+ {
131
+ "epoch": 0.03,
132
+ "learning_rate": 1.7500000000000002e-05,
133
+ "loss": 2.3086,
134
+ "step": 21
135
+ },
136
+ {
137
+ "epoch": 0.03,
138
+ "learning_rate": 1.8333333333333333e-05,
139
+ "loss": 2.3511,
140
+ "step": 22
141
+ },
142
+ {
143
+ "epoch": 0.03,
144
+ "learning_rate": 1.916666666666667e-05,
145
+ "loss": 2.3682,
146
+ "step": 23
147
+ },
148
+ {
149
+ "epoch": 0.03,
150
+ "learning_rate": 2e-05,
151
+ "loss": 2.1953,
152
+ "step": 24
153
+ },
154
+ {
155
+ "epoch": 0.03,
156
+ "learning_rate": 1.9999913885285243e-05,
157
+ "loss": 2.209,
158
+ "step": 25
159
+ },
160
+ {
161
+ "epoch": 0.03,
162
+ "learning_rate": 1.9999655542624107e-05,
163
+ "loss": 2.2524,
164
+ "step": 26
165
+ },
166
+ {
167
+ "epoch": 0.03,
168
+ "learning_rate": 1.999922497646602e-05,
169
+ "loss": 2.2725,
170
+ "step": 27
171
+ },
172
+ {
173
+ "epoch": 0.04,
174
+ "learning_rate": 1.9998622194226605e-05,
175
+ "loss": 2.1963,
176
+ "step": 28
177
+ },
178
+ {
179
+ "epoch": 0.04,
180
+ "learning_rate": 1.9997847206287534e-05,
181
+ "loss": 2.2461,
182
+ "step": 29
183
+ },
184
+ {
185
+ "epoch": 0.04,
186
+ "learning_rate": 1.9996900025996385e-05,
187
+ "loss": 2.3765,
188
+ "step": 30
189
+ },
190
+ {
191
+ "epoch": 0.04,
192
+ "learning_rate": 1.999578066966639e-05,
193
+ "loss": 2.2871,
194
+ "step": 31
195
+ },
196
+ {
197
+ "epoch": 0.04,
198
+ "learning_rate": 1.9994489156576158e-05,
199
+ "loss": 2.333,
200
+ "step": 32
201
+ },
202
+ {
203
+ "epoch": 0.04,
204
+ "learning_rate": 1.999302550896935e-05,
205
+ "loss": 2.2876,
206
+ "step": 33
207
+ },
208
+ {
209
+ "epoch": 0.04,
210
+ "learning_rate": 1.999138975205428e-05,
211
+ "loss": 2.1973,
212
+ "step": 34
213
+ },
214
+ {
215
+ "epoch": 0.04,
216
+ "learning_rate": 1.9989581914003496e-05,
217
+ "loss": 2.2651,
218
+ "step": 35
219
+ },
220
+ {
221
+ "epoch": 0.05,
222
+ "learning_rate": 1.9987602025953293e-05,
223
+ "loss": 2.3105,
224
+ "step": 36
225
+ },
226
+ {
227
+ "epoch": 0.05,
228
+ "learning_rate": 1.998545012200317e-05,
229
+ "loss": 2.2871,
230
+ "step": 37
231
+ },
232
+ {
233
+ "epoch": 0.05,
234
+ "learning_rate": 1.9983126239215243e-05,
235
+ "loss": 2.2607,
236
+ "step": 38
237
+ },
238
+ {
239
+ "epoch": 0.05,
240
+ "learning_rate": 1.9980630417613615e-05,
241
+ "loss": 2.2407,
242
+ "step": 39
243
+ },
244
+ {
245
+ "epoch": 0.05,
246
+ "learning_rate": 1.997796270018368e-05,
247
+ "loss": 2.2241,
248
+ "step": 40
249
+ },
250
+ {
251
+ "epoch": 0.05,
252
+ "learning_rate": 1.9975123132871376e-05,
253
+ "loss": 2.2578,
254
+ "step": 41
255
+ },
256
+ {
257
+ "epoch": 0.05,
258
+ "learning_rate": 1.9972111764582416e-05,
259
+ "loss": 2.1167,
260
+ "step": 42
261
+ },
262
+ {
263
+ "epoch": 0.06,
264
+ "learning_rate": 1.9968928647181423e-05,
265
+ "loss": 2.2305,
266
+ "step": 43
267
+ },
268
+ {
269
+ "epoch": 0.06,
270
+ "learning_rate": 1.9965573835491046e-05,
271
+ "loss": 2.2188,
272
+ "step": 44
273
+ },
274
+ {
275
+ "epoch": 0.06,
276
+ "learning_rate": 1.9962047387291017e-05,
277
+ "loss": 2.2451,
278
+ "step": 45
279
+ },
280
+ {
281
+ "epoch": 0.06,
282
+ "learning_rate": 1.995834936331715e-05,
283
+ "loss": 2.2041,
284
+ "step": 46
285
+ },
286
+ {
287
+ "epoch": 0.06,
288
+ "learning_rate": 1.9954479827260297e-05,
289
+ "loss": 2.1416,
290
+ "step": 47
291
+ },
292
+ {
293
+ "epoch": 0.06,
294
+ "learning_rate": 1.9950438845765263e-05,
295
+ "loss": 2.165,
296
+ "step": 48
297
+ },
298
+ {
299
+ "epoch": 0.06,
300
+ "learning_rate": 1.994622648842964e-05,
301
+ "loss": 2.2251,
302
+ "step": 49
303
+ },
304
+ {
305
+ "epoch": 0.06,
306
+ "learning_rate": 1.994184282780262e-05,
307
+ "loss": 2.1714,
308
+ "step": 50
309
+ },
310
+ {
311
+ "epoch": 0.07,
312
+ "learning_rate": 1.9937287939383737e-05,
313
+ "loss": 2.186,
314
+ "step": 51
315
+ },
316
+ {
317
+ "epoch": 0.07,
318
+ "learning_rate": 1.9932561901621574e-05,
319
+ "loss": 2.2227,
320
+ "step": 52
321
+ },
322
+ {
323
+ "epoch": 0.07,
324
+ "learning_rate": 1.9927664795912412e-05,
325
+ "loss": 2.144,
326
+ "step": 53
327
+ },
328
+ {
329
+ "epoch": 0.07,
330
+ "learning_rate": 1.992259670659882e-05,
331
+ "loss": 2.1797,
332
+ "step": 54
333
+ },
334
+ {
335
+ "epoch": 0.07,
336
+ "learning_rate": 1.9917357720968213e-05,
337
+ "loss": 2.1357,
338
+ "step": 55
339
+ },
340
+ {
341
+ "epoch": 0.07,
342
+ "learning_rate": 1.9911947929251342e-05,
343
+ "loss": 2.229,
344
+ "step": 56
345
+ },
346
+ {
347
+ "epoch": 0.07,
348
+ "learning_rate": 1.9906367424620747e-05,
349
+ "loss": 2.1748,
350
+ "step": 57
351
+ },
352
+ {
353
+ "epoch": 0.07,
354
+ "learning_rate": 1.9900616303189126e-05,
355
+ "loss": 2.1392,
356
+ "step": 58
357
+ },
358
+ {
359
+ "epoch": 0.08,
360
+ "learning_rate": 1.989469466400773e-05,
361
+ "loss": 2.105,
362
+ "step": 59
363
+ },
364
+ {
365
+ "epoch": 0.08,
366
+ "learning_rate": 1.9888602609064604e-05,
367
+ "loss": 2.1284,
368
+ "step": 60
369
+ },
370
+ {
371
+ "epoch": 0.08,
372
+ "learning_rate": 1.988234024328287e-05,
373
+ "loss": 2.103,
374
+ "step": 61
375
+ },
376
+ {
377
+ "epoch": 0.08,
378
+ "learning_rate": 1.987590767451889e-05,
379
+ "loss": 2.1504,
380
+ "step": 62
381
+ },
382
+ {
383
+ "epoch": 0.08,
384
+ "learning_rate": 1.9869305013560428e-05,
385
+ "loss": 2.1836,
386
+ "step": 63
387
+ },
388
+ {
389
+ "epoch": 0.08,
390
+ "learning_rate": 1.9862532374124745e-05,
391
+ "loss": 2.1982,
392
+ "step": 64
393
+ },
394
+ {
395
+ "epoch": 0.08,
396
+ "learning_rate": 1.9855589872856617e-05,
397
+ "loss": 2.1484,
398
+ "step": 65
399
+ },
400
+ {
401
+ "epoch": 0.08,
402
+ "learning_rate": 1.984847762932635e-05,
403
+ "loss": 2.189,
404
+ "step": 66
405
+ },
406
+ {
407
+ "epoch": 0.09,
408
+ "learning_rate": 1.984119576602771e-05,
409
+ "loss": 2.2363,
410
+ "step": 67
411
+ },
412
+ {
413
+ "epoch": 0.09,
414
+ "learning_rate": 1.9833744408375804e-05,
415
+ "loss": 2.1582,
416
+ "step": 68
417
+ },
418
+ {
419
+ "epoch": 0.09,
420
+ "learning_rate": 1.982612368470495e-05,
421
+ "loss": 2.1382,
422
+ "step": 69
423
+ },
424
+ {
425
+ "epoch": 0.09,
426
+ "learning_rate": 1.9818333726266435e-05,
427
+ "loss": 2.1602,
428
+ "step": 70
429
+ },
430
+ {
431
+ "epoch": 0.09,
432
+ "learning_rate": 1.981037466722627e-05,
433
+ "loss": 2.1313,
434
+ "step": 71
435
+ },
436
+ {
437
+ "epoch": 0.09,
438
+ "learning_rate": 1.980224664466287e-05,
439
+ "loss": 2.1372,
440
+ "step": 72
441
+ },
442
+ {
443
+ "epoch": 0.09,
444
+ "learning_rate": 1.9793949798564703e-05,
445
+ "loss": 2.1133,
446
+ "step": 73
447
+ },
448
+ {
449
+ "epoch": 0.09,
450
+ "learning_rate": 1.9785484271827884e-05,
451
+ "loss": 2.1406,
452
+ "step": 74
453
+ },
454
+ {
455
+ "epoch": 0.1,
456
+ "learning_rate": 1.977685021025369e-05,
457
+ "loss": 2.0957,
458
+ "step": 75
459
+ },
460
+ {
461
+ "epoch": 0.1,
462
+ "learning_rate": 1.9768047762546076e-05,
463
+ "loss": 2.1333,
464
+ "step": 76
465
+ },
466
+ {
467
+ "epoch": 0.1,
468
+ "learning_rate": 1.975907708030909e-05,
469
+ "loss": 2.1313,
470
+ "step": 77
471
+ },
472
+ {
473
+ "epoch": 0.1,
474
+ "learning_rate": 1.9749938318044286e-05,
475
+ "loss": 2.123,
476
+ "step": 78
477
+ },
478
+ {
479
+ "epoch": 0.1,
480
+ "learning_rate": 1.9740631633148045e-05,
481
+ "loss": 2.1191,
482
+ "step": 79
483
+ },
484
+ {
485
+ "epoch": 0.1,
486
+ "learning_rate": 1.9731157185908866e-05,
487
+ "loss": 2.165,
488
+ "step": 80
489
+ },
490
+ {
491
+ "epoch": 0.1,
492
+ "learning_rate": 1.9721515139504617e-05,
493
+ "loss": 2.1245,
494
+ "step": 81
495
+ },
496
+ {
497
+ "epoch": 0.1,
498
+ "learning_rate": 1.9711705659999714e-05,
499
+ "loss": 2.0884,
500
+ "step": 82
501
+ },
502
+ {
503
+ "epoch": 0.11,
504
+ "learning_rate": 1.970172891634226e-05,
505
+ "loss": 2.1914,
506
+ "step": 83
507
+ },
508
+ {
509
+ "epoch": 0.11,
510
+ "learning_rate": 1.9691585080361138e-05,
511
+ "loss": 2.1206,
512
+ "step": 84
513
+ },
514
+ {
515
+ "epoch": 0.11,
516
+ "learning_rate": 1.9681274326763068e-05,
517
+ "loss": 2.1738,
518
+ "step": 85
519
+ },
520
+ {
521
+ "epoch": 0.11,
522
+ "learning_rate": 1.9670796833129563e-05,
523
+ "loss": 2.1753,
524
+ "step": 86
525
+ },
526
+ {
527
+ "epoch": 0.11,
528
+ "learning_rate": 1.9660152779913898e-05,
529
+ "loss": 2.1592,
530
+ "step": 87
531
+ },
532
+ {
533
+ "epoch": 0.11,
534
+ "learning_rate": 1.9649342350437998e-05,
535
+ "loss": 2.1284,
536
+ "step": 88
537
+ },
538
+ {
539
+ "epoch": 0.11,
540
+ "learning_rate": 1.9638365730889266e-05,
541
+ "loss": 2.0859,
542
+ "step": 89
543
+ },
544
+ {
545
+ "epoch": 0.12,
546
+ "learning_rate": 1.9627223110317402e-05,
547
+ "loss": 2.1558,
548
+ "step": 90
549
+ },
550
+ {
551
+ "epoch": 0.12,
552
+ "learning_rate": 1.961591468063112e-05,
553
+ "loss": 2.2393,
554
+ "step": 91
555
+ },
556
+ {
557
+ "epoch": 0.12,
558
+ "learning_rate": 1.9604440636594863e-05,
559
+ "loss": 2.1064,
560
+ "step": 92
561
+ },
562
+ {
563
+ "epoch": 0.12,
564
+ "learning_rate": 1.9592801175825435e-05,
565
+ "loss": 2.0718,
566
+ "step": 93
567
+ },
568
+ {
569
+ "epoch": 0.12,
570
+ "learning_rate": 1.9580996498788602e-05,
571
+ "loss": 2.0894,
572
+ "step": 94
573
+ },
574
+ {
575
+ "epoch": 0.12,
576
+ "learning_rate": 1.9569026808795647e-05,
577
+ "loss": 2.1514,
578
+ "step": 95
579
+ },
580
+ {
581
+ "epoch": 0.12,
582
+ "learning_rate": 1.955689231199986e-05,
583
+ "loss": 2.1646,
584
+ "step": 96
585
+ },
586
+ {
587
+ "epoch": 0.12,
588
+ "learning_rate": 1.954459321739298e-05,
589
+ "loss": 2.124,
590
+ "step": 97
591
+ },
592
+ {
593
+ "epoch": 0.13,
594
+ "learning_rate": 1.9532129736801616e-05,
595
+ "loss": 2.0737,
596
+ "step": 98
597
+ },
598
+ {
599
+ "epoch": 0.13,
600
+ "learning_rate": 1.9519502084883585e-05,
601
+ "loss": 2.0039,
602
+ "step": 99
603
+ },
604
+ {
605
+ "epoch": 0.13,
606
+ "learning_rate": 1.9506710479124212e-05,
607
+ "loss": 2.1216,
608
+ "step": 100
609
+ },
610
+ {
611
+ "epoch": 0.13,
612
+ "learning_rate": 1.9493755139832593e-05,
613
+ "loss": 2.1719,
614
+ "step": 101
615
+ },
616
+ {
617
+ "epoch": 0.13,
618
+ "learning_rate": 1.9480636290137803e-05,
619
+ "loss": 2.1074,
620
+ "step": 102
621
+ },
622
+ {
623
+ "epoch": 0.13,
624
+ "learning_rate": 1.946735415598503e-05,
625
+ "loss": 2.1665,
626
+ "step": 103
627
+ },
628
+ {
629
+ "epoch": 0.13,
630
+ "learning_rate": 1.9453908966131728e-05,
631
+ "loss": 2.0825,
632
+ "step": 104
633
+ },
634
+ {
635
+ "epoch": 0.13,
636
+ "learning_rate": 1.9440300952143628e-05,
637
+ "loss": 2.1929,
638
+ "step": 105
639
+ },
640
+ {
641
+ "epoch": 0.14,
642
+ "learning_rate": 1.9426530348390777e-05,
643
+ "loss": 2.1001,
644
+ "step": 106
645
+ },
646
+ {
647
+ "epoch": 0.14,
648
+ "learning_rate": 1.94125973920435e-05,
649
+ "loss": 2.1572,
650
+ "step": 107
651
+ },
652
+ {
653
+ "epoch": 0.14,
654
+ "learning_rate": 1.9398502323068306e-05,
655
+ "loss": 2.1309,
656
+ "step": 108
657
+ },
658
+ {
659
+ "epoch": 0.14,
660
+ "learning_rate": 1.9384245384223767e-05,
661
+ "loss": 2.124,
662
+ "step": 109
663
+ },
664
+ {
665
+ "epoch": 0.14,
666
+ "learning_rate": 1.9369826821056327e-05,
667
+ "loss": 2.1812,
668
+ "step": 110
669
+ },
670
+ {
671
+ "epoch": 0.14,
672
+ "learning_rate": 1.9355246881896078e-05,
673
+ "loss": 2.1392,
674
+ "step": 111
675
+ },
676
+ {
677
+ "epoch": 0.14,
678
+ "learning_rate": 1.934050581785248e-05,
679
+ "loss": 2.1357,
680
+ "step": 112
681
+ },
682
+ {
683
+ "epoch": 0.14,
684
+ "learning_rate": 1.932560388281003e-05,
685
+ "loss": 2.1147,
686
+ "step": 113
687
+ },
688
+ {
689
+ "epoch": 0.15,
690
+ "learning_rate": 1.931054133342392e-05,
691
+ "loss": 2.1904,
692
+ "step": 114
693
+ },
694
+ {
695
+ "epoch": 0.15,
696
+ "learning_rate": 1.929531842911557e-05,
697
+ "loss": 2.1313,
698
+ "step": 115
699
+ },
700
+ {
701
+ "epoch": 0.15,
702
+ "learning_rate": 1.927993543206819e-05,
703
+ "loss": 2.1089,
704
+ "step": 116
705
+ },
706
+ {
707
+ "epoch": 0.15,
708
+ "learning_rate": 1.9264392607222263e-05,
709
+ "loss": 2.022,
710
+ "step": 117
711
+ },
712
+ {
713
+ "epoch": 0.15,
714
+ "learning_rate": 1.9248690222270983e-05,
715
+ "loss": 2.0171,
716
+ "step": 118
717
+ },
718
+ {
719
+ "epoch": 0.15,
720
+ "learning_rate": 1.9232828547655616e-05,
721
+ "loss": 2.1177,
722
+ "step": 119
723
+ },
724
+ {
725
+ "epoch": 0.15,
726
+ "learning_rate": 1.921680785656089e-05,
727
+ "loss": 2.0728,
728
+ "step": 120
729
+ },
730
+ {
731
+ "epoch": 0.15,
732
+ "learning_rate": 1.920062842491025e-05,
733
+ "loss": 2.0376,
734
+ "step": 121
735
+ },
736
+ {
737
+ "epoch": 0.16,
738
+ "learning_rate": 1.918429053136112e-05,
739
+ "loss": 2.0513,
740
+ "step": 122
741
+ },
742
+ {
743
+ "epoch": 0.16,
744
+ "learning_rate": 1.916779445730011e-05,
745
+ "loss": 2.0405,
746
+ "step": 123
747
+ },
748
+ {
749
+ "epoch": 0.16,
750
+ "learning_rate": 1.9151140486838173e-05,
751
+ "loss": 2.0898,
752
+ "step": 124
753
+ },
754
+ {
755
+ "epoch": 0.16,
756
+ "learning_rate": 1.913432890680568e-05,
757
+ "loss": 2.1895,
758
+ "step": 125
759
+ },
760
+ {
761
+ "epoch": 0.16,
762
+ "learning_rate": 1.9117360006747524e-05,
763
+ "loss": 2.0532,
764
+ "step": 126
765
+ },
766
+ {
767
+ "epoch": 0.16,
768
+ "learning_rate": 1.910023407891809e-05,
769
+ "loss": 2.1719,
770
+ "step": 127
771
+ },
772
+ {
773
+ "epoch": 0.16,
774
+ "learning_rate": 1.908295141827627e-05,
775
+ "loss": 2.1235,
776
+ "step": 128
777
+ },
778
+ {
779
+ "epoch": 0.17,
780
+ "learning_rate": 1.9065512322480332e-05,
781
+ "loss": 2.0361,
782
+ "step": 129
783
+ },
784
+ {
785
+ "epoch": 0.17,
786
+ "learning_rate": 1.9047917091882835e-05,
787
+ "loss": 2.1084,
788
+ "step": 130
789
+ },
790
+ {
791
+ "epoch": 0.17,
792
+ "learning_rate": 1.9030166029525427e-05,
793
+ "loss": 2.0547,
794
+ "step": 131
795
+ },
796
+ {
797
+ "epoch": 0.17,
798
+ "learning_rate": 1.9012259441133646e-05,
799
+ "loss": 2.1025,
800
+ "step": 132
801
+ },
802
+ {
803
+ "epoch": 0.17,
804
+ "learning_rate": 1.8994197635111637e-05,
805
+ "loss": 2.1279,
806
+ "step": 133
807
+ },
808
+ {
809
+ "epoch": 0.17,
810
+ "learning_rate": 1.8975980922536862e-05,
811
+ "loss": 2.1646,
812
+ "step": 134
813
+ },
814
+ {
815
+ "epoch": 0.17,
816
+ "learning_rate": 1.8957609617154718e-05,
817
+ "loss": 2.1016,
818
+ "step": 135
819
+ },
820
+ {
821
+ "epoch": 0.17,
822
+ "learning_rate": 1.893908403537315e-05,
823
+ "loss": 2.0938,
824
+ "step": 136
825
+ },
826
+ {
827
+ "epoch": 0.18,
828
+ "learning_rate": 1.8920404496257197e-05,
829
+ "loss": 2.0996,
830
+ "step": 137
831
+ },
832
+ {
833
+ "epoch": 0.18,
834
+ "learning_rate": 1.8901571321523494e-05,
835
+ "loss": 2.0811,
836
+ "step": 138
837
+ },
838
+ {
839
+ "epoch": 0.18,
840
+ "learning_rate": 1.888258483553474e-05,
841
+ "loss": 2.0811,
842
+ "step": 139
843
+ },
844
+ {
845
+ "epoch": 0.18,
846
+ "learning_rate": 1.8863445365294088e-05,
847
+ "loss": 2.0747,
848
+ "step": 140
849
+ },
850
+ {
851
+ "epoch": 0.18,
852
+ "learning_rate": 1.8844153240439555e-05,
853
+ "loss": 2.1177,
854
+ "step": 141
855
+ },
856
+ {
857
+ "epoch": 0.18,
858
+ "learning_rate": 1.8824708793238303e-05,
859
+ "loss": 2.0962,
860
+ "step": 142
861
+ },
862
+ {
863
+ "epoch": 0.18,
864
+ "learning_rate": 1.8805112358580936e-05,
865
+ "loss": 1.9995,
866
+ "step": 143
867
+ },
868
+ {
869
+ "epoch": 0.18,
870
+ "learning_rate": 1.878536427397573e-05,
871
+ "loss": 2.0405,
872
+ "step": 144
873
+ },
874
+ {
875
+ "epoch": 0.19,
876
+ "learning_rate": 1.8765464879542817e-05,
877
+ "loss": 2.0479,
878
+ "step": 145
879
+ },
880
+ {
881
+ "epoch": 0.19,
882
+ "learning_rate": 1.8745414518008335e-05,
883
+ "loss": 2.0933,
884
+ "step": 146
885
+ },
886
+ {
887
+ "epoch": 0.19,
888
+ "learning_rate": 1.872521353469852e-05,
889
+ "loss": 2.021,
890
+ "step": 147
891
+ },
892
+ {
893
+ "epoch": 0.19,
894
+ "learning_rate": 1.870486227753375e-05,
895
+ "loss": 2.0752,
896
+ "step": 148
897
+ },
898
+ {
899
+ "epoch": 0.19,
900
+ "learning_rate": 1.8684361097022566e-05,
901
+ "loss": 2.1606,
902
+ "step": 149
903
+ },
904
+ {
905
+ "epoch": 0.19,
906
+ "learning_rate": 1.8663710346255633e-05,
907
+ "loss": 2.0,
908
+ "step": 150
909
+ },
910
+ {
911
+ "epoch": 0.19,
912
+ "learning_rate": 1.8642910380899656e-05,
913
+ "loss": 2.0757,
914
+ "step": 151
915
+ },
916
+ {
917
+ "epoch": 0.19,
918
+ "learning_rate": 1.8621961559191245e-05,
919
+ "loss": 2.0078,
920
+ "step": 152
921
+ },
922
+ {
923
+ "epoch": 0.2,
924
+ "learning_rate": 1.860086424193077e-05,
925
+ "loss": 2.0132,
926
+ "step": 153
927
+ },
928
+ {
929
+ "epoch": 0.2,
930
+ "learning_rate": 1.8579618792476107e-05,
931
+ "loss": 2.0513,
932
+ "step": 154
933
+ },
934
+ {
935
+ "epoch": 0.2,
936
+ "learning_rate": 1.8558225576736436e-05,
937
+ "loss": 2.0806,
938
+ "step": 155
939
+ },
940
+ {
941
+ "epoch": 0.2,
942
+ "learning_rate": 1.853668496316589e-05,
943
+ "loss": 2.0444,
944
+ "step": 156
945
+ },
946
+ {
947
+ "epoch": 0.2,
948
+ "learning_rate": 1.8514997322757215e-05,
949
+ "loss": 2.0415,
950
+ "step": 157
951
+ },
952
+ {
953
+ "epoch": 0.2,
954
+ "learning_rate": 1.849316302903542e-05,
955
+ "loss": 1.9946,
956
+ "step": 158
957
+ },
958
+ {
959
+ "epoch": 0.2,
960
+ "learning_rate": 1.8471182458051285e-05,
961
+ "loss": 2.1104,
962
+ "step": 159
963
+ },
964
+ {
965
+ "epoch": 0.2,
966
+ "learning_rate": 1.8449055988374943e-05,
967
+ "loss": 2.0034,
968
+ "step": 160
969
+ },
970
+ {
971
+ "epoch": 0.21,
972
+ "learning_rate": 1.8426784001089312e-05,
973
+ "loss": 2.0327,
974
+ "step": 161
975
+ },
976
+ {
977
+ "epoch": 0.21,
978
+ "learning_rate": 1.840436687978356e-05,
979
+ "loss": 2.0664,
980
+ "step": 162
981
+ },
982
+ {
983
+ "epoch": 0.21,
984
+ "learning_rate": 1.8381805010546493e-05,
985
+ "loss": 2.1455,
986
+ "step": 163
987
+ },
988
+ {
989
+ "epoch": 0.21,
990
+ "learning_rate": 1.835909878195989e-05,
991
+ "loss": 2.0742,
992
+ "step": 164
993
+ },
994
+ {
995
+ "epoch": 0.21,
996
+ "learning_rate": 1.8336248585091834e-05,
997
+ "loss": 2.1191,
998
+ "step": 165
999
+ },
1000
+ {
1001
+ "epoch": 0.21,
1002
+ "learning_rate": 1.831325481348996e-05,
1003
+ "loss": 2.0776,
1004
+ "step": 166
1005
+ },
1006
+ {
1007
+ "epoch": 0.21,
1008
+ "learning_rate": 1.829011786317469e-05,
1009
+ "loss": 2.0229,
1010
+ "step": 167
1011
+ },
1012
+ {
1013
+ "epoch": 0.21,
1014
+ "learning_rate": 1.8266838132632396e-05,
1015
+ "loss": 2.083,
1016
+ "step": 168
1017
+ },
1018
+ {
1019
+ "epoch": 0.22,
1020
+ "learning_rate": 1.824341602280855e-05,
1021
+ "loss": 2.0801,
1022
+ "step": 169
1023
+ },
1024
+ {
1025
+ "epoch": 0.22,
1026
+ "learning_rate": 1.821985193710081e-05,
1027
+ "loss": 1.9814,
1028
+ "step": 170
1029
+ },
1030
+ {
1031
+ "epoch": 0.22,
1032
+ "learning_rate": 1.8196146281352086e-05,
1033
+ "loss": 2.0107,
1034
+ "step": 171
1035
+ },
1036
+ {
1037
+ "epoch": 0.22,
1038
+ "learning_rate": 1.8172299463843525e-05,
1039
+ "loss": 2.0088,
1040
+ "step": 172
1041
+ },
1042
+ {
1043
+ "epoch": 0.22,
1044
+ "learning_rate": 1.8148311895287513e-05,
1045
+ "loss": 2.0024,
1046
+ "step": 173
1047
+ },
1048
+ {
1049
+ "epoch": 0.22,
1050
+ "learning_rate": 1.8124183988820572e-05,
1051
+ "loss": 1.9897,
1052
+ "step": 174
1053
+ },
1054
+ {
1055
+ "epoch": 0.22,
1056
+ "learning_rate": 1.8099916159996263e-05,
1057
+ "loss": 2.103,
1058
+ "step": 175
1059
+ },
1060
+ {
1061
+ "epoch": 0.23,
1062
+ "learning_rate": 1.8075508826778013e-05,
1063
+ "loss": 2.0537,
1064
+ "step": 176
1065
+ },
1066
+ {
1067
+ "epoch": 0.23,
1068
+ "learning_rate": 1.8050962409531926e-05,
1069
+ "loss": 2.0156,
1070
+ "step": 177
1071
+ },
1072
+ {
1073
+ "epoch": 0.23,
1074
+ "learning_rate": 1.8026277331019552e-05,
1075
+ "loss": 2.0649,
1076
+ "step": 178
1077
+ },
1078
+ {
1079
+ "epoch": 0.23,
1080
+ "learning_rate": 1.8001454016390587e-05,
1081
+ "loss": 2.064,
1082
+ "step": 179
1083
+ },
1084
+ {
1085
+ "epoch": 0.23,
1086
+ "learning_rate": 1.7976492893175563e-05,
1087
+ "loss": 2.0864,
1088
+ "step": 180
1089
+ },
1090
+ {
1091
+ "epoch": 0.23,
1092
+ "learning_rate": 1.7951394391278484e-05,
1093
+ "loss": 2.0498,
1094
+ "step": 181
1095
+ },
1096
+ {
1097
+ "epoch": 0.23,
1098
+ "learning_rate": 1.7926158942969413e-05,
1099
+ "loss": 2.1182,
1100
+ "step": 182
1101
+ },
1102
+ {
1103
+ "epoch": 0.23,
1104
+ "learning_rate": 1.7900786982877037e-05,
1105
+ "loss": 2.0962,
1106
+ "step": 183
1107
+ },
1108
+ {
1109
+ "epoch": 0.24,
1110
+ "learning_rate": 1.787527894798118e-05,
1111
+ "loss": 2.1284,
1112
+ "step": 184
1113
+ },
1114
+ {
1115
+ "epoch": 0.24,
1116
+ "learning_rate": 1.784963527760527e-05,
1117
+ "loss": 2.0625,
1118
+ "step": 185
1119
+ },
1120
+ {
1121
+ "epoch": 0.24,
1122
+ "learning_rate": 1.7823856413408775e-05,
1123
+ "loss": 1.8911,
1124
+ "step": 186
1125
+ },
1126
+ {
1127
+ "epoch": 0.24,
1128
+ "learning_rate": 1.7797942799379608e-05,
1129
+ "loss": 2.043,
1130
+ "step": 187
1131
+ },
1132
+ {
1133
+ "epoch": 0.24,
1134
+ "learning_rate": 1.7771894881826464e-05,
1135
+ "loss": 2.0127,
1136
+ "step": 188
1137
+ },
1138
+ {
1139
+ "epoch": 0.24,
1140
+ "learning_rate": 1.7745713109371138e-05,
1141
+ "loss": 2.123,
1142
+ "step": 189
1143
+ },
1144
+ {
1145
+ "epoch": 0.24,
1146
+ "learning_rate": 1.7719397932940808e-05,
1147
+ "loss": 2.0132,
1148
+ "step": 190
1149
+ },
1150
+ {
1151
+ "epoch": 0.24,
1152
+ "learning_rate": 1.7692949805760256e-05,
1153
+ "loss": 1.877,
1154
+ "step": 191
1155
+ },
1156
+ {
1157
+ "epoch": 0.25,
1158
+ "learning_rate": 1.7666369183344065e-05,
1159
+ "loss": 1.9692,
1160
+ "step": 192
1161
+ },
1162
+ {
1163
+ "epoch": 0.25,
1164
+ "learning_rate": 1.7639656523488777e-05,
1165
+ "loss": 2.04,
1166
+ "step": 193
1167
+ },
1168
+ {
1169
+ "epoch": 0.25,
1170
+ "learning_rate": 1.7612812286265015e-05,
1171
+ "loss": 2.1187,
1172
+ "step": 194
1173
+ },
1174
+ {
1175
+ "epoch": 0.25,
1176
+ "learning_rate": 1.7585836934009535e-05,
1177
+ "loss": 2.0469,
1178
+ "step": 195
1179
+ },
1180
+ {
1181
+ "epoch": 0.25,
1182
+ "learning_rate": 1.75587309313173e-05,
1183
+ "loss": 2.0776,
1184
+ "step": 196
1185
+ },
1186
+ {
1187
+ "epoch": 0.25,
1188
+ "learning_rate": 1.7531494745033446e-05,
1189
+ "loss": 2.0845,
1190
+ "step": 197
1191
+ },
1192
+ {
1193
+ "epoch": 0.25,
1194
+ "learning_rate": 1.750412884424525e-05,
1195
+ "loss": 2.0469,
1196
+ "step": 198
1197
+ },
1198
+ {
1199
+ "epoch": 0.25,
1200
+ "learning_rate": 1.747663370027407e-05,
1201
+ "loss": 2.0996,
1202
+ "step": 199
1203
+ },
1204
+ {
1205
+ "epoch": 0.26,
1206
+ "learning_rate": 1.7449009786667192e-05,
1207
+ "loss": 1.9468,
1208
+ "step": 200
1209
+ },
1210
+ {
1211
+ "epoch": 0.26,
1212
+ "learning_rate": 1.742125757918971e-05,
1213
+ "loss": 1.9131,
1214
+ "step": 201
1215
+ },
1216
+ {
1217
+ "epoch": 0.26,
1218
+ "learning_rate": 1.7393377555816307e-05,
1219
+ "loss": 2.1069,
1220
+ "step": 202
1221
+ },
1222
+ {
1223
+ "epoch": 0.26,
1224
+ "learning_rate": 1.7365370196723035e-05,
1225
+ "loss": 2.0781,
1226
+ "step": 203
1227
+ },
1228
+ {
1229
+ "epoch": 0.26,
1230
+ "learning_rate": 1.7337235984279047e-05,
1231
+ "loss": 2.0801,
1232
+ "step": 204
1233
+ },
1234
+ {
1235
+ "epoch": 0.26,
1236
+ "learning_rate": 1.7308975403038278e-05,
1237
+ "loss": 2.0181,
1238
+ "step": 205
1239
+ },
1240
+ {
1241
+ "epoch": 0.26,
1242
+ "learning_rate": 1.7280588939731097e-05,
1243
+ "loss": 1.9966,
1244
+ "step": 206
1245
+ },
1246
+ {
1247
+ "epoch": 0.26,
1248
+ "learning_rate": 1.7252077083255953e-05,
1249
+ "loss": 2.0308,
1250
+ "step": 207
1251
+ },
1252
+ {
1253
+ "epoch": 0.27,
1254
+ "learning_rate": 1.7223440324670918e-05,
1255
+ "loss": 2.064,
1256
+ "step": 208
1257
+ },
1258
+ {
1259
+ "epoch": 0.27,
1260
+ "learning_rate": 1.719467915718526e-05,
1261
+ "loss": 2.0063,
1262
+ "step": 209
1263
+ },
1264
+ {
1265
+ "epoch": 0.27,
1266
+ "learning_rate": 1.716579407615091e-05,
1267
+ "loss": 2.1196,
1268
+ "step": 210
1269
+ },
1270
+ {
1271
+ "epoch": 0.27,
1272
+ "learning_rate": 1.7136785579053983e-05,
1273
+ "loss": 1.998,
1274
+ "step": 211
1275
+ },
1276
+ {
1277
+ "epoch": 0.27,
1278
+ "learning_rate": 1.710765416550617e-05,
1279
+ "loss": 2.166,
1280
+ "step": 212
1281
+ },
1282
+ {
1283
+ "epoch": 0.27,
1284
+ "learning_rate": 1.7078400337236135e-05,
1285
+ "loss": 2.1338,
1286
+ "step": 213
1287
+ },
1288
+ {
1289
+ "epoch": 0.27,
1290
+ "learning_rate": 1.7049024598080902e-05,
1291
+ "loss": 2.1411,
1292
+ "step": 214
1293
+ },
1294
+ {
1295
+ "epoch": 0.28,
1296
+ "learning_rate": 1.7019527453977146e-05,
1297
+ "loss": 1.9976,
1298
+ "step": 215
1299
+ },
1300
+ {
1301
+ "epoch": 0.28,
1302
+ "learning_rate": 1.69899094129525e-05,
1303
+ "loss": 2.0835,
1304
+ "step": 216
1305
+ },
1306
+ {
1307
+ "epoch": 0.28,
1308
+ "learning_rate": 1.6960170985116796e-05,
1309
+ "loss": 2.0518,
1310
+ "step": 217
1311
+ },
1312
+ {
1313
+ "epoch": 0.28,
1314
+ "learning_rate": 1.6930312682653277e-05,
1315
+ "loss": 2.1416,
1316
+ "step": 218
1317
+ },
1318
+ {
1319
+ "epoch": 0.28,
1320
+ "learning_rate": 1.6900335019809784e-05,
1321
+ "loss": 2.0566,
1322
+ "step": 219
1323
+ },
1324
+ {
1325
+ "epoch": 0.28,
1326
+ "learning_rate": 1.6870238512889895e-05,
1327
+ "loss": 1.8994,
1328
+ "step": 220
1329
+ },
1330
+ {
1331
+ "epoch": 0.28,
1332
+ "learning_rate": 1.6840023680244033e-05,
1333
+ "loss": 2.2329,
1334
+ "step": 221
1335
+ },
1336
+ {
1337
+ "epoch": 0.28,
1338
+ "learning_rate": 1.6809691042260535e-05,
1339
+ "loss": 1.9653,
1340
+ "step": 222
1341
+ },
1342
+ {
1343
+ "epoch": 0.29,
1344
+ "learning_rate": 1.677924112135669e-05,
1345
+ "loss": 2.0039,
1346
+ "step": 223
1347
+ },
1348
+ {
1349
+ "epoch": 0.29,
1350
+ "learning_rate": 1.6748674441969757e-05,
1351
+ "loss": 1.9346,
1352
+ "step": 224
1353
+ },
1354
+ {
1355
+ "epoch": 0.29,
1356
+ "learning_rate": 1.6717991530547906e-05,
1357
+ "loss": 2.0918,
1358
+ "step": 225
1359
+ },
1360
+ {
1361
+ "epoch": 0.29,
1362
+ "learning_rate": 1.6687192915541172e-05,
1363
+ "loss": 2.063,
1364
+ "step": 226
1365
+ },
1366
+ {
1367
+ "epoch": 0.29,
1368
+ "learning_rate": 1.6656279127392343e-05,
1369
+ "loss": 2.0088,
1370
+ "step": 227
1371
+ },
1372
+ {
1373
+ "epoch": 0.29,
1374
+ "learning_rate": 1.662525069852783e-05,
1375
+ "loss": 1.9795,
1376
+ "step": 228
1377
+ },
1378
+ {
1379
+ "epoch": 0.29,
1380
+ "learning_rate": 1.6594108163348494e-05,
1381
+ "loss": 1.9824,
1382
+ "step": 229
1383
+ },
1384
+ {
1385
+ "epoch": 0.29,
1386
+ "learning_rate": 1.6562852058220436e-05,
1387
+ "loss": 2.0737,
1388
+ "step": 230
1389
+ },
1390
+ {
1391
+ "epoch": 0.3,
1392
+ "learning_rate": 1.653148292146578e-05,
1393
+ "loss": 2.0659,
1394
+ "step": 231
1395
+ },
1396
+ {
1397
+ "epoch": 0.3,
1398
+ "learning_rate": 1.6500001293353375e-05,
1399
+ "loss": 2.0552,
1400
+ "step": 232
1401
+ },
1402
+ {
1403
+ "epoch": 0.3,
1404
+ "learning_rate": 1.6468407716089503e-05,
1405
+ "loss": 1.9307,
1406
+ "step": 233
1407
+ },
1408
+ {
1409
+ "epoch": 0.3,
1410
+ "learning_rate": 1.6436702733808546e-05,
1411
+ "loss": 2.0146,
1412
+ "step": 234
1413
+ },
1414
+ {
1415
+ "epoch": 0.3,
1416
+ "learning_rate": 1.640488689256361e-05,
1417
+ "loss": 2.0645,
1418
+ "step": 235
1419
+ },
1420
+ {
1421
+ "epoch": 0.3,
1422
+ "learning_rate": 1.6372960740317104e-05,
1423
+ "loss": 1.9868,
1424
+ "step": 236
1425
+ },
1426
+ {
1427
+ "epoch": 0.3,
1428
+ "learning_rate": 1.634092482693133e-05,
1429
+ "loss": 1.9507,
1430
+ "step": 237
1431
+ },
1432
+ {
1433
+ "epoch": 0.3,
1434
+ "learning_rate": 1.6308779704159002e-05,
1435
+ "loss": 1.9897,
1436
+ "step": 238
1437
+ },
1438
+ {
1439
+ "epoch": 0.31,
1440
+ "learning_rate": 1.627652592563373e-05,
1441
+ "loss": 2.0942,
1442
+ "step": 239
1443
+ },
1444
+ {
1445
+ "epoch": 0.31,
1446
+ "learning_rate": 1.6244164046860503e-05,
1447
+ "loss": 2.0078,
1448
+ "step": 240
1449
+ },
1450
+ {
1451
+ "epoch": 0.31,
1452
+ "learning_rate": 1.6211694625206114e-05,
1453
+ "loss": 1.9844,
1454
+ "step": 241
1455
+ },
1456
+ {
1457
+ "epoch": 0.31,
1458
+ "learning_rate": 1.6179118219889555e-05,
1459
+ "loss": 2.0039,
1460
+ "step": 242
1461
+ },
1462
+ {
1463
+ "epoch": 0.31,
1464
+ "learning_rate": 1.6146435391972405e-05,
1465
+ "loss": 1.9844,
1466
+ "step": 243
1467
+ },
1468
+ {
1469
+ "epoch": 0.31,
1470
+ "learning_rate": 1.611364670434914e-05,
1471
+ "loss": 1.9854,
1472
+ "step": 244
1473
+ },
1474
+ {
1475
+ "epoch": 0.31,
1476
+ "learning_rate": 1.6080752721737454e-05,
1477
+ "loss": 1.9741,
1478
+ "step": 245
1479
+ },
1480
+ {
1481
+ "epoch": 0.31,
1482
+ "learning_rate": 1.6047754010668536e-05,
1483
+ "loss": 2.0376,
1484
+ "step": 246
1485
+ },
1486
+ {
1487
+ "epoch": 0.32,
1488
+ "learning_rate": 1.6014651139477307e-05,
1489
+ "loss": 1.915,
1490
+ "step": 247
1491
+ },
1492
+ {
1493
+ "epoch": 0.32,
1494
+ "learning_rate": 1.598144467829262e-05,
1495
+ "loss": 1.9546,
1496
+ "step": 248
1497
+ },
1498
+ {
1499
+ "epoch": 0.32,
1500
+ "learning_rate": 1.5948135199027474e-05,
1501
+ "loss": 1.9824,
1502
+ "step": 249
1503
+ },
1504
+ {
1505
+ "epoch": 0.32,
1506
+ "learning_rate": 1.591472327536912e-05,
1507
+ "loss": 2.0122,
1508
+ "step": 250
1509
+ },
1510
+ {
1511
+ "epoch": 0.32,
1512
+ "learning_rate": 1.5881209482769214e-05,
1513
+ "loss": 2.0542,
1514
+ "step": 251
1515
+ },
1516
+ {
1517
+ "epoch": 0.32,
1518
+ "learning_rate": 1.5847594398433904e-05,
1519
+ "loss": 2.0176,
1520
+ "step": 252
1521
+ },
1522
+ {
1523
+ "epoch": 0.32,
1524
+ "learning_rate": 1.5813878601313857e-05,
1525
+ "loss": 2.0449,
1526
+ "step": 253
1527
+ },
1528
+ {
1529
+ "epoch": 0.33,
1530
+ "learning_rate": 1.5780062672094332e-05,
1531
+ "loss": 2.0352,
1532
+ "step": 254
1533
+ },
1534
+ {
1535
+ "epoch": 0.33,
1536
+ "learning_rate": 1.5746147193185142e-05,
1537
+ "loss": 1.9858,
1538
+ "step": 255
1539
+ },
1540
+ {
1541
+ "epoch": 0.33,
1542
+ "learning_rate": 1.5712132748710653e-05,
1543
+ "loss": 1.957,
1544
+ "step": 256
1545
+ },
1546
+ {
1547
+ "epoch": 0.33,
1548
+ "learning_rate": 1.5678019924499697e-05,
1549
+ "loss": 1.9795,
1550
+ "step": 257
1551
+ },
1552
+ {
1553
+ "epoch": 0.33,
1554
+ "learning_rate": 1.56438093080755e-05,
1555
+ "loss": 2.0571,
1556
+ "step": 258
1557
+ },
1558
+ {
1559
+ "epoch": 0.33,
1560
+ "learning_rate": 1.5609501488645555e-05,
1561
+ "loss": 1.9717,
1562
+ "step": 259
1563
+ },
1564
+ {
1565
+ "epoch": 0.33,
1566
+ "learning_rate": 1.5575097057091485e-05,
1567
+ "loss": 2.0132,
1568
+ "step": 260
1569
+ },
1570
+ {
1571
+ "epoch": 0.33,
1572
+ "learning_rate": 1.5540596605958843e-05,
1573
+ "loss": 2.0527,
1574
+ "step": 261
1575
+ },
1576
+ {
1577
+ "epoch": 0.34,
1578
+ "learning_rate": 1.5506000729446943e-05,
1579
+ "loss": 1.9658,
1580
+ "step": 262
1581
+ },
1582
+ {
1583
+ "epoch": 0.34,
1584
+ "learning_rate": 1.547131002339858e-05,
1585
+ "loss": 2.0225,
1586
+ "step": 263
1587
+ },
1588
+ {
1589
+ "epoch": 0.34,
1590
+ "learning_rate": 1.5436525085289813e-05,
1591
+ "loss": 2.0601,
1592
+ "step": 264
1593
+ },
1594
+ {
1595
+ "epoch": 0.34,
1596
+ "learning_rate": 1.5401646514219642e-05,
1597
+ "loss": 2.1587,
1598
+ "step": 265
1599
+ },
1600
+ {
1601
+ "epoch": 0.34,
1602
+ "learning_rate": 1.536667491089971e-05,
1603
+ "loss": 1.9219,
1604
+ "step": 266
1605
+ },
1606
+ {
1607
+ "epoch": 0.34,
1608
+ "learning_rate": 1.5331610877643947e-05,
1609
+ "loss": 2.0254,
1610
+ "step": 267
1611
+ },
1612
+ {
1613
+ "epoch": 0.34,
1614
+ "learning_rate": 1.529645501835819e-05,
1615
+ "loss": 1.9697,
1616
+ "step": 268
1617
+ },
1618
+ {
1619
+ "epoch": 0.34,
1620
+ "learning_rate": 1.526120793852981e-05,
1621
+ "loss": 2.0352,
1622
+ "step": 269
1623
+ },
1624
+ {
1625
+ "epoch": 0.35,
1626
+ "learning_rate": 1.5225870245217245e-05,
1627
+ "loss": 1.999,
1628
+ "step": 270
1629
+ },
1630
+ {
1631
+ "epoch": 0.35,
1632
+ "learning_rate": 1.5190442547039567e-05,
1633
+ "loss": 2.0054,
1634
+ "step": 271
1635
+ },
1636
+ {
1637
+ "epoch": 0.35,
1638
+ "learning_rate": 1.5154925454166005e-05,
1639
+ "loss": 2.1294,
1640
+ "step": 272
1641
+ },
1642
+ {
1643
+ "epoch": 0.35,
1644
+ "learning_rate": 1.5119319578305427e-05,
1645
+ "loss": 1.9604,
1646
+ "step": 273
1647
+ },
1648
+ {
1649
+ "epoch": 0.35,
1650
+ "learning_rate": 1.5083625532695796e-05,
1651
+ "loss": 2.0996,
1652
+ "step": 274
1653
+ },
1654
+ {
1655
+ "epoch": 0.35,
1656
+ "learning_rate": 1.5047843932093627e-05,
1657
+ "loss": 2.0474,
1658
+ "step": 275
1659
+ },
1660
+ {
1661
+ "epoch": 0.35,
1662
+ "learning_rate": 1.5011975392763383e-05,
1663
+ "loss": 1.9824,
1664
+ "step": 276
1665
+ },
1666
+ {
1667
+ "epoch": 0.35,
1668
+ "learning_rate": 1.4976020532466872e-05,
1669
+ "loss": 2.0278,
1670
+ "step": 277
1671
+ },
1672
+ {
1673
+ "epoch": 0.36,
1674
+ "learning_rate": 1.49399799704526e-05,
1675
+ "loss": 2.0366,
1676
+ "step": 278
1677
+ },
1678
+ {
1679
+ "epoch": 0.36,
1680
+ "learning_rate": 1.4903854327445115e-05,
1681
+ "loss": 2.0146,
1682
+ "step": 279
1683
+ },
1684
+ {
1685
+ "epoch": 0.36,
1686
+ "learning_rate": 1.4867644225634302e-05,
1687
+ "loss": 1.9131,
1688
+ "step": 280
1689
+ },
1690
+ {
1691
+ "epoch": 0.36,
1692
+ "learning_rate": 1.483135028866468e-05,
1693
+ "loss": 2.0068,
1694
+ "step": 281
1695
+ },
1696
+ {
1697
+ "epoch": 0.36,
1698
+ "learning_rate": 1.4794973141624655e-05,
1699
+ "loss": 2.0952,
1700
+ "step": 282
1701
+ },
1702
+ {
1703
+ "epoch": 0.36,
1704
+ "learning_rate": 1.4758513411035753e-05,
1705
+ "loss": 2.0737,
1706
+ "step": 283
1707
+ },
1708
+ {
1709
+ "epoch": 0.36,
1710
+ "learning_rate": 1.4721971724841837e-05,
1711
+ "loss": 1.9829,
1712
+ "step": 284
1713
+ },
1714
+ {
1715
+ "epoch": 0.36,
1716
+ "learning_rate": 1.4685348712398281e-05,
1717
+ "loss": 1.9634,
1718
+ "step": 285
1719
+ },
1720
+ {
1721
+ "epoch": 0.37,
1722
+ "learning_rate": 1.4648645004461142e-05,
1723
+ "loss": 1.9995,
1724
+ "step": 286
1725
+ },
1726
+ {
1727
+ "epoch": 0.37,
1728
+ "learning_rate": 1.4611861233176288e-05,
1729
+ "loss": 2.1074,
1730
+ "step": 287
1731
+ },
1732
+ {
1733
+ "epoch": 0.37,
1734
+ "learning_rate": 1.457499803206851e-05,
1735
+ "loss": 1.9814,
1736
+ "step": 288
1737
+ },
1738
+ {
1739
+ "epoch": 0.37,
1740
+ "learning_rate": 1.4538056036030622e-05,
1741
+ "loss": 2.0332,
1742
+ "step": 289
1743
+ },
1744
+ {
1745
+ "epoch": 0.37,
1746
+ "learning_rate": 1.4501035881312506e-05,
1747
+ "loss": 1.9326,
1748
+ "step": 290
1749
+ },
1750
+ {
1751
+ "epoch": 0.37,
1752
+ "learning_rate": 1.4463938205510186e-05,
1753
+ "loss": 1.9854,
1754
+ "step": 291
1755
+ },
1756
+ {
1757
+ "epoch": 0.37,
1758
+ "learning_rate": 1.4426763647554809e-05,
1759
+ "loss": 1.9429,
1760
+ "step": 292
1761
+ },
1762
+ {
1763
+ "epoch": 0.37,
1764
+ "learning_rate": 1.4389512847701666e-05,
1765
+ "loss": 2.0698,
1766
+ "step": 293
1767
+ },
1768
+ {
1769
+ "epoch": 0.38,
1770
+ "learning_rate": 1.4352186447519163e-05,
1771
+ "loss": 1.9658,
1772
+ "step": 294
1773
+ },
1774
+ {
1775
+ "epoch": 0.38,
1776
+ "learning_rate": 1.4314785089877754e-05,
1777
+ "loss": 1.9731,
1778
+ "step": 295
1779
+ },
1780
+ {
1781
+ "epoch": 0.38,
1782
+ "learning_rate": 1.4277309418938892e-05,
1783
+ "loss": 1.9824,
1784
+ "step": 296
1785
+ },
1786
+ {
1787
+ "epoch": 0.38,
1788
+ "learning_rate": 1.423976008014392e-05,
1789
+ "loss": 1.9312,
1790
+ "step": 297
1791
+ },
1792
+ {
1793
+ "epoch": 0.38,
1794
+ "learning_rate": 1.4202137720202957e-05,
1795
+ "loss": 2.0278,
1796
+ "step": 298
1797
+ },
1798
+ {
1799
+ "epoch": 0.38,
1800
+ "learning_rate": 1.4164442987083763e-05,
1801
+ "loss": 1.814,
1802
+ "step": 299
1803
+ },
1804
+ {
1805
+ "epoch": 0.38,
1806
+ "learning_rate": 1.4126676530000571e-05,
1807
+ "loss": 1.9238,
1808
+ "step": 300
1809
+ },
1810
+ {
1811
+ "epoch": 0.39,
1812
+ "learning_rate": 1.4088838999402924e-05,
1813
+ "loss": 2.083,
1814
+ "step": 301
1815
+ },
1816
+ {
1817
+ "epoch": 0.39,
1818
+ "learning_rate": 1.405093104696445e-05,
1819
+ "loss": 1.9365,
1820
+ "step": 302
1821
+ },
1822
+ {
1823
+ "epoch": 0.39,
1824
+ "learning_rate": 1.401295332557165e-05,
1825
+ "loss": 1.9995,
1826
+ "step": 303
1827
+ },
1828
+ {
1829
+ "epoch": 0.39,
1830
+ "learning_rate": 1.3974906489312657e-05,
1831
+ "loss": 2.083,
1832
+ "step": 304
1833
+ },
1834
+ {
1835
+ "epoch": 0.39,
1836
+ "learning_rate": 1.3936791193465954e-05,
1837
+ "loss": 2.0029,
1838
+ "step": 305
1839
+ },
1840
+ {
1841
+ "epoch": 0.39,
1842
+ "learning_rate": 1.3898608094489115e-05,
1843
+ "loss": 2.063,
1844
+ "step": 306
1845
+ },
1846
+ {
1847
+ "epoch": 0.39,
1848
+ "learning_rate": 1.386035785000747e-05,
1849
+ "loss": 2.0127,
1850
+ "step": 307
1851
+ },
1852
+ {
1853
+ "epoch": 0.39,
1854
+ "learning_rate": 1.3822041118802798e-05,
1855
+ "loss": 2.0518,
1856
+ "step": 308
1857
+ },
1858
+ {
1859
+ "epoch": 0.4,
1860
+ "learning_rate": 1.378365856080198e-05,
1861
+ "loss": 1.8789,
1862
+ "step": 309
1863
+ },
1864
+ {
1865
+ "epoch": 0.4,
1866
+ "learning_rate": 1.3745210837065617e-05,
1867
+ "loss": 1.9512,
1868
+ "step": 310
1869
+ },
1870
+ {
1871
+ "epoch": 0.4,
1872
+ "learning_rate": 1.3706698609776662e-05,
1873
+ "loss": 2.0664,
1874
+ "step": 311
1875
+ },
1876
+ {
1877
+ "epoch": 0.4,
1878
+ "learning_rate": 1.3668122542229014e-05,
1879
+ "loss": 1.9902,
1880
+ "step": 312
1881
+ },
1882
+ {
1883
+ "epoch": 0.4,
1884
+ "learning_rate": 1.3629483298816077e-05,
1885
+ "loss": 2.0483,
1886
+ "step": 313
1887
+ },
1888
+ {
1889
+ "epoch": 0.4,
1890
+ "learning_rate": 1.359078154501934e-05,
1891
+ "loss": 2.0166,
1892
+ "step": 314
1893
+ },
1894
+ {
1895
+ "epoch": 0.4,
1896
+ "learning_rate": 1.3552017947396898e-05,
1897
+ "loss": 1.9863,
1898
+ "step": 315
1899
+ },
1900
+ {
1901
+ "epoch": 0.4,
1902
+ "learning_rate": 1.3513193173571983e-05,
1903
+ "loss": 2.0244,
1904
+ "step": 316
1905
+ },
1906
+ {
1907
+ "epoch": 0.41,
1908
+ "learning_rate": 1.3474307892221464e-05,
1909
+ "loss": 1.9409,
1910
+ "step": 317
1911
+ },
1912
+ {
1913
+ "epoch": 0.41,
1914
+ "learning_rate": 1.3435362773064316e-05,
1915
+ "loss": 2.0181,
1916
+ "step": 318
1917
+ },
1918
+ {
1919
+ "epoch": 0.41,
1920
+ "learning_rate": 1.3396358486850106e-05,
1921
+ "loss": 1.9707,
1922
+ "step": 319
1923
+ },
1924
+ {
1925
+ "epoch": 0.41,
1926
+ "learning_rate": 1.3357295705347433e-05,
1927
+ "loss": 1.9966,
1928
+ "step": 320
1929
+ },
1930
+ {
1931
+ "epoch": 0.41,
1932
+ "learning_rate": 1.3318175101332353e-05,
1933
+ "loss": 1.8218,
1934
+ "step": 321
1935
+ },
1936
+ {
1937
+ "epoch": 0.41,
1938
+ "learning_rate": 1.32789973485768e-05,
1939
+ "loss": 1.9785,
1940
+ "step": 322
1941
+ },
1942
+ {
1943
+ "epoch": 0.41,
1944
+ "learning_rate": 1.323976312183697e-05,
1945
+ "loss": 1.998,
1946
+ "step": 323
1947
+ },
1948
+ {
1949
+ "epoch": 0.41,
1950
+ "learning_rate": 1.3200473096841715e-05,
1951
+ "loss": 1.9785,
1952
+ "step": 324
1953
+ },
1954
+ {
1955
+ "epoch": 0.42,
1956
+ "learning_rate": 1.3161127950280894e-05,
1957
+ "loss": 2.0083,
1958
+ "step": 325
1959
+ },
1960
+ {
1961
+ "epoch": 0.42,
1962
+ "learning_rate": 1.3121728359793719e-05,
1963
+ "loss": 1.8789,
1964
+ "step": 326
1965
+ },
1966
+ {
1967
+ "epoch": 0.42,
1968
+ "learning_rate": 1.3082275003957094e-05,
1969
+ "loss": 1.9678,
1970
+ "step": 327
1971
+ },
1972
+ {
1973
+ "epoch": 0.42,
1974
+ "learning_rate": 1.304276856227391e-05,
1975
+ "loss": 2.0073,
1976
+ "step": 328
1977
+ },
1978
+ {
1979
+ "epoch": 0.42,
1980
+ "learning_rate": 1.3003209715161362e-05,
1981
+ "loss": 2.1016,
1982
+ "step": 329
1983
+ },
1984
+ {
1985
+ "epoch": 0.42,
1986
+ "learning_rate": 1.296359914393922e-05,
1987
+ "loss": 2.0684,
1988
+ "step": 330
1989
+ },
1990
+ {
1991
+ "epoch": 0.42,
1992
+ "learning_rate": 1.2923937530818082e-05,
1993
+ "loss": 1.8833,
1994
+ "step": 331
1995
+ },
1996
+ {
1997
+ "epoch": 0.42,
1998
+ "learning_rate": 1.2884225558887664e-05,
1999
+ "loss": 2.0171,
2000
+ "step": 332
2001
+ },
2002
+ {
2003
+ "epoch": 0.43,
2004
+ "learning_rate": 1.2844463912104979e-05,
2005
+ "loss": 1.9766,
2006
+ "step": 333
2007
+ },
2008
+ {
2009
+ "epoch": 0.43,
2010
+ "learning_rate": 1.2804653275282605e-05,
2011
+ "loss": 1.9922,
2012
+ "step": 334
2013
+ },
2014
+ {
2015
+ "epoch": 0.43,
2016
+ "learning_rate": 1.2764794334076876e-05,
2017
+ "loss": 1.936,
2018
+ "step": 335
2019
+ },
2020
+ {
2021
+ "epoch": 0.43,
2022
+ "learning_rate": 1.2724887774976051e-05,
2023
+ "loss": 1.9175,
2024
+ "step": 336
2025
+ },
2026
+ {
2027
+ "epoch": 0.43,
2028
+ "learning_rate": 1.2684934285288527e-05,
2029
+ "loss": 1.8794,
2030
+ "step": 337
2031
+ },
2032
+ {
2033
+ "epoch": 0.43,
2034
+ "learning_rate": 1.2644934553130981e-05,
2035
+ "loss": 1.9541,
2036
+ "step": 338
2037
+ },
2038
+ {
2039
+ "epoch": 0.43,
2040
+ "learning_rate": 1.2604889267416511e-05,
2041
+ "loss": 1.8853,
2042
+ "step": 339
2043
+ },
2044
+ {
2045
+ "epoch": 0.44,
2046
+ "learning_rate": 1.2564799117842796e-05,
2047
+ "loss": 1.9663,
2048
+ "step": 340
2049
+ },
2050
+ {
2051
+ "epoch": 0.44,
2052
+ "learning_rate": 1.2524664794880189e-05,
2053
+ "loss": 1.9648,
2054
+ "step": 341
2055
+ },
2056
+ {
2057
+ "epoch": 0.44,
2058
+ "learning_rate": 1.2484486989759846e-05,
2059
+ "loss": 2.0068,
2060
+ "step": 342
2061
+ },
2062
+ {
2063
+ "epoch": 0.44,
2064
+ "learning_rate": 1.2444266394461808e-05,
2065
+ "loss": 2.0991,
2066
+ "step": 343
2067
+ },
2068
+ {
2069
+ "epoch": 0.44,
2070
+ "learning_rate": 1.2404003701703102e-05,
2071
+ "loss": 1.8794,
2072
+ "step": 344
2073
+ },
2074
+ {
2075
+ "epoch": 0.44,
2076
+ "learning_rate": 1.236369960492579e-05,
2077
+ "loss": 1.9502,
2078
+ "step": 345
2079
+ },
2080
+ {
2081
+ "epoch": 0.44,
2082
+ "learning_rate": 1.232335479828502e-05,
2083
+ "loss": 2.0928,
2084
+ "step": 346
2085
+ },
2086
+ {
2087
+ "epoch": 0.44,
2088
+ "learning_rate": 1.2282969976637107e-05,
2089
+ "loss": 1.8931,
2090
+ "step": 347
2091
+ },
2092
+ {
2093
+ "epoch": 0.45,
2094
+ "learning_rate": 1.2242545835527519e-05,
2095
+ "loss": 2.0435,
2096
+ "step": 348
2097
+ },
2098
+ {
2099
+ "epoch": 0.45,
2100
+ "learning_rate": 1.2202083071178938e-05,
2101
+ "loss": 1.9668,
2102
+ "step": 349
2103
+ },
2104
+ {
2105
+ "epoch": 0.45,
2106
+ "learning_rate": 1.2161582380479251e-05,
2107
+ "loss": 2.04,
2108
+ "step": 350
2109
+ },
2110
+ {
2111
+ "epoch": 0.45,
2112
+ "learning_rate": 1.2121044460969538e-05,
2113
+ "loss": 1.9336,
2114
+ "step": 351
2115
+ },
2116
+ {
2117
+ "epoch": 0.45,
2118
+ "learning_rate": 1.2080470010832074e-05,
2119
+ "loss": 2.0298,
2120
+ "step": 352
2121
+ },
2122
+ {
2123
+ "epoch": 0.45,
2124
+ "learning_rate": 1.2039859728878298e-05,
2125
+ "loss": 1.9668,
2126
+ "step": 353
2127
+ },
2128
+ {
2129
+ "epoch": 0.45,
2130
+ "learning_rate": 1.1999214314536783e-05,
2131
+ "loss": 1.9595,
2132
+ "step": 354
2133
+ },
2134
+ {
2135
+ "epoch": 0.45,
2136
+ "learning_rate": 1.1958534467841178e-05,
2137
+ "loss": 2.0029,
2138
+ "step": 355
2139
+ },
2140
+ {
2141
+ "epoch": 0.46,
2142
+ "learning_rate": 1.1917820889418165e-05,
2143
+ "loss": 2.0039,
2144
+ "step": 356
2145
+ },
2146
+ {
2147
+ "epoch": 0.46,
2148
+ "learning_rate": 1.1877074280475383e-05,
2149
+ "loss": 2.0479,
2150
+ "step": 357
2151
+ },
2152
+ {
2153
+ "epoch": 0.46,
2154
+ "learning_rate": 1.1836295342789348e-05,
2155
+ "loss": 1.9512,
2156
+ "step": 358
2157
+ },
2158
+ {
2159
+ "epoch": 0.46,
2160
+ "learning_rate": 1.1795484778693382e-05,
2161
+ "loss": 2.0303,
2162
+ "step": 359
2163
+ },
2164
+ {
2165
+ "epoch": 0.46,
2166
+ "learning_rate": 1.1754643291065503e-05,
2167
+ "loss": 1.8911,
2168
+ "step": 360
2169
+ },
2170
+ {
2171
+ "epoch": 0.46,
2172
+ "learning_rate": 1.1713771583316315e-05,
2173
+ "loss": 1.9609,
2174
+ "step": 361
2175
+ },
2176
+ {
2177
+ "epoch": 0.46,
2178
+ "learning_rate": 1.1672870359376922e-05,
2179
+ "loss": 1.938,
2180
+ "step": 362
2181
+ },
2182
+ {
2183
+ "epoch": 0.46,
2184
+ "learning_rate": 1.1631940323686758e-05,
2185
+ "loss": 1.9536,
2186
+ "step": 363
2187
+ },
2188
+ {
2189
+ "epoch": 0.47,
2190
+ "learning_rate": 1.15909821811815e-05,
2191
+ "loss": 1.8066,
2192
+ "step": 364
2193
+ },
2194
+ {
2195
+ "epoch": 0.47,
2196
+ "learning_rate": 1.15499966372809e-05,
2197
+ "loss": 2.0933,
2198
+ "step": 365
2199
+ },
2200
+ {
2201
+ "epoch": 0.47,
2202
+ "learning_rate": 1.1508984397876636e-05,
2203
+ "loss": 1.9395,
2204
+ "step": 366
2205
+ },
2206
+ {
2207
+ "epoch": 0.47,
2208
+ "learning_rate": 1.1467946169320175e-05,
2209
+ "loss": 1.9658,
2210
+ "step": 367
2211
+ },
2212
+ {
2213
+ "epoch": 0.47,
2214
+ "learning_rate": 1.1426882658410578e-05,
2215
+ "loss": 1.9121,
2216
+ "step": 368
2217
+ },
2218
+ {
2219
+ "epoch": 0.47,
2220
+ "learning_rate": 1.1385794572382359e-05,
2221
+ "loss": 2.0469,
2222
+ "step": 369
2223
+ },
2224
+ {
2225
+ "epoch": 0.47,
2226
+ "learning_rate": 1.1344682618893279e-05,
2227
+ "loss": 1.9639,
2228
+ "step": 370
2229
+ },
2230
+ {
2231
+ "epoch": 0.47,
2232
+ "learning_rate": 1.1303547506012163e-05,
2233
+ "loss": 1.9038,
2234
+ "step": 371
2235
+ },
2236
+ {
2237
+ "epoch": 0.48,
2238
+ "learning_rate": 1.1262389942206713e-05,
2239
+ "loss": 2.0273,
2240
+ "step": 372
2241
+ },
2242
+ {
2243
+ "epoch": 0.48,
2244
+ "learning_rate": 1.1221210636331307e-05,
2245
+ "loss": 1.9492,
2246
+ "step": 373
2247
+ },
2248
+ {
2249
+ "epoch": 0.48,
2250
+ "learning_rate": 1.1180010297614779e-05,
2251
+ "loss": 2.0322,
2252
+ "step": 374
2253
+ },
2254
+ {
2255
+ "epoch": 0.48,
2256
+ "learning_rate": 1.1138789635648212e-05,
2257
+ "loss": 1.9922,
2258
+ "step": 375
2259
+ },
2260
+ {
2261
+ "epoch": 0.48,
2262
+ "learning_rate": 1.1097549360372718e-05,
2263
+ "loss": 1.8462,
2264
+ "step": 376
2265
+ },
2266
+ {
2267
+ "epoch": 0.48,
2268
+ "learning_rate": 1.10562901820672e-05,
2269
+ "loss": 2.0283,
2270
+ "step": 377
2271
+ },
2272
+ {
2273
+ "epoch": 0.48,
2274
+ "learning_rate": 1.101501281133614e-05,
2275
+ "loss": 1.9414,
2276
+ "step": 378
2277
+ },
2278
+ {
2279
+ "epoch": 0.48,
2280
+ "learning_rate": 1.0973717959097327e-05,
2281
+ "loss": 1.9673,
2282
+ "step": 379
2283
+ },
2284
+ {
2285
+ "epoch": 0.49,
2286
+ "learning_rate": 1.0932406336569658e-05,
2287
+ "loss": 1.8564,
2288
+ "step": 380
2289
+ },
2290
+ {
2291
+ "epoch": 0.49,
2292
+ "learning_rate": 1.0891078655260845e-05,
2293
+ "loss": 1.9932,
2294
+ "step": 381
2295
+ },
2296
+ {
2297
+ "epoch": 0.49,
2298
+ "learning_rate": 1.0849735626955186e-05,
2299
+ "loss": 2.0469,
2300
+ "step": 382
2301
+ },
2302
+ {
2303
+ "epoch": 0.49,
2304
+ "learning_rate": 1.0808377963701302e-05,
2305
+ "loss": 1.9185,
2306
+ "step": 383
2307
+ },
2308
+ {
2309
+ "epoch": 0.49,
2310
+ "learning_rate": 1.0767006377799864e-05,
2311
+ "loss": 1.8628,
2312
+ "step": 384
2313
+ },
2314
+ {
2315
+ "epoch": 0.49,
2316
+ "learning_rate": 1.0725621581791336e-05,
2317
+ "loss": 1.979,
2318
+ "step": 385
2319
+ },
2320
+ {
2321
+ "epoch": 0.49,
2322
+ "learning_rate": 1.0684224288443701e-05,
2323
+ "loss": 1.9805,
2324
+ "step": 386
2325
+ },
2326
+ {
2327
+ "epoch": 0.5,
2328
+ "learning_rate": 1.0642815210740181e-05,
2329
+ "loss": 1.979,
2330
+ "step": 387
2331
+ },
2332
+ {
2333
+ "epoch": 0.5,
2334
+ "learning_rate": 1.0601395061866957e-05,
2335
+ "loss": 1.8887,
2336
+ "step": 388
2337
+ },
2338
+ {
2339
+ "epoch": 0.5,
2340
+ "learning_rate": 1.055996455520089e-05,
2341
+ "loss": 1.8184,
2342
+ "step": 389
2343
+ },
2344
+ {
2345
+ "epoch": 0.5,
2346
+ "learning_rate": 1.0518524404297232e-05,
2347
+ "loss": 1.9312,
2348
+ "step": 390
2349
+ },
2350
+ {
2351
+ "epoch": 0.5,
2352
+ "learning_rate": 1.0477075322877342e-05,
2353
+ "loss": 1.9297,
2354
+ "step": 391
2355
+ },
2356
+ {
2357
+ "epoch": 0.5,
2358
+ "learning_rate": 1.0435618024816381e-05,
2359
+ "loss": 1.9692,
2360
+ "step": 392
2361
+ },
2362
+ {
2363
+ "epoch": 0.5,
2364
+ "learning_rate": 1.0394153224131031e-05,
2365
+ "loss": 1.9863,
2366
+ "step": 393
2367
+ },
2368
+ {
2369
+ "epoch": 0.5,
2370
+ "learning_rate": 1.0352681634967185e-05,
2371
+ "loss": 1.9209,
2372
+ "step": 394
2373
+ },
2374
+ {
2375
+ "epoch": 0.51,
2376
+ "learning_rate": 1.0311203971587662e-05,
2377
+ "loss": 1.9927,
2378
+ "step": 395
2379
+ },
2380
+ {
2381
+ "epoch": 0.51,
2382
+ "learning_rate": 1.0269720948359891e-05,
2383
+ "loss": 1.9634,
2384
+ "step": 396
2385
+ },
2386
+ {
2387
+ "epoch": 0.51,
2388
+ "learning_rate": 1.0228233279743608e-05,
2389
+ "loss": 1.9648,
2390
+ "step": 397
2391
+ },
2392
+ {
2393
+ "epoch": 0.51,
2394
+ "learning_rate": 1.0186741680278573e-05,
2395
+ "loss": 1.8765,
2396
+ "step": 398
2397
+ },
2398
+ {
2399
+ "epoch": 0.51,
2400
+ "learning_rate": 1.0145246864572231e-05,
2401
+ "loss": 1.8594,
2402
+ "step": 399
2403
+ },
2404
+ {
2405
+ "epoch": 0.51,
2406
+ "learning_rate": 1.0103749547287425e-05,
2407
+ "loss": 1.8989,
2408
+ "step": 400
2409
+ },
2410
+ {
2411
+ "epoch": 0.51,
2412
+ "learning_rate": 1.0062250443130087e-05,
2413
+ "loss": 1.9629,
2414
+ "step": 401
2415
+ },
2416
+ {
2417
+ "epoch": 0.51,
2418
+ "learning_rate": 1.0020750266836919e-05,
2419
+ "loss": 2.0103,
2420
+ "step": 402
2421
+ },
2422
+ {
2423
+ "epoch": 0.52,
2424
+ "learning_rate": 9.979249733163088e-06,
2425
+ "loss": 1.8828,
2426
+ "step": 403
2427
+ },
2428
+ {
2429
+ "epoch": 0.52,
2430
+ "learning_rate": 9.937749556869914e-06,
2431
+ "loss": 1.9453,
2432
+ "step": 404
2433
+ },
2434
+ {
2435
+ "epoch": 0.52,
2436
+ "learning_rate": 9.896250452712576e-06,
2437
+ "loss": 1.9536,
2438
+ "step": 405
2439
+ },
2440
+ {
2441
+ "epoch": 0.52,
2442
+ "learning_rate": 9.854753135427772e-06,
2443
+ "loss": 1.979,
2444
+ "step": 406
2445
+ },
2446
+ {
2447
+ "epoch": 0.52,
2448
+ "learning_rate": 9.813258319721425e-06,
2449
+ "loss": 1.9497,
2450
+ "step": 407
2451
+ },
2452
+ {
2453
+ "epoch": 0.52,
2454
+ "learning_rate": 9.771766720256393e-06,
2455
+ "loss": 1.9351,
2456
+ "step": 408
2457
+ },
2458
+ {
2459
+ "epoch": 0.52,
2460
+ "learning_rate": 9.73027905164011e-06,
2461
+ "loss": 1.9131,
2462
+ "step": 409
2463
+ },
2464
+ {
2465
+ "epoch": 0.52,
2466
+ "learning_rate": 9.68879602841234e-06,
2467
+ "loss": 1.9775,
2468
+ "step": 410
2469
+ },
2470
+ {
2471
+ "epoch": 0.53,
2472
+ "learning_rate": 9.647318365032817e-06,
2473
+ "loss": 1.8779,
2474
+ "step": 411
2475
+ },
2476
+ {
2477
+ "epoch": 0.53,
2478
+ "learning_rate": 9.605846775868969e-06,
2479
+ "loss": 1.9995,
2480
+ "step": 412
2481
+ },
2482
+ {
2483
+ "epoch": 0.53,
2484
+ "learning_rate": 9.564381975183622e-06,
2485
+ "loss": 2.083,
2486
+ "step": 413
2487
+ },
2488
+ {
2489
+ "epoch": 0.53,
2490
+ "learning_rate": 9.522924677122658e-06,
2491
+ "loss": 2.0142,
2492
+ "step": 414
2493
+ },
2494
+ {
2495
+ "epoch": 0.53,
2496
+ "learning_rate": 9.48147559570277e-06,
2497
+ "loss": 1.8374,
2498
+ "step": 415
2499
+ },
2500
+ {
2501
+ "epoch": 0.53,
2502
+ "learning_rate": 9.440035444799113e-06,
2503
+ "loss": 1.9844,
2504
+ "step": 416
2505
+ },
2506
+ {
2507
+ "epoch": 0.53,
2508
+ "learning_rate": 9.398604938133048e-06,
2509
+ "loss": 1.8804,
2510
+ "step": 417
2511
+ },
2512
+ {
2513
+ "epoch": 0.53,
2514
+ "learning_rate": 9.357184789259824e-06,
2515
+ "loss": 2.0132,
2516
+ "step": 418
2517
+ },
2518
+ {
2519
+ "epoch": 0.54,
2520
+ "learning_rate": 9.315775711556299e-06,
2521
+ "loss": 2.0073,
2522
+ "step": 419
2523
+ },
2524
+ {
2525
+ "epoch": 0.54,
2526
+ "learning_rate": 9.274378418208666e-06,
2527
+ "loss": 1.9912,
2528
+ "step": 420
2529
+ },
2530
+ {
2531
+ "epoch": 0.54,
2532
+ "learning_rate": 9.232993622200139e-06,
2533
+ "loss": 1.9824,
2534
+ "step": 421
2535
+ },
2536
+ {
2537
+ "epoch": 0.54,
2538
+ "learning_rate": 9.191622036298701e-06,
2539
+ "loss": 1.9194,
2540
+ "step": 422
2541
+ },
2542
+ {
2543
+ "epoch": 0.54,
2544
+ "learning_rate": 9.150264373044815e-06,
2545
+ "loss": 2.1201,
2546
+ "step": 423
2547
+ },
2548
+ {
2549
+ "epoch": 0.54,
2550
+ "learning_rate": 9.108921344739156e-06,
2551
+ "loss": 1.9688,
2552
+ "step": 424
2553
+ },
2554
+ {
2555
+ "epoch": 0.54,
2556
+ "learning_rate": 9.067593663430343e-06,
2557
+ "loss": 1.96,
2558
+ "step": 425
2559
+ },
2560
+ {
2561
+ "epoch": 0.55,
2562
+ "learning_rate": 9.026282040902675e-06,
2563
+ "loss": 1.9531,
2564
+ "step": 426
2565
+ },
2566
+ {
2567
+ "epoch": 0.55,
2568
+ "learning_rate": 8.984987188663867e-06,
2569
+ "loss": 1.9429,
2570
+ "step": 427
2571
+ },
2572
+ {
2573
+ "epoch": 0.55,
2574
+ "learning_rate": 8.943709817932803e-06,
2575
+ "loss": 1.9868,
2576
+ "step": 428
2577
+ },
2578
+ {
2579
+ "epoch": 0.55,
2580
+ "learning_rate": 8.902450639627287e-06,
2581
+ "loss": 1.9922,
2582
+ "step": 429
2583
+ },
2584
+ {
2585
+ "epoch": 0.55,
2586
+ "learning_rate": 8.86121036435179e-06,
2587
+ "loss": 1.8721,
2588
+ "step": 430
2589
+ },
2590
+ {
2591
+ "epoch": 0.55,
2592
+ "learning_rate": 8.819989702385222e-06,
2593
+ "loss": 1.9849,
2594
+ "step": 431
2595
+ },
2596
+ {
2597
+ "epoch": 0.55,
2598
+ "learning_rate": 8.778789363668696e-06,
2599
+ "loss": 1.8711,
2600
+ "step": 432
2601
+ },
2602
+ {
2603
+ "epoch": 0.55,
2604
+ "learning_rate": 8.73761005779329e-06,
2605
+ "loss": 1.9146,
2606
+ "step": 433
2607
+ },
2608
+ {
2609
+ "epoch": 0.56,
2610
+ "learning_rate": 8.696452493987844e-06,
2611
+ "loss": 1.8477,
2612
+ "step": 434
2613
+ },
2614
+ {
2615
+ "epoch": 0.56,
2616
+ "learning_rate": 8.655317381106725e-06,
2617
+ "loss": 1.8242,
2618
+ "step": 435
2619
+ },
2620
+ {
2621
+ "epoch": 0.56,
2622
+ "learning_rate": 8.614205427617641e-06,
2623
+ "loss": 1.9556,
2624
+ "step": 436
2625
+ },
2626
+ {
2627
+ "epoch": 0.56,
2628
+ "learning_rate": 8.573117341589423e-06,
2629
+ "loss": 2.0742,
2630
+ "step": 437
2631
+ },
2632
+ {
2633
+ "epoch": 0.56,
2634
+ "learning_rate": 8.53205383067983e-06,
2635
+ "loss": 2.0029,
2636
+ "step": 438
2637
+ },
2638
+ {
2639
+ "epoch": 0.56,
2640
+ "learning_rate": 8.491015602123369e-06,
2641
+ "loss": 2.0884,
2642
+ "step": 439
2643
+ },
2644
+ {
2645
+ "epoch": 0.56,
2646
+ "learning_rate": 8.450003362719104e-06,
2647
+ "loss": 1.9546,
2648
+ "step": 440
2649
+ },
2650
+ {
2651
+ "epoch": 0.56,
2652
+ "learning_rate": 8.409017818818505e-06,
2653
+ "loss": 1.877,
2654
+ "step": 441
2655
+ },
2656
+ {
2657
+ "epoch": 0.57,
2658
+ "learning_rate": 8.368059676313243e-06,
2659
+ "loss": 1.978,
2660
+ "step": 442
2661
+ },
2662
+ {
2663
+ "epoch": 0.57,
2664
+ "learning_rate": 8.327129640623081e-06,
2665
+ "loss": 1.9209,
2666
+ "step": 443
2667
+ },
2668
+ {
2669
+ "epoch": 0.57,
2670
+ "learning_rate": 8.286228416683686e-06,
2671
+ "loss": 1.9639,
2672
+ "step": 444
2673
+ },
2674
+ {
2675
+ "epoch": 0.57,
2676
+ "learning_rate": 8.2453567089345e-06,
2677
+ "loss": 1.9077,
2678
+ "step": 445
2679
+ },
2680
+ {
2681
+ "epoch": 0.57,
2682
+ "learning_rate": 8.204515221306623e-06,
2683
+ "loss": 1.8823,
2684
+ "step": 446
2685
+ },
2686
+ {
2687
+ "epoch": 0.57,
2688
+ "learning_rate": 8.163704657210653e-06,
2689
+ "loss": 1.8364,
2690
+ "step": 447
2691
+ },
2692
+ {
2693
+ "epoch": 0.57,
2694
+ "learning_rate": 8.12292571952462e-06,
2695
+ "loss": 1.9404,
2696
+ "step": 448
2697
+ },
2698
+ {
2699
+ "epoch": 0.57,
2700
+ "learning_rate": 8.082179110581838e-06,
2701
+ "loss": 2.0288,
2702
+ "step": 449
2703
+ },
2704
+ {
2705
+ "epoch": 0.58,
2706
+ "learning_rate": 8.041465532158822e-06,
2707
+ "loss": 1.8667,
2708
+ "step": 450
2709
+ },
2710
+ {
2711
+ "epoch": 0.58,
2712
+ "learning_rate": 8.000785685463222e-06,
2713
+ "loss": 1.8896,
2714
+ "step": 451
2715
+ },
2716
+ {
2717
+ "epoch": 0.58,
2718
+ "learning_rate": 7.960140271121704e-06,
2719
+ "loss": 1.9355,
2720
+ "step": 452
2721
+ },
2722
+ {
2723
+ "epoch": 0.58,
2724
+ "learning_rate": 7.91952998916793e-06,
2725
+ "loss": 2.0508,
2726
+ "step": 453
2727
+ },
2728
+ {
2729
+ "epoch": 0.58,
2730
+ "learning_rate": 7.878955539030467e-06,
2731
+ "loss": 1.877,
2732
+ "step": 454
2733
+ },
2734
+ {
2735
+ "epoch": 0.58,
2736
+ "learning_rate": 7.838417619520749e-06,
2737
+ "loss": 1.9707,
2738
+ "step": 455
2739
+ },
2740
+ {
2741
+ "epoch": 0.58,
2742
+ "learning_rate": 7.797916928821064e-06,
2743
+ "loss": 1.9692,
2744
+ "step": 456
2745
+ },
2746
+ {
2747
+ "epoch": 0.58,
2748
+ "learning_rate": 7.757454164472483e-06,
2749
+ "loss": 1.8623,
2750
+ "step": 457
2751
+ },
2752
+ {
2753
+ "epoch": 0.59,
2754
+ "learning_rate": 7.7170300233629e-06,
2755
+ "loss": 1.9805,
2756
+ "step": 458
2757
+ },
2758
+ {
2759
+ "epoch": 0.59,
2760
+ "learning_rate": 7.676645201714982e-06,
2761
+ "loss": 1.8755,
2762
+ "step": 459
2763
+ },
2764
+ {
2765
+ "epoch": 0.59,
2766
+ "learning_rate": 7.636300395074212e-06,
2767
+ "loss": 1.9907,
2768
+ "step": 460
2769
+ },
2770
+ {
2771
+ "epoch": 0.59,
2772
+ "learning_rate": 7.595996298296899e-06,
2773
+ "loss": 1.8579,
2774
+ "step": 461
2775
+ },
2776
+ {
2777
+ "epoch": 0.59,
2778
+ "learning_rate": 7.555733605538191e-06,
2779
+ "loss": 1.9761,
2780
+ "step": 462
2781
+ },
2782
+ {
2783
+ "epoch": 0.59,
2784
+ "learning_rate": 7.51551301024016e-06,
2785
+ "loss": 1.9346,
2786
+ "step": 463
2787
+ },
2788
+ {
2789
+ "epoch": 0.59,
2790
+ "learning_rate": 7.475335205119816e-06,
2791
+ "loss": 1.7603,
2792
+ "step": 464
2793
+ },
2794
+ {
2795
+ "epoch": 0.6,
2796
+ "learning_rate": 7.435200882157209e-06,
2797
+ "loss": 1.9668,
2798
+ "step": 465
2799
+ },
2800
+ {
2801
+ "epoch": 0.6,
2802
+ "learning_rate": 7.395110732583491e-06,
2803
+ "loss": 1.9907,
2804
+ "step": 466
2805
+ },
2806
+ {
2807
+ "epoch": 0.6,
2808
+ "learning_rate": 7.35506544686902e-06,
2809
+ "loss": 1.894,
2810
+ "step": 467
2811
+ },
2812
+ {
2813
+ "epoch": 0.6,
2814
+ "learning_rate": 7.315065714711473e-06,
2815
+ "loss": 1.8774,
2816
+ "step": 468
2817
+ },
2818
+ {
2819
+ "epoch": 0.6,
2820
+ "learning_rate": 7.2751122250239505e-06,
2821
+ "loss": 1.8784,
2822
+ "step": 469
2823
+ },
2824
+ {
2825
+ "epoch": 0.6,
2826
+ "learning_rate": 7.235205665923131e-06,
2827
+ "loss": 2.002,
2828
+ "step": 470
2829
+ },
2830
+ {
2831
+ "epoch": 0.6,
2832
+ "learning_rate": 7.195346724717397e-06,
2833
+ "loss": 2.019,
2834
+ "step": 471
2835
+ },
2836
+ {
2837
+ "epoch": 0.6,
2838
+ "learning_rate": 7.155536087895023e-06,
2839
+ "loss": 1.8442,
2840
+ "step": 472
2841
+ },
2842
+ {
2843
+ "epoch": 0.61,
2844
+ "learning_rate": 7.11577444111234e-06,
2845
+ "loss": 1.9863,
2846
+ "step": 473
2847
+ },
2848
+ {
2849
+ "epoch": 0.61,
2850
+ "learning_rate": 7.0760624691819164e-06,
2851
+ "loss": 1.8501,
2852
+ "step": 474
2853
+ },
2854
+ {
2855
+ "epoch": 0.61,
2856
+ "learning_rate": 7.0364008560607854e-06,
2857
+ "loss": 1.9819,
2858
+ "step": 475
2859
+ },
2860
+ {
2861
+ "epoch": 0.61,
2862
+ "learning_rate": 6.996790284838639e-06,
2863
+ "loss": 1.9263,
2864
+ "step": 476
2865
+ },
2866
+ {
2867
+ "epoch": 0.61,
2868
+ "learning_rate": 6.957231437726094e-06,
2869
+ "loss": 1.9229,
2870
+ "step": 477
2871
+ },
2872
+ {
2873
+ "epoch": 0.61,
2874
+ "learning_rate": 6.917724996042909e-06,
2875
+ "loss": 1.9482,
2876
+ "step": 478
2877
+ },
2878
+ {
2879
+ "epoch": 0.61,
2880
+ "learning_rate": 6.8782716402062825e-06,
2881
+ "loss": 1.8799,
2882
+ "step": 479
2883
+ },
2884
+ {
2885
+ "epoch": 0.61,
2886
+ "learning_rate": 6.838872049719109e-06,
2887
+ "loss": 2.0112,
2888
+ "step": 480
2889
+ },
2890
+ {
2891
+ "epoch": 0.62,
2892
+ "learning_rate": 6.799526903158288e-06,
2893
+ "loss": 1.8721,
2894
+ "step": 481
2895
+ },
2896
+ {
2897
+ "epoch": 0.62,
2898
+ "learning_rate": 6.760236878163035e-06,
2899
+ "loss": 1.8999,
2900
+ "step": 482
2901
+ },
2902
+ {
2903
+ "epoch": 0.62,
2904
+ "learning_rate": 6.721002651423203e-06,
2905
+ "loss": 1.9355,
2906
+ "step": 483
2907
+ },
2908
+ {
2909
+ "epoch": 0.62,
2910
+ "learning_rate": 6.681824898667648e-06,
2911
+ "loss": 1.9556,
2912
+ "step": 484
2913
+ },
2914
+ {
2915
+ "epoch": 0.62,
2916
+ "learning_rate": 6.64270429465257e-06,
2917
+ "loss": 1.8853,
2918
+ "step": 485
2919
+ },
2920
+ {
2921
+ "epoch": 0.62,
2922
+ "learning_rate": 6.603641513149897e-06,
2923
+ "loss": 2.0312,
2924
+ "step": 486
2925
+ },
2926
+ {
2927
+ "epoch": 0.62,
2928
+ "learning_rate": 6.564637226935691e-06,
2929
+ "loss": 1.8989,
2930
+ "step": 487
2931
+ },
2932
+ {
2933
+ "epoch": 0.62,
2934
+ "learning_rate": 6.52569210777854e-06,
2935
+ "loss": 1.9609,
2936
+ "step": 488
2937
+ },
2938
+ {
2939
+ "epoch": 0.63,
2940
+ "learning_rate": 6.486806826428017e-06,
2941
+ "loss": 2.0303,
2942
+ "step": 489
2943
+ },
2944
+ {
2945
+ "epoch": 0.63,
2946
+ "learning_rate": 6.447982052603105e-06,
2947
+ "loss": 1.9087,
2948
+ "step": 490
2949
+ },
2950
+ {
2951
+ "epoch": 0.63,
2952
+ "learning_rate": 6.409218454980663e-06,
2953
+ "loss": 1.9595,
2954
+ "step": 491
2955
+ },
2956
+ {
2957
+ "epoch": 0.63,
2958
+ "learning_rate": 6.370516701183926e-06,
2959
+ "loss": 1.9961,
2960
+ "step": 492
2961
+ },
2962
+ {
2963
+ "epoch": 0.63,
2964
+ "learning_rate": 6.331877457770987e-06,
2965
+ "loss": 1.9521,
2966
+ "step": 493
2967
+ },
2968
+ {
2969
+ "epoch": 0.63,
2970
+ "learning_rate": 6.293301390223339e-06,
2971
+ "loss": 1.9722,
2972
+ "step": 494
2973
+ },
2974
+ {
2975
+ "epoch": 0.63,
2976
+ "learning_rate": 6.254789162934386e-06,
2977
+ "loss": 1.9175,
2978
+ "step": 495
2979
+ },
2980
+ {
2981
+ "epoch": 0.63,
2982
+ "learning_rate": 6.216341439198023e-06,
2983
+ "loss": 1.9355,
2984
+ "step": 496
2985
+ },
2986
+ {
2987
+ "epoch": 0.64,
2988
+ "learning_rate": 6.177958881197205e-06,
2989
+ "loss": 1.9277,
2990
+ "step": 497
2991
+ },
2992
+ {
2993
+ "epoch": 0.64,
2994
+ "learning_rate": 6.139642149992532e-06,
2995
+ "loss": 1.8589,
2996
+ "step": 498
2997
+ },
2998
+ {
2999
+ "epoch": 0.64,
3000
+ "learning_rate": 6.101391905510889e-06,
3001
+ "loss": 2.0146,
3002
+ "step": 499
3003
+ },
3004
+ {
3005
+ "epoch": 0.64,
3006
+ "learning_rate": 6.063208806534049e-06,
3007
+ "loss": 1.79,
3008
+ "step": 500
3009
+ },
3010
+ {
3011
+ "epoch": 0.64,
3012
+ "learning_rate": 6.0250935106873474e-06,
3013
+ "loss": 1.9204,
3014
+ "step": 501
3015
+ },
3016
+ {
3017
+ "epoch": 0.64,
3018
+ "learning_rate": 5.987046674428354e-06,
3019
+ "loss": 1.9365,
3020
+ "step": 502
3021
+ },
3022
+ {
3023
+ "epoch": 0.64,
3024
+ "learning_rate": 5.949068953035551e-06,
3025
+ "loss": 1.896,
3026
+ "step": 503
3027
+ },
3028
+ {
3029
+ "epoch": 0.64,
3030
+ "learning_rate": 5.911161000597079e-06,
3031
+ "loss": 1.9868,
3032
+ "step": 504
3033
+ },
3034
+ {
3035
+ "epoch": 0.65,
3036
+ "learning_rate": 5.87332346999943e-06,
3037
+ "loss": 1.8936,
3038
+ "step": 505
3039
+ },
3040
+ {
3041
+ "epoch": 0.65,
3042
+ "learning_rate": 5.835557012916242e-06,
3043
+ "loss": 1.9629,
3044
+ "step": 506
3045
+ },
3046
+ {
3047
+ "epoch": 0.65,
3048
+ "learning_rate": 5.797862279797046e-06,
3049
+ "loss": 1.8838,
3050
+ "step": 507
3051
+ },
3052
+ {
3053
+ "epoch": 0.65,
3054
+ "learning_rate": 5.7602399198560835e-06,
3055
+ "loss": 2.0098,
3056
+ "step": 508
3057
+ },
3058
+ {
3059
+ "epoch": 0.65,
3060
+ "learning_rate": 5.72269058106111e-06,
3061
+ "loss": 1.8672,
3062
+ "step": 509
3063
+ },
3064
+ {
3065
+ "epoch": 0.65,
3066
+ "learning_rate": 5.6852149101222475e-06,
3067
+ "loss": 1.834,
3068
+ "step": 510
3069
+ },
3070
+ {
3071
+ "epoch": 0.65,
3072
+ "learning_rate": 5.647813552480843e-06,
3073
+ "loss": 1.918,
3074
+ "step": 511
3075
+ },
3076
+ {
3077
+ "epoch": 0.66,
3078
+ "learning_rate": 5.610487152298334e-06,
3079
+ "loss": 1.9829,
3080
+ "step": 512
3081
+ },
3082
+ {
3083
+ "epoch": 0.66,
3084
+ "learning_rate": 5.573236352445193e-06,
3085
+ "loss": 1.8784,
3086
+ "step": 513
3087
+ },
3088
+ {
3089
+ "epoch": 0.66,
3090
+ "learning_rate": 5.536061794489818e-06,
3091
+ "loss": 2.0059,
3092
+ "step": 514
3093
+ },
3094
+ {
3095
+ "epoch": 0.66,
3096
+ "learning_rate": 5.498964118687497e-06,
3097
+ "loss": 2.0068,
3098
+ "step": 515
3099
+ },
3100
+ {
3101
+ "epoch": 0.66,
3102
+ "learning_rate": 5.461943963969384e-06,
3103
+ "loss": 1.9331,
3104
+ "step": 516
3105
+ },
3106
+ {
3107
+ "epoch": 0.66,
3108
+ "learning_rate": 5.425001967931494e-06,
3109
+ "loss": 1.873,
3110
+ "step": 517
3111
+ },
3112
+ {
3113
+ "epoch": 0.66,
3114
+ "learning_rate": 5.388138766823714e-06,
3115
+ "loss": 1.8662,
3116
+ "step": 518
3117
+ },
3118
+ {
3119
+ "epoch": 0.66,
3120
+ "learning_rate": 5.351354995538859e-06,
3121
+ "loss": 1.9282,
3122
+ "step": 519
3123
+ },
3124
+ {
3125
+ "epoch": 0.67,
3126
+ "learning_rate": 5.314651287601719e-06,
3127
+ "loss": 1.9414,
3128
+ "step": 520
3129
+ },
3130
+ {
3131
+ "epoch": 0.67,
3132
+ "learning_rate": 5.278028275158167e-06,
3133
+ "loss": 1.855,
3134
+ "step": 521
3135
+ },
3136
+ {
3137
+ "epoch": 0.67,
3138
+ "learning_rate": 5.241486588964246e-06,
3139
+ "loss": 1.9526,
3140
+ "step": 522
3141
+ },
3142
+ {
3143
+ "epoch": 0.67,
3144
+ "learning_rate": 5.205026858375349e-06,
3145
+ "loss": 1.9907,
3146
+ "step": 523
3147
+ },
3148
+ {
3149
+ "epoch": 0.67,
3150
+ "learning_rate": 5.168649711335323e-06,
3151
+ "loss": 1.9683,
3152
+ "step": 524
3153
+ },
3154
+ {
3155
+ "epoch": 0.67,
3156
+ "learning_rate": 5.132355774365701e-06,
3157
+ "loss": 1.8896,
3158
+ "step": 525
3159
+ },
3160
+ {
3161
+ "epoch": 0.67,
3162
+ "learning_rate": 5.096145672554886e-06,
3163
+ "loss": 1.9541,
3164
+ "step": 526
3165
+ },
3166
+ {
3167
+ "epoch": 0.67,
3168
+ "learning_rate": 5.060020029547401e-06,
3169
+ "loss": 1.9482,
3170
+ "step": 527
3171
+ },
3172
+ {
3173
+ "epoch": 0.68,
3174
+ "learning_rate": 5.02397946753313e-06,
3175
+ "loss": 1.9028,
3176
+ "step": 528
3177
+ },
3178
+ {
3179
+ "epoch": 0.68,
3180
+ "learning_rate": 4.988024607236619e-06,
3181
+ "loss": 1.9028,
3182
+ "step": 529
3183
+ },
3184
+ {
3185
+ "epoch": 0.68,
3186
+ "learning_rate": 4.952156067906378e-06,
3187
+ "loss": 1.9551,
3188
+ "step": 530
3189
+ },
3190
+ {
3191
+ "epoch": 0.68,
3192
+ "learning_rate": 4.916374467304208e-06,
3193
+ "loss": 1.9941,
3194
+ "step": 531
3195
+ },
3196
+ {
3197
+ "epoch": 0.68,
3198
+ "learning_rate": 4.880680421694573e-06,
3199
+ "loss": 1.9458,
3200
+ "step": 532
3201
+ },
3202
+ {
3203
+ "epoch": 0.68,
3204
+ "learning_rate": 4.845074545833996e-06,
3205
+ "loss": 1.9141,
3206
+ "step": 533
3207
+ },
3208
+ {
3209
+ "epoch": 0.68,
3210
+ "learning_rate": 4.809557452960436e-06,
3211
+ "loss": 2.0146,
3212
+ "step": 534
3213
+ },
3214
+ {
3215
+ "epoch": 0.68,
3216
+ "learning_rate": 4.774129754782759e-06,
3217
+ "loss": 1.9121,
3218
+ "step": 535
3219
+ },
3220
+ {
3221
+ "epoch": 0.69,
3222
+ "learning_rate": 4.738792061470191e-06,
3223
+ "loss": 1.9287,
3224
+ "step": 536
3225
+ },
3226
+ {
3227
+ "epoch": 0.69,
3228
+ "learning_rate": 4.703544981641807e-06,
3229
+ "loss": 1.8999,
3230
+ "step": 537
3231
+ },
3232
+ {
3233
+ "epoch": 0.69,
3234
+ "learning_rate": 4.668389122356054e-06,
3235
+ "loss": 1.9165,
3236
+ "step": 538
3237
+ },
3238
+ {
3239
+ "epoch": 0.69,
3240
+ "learning_rate": 4.633325089100289e-06,
3241
+ "loss": 1.9824,
3242
+ "step": 539
3243
+ },
3244
+ {
3245
+ "epoch": 0.69,
3246
+ "learning_rate": 4.598353485780362e-06,
3247
+ "loss": 1.8184,
3248
+ "step": 540
3249
+ },
3250
+ {
3251
+ "epoch": 0.69,
3252
+ "learning_rate": 4.563474914710191e-06,
3253
+ "loss": 1.8813,
3254
+ "step": 541
3255
+ },
3256
+ {
3257
+ "epoch": 0.69,
3258
+ "learning_rate": 4.528689976601423e-06,
3259
+ "loss": 1.8662,
3260
+ "step": 542
3261
+ },
3262
+ {
3263
+ "epoch": 0.69,
3264
+ "learning_rate": 4.4939992705530615e-06,
3265
+ "loss": 1.8066,
3266
+ "step": 543
3267
+ },
3268
+ {
3269
+ "epoch": 0.7,
3270
+ "learning_rate": 4.459403394041156e-06,
3271
+ "loss": 2.0332,
3272
+ "step": 544
3273
+ },
3274
+ {
3275
+ "epoch": 0.7,
3276
+ "learning_rate": 4.424902942908518e-06,
3277
+ "loss": 1.8779,
3278
+ "step": 545
3279
+ },
3280
+ {
3281
+ "epoch": 0.7,
3282
+ "learning_rate": 4.390498511354445e-06,
3283
+ "loss": 1.8657,
3284
+ "step": 546
3285
+ },
3286
+ {
3287
+ "epoch": 0.7,
3288
+ "learning_rate": 4.356190691924505e-06,
3289
+ "loss": 1.9136,
3290
+ "step": 547
3291
+ },
3292
+ {
3293
+ "epoch": 0.7,
3294
+ "learning_rate": 4.321980075500305e-06,
3295
+ "loss": 2.0371,
3296
+ "step": 548
3297
+ },
3298
+ {
3299
+ "epoch": 0.7,
3300
+ "learning_rate": 4.287867251289348e-06,
3301
+ "loss": 1.834,
3302
+ "step": 549
3303
+ },
3304
+ {
3305
+ "epoch": 0.7,
3306
+ "learning_rate": 4.25385280681486e-06,
3307
+ "loss": 2.0015,
3308
+ "step": 550
3309
+ },
3310
+ {
3311
+ "epoch": 0.71,
3312
+ "learning_rate": 4.219937327905672e-06,
3313
+ "loss": 1.9365,
3314
+ "step": 551
3315
+ },
3316
+ {
3317
+ "epoch": 0.71,
3318
+ "learning_rate": 4.1861213986861445e-06,
3319
+ "loss": 1.8994,
3320
+ "step": 552
3321
+ },
3322
+ {
3323
+ "epoch": 0.71,
3324
+ "learning_rate": 4.1524056015661e-06,
3325
+ "loss": 1.8975,
3326
+ "step": 553
3327
+ },
3328
+ {
3329
+ "epoch": 0.71,
3330
+ "learning_rate": 4.118790517230788e-06,
3331
+ "loss": 1.8511,
3332
+ "step": 554
3333
+ },
3334
+ {
3335
+ "epoch": 0.71,
3336
+ "learning_rate": 4.085276724630882e-06,
3337
+ "loss": 1.8867,
3338
+ "step": 555
3339
+ },
3340
+ {
3341
+ "epoch": 0.71,
3342
+ "learning_rate": 4.051864800972528e-06,
3343
+ "loss": 1.9102,
3344
+ "step": 556
3345
+ },
3346
+ {
3347
+ "epoch": 0.71,
3348
+ "learning_rate": 4.018555321707382e-06,
3349
+ "loss": 1.9824,
3350
+ "step": 557
3351
+ },
3352
+ {
3353
+ "epoch": 0.71,
3354
+ "learning_rate": 3.985348860522698e-06,
3355
+ "loss": 1.9912,
3356
+ "step": 558
3357
+ },
3358
+ {
3359
+ "epoch": 0.72,
3360
+ "learning_rate": 3.952245989331466e-06,
3361
+ "loss": 1.9609,
3362
+ "step": 559
3363
+ },
3364
+ {
3365
+ "epoch": 0.72,
3366
+ "learning_rate": 3.919247278262548e-06,
3367
+ "loss": 1.9033,
3368
+ "step": 560
3369
+ },
3370
+ {
3371
+ "epoch": 0.72,
3372
+ "learning_rate": 3.886353295650863e-06,
3373
+ "loss": 1.8657,
3374
+ "step": 561
3375
+ },
3376
+ {
3377
+ "epoch": 0.72,
3378
+ "learning_rate": 3.8535646080275965e-06,
3379
+ "loss": 1.979,
3380
+ "step": 562
3381
+ },
3382
+ {
3383
+ "epoch": 0.72,
3384
+ "learning_rate": 3.820881780110445e-06,
3385
+ "loss": 1.9585,
3386
+ "step": 563
3387
+ },
3388
+ {
3389
+ "epoch": 0.72,
3390
+ "learning_rate": 3.7883053747938927e-06,
3391
+ "loss": 1.9097,
3392
+ "step": 564
3393
+ },
3394
+ {
3395
+ "epoch": 0.72,
3396
+ "learning_rate": 3.7558359531394983e-06,
3397
+ "loss": 1.916,
3398
+ "step": 565
3399
+ },
3400
+ {
3401
+ "epoch": 0.72,
3402
+ "learning_rate": 3.7234740743662746e-06,
3403
+ "loss": 1.9028,
3404
+ "step": 566
3405
+ },
3406
+ {
3407
+ "epoch": 0.73,
3408
+ "learning_rate": 3.6912202958410024e-06,
3409
+ "loss": 1.9531,
3410
+ "step": 567
3411
+ },
3412
+ {
3413
+ "epoch": 0.73,
3414
+ "learning_rate": 3.6590751730686726e-06,
3415
+ "loss": 1.9165,
3416
+ "step": 568
3417
+ },
3418
+ {
3419
+ "epoch": 0.73,
3420
+ "learning_rate": 3.6270392596828995e-06,
3421
+ "loss": 1.9932,
3422
+ "step": 569
3423
+ },
3424
+ {
3425
+ "epoch": 0.73,
3426
+ "learning_rate": 3.5951131074363942e-06,
3427
+ "loss": 1.9619,
3428
+ "step": 570
3429
+ },
3430
+ {
3431
+ "epoch": 0.73,
3432
+ "learning_rate": 3.5632972661914543e-06,
3433
+ "loss": 1.9092,
3434
+ "step": 571
3435
+ },
3436
+ {
3437
+ "epoch": 0.73,
3438
+ "learning_rate": 3.5315922839104976e-06,
3439
+ "loss": 1.9297,
3440
+ "step": 572
3441
+ },
3442
+ {
3443
+ "epoch": 0.73,
3444
+ "learning_rate": 3.4999987066466268e-06,
3445
+ "loss": 2.0015,
3446
+ "step": 573
3447
+ },
3448
+ {
3449
+ "epoch": 0.73,
3450
+ "learning_rate": 3.4685170785342235e-06,
3451
+ "loss": 1.915,
3452
+ "step": 574
3453
+ },
3454
+ {
3455
+ "epoch": 0.74,
3456
+ "learning_rate": 3.437147941779564e-06,
3457
+ "loss": 1.9634,
3458
+ "step": 575
3459
+ },
3460
+ {
3461
+ "epoch": 0.74,
3462
+ "learning_rate": 3.4058918366515114e-06,
3463
+ "loss": 1.9263,
3464
+ "step": 576
3465
+ },
3466
+ {
3467
+ "epoch": 0.74,
3468
+ "learning_rate": 3.374749301472173e-06,
3469
+ "loss": 1.9575,
3470
+ "step": 577
3471
+ },
3472
+ {
3473
+ "epoch": 0.74,
3474
+ "learning_rate": 3.34372087260766e-06,
3475
+ "loss": 2.022,
3476
+ "step": 578
3477
+ },
3478
+ {
3479
+ "epoch": 0.74,
3480
+ "learning_rate": 3.312807084458831e-06,
3481
+ "loss": 1.9121,
3482
+ "step": 579
3483
+ },
3484
+ {
3485
+ "epoch": 0.74,
3486
+ "learning_rate": 3.2820084694520973e-06,
3487
+ "loss": 1.9155,
3488
+ "step": 580
3489
+ },
3490
+ {
3491
+ "epoch": 0.74,
3492
+ "learning_rate": 3.251325558030245e-06,
3493
+ "loss": 1.9087,
3494
+ "step": 581
3495
+ },
3496
+ {
3497
+ "epoch": 0.74,
3498
+ "learning_rate": 3.2207588786433108e-06,
3499
+ "loss": 1.8374,
3500
+ "step": 582
3501
+ },
3502
+ {
3503
+ "epoch": 0.75,
3504
+ "learning_rate": 3.1903089577394707e-06,
3505
+ "loss": 1.8921,
3506
+ "step": 583
3507
+ },
3508
+ {
3509
+ "epoch": 0.75,
3510
+ "learning_rate": 3.1599763197559705e-06,
3511
+ "loss": 1.9136,
3512
+ "step": 584
3513
+ },
3514
+ {
3515
+ "epoch": 0.75,
3516
+ "learning_rate": 3.1297614871101035e-06,
3517
+ "loss": 1.8564,
3518
+ "step": 585
3519
+ },
3520
+ {
3521
+ "epoch": 0.75,
3522
+ "learning_rate": 3.0996649801902167e-06,
3523
+ "loss": 1.9043,
3524
+ "step": 586
3525
+ },
3526
+ {
3527
+ "epoch": 0.75,
3528
+ "learning_rate": 3.0696873173467245e-06,
3529
+ "loss": 1.9072,
3530
+ "step": 587
3531
+ },
3532
+ {
3533
+ "epoch": 0.75,
3534
+ "learning_rate": 3.0398290148832056e-06,
3535
+ "loss": 1.8618,
3536
+ "step": 588
3537
+ },
3538
+ {
3539
+ "epoch": 0.75,
3540
+ "learning_rate": 3.0100905870475004e-06,
3541
+ "loss": 1.9229,
3542
+ "step": 589
3543
+ },
3544
+ {
3545
+ "epoch": 0.75,
3546
+ "learning_rate": 2.980472546022859e-06,
3547
+ "loss": 1.8867,
3548
+ "step": 590
3549
+ },
3550
+ {
3551
+ "epoch": 0.76,
3552
+ "learning_rate": 2.9509754019191017e-06,
3553
+ "loss": 1.9492,
3554
+ "step": 591
3555
+ },
3556
+ {
3557
+ "epoch": 0.76,
3558
+ "learning_rate": 2.921599662763868e-06,
3559
+ "loss": 1.8779,
3560
+ "step": 592
3561
+ },
3562
+ {
3563
+ "epoch": 0.76,
3564
+ "learning_rate": 2.8923458344938383e-06,
3565
+ "loss": 1.793,
3566
+ "step": 593
3567
+ },
3568
+ {
3569
+ "epoch": 0.76,
3570
+ "learning_rate": 2.863214420946021e-06,
3571
+ "loss": 1.8242,
3572
+ "step": 594
3573
+ },
3574
+ {
3575
+ "epoch": 0.76,
3576
+ "learning_rate": 2.8342059238490937e-06,
3577
+ "loss": 1.8652,
3578
+ "step": 595
3579
+ },
3580
+ {
3581
+ "epoch": 0.76,
3582
+ "learning_rate": 2.8053208428147462e-06,
3583
+ "loss": 1.9077,
3584
+ "step": 596
3585
+ },
3586
+ {
3587
+ "epoch": 0.76,
3588
+ "learning_rate": 2.7765596753290826e-06,
3589
+ "loss": 1.8579,
3590
+ "step": 597
3591
+ },
3592
+ {
3593
+ "epoch": 0.77,
3594
+ "learning_rate": 2.747922916744048e-06,
3595
+ "loss": 1.9146,
3596
+ "step": 598
3597
+ },
3598
+ {
3599
+ "epoch": 0.77,
3600
+ "learning_rate": 2.719411060268903e-06,
3601
+ "loss": 1.8804,
3602
+ "step": 599
3603
+ },
3604
+ {
3605
+ "epoch": 0.77,
3606
+ "learning_rate": 2.691024596961729e-06,
3607
+ "loss": 1.8955,
3608
+ "step": 600
3609
+ },
3610
+ {
3611
+ "epoch": 0.77,
3612
+ "learning_rate": 2.6627640157209523e-06,
3613
+ "loss": 1.8413,
3614
+ "step": 601
3615
+ },
3616
+ {
3617
+ "epoch": 0.77,
3618
+ "learning_rate": 2.6346298032769657e-06,
3619
+ "loss": 1.9165,
3620
+ "step": 602
3621
+ },
3622
+ {
3623
+ "epoch": 0.77,
3624
+ "learning_rate": 2.6066224441836963e-06,
3625
+ "loss": 1.9927,
3626
+ "step": 603
3627
+ },
3628
+ {
3629
+ "epoch": 0.77,
3630
+ "learning_rate": 2.578742420810294e-06,
3631
+ "loss": 1.9009,
3632
+ "step": 604
3633
+ },
3634
+ {
3635
+ "epoch": 0.77,
3636
+ "learning_rate": 2.5509902133328113e-06,
3637
+ "loss": 1.9141,
3638
+ "step": 605
3639
+ },
3640
+ {
3641
+ "epoch": 0.78,
3642
+ "learning_rate": 2.5233662997259336e-06,
3643
+ "loss": 1.9541,
3644
+ "step": 606
3645
+ },
3646
+ {
3647
+ "epoch": 0.78,
3648
+ "learning_rate": 2.4958711557547523e-06,
3649
+ "loss": 1.9219,
3650
+ "step": 607
3651
+ },
3652
+ {
3653
+ "epoch": 0.78,
3654
+ "learning_rate": 2.4685052549665556e-06,
3655
+ "loss": 1.9409,
3656
+ "step": 608
3657
+ },
3658
+ {
3659
+ "epoch": 0.78,
3660
+ "learning_rate": 2.4412690686827e-06,
3661
+ "loss": 1.9028,
3662
+ "step": 609
3663
+ },
3664
+ {
3665
+ "epoch": 0.78,
3666
+ "learning_rate": 2.414163065990468e-06,
3667
+ "loss": 1.8433,
3668
+ "step": 610
3669
+ },
3670
+ {
3671
+ "epoch": 0.78,
3672
+ "learning_rate": 2.3871877137349887e-06,
3673
+ "loss": 1.8901,
3674
+ "step": 611
3675
+ },
3676
+ {
3677
+ "epoch": 0.78,
3678
+ "learning_rate": 2.3603434765112266e-06,
3679
+ "loss": 1.915,
3680
+ "step": 612
3681
+ },
3682
+ {
3683
+ "epoch": 0.78,
3684
+ "learning_rate": 2.3336308166559385e-06,
3685
+ "loss": 1.8887,
3686
+ "step": 613
3687
+ },
3688
+ {
3689
+ "epoch": 0.79,
3690
+ "learning_rate": 2.307050194239746e-06,
3691
+ "loss": 1.9546,
3692
+ "step": 614
3693
+ },
3694
+ {
3695
+ "epoch": 0.79,
3696
+ "learning_rate": 2.280602067059192e-06,
3697
+ "loss": 2.0059,
3698
+ "step": 615
3699
+ },
3700
+ {
3701
+ "epoch": 0.79,
3702
+ "learning_rate": 2.2542868906288617e-06,
3703
+ "loss": 1.9775,
3704
+ "step": 616
3705
+ },
3706
+ {
3707
+ "epoch": 0.79,
3708
+ "learning_rate": 2.22810511817354e-06,
3709
+ "loss": 1.9321,
3710
+ "step": 617
3711
+ },
3712
+ {
3713
+ "epoch": 0.79,
3714
+ "learning_rate": 2.2020572006203934e-06,
3715
+ "loss": 1.8691,
3716
+ "step": 618
3717
+ },
3718
+ {
3719
+ "epoch": 0.79,
3720
+ "learning_rate": 2.1761435865912296e-06,
3721
+ "loss": 1.9443,
3722
+ "step": 619
3723
+ },
3724
+ {
3725
+ "epoch": 0.79,
3726
+ "learning_rate": 2.1503647223947355e-06,
3727
+ "loss": 1.8301,
3728
+ "step": 620
3729
+ },
3730
+ {
3731
+ "epoch": 0.79,
3732
+ "learning_rate": 2.124721052018821e-06,
3733
+ "loss": 1.939,
3734
+ "step": 621
3735
+ },
3736
+ {
3737
+ "epoch": 0.8,
3738
+ "learning_rate": 2.0992130171229653e-06,
3739
+ "loss": 1.8184,
3740
+ "step": 622
3741
+ },
3742
+ {
3743
+ "epoch": 0.8,
3744
+ "learning_rate": 2.07384105703059e-06,
3745
+ "loss": 1.8618,
3746
+ "step": 623
3747
+ },
3748
+ {
3749
+ "epoch": 0.8,
3750
+ "learning_rate": 2.0486056087215177e-06,
3751
+ "loss": 1.9854,
3752
+ "step": 624
3753
+ },
3754
+ {
3755
+ "epoch": 0.8,
3756
+ "learning_rate": 2.0235071068244383e-06,
3757
+ "loss": 1.8643,
3758
+ "step": 625
3759
+ },
3760
+ {
3761
+ "epoch": 0.8,
3762
+ "learning_rate": 1.998545983609417e-06,
3763
+ "loss": 1.9487,
3764
+ "step": 626
3765
+ },
3766
+ {
3767
+ "epoch": 0.8,
3768
+ "learning_rate": 1.9737226689804524e-06,
3769
+ "loss": 1.9106,
3770
+ "step": 627
3771
+ },
3772
+ {
3773
+ "epoch": 0.8,
3774
+ "learning_rate": 1.9490375904680747e-06,
3775
+ "loss": 1.8037,
3776
+ "step": 628
3777
+ },
3778
+ {
3779
+ "epoch": 0.8,
3780
+ "learning_rate": 1.9244911732219917e-06,
3781
+ "loss": 1.8472,
3782
+ "step": 629
3783
+ },
3784
+ {
3785
+ "epoch": 0.81,
3786
+ "learning_rate": 1.900083840003739e-06,
3787
+ "loss": 1.8271,
3788
+ "step": 630
3789
+ },
3790
+ {
3791
+ "epoch": 0.81,
3792
+ "learning_rate": 1.8758160111794266e-06,
3793
+ "loss": 1.96,
3794
+ "step": 631
3795
+ },
3796
+ {
3797
+ "epoch": 0.81,
3798
+ "learning_rate": 1.8516881047124869e-06,
3799
+ "loss": 1.9526,
3800
+ "step": 632
3801
+ },
3802
+ {
3803
+ "epoch": 0.81,
3804
+ "learning_rate": 1.827700536156476e-06,
3805
+ "loss": 1.8633,
3806
+ "step": 633
3807
+ },
3808
+ {
3809
+ "epoch": 0.81,
3810
+ "learning_rate": 1.8038537186479178e-06,
3811
+ "loss": 1.8286,
3812
+ "step": 634
3813
+ },
3814
+ {
3815
+ "epoch": 0.81,
3816
+ "learning_rate": 1.7801480628991907e-06,
3817
+ "loss": 1.9351,
3818
+ "step": 635
3819
+ },
3820
+ {
3821
+ "epoch": 0.81,
3822
+ "learning_rate": 1.7565839771914551e-06,
3823
+ "loss": 1.8481,
3824
+ "step": 636
3825
+ },
3826
+ {
3827
+ "epoch": 0.82,
3828
+ "learning_rate": 1.7331618673676064e-06,
3829
+ "loss": 1.7974,
3830
+ "step": 637
3831
+ },
3832
+ {
3833
+ "epoch": 0.82,
3834
+ "learning_rate": 1.7098821368253094e-06,
3835
+ "loss": 1.9756,
3836
+ "step": 638
3837
+ },
3838
+ {
3839
+ "epoch": 0.82,
3840
+ "learning_rate": 1.6867451865100414e-06,
3841
+ "loss": 1.9395,
3842
+ "step": 639
3843
+ },
3844
+ {
3845
+ "epoch": 0.82,
3846
+ "learning_rate": 1.663751414908169e-06,
3847
+ "loss": 1.8945,
3848
+ "step": 640
3849
+ },
3850
+ {
3851
+ "epoch": 0.82,
3852
+ "learning_rate": 1.640901218040113e-06,
3853
+ "loss": 1.8223,
3854
+ "step": 641
3855
+ },
3856
+ {
3857
+ "epoch": 0.82,
3858
+ "learning_rate": 1.618194989453511e-06,
3859
+ "loss": 1.8765,
3860
+ "step": 642
3861
+ },
3862
+ {
3863
+ "epoch": 0.82,
3864
+ "learning_rate": 1.5956331202164422e-06,
3865
+ "loss": 1.9194,
3866
+ "step": 643
3867
+ },
3868
+ {
3869
+ "epoch": 0.82,
3870
+ "learning_rate": 1.5732159989106888e-06,
3871
+ "loss": 1.9219,
3872
+ "step": 644
3873
+ },
3874
+ {
3875
+ "epoch": 0.83,
3876
+ "learning_rate": 1.5509440116250586e-06,
3877
+ "loss": 1.9521,
3878
+ "step": 645
3879
+ },
3880
+ {
3881
+ "epoch": 0.83,
3882
+ "learning_rate": 1.5288175419487172e-06,
3883
+ "loss": 1.8164,
3884
+ "step": 646
3885
+ },
3886
+ {
3887
+ "epoch": 0.83,
3888
+ "learning_rate": 1.5068369709645859e-06,
3889
+ "loss": 1.9409,
3890
+ "step": 647
3891
+ },
3892
+ {
3893
+ "epoch": 0.83,
3894
+ "learning_rate": 1.4850026772427862e-06,
3895
+ "loss": 1.9082,
3896
+ "step": 648
3897
+ },
3898
+ {
3899
+ "epoch": 0.83,
3900
+ "learning_rate": 1.4633150368341154e-06,
3901
+ "loss": 1.9541,
3902
+ "step": 649
3903
+ },
3904
+ {
3905
+ "epoch": 0.83,
3906
+ "learning_rate": 1.4417744232635644e-06,
3907
+ "loss": 1.8682,
3908
+ "step": 650
3909
+ },
3910
+ {
3911
+ "epoch": 0.83,
3912
+ "learning_rate": 1.4203812075238932e-06,
3913
+ "loss": 1.9004,
3914
+ "step": 651
3915
+ },
3916
+ {
3917
+ "epoch": 0.83,
3918
+ "learning_rate": 1.3991357580692356e-06,
3919
+ "loss": 1.8926,
3920
+ "step": 652
3921
+ },
3922
+ {
3923
+ "epoch": 0.84,
3924
+ "learning_rate": 1.3780384408087577e-06,
3925
+ "loss": 1.9297,
3926
+ "step": 653
3927
+ },
3928
+ {
3929
+ "epoch": 0.84,
3930
+ "learning_rate": 1.3570896191003457e-06,
3931
+ "loss": 1.9326,
3932
+ "step": 654
3933
+ },
3934
+ {
3935
+ "epoch": 0.84,
3936
+ "learning_rate": 1.3362896537443682e-06,
3937
+ "loss": 1.8862,
3938
+ "step": 655
3939
+ },
3940
+ {
3941
+ "epoch": 0.84,
3942
+ "learning_rate": 1.3156389029774352e-06,
3943
+ "loss": 1.8867,
3944
+ "step": 656
3945
+ },
3946
+ {
3947
+ "epoch": 0.84,
3948
+ "learning_rate": 1.295137722466253e-06,
3949
+ "loss": 1.8989,
3950
+ "step": 657
3951
+ },
3952
+ {
3953
+ "epoch": 0.84,
3954
+ "learning_rate": 1.274786465301483e-06,
3955
+ "loss": 1.9619,
3956
+ "step": 658
3957
+ },
3958
+ {
3959
+ "epoch": 0.84,
3960
+ "learning_rate": 1.2545854819916646e-06,
3961
+ "loss": 1.9019,
3962
+ "step": 659
3963
+ },
3964
+ {
3965
+ "epoch": 0.84,
3966
+ "learning_rate": 1.2345351204571866e-06,
3967
+ "loss": 1.9165,
3968
+ "step": 660
3969
+ },
3970
+ {
3971
+ "epoch": 0.85,
3972
+ "learning_rate": 1.2146357260242726e-06,
3973
+ "loss": 1.9258,
3974
+ "step": 661
3975
+ },
3976
+ {
3977
+ "epoch": 0.85,
3978
+ "learning_rate": 1.1948876414190657e-06,
3979
+ "loss": 1.8911,
3980
+ "step": 662
3981
+ },
3982
+ {
3983
+ "epoch": 0.85,
3984
+ "learning_rate": 1.1752912067616995e-06,
3985
+ "loss": 1.9722,
3986
+ "step": 663
3987
+ },
3988
+ {
3989
+ "epoch": 0.85,
3990
+ "learning_rate": 1.155846759560444e-06,
3991
+ "loss": 1.9258,
3992
+ "step": 664
3993
+ },
3994
+ {
3995
+ "epoch": 0.85,
3996
+ "learning_rate": 1.136554634705913e-06,
3997
+ "loss": 1.9492,
3998
+ "step": 665
3999
+ },
4000
+ {
4001
+ "epoch": 0.85,
4002
+ "learning_rate": 1.1174151644652653e-06,
4003
+ "loss": 1.8486,
4004
+ "step": 666
4005
+ },
4006
+ {
4007
+ "epoch": 0.85,
4008
+ "learning_rate": 1.0984286784765075e-06,
4009
+ "loss": 1.8589,
4010
+ "step": 667
4011
+ },
4012
+ {
4013
+ "epoch": 0.85,
4014
+ "learning_rate": 1.0795955037428052e-06,
4015
+ "loss": 1.9229,
4016
+ "step": 668
4017
+ },
4018
+ {
4019
+ "epoch": 0.86,
4020
+ "learning_rate": 1.0609159646268507e-06,
4021
+ "loss": 1.8252,
4022
+ "step": 669
4023
+ },
4024
+ {
4025
+ "epoch": 0.86,
4026
+ "learning_rate": 1.042390382845284e-06,
4027
+ "loss": 1.8135,
4028
+ "step": 670
4029
+ },
4030
+ {
4031
+ "epoch": 0.86,
4032
+ "learning_rate": 1.0240190774631376e-06,
4033
+ "loss": 1.9185,
4034
+ "step": 671
4035
+ },
4036
+ {
4037
+ "epoch": 0.86,
4038
+ "learning_rate": 1.0058023648883642e-06,
4039
+ "loss": 1.7939,
4040
+ "step": 672
4041
+ },
4042
+ {
4043
+ "epoch": 0.86,
4044
+ "learning_rate": 9.87740558866357e-07,
4045
+ "loss": 1.8838,
4046
+ "step": 673
4047
+ },
4048
+ {
4049
+ "epoch": 0.86,
4050
+ "learning_rate": 9.698339704745729e-07,
4051
+ "loss": 2.0044,
4052
+ "step": 674
4053
+ },
4054
+ {
4055
+ "epoch": 0.86,
4056
+ "learning_rate": 9.520829081171678e-07,
4057
+ "loss": 1.9146,
4058
+ "step": 675
4059
+ },
4060
+ {
4061
+ "epoch": 0.87,
4062
+ "learning_rate": 9.344876775196688e-07,
4063
+ "loss": 1.9492,
4064
+ "step": 676
4065
+ },
4066
+ {
4067
+ "epoch": 0.87,
4068
+ "learning_rate": 9.170485817237329e-07,
4069
+ "loss": 1.9033,
4070
+ "step": 677
4071
+ },
4072
+ {
4073
+ "epoch": 0.87,
4074
+ "learning_rate": 8.997659210819099e-07,
4075
+ "loss": 1.8765,
4076
+ "step": 678
4077
+ },
4078
+ {
4079
+ "epoch": 0.87,
4080
+ "learning_rate": 8.82639993252482e-07,
4081
+ "loss": 1.7529,
4082
+ "step": 679
4083
+ },
4084
+ {
4085
+ "epoch": 0.87,
4086
+ "learning_rate": 8.656710931943213e-07,
4087
+ "loss": 1.8853,
4088
+ "step": 680
4089
+ },
4090
+ {
4091
+ "epoch": 0.87,
4092
+ "learning_rate": 8.488595131618272e-07,
4093
+ "loss": 1.8647,
4094
+ "step": 681
4095
+ },
4096
+ {
4097
+ "epoch": 0.87,
4098
+ "learning_rate": 8.322055426998888e-07,
4099
+ "loss": 1.7764,
4100
+ "step": 682
4101
+ },
4102
+ {
4103
+ "epoch": 0.87,
4104
+ "learning_rate": 8.157094686388833e-07,
4105
+ "loss": 1.8828,
4106
+ "step": 683
4107
+ },
4108
+ {
4109
+ "epoch": 0.88,
4110
+ "learning_rate": 7.993715750897557e-07,
4111
+ "loss": 1.8081,
4112
+ "step": 684
4113
+ },
4114
+ {
4115
+ "epoch": 0.88,
4116
+ "learning_rate": 7.831921434391132e-07,
4117
+ "loss": 1.8779,
4118
+ "step": 685
4119
+ },
4120
+ {
4121
+ "epoch": 0.88,
4122
+ "learning_rate": 7.671714523443852e-07,
4123
+ "loss": 1.9888,
4124
+ "step": 686
4125
+ },
4126
+ {
4127
+ "epoch": 0.88,
4128
+ "learning_rate": 7.513097777290201e-07,
4129
+ "loss": 2.0195,
4130
+ "step": 687
4131
+ },
4132
+ {
4133
+ "epoch": 0.88,
4134
+ "learning_rate": 7.356073927777352e-07,
4135
+ "loss": 1.856,
4136
+ "step": 688
4137
+ },
4138
+ {
4139
+ "epoch": 0.88,
4140
+ "learning_rate": 7.20064567931813e-07,
4141
+ "loss": 1.8218,
4142
+ "step": 689
4143
+ },
4144
+ {
4145
+ "epoch": 0.88,
4146
+ "learning_rate": 7.046815708844346e-07,
4147
+ "loss": 1.8584,
4148
+ "step": 690
4149
+ },
4150
+ {
4151
+ "epoch": 0.88,
4152
+ "learning_rate": 6.894586665760827e-07,
4153
+ "loss": 1.98,
4154
+ "step": 691
4155
+ },
4156
+ {
4157
+ "epoch": 0.89,
4158
+ "learning_rate": 6.743961171899693e-07,
4159
+ "loss": 1.8188,
4160
+ "step": 692
4161
+ },
4162
+ {
4163
+ "epoch": 0.89,
4164
+ "learning_rate": 6.594941821475243e-07,
4165
+ "loss": 1.9395,
4166
+ "step": 693
4167
+ },
4168
+ {
4169
+ "epoch": 0.89,
4170
+ "learning_rate": 6.447531181039246e-07,
4171
+ "loss": 1.874,
4172
+ "step": 694
4173
+ },
4174
+ {
4175
+ "epoch": 0.89,
4176
+ "learning_rate": 6.301731789436738e-07,
4177
+ "loss": 1.937,
4178
+ "step": 695
4179
+ },
4180
+ {
4181
+ "epoch": 0.89,
4182
+ "learning_rate": 6.157546157762361e-07,
4183
+ "loss": 1.832,
4184
+ "step": 696
4185
+ },
4186
+ {
4187
+ "epoch": 0.89,
4188
+ "learning_rate": 6.01497676931696e-07,
4189
+ "loss": 1.9634,
4190
+ "step": 697
4191
+ },
4192
+ {
4193
+ "epoch": 0.89,
4194
+ "learning_rate": 5.87402607956502e-07,
4195
+ "loss": 1.8179,
4196
+ "step": 698
4197
+ },
4198
+ {
4199
+ "epoch": 0.89,
4200
+ "learning_rate": 5.734696516092253e-07,
4201
+ "loss": 1.9551,
4202
+ "step": 699
4203
+ },
4204
+ {
4205
+ "epoch": 0.9,
4206
+ "learning_rate": 5.596990478563735e-07,
4207
+ "loss": 1.8711,
4208
+ "step": 700
4209
+ },
4210
+ {
4211
+ "epoch": 0.9,
4212
+ "learning_rate": 5.460910338682723e-07,
4213
+ "loss": 1.9565,
4214
+ "step": 701
4215
+ },
4216
+ {
4217
+ "epoch": 0.9,
4218
+ "learning_rate": 5.326458440149684e-07,
4219
+ "loss": 1.9307,
4220
+ "step": 702
4221
+ },
4222
+ {
4223
+ "epoch": 0.9,
4224
+ "learning_rate": 5.193637098622007e-07,
4225
+ "loss": 1.9829,
4226
+ "step": 703
4227
+ },
4228
+ {
4229
+ "epoch": 0.9,
4230
+ "learning_rate": 5.062448601674086e-07,
4231
+ "loss": 1.8389,
4232
+ "step": 704
4233
+ },
4234
+ {
4235
+ "epoch": 0.9,
4236
+ "learning_rate": 4.932895208757905e-07,
4237
+ "loss": 1.7654,
4238
+ "step": 705
4239
+ },
4240
+ {
4241
+ "epoch": 0.9,
4242
+ "learning_rate": 4.804979151164191e-07,
4243
+ "loss": 1.8574,
4244
+ "step": 706
4245
+ },
4246
+ {
4247
+ "epoch": 0.9,
4248
+ "learning_rate": 4.6787026319838445e-07,
4249
+ "loss": 1.9209,
4250
+ "step": 707
4251
+ },
4252
+ {
4253
+ "epoch": 0.91,
4254
+ "learning_rate": 4.55406782607023e-07,
4255
+ "loss": 1.7476,
4256
+ "step": 708
4257
+ },
4258
+ {
4259
+ "epoch": 0.91,
4260
+ "learning_rate": 4.431076880001439e-07,
4261
+ "loss": 1.9224,
4262
+ "step": 709
4263
+ },
4264
+ {
4265
+ "epoch": 0.91,
4266
+ "learning_rate": 4.3097319120435356e-07,
4267
+ "loss": 1.9302,
4268
+ "step": 710
4269
+ },
4270
+ {
4271
+ "epoch": 0.91,
4272
+ "learning_rate": 4.190035012113991e-07,
4273
+ "loss": 1.9175,
4274
+ "step": 711
4275
+ },
4276
+ {
4277
+ "epoch": 0.91,
4278
+ "learning_rate": 4.0719882417456726e-07,
4279
+ "loss": 1.9766,
4280
+ "step": 712
4281
+ },
4282
+ {
4283
+ "epoch": 0.91,
4284
+ "learning_rate": 3.9555936340514025e-07,
4285
+ "loss": 1.8901,
4286
+ "step": 713
4287
+ },
4288
+ {
4289
+ "epoch": 0.91,
4290
+ "learning_rate": 3.8408531936888095e-07,
4291
+ "loss": 1.9224,
4292
+ "step": 714
4293
+ },
4294
+ {
4295
+ "epoch": 0.91,
4296
+ "learning_rate": 3.7277688968260117e-07,
4297
+ "loss": 1.9058,
4298
+ "step": 715
4299
+ },
4300
+ {
4301
+ "epoch": 0.92,
4302
+ "learning_rate": 3.616342691107366e-07,
4303
+ "loss": 1.9355,
4304
+ "step": 716
4305
+ },
4306
+ {
4307
+ "epoch": 0.92,
4308
+ "learning_rate": 3.506576495620062e-07,
4309
+ "loss": 1.8789,
4310
+ "step": 717
4311
+ },
4312
+ {
4313
+ "epoch": 0.92,
4314
+ "learning_rate": 3.398472200861047e-07,
4315
+ "loss": 1.8716,
4316
+ "step": 718
4317
+ },
4318
+ {
4319
+ "epoch": 0.92,
4320
+ "learning_rate": 3.2920316687043986e-07,
4321
+ "loss": 1.9707,
4322
+ "step": 719
4323
+ },
4324
+ {
4325
+ "epoch": 0.92,
4326
+ "learning_rate": 3.187256732369337e-07,
4327
+ "loss": 1.8418,
4328
+ "step": 720
4329
+ },
4330
+ {
4331
+ "epoch": 0.92,
4332
+ "learning_rate": 3.0841491963886083e-07,
4333
+ "loss": 1.7754,
4334
+ "step": 721
4335
+ },
4336
+ {
4337
+ "epoch": 0.92,
4338
+ "learning_rate": 2.9827108365774294e-07,
4339
+ "loss": 1.8809,
4340
+ "step": 722
4341
+ },
4342
+ {
4343
+ "epoch": 0.93,
4344
+ "learning_rate": 2.882943400002891e-07,
4345
+ "loss": 1.8335,
4346
+ "step": 723
4347
+ },
4348
+ {
4349
+ "epoch": 0.93,
4350
+ "learning_rate": 2.784848604953827e-07,
4351
+ "loss": 1.8765,
4352
+ "step": 724
4353
+ },
4354
+ {
4355
+ "epoch": 0.93,
4356
+ "learning_rate": 2.688428140911359e-07,
4357
+ "loss": 1.98,
4358
+ "step": 725
4359
+ },
4360
+ {
4361
+ "epoch": 0.93,
4362
+ "learning_rate": 2.593683668519575e-07,
4363
+ "loss": 1.8096,
4364
+ "step": 726
4365
+ },
4366
+ {
4367
+ "epoch": 0.93,
4368
+ "learning_rate": 2.5006168195571555e-07,
4369
+ "loss": 1.8789,
4370
+ "step": 727
4371
+ },
4372
+ {
4373
+ "epoch": 0.93,
4374
+ "learning_rate": 2.409229196909113e-07,
4375
+ "loss": 1.9224,
4376
+ "step": 728
4377
+ },
4378
+ {
4379
+ "epoch": 0.93,
4380
+ "learning_rate": 2.3195223745392736e-07,
4381
+ "loss": 1.9131,
4382
+ "step": 729
4383
+ },
4384
+ {
4385
+ "epoch": 0.93,
4386
+ "learning_rate": 2.2314978974631084e-07,
4387
+ "loss": 1.9897,
4388
+ "step": 730
4389
+ },
4390
+ {
4391
+ "epoch": 0.94,
4392
+ "learning_rate": 2.1451572817211774e-07,
4393
+ "loss": 1.8564,
4394
+ "step": 731
4395
+ },
4396
+ {
4397
+ "epoch": 0.94,
4398
+ "learning_rate": 2.060502014352972e-07,
4399
+ "loss": 1.8457,
4400
+ "step": 732
4401
+ },
4402
+ {
4403
+ "epoch": 0.94,
4404
+ "learning_rate": 1.9775335533713357e-07,
4405
+ "loss": 1.8062,
4406
+ "step": 733
4407
+ },
4408
+ {
4409
+ "epoch": 0.94,
4410
+ "learning_rate": 1.8962533277373185e-07,
4411
+ "loss": 1.9756,
4412
+ "step": 734
4413
+ },
4414
+ {
4415
+ "epoch": 0.94,
4416
+ "learning_rate": 1.8166627373356504e-07,
4417
+ "loss": 1.9424,
4418
+ "step": 735
4419
+ },
4420
+ {
4421
+ "epoch": 0.94,
4422
+ "learning_rate": 1.7387631529504955e-07,
4423
+ "loss": 1.8335,
4424
+ "step": 736
4425
+ },
4426
+ {
4427
+ "epoch": 0.94,
4428
+ "learning_rate": 1.6625559162419703e-07,
4429
+ "loss": 1.918,
4430
+ "step": 737
4431
+ },
4432
+ {
4433
+ "epoch": 0.94,
4434
+ "learning_rate": 1.5880423397229506e-07,
4435
+ "loss": 1.8936,
4436
+ "step": 738
4437
+ },
4438
+ {
4439
+ "epoch": 0.95,
4440
+ "learning_rate": 1.515223706736524e-07,
4441
+ "loss": 1.8784,
4442
+ "step": 739
4443
+ },
4444
+ {
4445
+ "epoch": 0.95,
4446
+ "learning_rate": 1.444101271433851e-07,
4447
+ "loss": 1.957,
4448
+ "step": 740
4449
+ },
4450
+ {
4451
+ "epoch": 0.95,
4452
+ "learning_rate": 1.3746762587525718e-07,
4453
+ "loss": 1.9399,
4454
+ "step": 741
4455
+ },
4456
+ {
4457
+ "epoch": 0.95,
4458
+ "learning_rate": 1.3069498643957235e-07,
4459
+ "loss": 1.8203,
4460
+ "step": 742
4461
+ },
4462
+ {
4463
+ "epoch": 0.95,
4464
+ "learning_rate": 1.2409232548111327e-07,
4465
+ "loss": 1.8911,
4466
+ "step": 743
4467
+ },
4468
+ {
4469
+ "epoch": 0.95,
4470
+ "learning_rate": 1.1765975671713336e-07,
4471
+ "loss": 1.8179,
4472
+ "step": 744
4473
+ },
4474
+ {
4475
+ "epoch": 0.95,
4476
+ "learning_rate": 1.1139739093539714e-07,
4477
+ "loss": 1.9487,
4478
+ "step": 745
4479
+ },
4480
+ {
4481
+ "epoch": 0.95,
4482
+ "learning_rate": 1.0530533599227177e-07,
4483
+ "loss": 1.9131,
4484
+ "step": 746
4485
+ },
4486
+ {
4487
+ "epoch": 0.96,
4488
+ "learning_rate": 9.938369681087412e-08,
4489
+ "loss": 1.9028,
4490
+ "step": 747
4491
+ },
4492
+ {
4493
+ "epoch": 0.96,
4494
+ "learning_rate": 9.363257537925774e-08,
4495
+ "loss": 1.9678,
4496
+ "step": 748
4497
+ },
4498
+ {
4499
+ "epoch": 0.96,
4500
+ "learning_rate": 8.805207074865874e-08,
4501
+ "loss": 1.854,
4502
+ "step": 749
4503
+ },
4504
+ {
4505
+ "epoch": 0.96,
4506
+ "learning_rate": 8.264227903178823e-08,
4507
+ "loss": 1.9575,
4508
+ "step": 750
4509
+ },
4510
+ {
4511
+ "epoch": 0.96,
4512
+ "learning_rate": 7.740329340118258e-08,
4513
+ "loss": 1.8394,
4514
+ "step": 751
4515
+ },
4516
+ {
4517
+ "epoch": 0.96,
4518
+ "learning_rate": 7.233520408759242e-08,
4519
+ "loss": 1.8657,
4520
+ "step": 752
4521
+ },
4522
+ {
4523
+ "epoch": 0.96,
4524
+ "learning_rate": 6.74380983784284e-08,
4525
+ "loss": 1.8408,
4526
+ "step": 753
4527
+ },
4528
+ {
4529
+ "epoch": 0.96,
4530
+ "learning_rate": 6.271206061626567e-08,
4531
+ "loss": 1.8823,
4532
+ "step": 754
4533
+ },
4534
+ {
4535
+ "epoch": 0.97,
4536
+ "learning_rate": 5.81571721973817e-08,
4537
+ "loss": 1.8818,
4538
+ "step": 755
4539
+ },
4540
+ {
4541
+ "epoch": 0.97,
4542
+ "learning_rate": 5.37735115703597e-08,
4543
+ "loss": 1.8403,
4544
+ "step": 756
4545
+ },
4546
+ {
4547
+ "epoch": 0.97,
4548
+ "learning_rate": 4.95611542347374e-08,
4549
+ "loss": 1.7656,
4550
+ "step": 757
4551
+ },
4552
+ {
4553
+ "epoch": 0.97,
4554
+ "learning_rate": 4.5520172739703664e-08,
4555
+ "loss": 1.8252,
4556
+ "step": 758
4557
+ },
4558
+ {
4559
+ "epoch": 0.97,
4560
+ "learning_rate": 4.1650636682853964e-08,
4561
+ "loss": 1.8237,
4562
+ "step": 759
4563
+ },
4564
+ {
4565
+ "epoch": 0.97,
4566
+ "learning_rate": 3.795261270898465e-08,
4567
+ "loss": 1.9946,
4568
+ "step": 760
4569
+ },
4570
+ {
4571
+ "epoch": 0.97,
4572
+ "learning_rate": 3.442616450895386e-08,
4573
+ "loss": 2.0083,
4574
+ "step": 761
4575
+ },
4576
+ {
4577
+ "epoch": 0.98,
4578
+ "learning_rate": 3.107135281857687e-08,
4579
+ "loss": 1.9736,
4580
+ "step": 762
4581
+ },
4582
+ {
4583
+ "epoch": 0.98,
4584
+ "learning_rate": 2.7888235417584676e-08,
4585
+ "loss": 1.8921,
4586
+ "step": 763
4587
+ },
4588
+ {
4589
+ "epoch": 0.98,
4590
+ "learning_rate": 2.4876867128625916e-08,
4591
+ "loss": 1.9229,
4592
+ "step": 764
4593
+ },
4594
+ {
4595
+ "epoch": 0.98,
4596
+ "learning_rate": 2.2037299816324297e-08,
4597
+ "loss": 1.8872,
4598
+ "step": 765
4599
+ },
4600
+ {
4601
+ "epoch": 0.98,
4602
+ "learning_rate": 1.9369582386387085e-08,
4603
+ "loss": 1.8955,
4604
+ "step": 766
4605
+ },
4606
+ {
4607
+ "epoch": 0.98,
4608
+ "learning_rate": 1.6873760784758e-08,
4609
+ "loss": 1.9434,
4610
+ "step": 767
4611
+ },
4612
+ {
4613
+ "epoch": 0.98,
4614
+ "learning_rate": 1.4549877996831162e-08,
4615
+ "loss": 1.8501,
4616
+ "step": 768
4617
+ },
4618
+ {
4619
+ "epoch": 0.98,
4620
+ "learning_rate": 1.2397974046707284e-08,
4621
+ "loss": 1.8843,
4622
+ "step": 769
4623
+ },
4624
+ {
4625
+ "epoch": 0.99,
4626
+ "learning_rate": 1.041808599650529e-08,
4627
+ "loss": 1.769,
4628
+ "step": 770
4629
+ },
4630
+ {
4631
+ "epoch": 0.99,
4632
+ "learning_rate": 8.610247945722849e-09,
4633
+ "loss": 1.8657,
4634
+ "step": 771
4635
+ },
4636
+ {
4637
+ "epoch": 0.99,
4638
+ "learning_rate": 6.974491030653507e-09,
4639
+ "loss": 1.8413,
4640
+ "step": 772
4641
+ },
4642
+ {
4643
+ "epoch": 0.99,
4644
+ "learning_rate": 5.510843423842671e-09,
4645
+ "loss": 1.8325,
4646
+ "step": 773
4647
+ },
4648
+ {
4649
+ "epoch": 0.99,
4650
+ "learning_rate": 4.219330333612437e-09,
4651
+ "loss": 1.9419,
4652
+ "step": 774
4653
+ },
4654
+ {
4655
+ "epoch": 0.99,
4656
+ "learning_rate": 3.099974003618611e-09,
4657
+ "loss": 1.8892,
4658
+ "step": 775
4659
+ },
4660
+ {
4661
+ "epoch": 0.99,
4662
+ "learning_rate": 2.1527937124687927e-09,
4663
+ "loss": 2.0103,
4664
+ "step": 776
4665
+ },
4666
+ {
4667
+ "epoch": 0.99,
4668
+ "learning_rate": 1.3778057733981886e-09,
4669
+ "loss": 1.9126,
4670
+ "step": 777
4671
+ },
4672
+ {
4673
+ "epoch": 1.0,
4674
+ "learning_rate": 7.750235339787359e-10,
4675
+ "loss": 1.917,
4676
+ "step": 778
4677
+ },
4678
+ {
4679
+ "epoch": 1.0,
4680
+ "learning_rate": 3.444573758937253e-10,
4681
+ "loss": 1.9062,
4682
+ "step": 779
4683
+ },
4684
+ {
4685
+ "epoch": 1.0,
4686
+ "learning_rate": 8.611471476016598e-11,
4687
+ "loss": 1.8867,
4688
+ "step": 780
4689
+ },
4690
+ {
4691
+ "epoch": 1.0,
4692
+ "learning_rate": 0.0,
4693
+ "loss": 1.9629,
4694
+ "step": 781
4695
+ },
4696
+ {
4697
+ "epoch": 1.0,
4698
+ "step": 781,
4699
+ "total_flos": 1.5635750350529495e+18,
4700
+ "train_loss": 2.002131619618278,
4701
+ "train_runtime": 19394.4142,
4702
+ "train_samples_per_second": 10.312,
4703
+ "train_steps_per_second": 0.04
4704
+ }
4705
+ ],
4706
+ "max_steps": 781,
4707
+ "num_train_epochs": 1,
4708
+ "total_flos": 1.5635750350529495e+18,
4709
+ "trial_name": null,
4710
+ "trial_params": null
4711
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f36498b2c51d184b391c60e8e601953aff2cc603229e40086245cffb0fdb490d
3
+ size 5499