Hugging Face's logo

qanastek
/
MedAlpaca-LLaMa2-7B

PEFT
English
medical
llama2
Model card Files Community
qanastek commited on
Commit
39ea18c
•
1 Parent(s): df1b5b3

Upload 12 files

Browse files
Files changed (12) hide show
  1. README.md +32 -0
  2. adapter_config.json +21 -0
  3. adapter_model.bin +3 -0
  4. optimizer.pt +3 -0
  5. rng_state.pth +3 -0
  6. scheduler.pt +3 -0
  7. special_tokens_map.json +24 -0
  8. tokenizer.json +0 -0
  9. tokenizer.model +3 -0
  10. tokenizer_config.json +32 -0
  11. trainer_state.json +5296 -0
  12. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - load_in_8bit: False
9
+ - load_in_4bit: True
10
+ - llm_int8_threshold: 6.0
11
+ - llm_int8_skip_modules: None
12
+ - llm_int8_enable_fp32_cpu_offload: False
13
+ - llm_int8_has_fp16_weight: False
14
+ - bnb_4bit_quant_type: nf4
15
+ - bnb_4bit_use_double_quant: False
16
+ - bnb_4bit_compute_dtype: float16
17
+
18
+ The following `bitsandbytes` quantization config was used during training:
19
+ - load_in_8bit: False
20
+ - load_in_4bit: True
21
+ - llm_int8_threshold: 6.0
22
+ - llm_int8_skip_modules: None
23
+ - llm_int8_enable_fp32_cpu_offload: False
24
+ - llm_int8_has_fp16_weight: False
25
+ - bnb_4bit_quant_type: nf4
26
+ - bnb_4bit_use_double_quant: False
27
+ - bnb_4bit_compute_dtype: float16
28
+ ### Framework versions
29
+
30
+ - PEFT 0.4.0
31
+
32
+ - PEFT 0.4.0
adapter_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16,
11
+ "lora_dropout": 0.1,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj"
19
+ ],
20
+ "task_type": "CAUSAL_LM"
21
+ }
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51eecc3e1cfd2cdf9266ba09f9ede9a0f77dc2f91f55f26378e5d369e7566073
3
+ size 134263757
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63b6c90a9a87dbfd42a6379ed07e44b5b7dd5fe0f82737ff46fec669dbc1619e
3
+ size 268514565
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fde307d76cfd35e2a9a26bef1ae2d80978f82863b447357a6377e201e838f54
3
+ size 14575
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c21e5e690d8e0cade5418c2fcaefb2d060d01ccdb64d9d6411a6d2dbbe882b0
3
+ size 627
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": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
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 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "__type": "AddedToken",
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ "clean_up_tokenization_spaces": false,
11
+ "eos_token": {
12
+ "__type": "AddedToken",
13
+ "content": "</s>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false
18
+ },
19
+ "legacy": false,
20
+ "model_max_length": 1000000000000000019884624838656,
21
+ "pad_token": null,
22
+ "sp_model_kwargs": {},
23
+ "tokenizer_class": "LlamaTokenizer",
24
+ "unk_token": {
25
+ "__type": "AddedToken",
26
+ "content": "<unk>",
27
+ "lstrip": false,
28
+ "normalized": false,
29
+ "rstrip": false,
30
+ "single_word": false
31
+ }
32
+ }
trainer_state.json ADDED
@@ -0,0 +1,5296 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0057142857142858,
5
+ "global_step": 8800,
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": 0.0002,
13
+ "loss": 1.7431,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 0.0002,
19
+ "loss": 1.5523,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.0,
24
+ "learning_rate": 0.0002,
25
+ "loss": 1.5362,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.0,
30
+ "learning_rate": 0.0002,
31
+ "loss": 1.5844,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 0.0002,
37
+ "loss": 1.2511,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 0.0002,
43
+ "loss": 1.363,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 0.0002,
49
+ "loss": 1.3879,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.01,
54
+ "learning_rate": 0.0002,
55
+ "loss": 1.5116,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.01,
60
+ "learning_rate": 0.0002,
61
+ "loss": 1.4911,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.01,
66
+ "learning_rate": 0.0002,
67
+ "loss": 1.2505,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.01,
72
+ "learning_rate": 0.0002,
73
+ "loss": 1.3797,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.01,
78
+ "learning_rate": 0.0002,
79
+ "loss": 1.3631,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.01,
84
+ "learning_rate": 0.0002,
85
+ "loss": 1.4894,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "learning_rate": 0.0002,
91
+ "loss": 1.5071,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.02,
96
+ "learning_rate": 0.0002,
97
+ "loss": 1.1832,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.02,
102
+ "learning_rate": 0.0002,
103
+ "loss": 1.376,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.02,
108
+ "learning_rate": 0.0002,
109
+ "loss": 1.395,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.02,
114
+ "learning_rate": 0.0002,
115
+ "loss": 1.4788,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.02,
120
+ "learning_rate": 0.0002,
121
+ "loss": 1.5353,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.02,
126
+ "learning_rate": 0.0002,
127
+ "loss": 1.1837,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.02,
132
+ "learning_rate": 0.0002,
133
+ "loss": 1.3746,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 0.03,
138
+ "learning_rate": 0.0002,
139
+ "loss": 1.4004,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 0.03,
144
+ "learning_rate": 0.0002,
145
+ "loss": 1.4537,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 0.03,
150
+ "learning_rate": 0.0002,
151
+ "loss": 1.5133,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 0.03,
156
+ "learning_rate": 0.0002,
157
+ "loss": 1.123,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 0.03,
162
+ "learning_rate": 0.0002,
163
+ "loss": 1.333,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 0.03,
168
+ "learning_rate": 0.0002,
169
+ "loss": 1.4077,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 0.03,
174
+ "learning_rate": 0.0002,
175
+ "loss": 1.5152,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 0.03,
180
+ "learning_rate": 0.0002,
181
+ "loss": 1.5137,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 0.03,
186
+ "learning_rate": 0.0002,
187
+ "loss": 1.1933,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 0.04,
192
+ "learning_rate": 0.0002,
193
+ "loss": 1.4112,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 0.04,
198
+ "learning_rate": 0.0002,
199
+ "loss": 1.4278,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 0.04,
204
+ "learning_rate": 0.0002,
205
+ "loss": 1.4328,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 0.04,
210
+ "learning_rate": 0.0002,
211
+ "loss": 1.5322,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 0.04,
216
+ "learning_rate": 0.0002,
217
+ "loss": 1.2536,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 0.04,
222
+ "learning_rate": 0.0002,
223
+ "loss": 1.3978,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 0.04,
228
+ "learning_rate": 0.0002,
229
+ "loss": 1.422,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 0.04,
234
+ "learning_rate": 0.0002,
235
+ "loss": 1.512,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 0.04,
240
+ "learning_rate": 0.0002,
241
+ "loss": 1.4934,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 0.05,
246
+ "learning_rate": 0.0002,
247
+ "loss": 1.1298,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 0.05,
252
+ "learning_rate": 0.0002,
253
+ "loss": 1.3553,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 0.05,
258
+ "learning_rate": 0.0002,
259
+ "loss": 1.4127,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 0.05,
264
+ "learning_rate": 0.0002,
265
+ "loss": 1.45,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 0.05,
270
+ "learning_rate": 0.0002,
271
+ "loss": 1.4496,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 0.05,
276
+ "learning_rate": 0.0002,
277
+ "loss": 1.1942,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 0.05,
282
+ "learning_rate": 0.0002,
283
+ "loss": 1.3397,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 0.05,
288
+ "learning_rate": 0.0002,
289
+ "loss": 1.3896,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 0.05,
294
+ "learning_rate": 0.0002,
295
+ "loss": 1.5023,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 0.06,
300
+ "learning_rate": 0.0002,
301
+ "loss": 1.5056,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 0.06,
306
+ "learning_rate": 0.0002,
307
+ "loss": 1.1338,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 0.06,
312
+ "learning_rate": 0.0002,
313
+ "loss": 1.3907,
314
+ "step": 510
315
+ },
316
+ {
317
+ "epoch": 0.06,
318
+ "learning_rate": 0.0002,
319
+ "loss": 1.4089,
320
+ "step": 520
321
+ },
322
+ {
323
+ "epoch": 0.06,
324
+ "learning_rate": 0.0002,
325
+ "loss": 1.4939,
326
+ "step": 530
327
+ },
328
+ {
329
+ "epoch": 0.06,
330
+ "learning_rate": 0.0002,
331
+ "loss": 1.5232,
332
+ "step": 540
333
+ },
334
+ {
335
+ "epoch": 0.06,
336
+ "learning_rate": 0.0002,
337
+ "loss": 1.2392,
338
+ "step": 550
339
+ },
340
+ {
341
+ "epoch": 0.06,
342
+ "learning_rate": 0.0002,
343
+ "loss": 1.3408,
344
+ "step": 560
345
+ },
346
+ {
347
+ "epoch": 0.07,
348
+ "learning_rate": 0.0002,
349
+ "loss": 1.4217,
350
+ "step": 570
351
+ },
352
+ {
353
+ "epoch": 0.07,
354
+ "learning_rate": 0.0002,
355
+ "loss": 1.4783,
356
+ "step": 580
357
+ },
358
+ {
359
+ "epoch": 0.07,
360
+ "learning_rate": 0.0002,
361
+ "loss": 1.4852,
362
+ "step": 590
363
+ },
364
+ {
365
+ "epoch": 0.07,
366
+ "learning_rate": 0.0002,
367
+ "loss": 1.2052,
368
+ "step": 600
369
+ },
370
+ {
371
+ "epoch": 0.07,
372
+ "learning_rate": 0.0002,
373
+ "loss": 1.3532,
374
+ "step": 610
375
+ },
376
+ {
377
+ "epoch": 0.07,
378
+ "learning_rate": 0.0002,
379
+ "loss": 1.4059,
380
+ "step": 620
381
+ },
382
+ {
383
+ "epoch": 0.07,
384
+ "learning_rate": 0.0002,
385
+ "loss": 1.5072,
386
+ "step": 630
387
+ },
388
+ {
389
+ "epoch": 0.07,
390
+ "learning_rate": 0.0002,
391
+ "loss": 1.4945,
392
+ "step": 640
393
+ },
394
+ {
395
+ "epoch": 0.07,
396
+ "learning_rate": 0.0002,
397
+ "loss": 1.1679,
398
+ "step": 650
399
+ },
400
+ {
401
+ "epoch": 0.08,
402
+ "learning_rate": 0.0002,
403
+ "loss": 1.3999,
404
+ "step": 660
405
+ },
406
+ {
407
+ "epoch": 0.08,
408
+ "learning_rate": 0.0002,
409
+ "loss": 1.3628,
410
+ "step": 670
411
+ },
412
+ {
413
+ "epoch": 0.08,
414
+ "learning_rate": 0.0002,
415
+ "loss": 1.4646,
416
+ "step": 680
417
+ },
418
+ {
419
+ "epoch": 0.08,
420
+ "learning_rate": 0.0002,
421
+ "loss": 1.5183,
422
+ "step": 690
423
+ },
424
+ {
425
+ "epoch": 0.08,
426
+ "learning_rate": 0.0002,
427
+ "loss": 1.1757,
428
+ "step": 700
429
+ },
430
+ {
431
+ "epoch": 0.08,
432
+ "learning_rate": 0.0002,
433
+ "loss": 1.333,
434
+ "step": 710
435
+ },
436
+ {
437
+ "epoch": 0.08,
438
+ "learning_rate": 0.0002,
439
+ "loss": 1.3804,
440
+ "step": 720
441
+ },
442
+ {
443
+ "epoch": 0.08,
444
+ "learning_rate": 0.0002,
445
+ "loss": 1.4815,
446
+ "step": 730
447
+ },
448
+ {
449
+ "epoch": 0.08,
450
+ "learning_rate": 0.0002,
451
+ "loss": 1.4817,
452
+ "step": 740
453
+ },
454
+ {
455
+ "epoch": 0.09,
456
+ "learning_rate": 0.0002,
457
+ "loss": 1.2177,
458
+ "step": 750
459
+ },
460
+ {
461
+ "epoch": 0.09,
462
+ "learning_rate": 0.0002,
463
+ "loss": 1.3734,
464
+ "step": 760
465
+ },
466
+ {
467
+ "epoch": 0.09,
468
+ "learning_rate": 0.0002,
469
+ "loss": 1.402,
470
+ "step": 770
471
+ },
472
+ {
473
+ "epoch": 0.09,
474
+ "learning_rate": 0.0002,
475
+ "loss": 1.4573,
476
+ "step": 780
477
+ },
478
+ {
479
+ "epoch": 0.09,
480
+ "learning_rate": 0.0002,
481
+ "loss": 1.5318,
482
+ "step": 790
483
+ },
484
+ {
485
+ "epoch": 0.09,
486
+ "learning_rate": 0.0002,
487
+ "loss": 1.2027,
488
+ "step": 800
489
+ },
490
+ {
491
+ "epoch": 0.09,
492
+ "learning_rate": 0.0002,
493
+ "loss": 1.3658,
494
+ "step": 810
495
+ },
496
+ {
497
+ "epoch": 0.09,
498
+ "learning_rate": 0.0002,
499
+ "loss": 1.3659,
500
+ "step": 820
501
+ },
502
+ {
503
+ "epoch": 0.09,
504
+ "learning_rate": 0.0002,
505
+ "loss": 1.5035,
506
+ "step": 830
507
+ },
508
+ {
509
+ "epoch": 0.1,
510
+ "learning_rate": 0.0002,
511
+ "loss": 1.4851,
512
+ "step": 840
513
+ },
514
+ {
515
+ "epoch": 0.1,
516
+ "learning_rate": 0.0002,
517
+ "loss": 1.1523,
518
+ "step": 850
519
+ },
520
+ {
521
+ "epoch": 0.1,
522
+ "learning_rate": 0.0002,
523
+ "loss": 1.3534,
524
+ "step": 860
525
+ },
526
+ {
527
+ "epoch": 0.1,
528
+ "learning_rate": 0.0002,
529
+ "loss": 1.4259,
530
+ "step": 870
531
+ },
532
+ {
533
+ "epoch": 0.1,
534
+ "learning_rate": 0.0002,
535
+ "loss": 1.5088,
536
+ "step": 880
537
+ },
538
+ {
539
+ "epoch": 0.1,
540
+ "learning_rate": 0.0002,
541
+ "loss": 1.505,
542
+ "step": 890
543
+ },
544
+ {
545
+ "epoch": 0.1,
546
+ "learning_rate": 0.0002,
547
+ "loss": 1.1439,
548
+ "step": 900
549
+ },
550
+ {
551
+ "epoch": 0.1,
552
+ "learning_rate": 0.0002,
553
+ "loss": 1.3234,
554
+ "step": 910
555
+ },
556
+ {
557
+ "epoch": 0.11,
558
+ "learning_rate": 0.0002,
559
+ "loss": 1.3883,
560
+ "step": 920
561
+ },
562
+ {
563
+ "epoch": 0.11,
564
+ "learning_rate": 0.0002,
565
+ "loss": 1.4831,
566
+ "step": 930
567
+ },
568
+ {
569
+ "epoch": 0.11,
570
+ "learning_rate": 0.0002,
571
+ "loss": 1.4565,
572
+ "step": 940
573
+ },
574
+ {
575
+ "epoch": 0.11,
576
+ "learning_rate": 0.0002,
577
+ "loss": 1.1611,
578
+ "step": 950
579
+ },
580
+ {
581
+ "epoch": 0.11,
582
+ "learning_rate": 0.0002,
583
+ "loss": 1.3911,
584
+ "step": 960
585
+ },
586
+ {
587
+ "epoch": 0.11,
588
+ "learning_rate": 0.0002,
589
+ "loss": 1.3923,
590
+ "step": 970
591
+ },
592
+ {
593
+ "epoch": 0.11,
594
+ "learning_rate": 0.0002,
595
+ "loss": 1.4645,
596
+ "step": 980
597
+ },
598
+ {
599
+ "epoch": 0.11,
600
+ "learning_rate": 0.0002,
601
+ "loss": 1.4959,
602
+ "step": 990
603
+ },
604
+ {
605
+ "epoch": 0.11,
606
+ "learning_rate": 0.0002,
607
+ "loss": 1.1621,
608
+ "step": 1000
609
+ },
610
+ {
611
+ "epoch": 0.12,
612
+ "learning_rate": 0.0002,
613
+ "loss": 1.3755,
614
+ "step": 1010
615
+ },
616
+ {
617
+ "epoch": 0.12,
618
+ "learning_rate": 0.0002,
619
+ "loss": 1.36,
620
+ "step": 1020
621
+ },
622
+ {
623
+ "epoch": 0.12,
624
+ "learning_rate": 0.0002,
625
+ "loss": 1.4925,
626
+ "step": 1030
627
+ },
628
+ {
629
+ "epoch": 0.12,
630
+ "learning_rate": 0.0002,
631
+ "loss": 1.4878,
632
+ "step": 1040
633
+ },
634
+ {
635
+ "epoch": 0.12,
636
+ "learning_rate": 0.0002,
637
+ "loss": 1.1575,
638
+ "step": 1050
639
+ },
640
+ {
641
+ "epoch": 0.12,
642
+ "learning_rate": 0.0002,
643
+ "loss": 1.3751,
644
+ "step": 1060
645
+ },
646
+ {
647
+ "epoch": 0.12,
648
+ "learning_rate": 0.0002,
649
+ "loss": 1.3347,
650
+ "step": 1070
651
+ },
652
+ {
653
+ "epoch": 0.12,
654
+ "learning_rate": 0.0002,
655
+ "loss": 1.5014,
656
+ "step": 1080
657
+ },
658
+ {
659
+ "epoch": 0.12,
660
+ "learning_rate": 0.0002,
661
+ "loss": 1.5137,
662
+ "step": 1090
663
+ },
664
+ {
665
+ "epoch": 0.13,
666
+ "learning_rate": 0.0002,
667
+ "loss": 1.2194,
668
+ "step": 1100
669
+ },
670
+ {
671
+ "epoch": 0.13,
672
+ "learning_rate": 0.0002,
673
+ "loss": 1.3787,
674
+ "step": 1110
675
+ },
676
+ {
677
+ "epoch": 0.13,
678
+ "learning_rate": 0.0002,
679
+ "loss": 1.3776,
680
+ "step": 1120
681
+ },
682
+ {
683
+ "epoch": 0.13,
684
+ "learning_rate": 0.0002,
685
+ "loss": 1.5055,
686
+ "step": 1130
687
+ },
688
+ {
689
+ "epoch": 0.13,
690
+ "learning_rate": 0.0002,
691
+ "loss": 1.4581,
692
+ "step": 1140
693
+ },
694
+ {
695
+ "epoch": 0.13,
696
+ "learning_rate": 0.0002,
697
+ "loss": 1.1616,
698
+ "step": 1150
699
+ },
700
+ {
701
+ "epoch": 0.13,
702
+ "learning_rate": 0.0002,
703
+ "loss": 1.33,
704
+ "step": 1160
705
+ },
706
+ {
707
+ "epoch": 0.13,
708
+ "learning_rate": 0.0002,
709
+ "loss": 1.414,
710
+ "step": 1170
711
+ },
712
+ {
713
+ "epoch": 0.13,
714
+ "learning_rate": 0.0002,
715
+ "loss": 1.5124,
716
+ "step": 1180
717
+ },
718
+ {
719
+ "epoch": 0.14,
720
+ "learning_rate": 0.0002,
721
+ "loss": 1.5139,
722
+ "step": 1190
723
+ },
724
+ {
725
+ "epoch": 0.14,
726
+ "learning_rate": 0.0002,
727
+ "loss": 1.1085,
728
+ "step": 1200
729
+ },
730
+ {
731
+ "epoch": 0.14,
732
+ "learning_rate": 0.0002,
733
+ "loss": 1.355,
734
+ "step": 1210
735
+ },
736
+ {
737
+ "epoch": 0.14,
738
+ "learning_rate": 0.0002,
739
+ "loss": 1.3597,
740
+ "step": 1220
741
+ },
742
+ {
743
+ "epoch": 0.14,
744
+ "learning_rate": 0.0002,
745
+ "loss": 1.4835,
746
+ "step": 1230
747
+ },
748
+ {
749
+ "epoch": 0.14,
750
+ "learning_rate": 0.0002,
751
+ "loss": 1.4739,
752
+ "step": 1240
753
+ },
754
+ {
755
+ "epoch": 0.14,
756
+ "learning_rate": 0.0002,
757
+ "loss": 1.1206,
758
+ "step": 1250
759
+ },
760
+ {
761
+ "epoch": 0.14,
762
+ "learning_rate": 0.0002,
763
+ "loss": 1.3581,
764
+ "step": 1260
765
+ },
766
+ {
767
+ "epoch": 0.15,
768
+ "learning_rate": 0.0002,
769
+ "loss": 1.388,
770
+ "step": 1270
771
+ },
772
+ {
773
+ "epoch": 0.15,
774
+ "learning_rate": 0.0002,
775
+ "loss": 1.4624,
776
+ "step": 1280
777
+ },
778
+ {
779
+ "epoch": 0.15,
780
+ "learning_rate": 0.0002,
781
+ "loss": 1.4926,
782
+ "step": 1290
783
+ },
784
+ {
785
+ "epoch": 0.15,
786
+ "learning_rate": 0.0002,
787
+ "loss": 1.1745,
788
+ "step": 1300
789
+ },
790
+ {
791
+ "epoch": 0.15,
792
+ "learning_rate": 0.0002,
793
+ "loss": 1.3422,
794
+ "step": 1310
795
+ },
796
+ {
797
+ "epoch": 0.15,
798
+ "learning_rate": 0.0002,
799
+ "loss": 1.3778,
800
+ "step": 1320
801
+ },
802
+ {
803
+ "epoch": 0.15,
804
+ "learning_rate": 0.0002,
805
+ "loss": 1.5218,
806
+ "step": 1330
807
+ },
808
+ {
809
+ "epoch": 0.15,
810
+ "learning_rate": 0.0002,
811
+ "loss": 1.5068,
812
+ "step": 1340
813
+ },
814
+ {
815
+ "epoch": 0.15,
816
+ "learning_rate": 0.0002,
817
+ "loss": 1.1988,
818
+ "step": 1350
819
+ },
820
+ {
821
+ "epoch": 0.16,
822
+ "learning_rate": 0.0002,
823
+ "loss": 1.3527,
824
+ "step": 1360
825
+ },
826
+ {
827
+ "epoch": 0.16,
828
+ "learning_rate": 0.0002,
829
+ "loss": 1.4118,
830
+ "step": 1370
831
+ },
832
+ {
833
+ "epoch": 0.16,
834
+ "learning_rate": 0.0002,
835
+ "loss": 1.46,
836
+ "step": 1380
837
+ },
838
+ {
839
+ "epoch": 0.16,
840
+ "learning_rate": 0.0002,
841
+ "loss": 1.5036,
842
+ "step": 1390
843
+ },
844
+ {
845
+ "epoch": 0.16,
846
+ "learning_rate": 0.0002,
847
+ "loss": 1.2142,
848
+ "step": 1400
849
+ },
850
+ {
851
+ "epoch": 0.16,
852
+ "learning_rate": 0.0002,
853
+ "loss": 1.3566,
854
+ "step": 1410
855
+ },
856
+ {
857
+ "epoch": 0.16,
858
+ "learning_rate": 0.0002,
859
+ "loss": 1.4106,
860
+ "step": 1420
861
+ },
862
+ {
863
+ "epoch": 0.16,
864
+ "learning_rate": 0.0002,
865
+ "loss": 1.4844,
866
+ "step": 1430
867
+ },
868
+ {
869
+ "epoch": 0.16,
870
+ "learning_rate": 0.0002,
871
+ "loss": 1.469,
872
+ "step": 1440
873
+ },
874
+ {
875
+ "epoch": 0.17,
876
+ "learning_rate": 0.0002,
877
+ "loss": 1.1471,
878
+ "step": 1450
879
+ },
880
+ {
881
+ "epoch": 0.17,
882
+ "learning_rate": 0.0002,
883
+ "loss": 1.3605,
884
+ "step": 1460
885
+ },
886
+ {
887
+ "epoch": 0.17,
888
+ "learning_rate": 0.0002,
889
+ "loss": 1.3821,
890
+ "step": 1470
891
+ },
892
+ {
893
+ "epoch": 0.17,
894
+ "learning_rate": 0.0002,
895
+ "loss": 1.4964,
896
+ "step": 1480
897
+ },
898
+ {
899
+ "epoch": 0.17,
900
+ "learning_rate": 0.0002,
901
+ "loss": 1.493,
902
+ "step": 1490
903
+ },
904
+ {
905
+ "epoch": 0.17,
906
+ "learning_rate": 0.0002,
907
+ "loss": 1.2144,
908
+ "step": 1500
909
+ },
910
+ {
911
+ "epoch": 0.17,
912
+ "learning_rate": 0.0002,
913
+ "loss": 1.3792,
914
+ "step": 1510
915
+ },
916
+ {
917
+ "epoch": 0.17,
918
+ "learning_rate": 0.0002,
919
+ "loss": 1.3787,
920
+ "step": 1520
921
+ },
922
+ {
923
+ "epoch": 0.17,
924
+ "learning_rate": 0.0002,
925
+ "loss": 1.4801,
926
+ "step": 1530
927
+ },
928
+ {
929
+ "epoch": 0.18,
930
+ "learning_rate": 0.0002,
931
+ "loss": 1.5004,
932
+ "step": 1540
933
+ },
934
+ {
935
+ "epoch": 0.18,
936
+ "learning_rate": 0.0002,
937
+ "loss": 1.1574,
938
+ "step": 1550
939
+ },
940
+ {
941
+ "epoch": 0.18,
942
+ "learning_rate": 0.0002,
943
+ "loss": 1.371,
944
+ "step": 1560
945
+ },
946
+ {
947
+ "epoch": 0.18,
948
+ "learning_rate": 0.0002,
949
+ "loss": 1.3763,
950
+ "step": 1570
951
+ },
952
+ {
953
+ "epoch": 0.18,
954
+ "learning_rate": 0.0002,
955
+ "loss": 1.4883,
956
+ "step": 1580
957
+ },
958
+ {
959
+ "epoch": 0.18,
960
+ "learning_rate": 0.0002,
961
+ "loss": 1.5189,
962
+ "step": 1590
963
+ },
964
+ {
965
+ "epoch": 0.18,
966
+ "learning_rate": 0.0002,
967
+ "loss": 1.0868,
968
+ "step": 1600
969
+ },
970
+ {
971
+ "epoch": 0.18,
972
+ "learning_rate": 0.0002,
973
+ "loss": 1.3933,
974
+ "step": 1610
975
+ },
976
+ {
977
+ "epoch": 0.19,
978
+ "learning_rate": 0.0002,
979
+ "loss": 1.4105,
980
+ "step": 1620
981
+ },
982
+ {
983
+ "epoch": 0.19,
984
+ "learning_rate": 0.0002,
985
+ "loss": 1.4881,
986
+ "step": 1630
987
+ },
988
+ {
989
+ "epoch": 0.19,
990
+ "learning_rate": 0.0002,
991
+ "loss": 1.4615,
992
+ "step": 1640
993
+ },
994
+ {
995
+ "epoch": 0.19,
996
+ "learning_rate": 0.0002,
997
+ "loss": 1.0866,
998
+ "step": 1650
999
+ },
1000
+ {
1001
+ "epoch": 0.19,
1002
+ "learning_rate": 0.0002,
1003
+ "loss": 1.3884,
1004
+ "step": 1660
1005
+ },
1006
+ {
1007
+ "epoch": 0.19,
1008
+ "learning_rate": 0.0002,
1009
+ "loss": 1.3647,
1010
+ "step": 1670
1011
+ },
1012
+ {
1013
+ "epoch": 0.19,
1014
+ "learning_rate": 0.0002,
1015
+ "loss": 1.5341,
1016
+ "step": 1680
1017
+ },
1018
+ {
1019
+ "epoch": 0.19,
1020
+ "learning_rate": 0.0002,
1021
+ "loss": 1.5178,
1022
+ "step": 1690
1023
+ },
1024
+ {
1025
+ "epoch": 0.19,
1026
+ "learning_rate": 0.0002,
1027
+ "loss": 1.1867,
1028
+ "step": 1700
1029
+ },
1030
+ {
1031
+ "epoch": 0.2,
1032
+ "learning_rate": 0.0002,
1033
+ "loss": 1.3494,
1034
+ "step": 1710
1035
+ },
1036
+ {
1037
+ "epoch": 0.2,
1038
+ "learning_rate": 0.0002,
1039
+ "loss": 1.3895,
1040
+ "step": 1720
1041
+ },
1042
+ {
1043
+ "epoch": 0.2,
1044
+ "learning_rate": 0.0002,
1045
+ "loss": 1.4585,
1046
+ "step": 1730
1047
+ },
1048
+ {
1049
+ "epoch": 0.2,
1050
+ "learning_rate": 0.0002,
1051
+ "loss": 1.4815,
1052
+ "step": 1740
1053
+ },
1054
+ {
1055
+ "epoch": 0.2,
1056
+ "learning_rate": 0.0002,
1057
+ "loss": 1.1065,
1058
+ "step": 1750
1059
+ },
1060
+ {
1061
+ "epoch": 0.2,
1062
+ "learning_rate": 0.0002,
1063
+ "loss": 1.3551,
1064
+ "step": 1760
1065
+ },
1066
+ {
1067
+ "epoch": 0.2,
1068
+ "learning_rate": 0.0002,
1069
+ "loss": 1.388,
1070
+ "step": 1770
1071
+ },
1072
+ {
1073
+ "epoch": 0.2,
1074
+ "learning_rate": 0.0002,
1075
+ "loss": 1.4888,
1076
+ "step": 1780
1077
+ },
1078
+ {
1079
+ "epoch": 0.2,
1080
+ "learning_rate": 0.0002,
1081
+ "loss": 1.4693,
1082
+ "step": 1790
1083
+ },
1084
+ {
1085
+ "epoch": 0.21,
1086
+ "learning_rate": 0.0002,
1087
+ "loss": 1.2007,
1088
+ "step": 1800
1089
+ },
1090
+ {
1091
+ "epoch": 0.21,
1092
+ "learning_rate": 0.0002,
1093
+ "loss": 1.3711,
1094
+ "step": 1810
1095
+ },
1096
+ {
1097
+ "epoch": 0.21,
1098
+ "learning_rate": 0.0002,
1099
+ "loss": 1.4145,
1100
+ "step": 1820
1101
+ },
1102
+ {
1103
+ "epoch": 0.21,
1104
+ "learning_rate": 0.0002,
1105
+ "loss": 1.4982,
1106
+ "step": 1830
1107
+ },
1108
+ {
1109
+ "epoch": 0.21,
1110
+ "learning_rate": 0.0002,
1111
+ "loss": 1.4874,
1112
+ "step": 1840
1113
+ },
1114
+ {
1115
+ "epoch": 0.21,
1116
+ "learning_rate": 0.0002,
1117
+ "loss": 1.1387,
1118
+ "step": 1850
1119
+ },
1120
+ {
1121
+ "epoch": 0.21,
1122
+ "learning_rate": 0.0002,
1123
+ "loss": 1.3761,
1124
+ "step": 1860
1125
+ },
1126
+ {
1127
+ "epoch": 0.21,
1128
+ "learning_rate": 0.0002,
1129
+ "loss": 1.3737,
1130
+ "step": 1870
1131
+ },
1132
+ {
1133
+ "epoch": 0.21,
1134
+ "learning_rate": 0.0002,
1135
+ "loss": 1.4931,
1136
+ "step": 1880
1137
+ },
1138
+ {
1139
+ "epoch": 0.22,
1140
+ "learning_rate": 0.0002,
1141
+ "loss": 1.4763,
1142
+ "step": 1890
1143
+ },
1144
+ {
1145
+ "epoch": 0.22,
1146
+ "learning_rate": 0.0002,
1147
+ "loss": 1.1276,
1148
+ "step": 1900
1149
+ },
1150
+ {
1151
+ "epoch": 0.22,
1152
+ "learning_rate": 0.0002,
1153
+ "loss": 1.3674,
1154
+ "step": 1910
1155
+ },
1156
+ {
1157
+ "epoch": 0.22,
1158
+ "learning_rate": 0.0002,
1159
+ "loss": 1.4133,
1160
+ "step": 1920
1161
+ },
1162
+ {
1163
+ "epoch": 0.22,
1164
+ "learning_rate": 0.0002,
1165
+ "loss": 1.451,
1166
+ "step": 1930
1167
+ },
1168
+ {
1169
+ "epoch": 0.22,
1170
+ "learning_rate": 0.0002,
1171
+ "loss": 1.5189,
1172
+ "step": 1940
1173
+ },
1174
+ {
1175
+ "epoch": 0.22,
1176
+ "learning_rate": 0.0002,
1177
+ "loss": 1.0636,
1178
+ "step": 1950
1179
+ },
1180
+ {
1181
+ "epoch": 0.22,
1182
+ "learning_rate": 0.0002,
1183
+ "loss": 1.3954,
1184
+ "step": 1960
1185
+ },
1186
+ {
1187
+ "epoch": 0.23,
1188
+ "learning_rate": 0.0002,
1189
+ "loss": 1.3651,
1190
+ "step": 1970
1191
+ },
1192
+ {
1193
+ "epoch": 0.23,
1194
+ "learning_rate": 0.0002,
1195
+ "loss": 1.4482,
1196
+ "step": 1980
1197
+ },
1198
+ {
1199
+ "epoch": 0.23,
1200
+ "learning_rate": 0.0002,
1201
+ "loss": 1.478,
1202
+ "step": 1990
1203
+ },
1204
+ {
1205
+ "epoch": 0.23,
1206
+ "learning_rate": 0.0002,
1207
+ "loss": 1.0778,
1208
+ "step": 2000
1209
+ },
1210
+ {
1211
+ "epoch": 0.23,
1212
+ "learning_rate": 0.0002,
1213
+ "loss": 1.3601,
1214
+ "step": 2010
1215
+ },
1216
+ {
1217
+ "epoch": 0.23,
1218
+ "learning_rate": 0.0002,
1219
+ "loss": 1.4251,
1220
+ "step": 2020
1221
+ },
1222
+ {
1223
+ "epoch": 0.23,
1224
+ "learning_rate": 0.0002,
1225
+ "loss": 1.4854,
1226
+ "step": 2030
1227
+ },
1228
+ {
1229
+ "epoch": 0.23,
1230
+ "learning_rate": 0.0002,
1231
+ "loss": 1.4664,
1232
+ "step": 2040
1233
+ },
1234
+ {
1235
+ "epoch": 0.23,
1236
+ "learning_rate": 0.0002,
1237
+ "loss": 1.141,
1238
+ "step": 2050
1239
+ },
1240
+ {
1241
+ "epoch": 0.24,
1242
+ "learning_rate": 0.0002,
1243
+ "loss": 1.3472,
1244
+ "step": 2060
1245
+ },
1246
+ {
1247
+ "epoch": 0.24,
1248
+ "learning_rate": 0.0002,
1249
+ "loss": 1.3278,
1250
+ "step": 2070
1251
+ },
1252
+ {
1253
+ "epoch": 0.24,
1254
+ "learning_rate": 0.0002,
1255
+ "loss": 1.5566,
1256
+ "step": 2080
1257
+ },
1258
+ {
1259
+ "epoch": 0.24,
1260
+ "learning_rate": 0.0002,
1261
+ "loss": 1.4889,
1262
+ "step": 2090
1263
+ },
1264
+ {
1265
+ "epoch": 0.24,
1266
+ "learning_rate": 0.0002,
1267
+ "loss": 1.2219,
1268
+ "step": 2100
1269
+ },
1270
+ {
1271
+ "epoch": 0.24,
1272
+ "learning_rate": 0.0002,
1273
+ "loss": 1.3655,
1274
+ "step": 2110
1275
+ },
1276
+ {
1277
+ "epoch": 0.24,
1278
+ "learning_rate": 0.0002,
1279
+ "loss": 1.4287,
1280
+ "step": 2120
1281
+ },
1282
+ {
1283
+ "epoch": 0.24,
1284
+ "learning_rate": 0.0002,
1285
+ "loss": 1.4809,
1286
+ "step": 2130
1287
+ },
1288
+ {
1289
+ "epoch": 0.24,
1290
+ "learning_rate": 0.0002,
1291
+ "loss": 1.5159,
1292
+ "step": 2140
1293
+ },
1294
+ {
1295
+ "epoch": 0.25,
1296
+ "learning_rate": 0.0002,
1297
+ "loss": 1.1543,
1298
+ "step": 2150
1299
+ },
1300
+ {
1301
+ "epoch": 0.25,
1302
+ "learning_rate": 0.0002,
1303
+ "loss": 1.3774,
1304
+ "step": 2160
1305
+ },
1306
+ {
1307
+ "epoch": 0.25,
1308
+ "learning_rate": 0.0002,
1309
+ "loss": 1.3806,
1310
+ "step": 2170
1311
+ },
1312
+ {
1313
+ "epoch": 0.25,
1314
+ "learning_rate": 0.0002,
1315
+ "loss": 1.4786,
1316
+ "step": 2180
1317
+ },
1318
+ {
1319
+ "epoch": 0.25,
1320
+ "learning_rate": 0.0002,
1321
+ "loss": 1.4945,
1322
+ "step": 2190
1323
+ },
1324
+ {
1325
+ "epoch": 0.25,
1326
+ "learning_rate": 0.0002,
1327
+ "loss": 1.1604,
1328
+ "step": 2200
1329
+ },
1330
+ {
1331
+ "epoch": 0.25,
1332
+ "learning_rate": 0.0002,
1333
+ "loss": 1.3667,
1334
+ "step": 2210
1335
+ },
1336
+ {
1337
+ "epoch": 0.25,
1338
+ "learning_rate": 0.0002,
1339
+ "loss": 1.3984,
1340
+ "step": 2220
1341
+ },
1342
+ {
1343
+ "epoch": 0.25,
1344
+ "learning_rate": 0.0002,
1345
+ "loss": 1.4627,
1346
+ "step": 2230
1347
+ },
1348
+ {
1349
+ "epoch": 0.26,
1350
+ "learning_rate": 0.0002,
1351
+ "loss": 1.4767,
1352
+ "step": 2240
1353
+ },
1354
+ {
1355
+ "epoch": 0.26,
1356
+ "learning_rate": 0.0002,
1357
+ "loss": 1.0874,
1358
+ "step": 2250
1359
+ },
1360
+ {
1361
+ "epoch": 0.26,
1362
+ "learning_rate": 0.0002,
1363
+ "loss": 1.3641,
1364
+ "step": 2260
1365
+ },
1366
+ {
1367
+ "epoch": 0.26,
1368
+ "learning_rate": 0.0002,
1369
+ "loss": 1.4196,
1370
+ "step": 2270
1371
+ },
1372
+ {
1373
+ "epoch": 0.26,
1374
+ "learning_rate": 0.0002,
1375
+ "loss": 1.5019,
1376
+ "step": 2280
1377
+ },
1378
+ {
1379
+ "epoch": 0.26,
1380
+ "learning_rate": 0.0002,
1381
+ "loss": 1.4729,
1382
+ "step": 2290
1383
+ },
1384
+ {
1385
+ "epoch": 0.26,
1386
+ "learning_rate": 0.0002,
1387
+ "loss": 1.0994,
1388
+ "step": 2300
1389
+ },
1390
+ {
1391
+ "epoch": 0.26,
1392
+ "learning_rate": 0.0002,
1393
+ "loss": 1.3485,
1394
+ "step": 2310
1395
+ },
1396
+ {
1397
+ "epoch": 0.27,
1398
+ "learning_rate": 0.0002,
1399
+ "loss": 1.353,
1400
+ "step": 2320
1401
+ },
1402
+ {
1403
+ "epoch": 0.27,
1404
+ "learning_rate": 0.0002,
1405
+ "loss": 1.4497,
1406
+ "step": 2330
1407
+ },
1408
+ {
1409
+ "epoch": 0.27,
1410
+ "learning_rate": 0.0002,
1411
+ "loss": 1.4204,
1412
+ "step": 2340
1413
+ },
1414
+ {
1415
+ "epoch": 0.27,
1416
+ "learning_rate": 0.0002,
1417
+ "loss": 1.1602,
1418
+ "step": 2350
1419
+ },
1420
+ {
1421
+ "epoch": 0.27,
1422
+ "learning_rate": 0.0002,
1423
+ "loss": 1.3371,
1424
+ "step": 2360
1425
+ },
1426
+ {
1427
+ "epoch": 0.27,
1428
+ "learning_rate": 0.0002,
1429
+ "loss": 1.404,
1430
+ "step": 2370
1431
+ },
1432
+ {
1433
+ "epoch": 0.27,
1434
+ "learning_rate": 0.0002,
1435
+ "loss": 1.5122,
1436
+ "step": 2380
1437
+ },
1438
+ {
1439
+ "epoch": 0.27,
1440
+ "learning_rate": 0.0002,
1441
+ "loss": 1.4756,
1442
+ "step": 2390
1443
+ },
1444
+ {
1445
+ "epoch": 0.27,
1446
+ "learning_rate": 0.0002,
1447
+ "loss": 1.1223,
1448
+ "step": 2400
1449
+ },
1450
+ {
1451
+ "epoch": 0.28,
1452
+ "learning_rate": 0.0002,
1453
+ "loss": 1.3523,
1454
+ "step": 2410
1455
+ },
1456
+ {
1457
+ "epoch": 0.28,
1458
+ "learning_rate": 0.0002,
1459
+ "loss": 1.433,
1460
+ "step": 2420
1461
+ },
1462
+ {
1463
+ "epoch": 0.28,
1464
+ "learning_rate": 0.0002,
1465
+ "loss": 1.5184,
1466
+ "step": 2430
1467
+ },
1468
+ {
1469
+ "epoch": 0.28,
1470
+ "learning_rate": 0.0002,
1471
+ "loss": 1.5016,
1472
+ "step": 2440
1473
+ },
1474
+ {
1475
+ "epoch": 0.28,
1476
+ "learning_rate": 0.0002,
1477
+ "loss": 1.1152,
1478
+ "step": 2450
1479
+ },
1480
+ {
1481
+ "epoch": 0.28,
1482
+ "learning_rate": 0.0002,
1483
+ "loss": 1.4008,
1484
+ "step": 2460
1485
+ },
1486
+ {
1487
+ "epoch": 0.28,
1488
+ "learning_rate": 0.0002,
1489
+ "loss": 1.405,
1490
+ "step": 2470
1491
+ },
1492
+ {
1493
+ "epoch": 0.28,
1494
+ "learning_rate": 0.0002,
1495
+ "loss": 1.4862,
1496
+ "step": 2480
1497
+ },
1498
+ {
1499
+ "epoch": 0.28,
1500
+ "learning_rate": 0.0002,
1501
+ "loss": 1.48,
1502
+ "step": 2490
1503
+ },
1504
+ {
1505
+ "epoch": 0.29,
1506
+ "learning_rate": 0.0002,
1507
+ "loss": 1.1708,
1508
+ "step": 2500
1509
+ },
1510
+ {
1511
+ "epoch": 0.29,
1512
+ "learning_rate": 0.0002,
1513
+ "loss": 1.3488,
1514
+ "step": 2510
1515
+ },
1516
+ {
1517
+ "epoch": 0.29,
1518
+ "learning_rate": 0.0002,
1519
+ "loss": 1.3866,
1520
+ "step": 2520
1521
+ },
1522
+ {
1523
+ "epoch": 0.29,
1524
+ "learning_rate": 0.0002,
1525
+ "loss": 1.4876,
1526
+ "step": 2530
1527
+ },
1528
+ {
1529
+ "epoch": 0.29,
1530
+ "learning_rate": 0.0002,
1531
+ "loss": 1.4691,
1532
+ "step": 2540
1533
+ },
1534
+ {
1535
+ "epoch": 0.29,
1536
+ "learning_rate": 0.0002,
1537
+ "loss": 1.1237,
1538
+ "step": 2550
1539
+ },
1540
+ {
1541
+ "epoch": 0.29,
1542
+ "learning_rate": 0.0002,
1543
+ "loss": 1.3791,
1544
+ "step": 2560
1545
+ },
1546
+ {
1547
+ "epoch": 0.29,
1548
+ "learning_rate": 0.0002,
1549
+ "loss": 1.3704,
1550
+ "step": 2570
1551
+ },
1552
+ {
1553
+ "epoch": 0.29,
1554
+ "learning_rate": 0.0002,
1555
+ "loss": 1.4423,
1556
+ "step": 2580
1557
+ },
1558
+ {
1559
+ "epoch": 0.3,
1560
+ "learning_rate": 0.0002,
1561
+ "loss": 1.5363,
1562
+ "step": 2590
1563
+ },
1564
+ {
1565
+ "epoch": 0.3,
1566
+ "learning_rate": 0.0002,
1567
+ "loss": 1.1361,
1568
+ "step": 2600
1569
+ },
1570
+ {
1571
+ "epoch": 0.3,
1572
+ "learning_rate": 0.0002,
1573
+ "loss": 1.3744,
1574
+ "step": 2610
1575
+ },
1576
+ {
1577
+ "epoch": 0.3,
1578
+ "learning_rate": 0.0002,
1579
+ "loss": 1.4153,
1580
+ "step": 2620
1581
+ },
1582
+ {
1583
+ "epoch": 0.3,
1584
+ "learning_rate": 0.0002,
1585
+ "loss": 1.5511,
1586
+ "step": 2630
1587
+ },
1588
+ {
1589
+ "epoch": 0.3,
1590
+ "learning_rate": 0.0002,
1591
+ "loss": 1.4789,
1592
+ "step": 2640
1593
+ },
1594
+ {
1595
+ "epoch": 0.3,
1596
+ "learning_rate": 0.0002,
1597
+ "loss": 1.1828,
1598
+ "step": 2650
1599
+ },
1600
+ {
1601
+ "epoch": 0.3,
1602
+ "learning_rate": 0.0002,
1603
+ "loss": 1.3552,
1604
+ "step": 2660
1605
+ },
1606
+ {
1607
+ "epoch": 0.31,
1608
+ "learning_rate": 0.0002,
1609
+ "loss": 1.3887,
1610
+ "step": 2670
1611
+ },
1612
+ {
1613
+ "epoch": 0.31,
1614
+ "learning_rate": 0.0002,
1615
+ "loss": 1.5207,
1616
+ "step": 2680
1617
+ },
1618
+ {
1619
+ "epoch": 0.31,
1620
+ "learning_rate": 0.0002,
1621
+ "loss": 1.4766,
1622
+ "step": 2690
1623
+ },
1624
+ {
1625
+ "epoch": 0.31,
1626
+ "learning_rate": 0.0002,
1627
+ "loss": 1.1612,
1628
+ "step": 2700
1629
+ },
1630
+ {
1631
+ "epoch": 0.31,
1632
+ "learning_rate": 0.0002,
1633
+ "loss": 1.3412,
1634
+ "step": 2710
1635
+ },
1636
+ {
1637
+ "epoch": 0.31,
1638
+ "learning_rate": 0.0002,
1639
+ "loss": 1.4014,
1640
+ "step": 2720
1641
+ },
1642
+ {
1643
+ "epoch": 0.31,
1644
+ "learning_rate": 0.0002,
1645
+ "loss": 1.4732,
1646
+ "step": 2730
1647
+ },
1648
+ {
1649
+ "epoch": 0.31,
1650
+ "learning_rate": 0.0002,
1651
+ "loss": 1.4645,
1652
+ "step": 2740
1653
+ },
1654
+ {
1655
+ "epoch": 0.31,
1656
+ "learning_rate": 0.0002,
1657
+ "loss": 1.1439,
1658
+ "step": 2750
1659
+ },
1660
+ {
1661
+ "epoch": 0.32,
1662
+ "learning_rate": 0.0002,
1663
+ "loss": 1.3987,
1664
+ "step": 2760
1665
+ },
1666
+ {
1667
+ "epoch": 0.32,
1668
+ "learning_rate": 0.0002,
1669
+ "loss": 1.4204,
1670
+ "step": 2770
1671
+ },
1672
+ {
1673
+ "epoch": 0.32,
1674
+ "learning_rate": 0.0002,
1675
+ "loss": 1.5018,
1676
+ "step": 2780
1677
+ },
1678
+ {
1679
+ "epoch": 0.32,
1680
+ "learning_rate": 0.0002,
1681
+ "loss": 1.5062,
1682
+ "step": 2790
1683
+ },
1684
+ {
1685
+ "epoch": 0.32,
1686
+ "learning_rate": 0.0002,
1687
+ "loss": 1.1205,
1688
+ "step": 2800
1689
+ },
1690
+ {
1691
+ "epoch": 0.32,
1692
+ "learning_rate": 0.0002,
1693
+ "loss": 1.3577,
1694
+ "step": 2810
1695
+ },
1696
+ {
1697
+ "epoch": 0.32,
1698
+ "learning_rate": 0.0002,
1699
+ "loss": 1.3845,
1700
+ "step": 2820
1701
+ },
1702
+ {
1703
+ "epoch": 0.32,
1704
+ "learning_rate": 0.0002,
1705
+ "loss": 1.4913,
1706
+ "step": 2830
1707
+ },
1708
+ {
1709
+ "epoch": 0.32,
1710
+ "learning_rate": 0.0002,
1711
+ "loss": 1.5165,
1712
+ "step": 2840
1713
+ },
1714
+ {
1715
+ "epoch": 0.33,
1716
+ "learning_rate": 0.0002,
1717
+ "loss": 1.1937,
1718
+ "step": 2850
1719
+ },
1720
+ {
1721
+ "epoch": 0.33,
1722
+ "learning_rate": 0.0002,
1723
+ "loss": 1.3885,
1724
+ "step": 2860
1725
+ },
1726
+ {
1727
+ "epoch": 0.33,
1728
+ "learning_rate": 0.0002,
1729
+ "loss": 1.3824,
1730
+ "step": 2870
1731
+ },
1732
+ {
1733
+ "epoch": 0.33,
1734
+ "learning_rate": 0.0002,
1735
+ "loss": 1.4714,
1736
+ "step": 2880
1737
+ },
1738
+ {
1739
+ "epoch": 0.33,
1740
+ "learning_rate": 0.0002,
1741
+ "loss": 1.4905,
1742
+ "step": 2890
1743
+ },
1744
+ {
1745
+ "epoch": 0.33,
1746
+ "learning_rate": 0.0002,
1747
+ "loss": 1.1824,
1748
+ "step": 2900
1749
+ },
1750
+ {
1751
+ "epoch": 0.33,
1752
+ "learning_rate": 0.0002,
1753
+ "loss": 1.3445,
1754
+ "step": 2910
1755
+ },
1756
+ {
1757
+ "epoch": 0.33,
1758
+ "learning_rate": 0.0002,
1759
+ "loss": 1.38,
1760
+ "step": 2920
1761
+ },
1762
+ {
1763
+ "epoch": 0.33,
1764
+ "learning_rate": 0.0002,
1765
+ "loss": 1.4518,
1766
+ "step": 2930
1767
+ },
1768
+ {
1769
+ "epoch": 0.34,
1770
+ "learning_rate": 0.0002,
1771
+ "loss": 1.4752,
1772
+ "step": 2940
1773
+ },
1774
+ {
1775
+ "epoch": 0.34,
1776
+ "learning_rate": 0.0002,
1777
+ "loss": 1.1911,
1778
+ "step": 2950
1779
+ },
1780
+ {
1781
+ "epoch": 0.34,
1782
+ "learning_rate": 0.0002,
1783
+ "loss": 1.3991,
1784
+ "step": 2960
1785
+ },
1786
+ {
1787
+ "epoch": 0.34,
1788
+ "learning_rate": 0.0002,
1789
+ "loss": 1.3773,
1790
+ "step": 2970
1791
+ },
1792
+ {
1793
+ "epoch": 0.34,
1794
+ "learning_rate": 0.0002,
1795
+ "loss": 1.4901,
1796
+ "step": 2980
1797
+ },
1798
+ {
1799
+ "epoch": 0.34,
1800
+ "learning_rate": 0.0002,
1801
+ "loss": 1.4659,
1802
+ "step": 2990
1803
+ },
1804
+ {
1805
+ "epoch": 0.34,
1806
+ "learning_rate": 0.0002,
1807
+ "loss": 1.1806,
1808
+ "step": 3000
1809
+ },
1810
+ {
1811
+ "epoch": 0.34,
1812
+ "learning_rate": 0.0002,
1813
+ "loss": 1.35,
1814
+ "step": 3010
1815
+ },
1816
+ {
1817
+ "epoch": 0.35,
1818
+ "learning_rate": 0.0002,
1819
+ "loss": 1.4208,
1820
+ "step": 3020
1821
+ },
1822
+ {
1823
+ "epoch": 0.35,
1824
+ "learning_rate": 0.0002,
1825
+ "loss": 1.4913,
1826
+ "step": 3030
1827
+ },
1828
+ {
1829
+ "epoch": 0.35,
1830
+ "learning_rate": 0.0002,
1831
+ "loss": 1.4977,
1832
+ "step": 3040
1833
+ },
1834
+ {
1835
+ "epoch": 0.35,
1836
+ "learning_rate": 0.0002,
1837
+ "loss": 1.2201,
1838
+ "step": 3050
1839
+ },
1840
+ {
1841
+ "epoch": 0.35,
1842
+ "learning_rate": 0.0002,
1843
+ "loss": 1.3578,
1844
+ "step": 3060
1845
+ },
1846
+ {
1847
+ "epoch": 0.35,
1848
+ "learning_rate": 0.0002,
1849
+ "loss": 1.3585,
1850
+ "step": 3070
1851
+ },
1852
+ {
1853
+ "epoch": 0.35,
1854
+ "learning_rate": 0.0002,
1855
+ "loss": 1.4842,
1856
+ "step": 3080
1857
+ },
1858
+ {
1859
+ "epoch": 0.35,
1860
+ "learning_rate": 0.0002,
1861
+ "loss": 1.461,
1862
+ "step": 3090
1863
+ },
1864
+ {
1865
+ "epoch": 0.35,
1866
+ "learning_rate": 0.0002,
1867
+ "loss": 1.1151,
1868
+ "step": 3100
1869
+ },
1870
+ {
1871
+ "epoch": 0.36,
1872
+ "learning_rate": 0.0002,
1873
+ "loss": 1.3658,
1874
+ "step": 3110
1875
+ },
1876
+ {
1877
+ "epoch": 0.36,
1878
+ "learning_rate": 0.0002,
1879
+ "loss": 1.4156,
1880
+ "step": 3120
1881
+ },
1882
+ {
1883
+ "epoch": 0.36,
1884
+ "learning_rate": 0.0002,
1885
+ "loss": 1.4507,
1886
+ "step": 3130
1887
+ },
1888
+ {
1889
+ "epoch": 0.36,
1890
+ "learning_rate": 0.0002,
1891
+ "loss": 1.4868,
1892
+ "step": 3140
1893
+ },
1894
+ {
1895
+ "epoch": 0.36,
1896
+ "learning_rate": 0.0002,
1897
+ "loss": 1.1919,
1898
+ "step": 3150
1899
+ },
1900
+ {
1901
+ "epoch": 0.36,
1902
+ "learning_rate": 0.0002,
1903
+ "loss": 1.3572,
1904
+ "step": 3160
1905
+ },
1906
+ {
1907
+ "epoch": 0.36,
1908
+ "learning_rate": 0.0002,
1909
+ "loss": 1.408,
1910
+ "step": 3170
1911
+ },
1912
+ {
1913
+ "epoch": 0.36,
1914
+ "learning_rate": 0.0002,
1915
+ "loss": 1.513,
1916
+ "step": 3180
1917
+ },
1918
+ {
1919
+ "epoch": 0.36,
1920
+ "learning_rate": 0.0002,
1921
+ "loss": 1.4418,
1922
+ "step": 3190
1923
+ },
1924
+ {
1925
+ "epoch": 0.37,
1926
+ "learning_rate": 0.0002,
1927
+ "loss": 1.1063,
1928
+ "step": 3200
1929
+ },
1930
+ {
1931
+ "epoch": 0.37,
1932
+ "learning_rate": 0.0002,
1933
+ "loss": 1.3842,
1934
+ "step": 3210
1935
+ },
1936
+ {
1937
+ "epoch": 0.37,
1938
+ "learning_rate": 0.0002,
1939
+ "loss": 1.4063,
1940
+ "step": 3220
1941
+ },
1942
+ {
1943
+ "epoch": 0.37,
1944
+ "learning_rate": 0.0002,
1945
+ "loss": 1.5104,
1946
+ "step": 3230
1947
+ },
1948
+ {
1949
+ "epoch": 0.37,
1950
+ "learning_rate": 0.0002,
1951
+ "loss": 1.4726,
1952
+ "step": 3240
1953
+ },
1954
+ {
1955
+ "epoch": 0.37,
1956
+ "learning_rate": 0.0002,
1957
+ "loss": 1.1022,
1958
+ "step": 3250
1959
+ },
1960
+ {
1961
+ "epoch": 0.37,
1962
+ "learning_rate": 0.0002,
1963
+ "loss": 1.3459,
1964
+ "step": 3260
1965
+ },
1966
+ {
1967
+ "epoch": 0.37,
1968
+ "learning_rate": 0.0002,
1969
+ "loss": 1.4093,
1970
+ "step": 3270
1971
+ },
1972
+ {
1973
+ "epoch": 0.37,
1974
+ "learning_rate": 0.0002,
1975
+ "loss": 1.4945,
1976
+ "step": 3280
1977
+ },
1978
+ {
1979
+ "epoch": 0.38,
1980
+ "learning_rate": 0.0002,
1981
+ "loss": 1.5151,
1982
+ "step": 3290
1983
+ },
1984
+ {
1985
+ "epoch": 0.38,
1986
+ "learning_rate": 0.0002,
1987
+ "loss": 1.1194,
1988
+ "step": 3300
1989
+ },
1990
+ {
1991
+ "epoch": 0.38,
1992
+ "learning_rate": 0.0002,
1993
+ "loss": 1.3495,
1994
+ "step": 3310
1995
+ },
1996
+ {
1997
+ "epoch": 0.38,
1998
+ "learning_rate": 0.0002,
1999
+ "loss": 1.348,
2000
+ "step": 3320
2001
+ },
2002
+ {
2003
+ "epoch": 0.38,
2004
+ "learning_rate": 0.0002,
2005
+ "loss": 1.5025,
2006
+ "step": 3330
2007
+ },
2008
+ {
2009
+ "epoch": 0.38,
2010
+ "learning_rate": 0.0002,
2011
+ "loss": 1.4858,
2012
+ "step": 3340
2013
+ },
2014
+ {
2015
+ "epoch": 0.38,
2016
+ "learning_rate": 0.0002,
2017
+ "loss": 1.1149,
2018
+ "step": 3350
2019
+ },
2020
+ {
2021
+ "epoch": 0.38,
2022
+ "learning_rate": 0.0002,
2023
+ "loss": 1.3526,
2024
+ "step": 3360
2025
+ },
2026
+ {
2027
+ "epoch": 0.39,
2028
+ "learning_rate": 0.0002,
2029
+ "loss": 1.4106,
2030
+ "step": 3370
2031
+ },
2032
+ {
2033
+ "epoch": 0.39,
2034
+ "learning_rate": 0.0002,
2035
+ "loss": 1.5304,
2036
+ "step": 3380
2037
+ },
2038
+ {
2039
+ "epoch": 0.39,
2040
+ "learning_rate": 0.0002,
2041
+ "loss": 1.462,
2042
+ "step": 3390
2043
+ },
2044
+ {
2045
+ "epoch": 0.39,
2046
+ "learning_rate": 0.0002,
2047
+ "loss": 1.0945,
2048
+ "step": 3400
2049
+ },
2050
+ {
2051
+ "epoch": 0.39,
2052
+ "learning_rate": 0.0002,
2053
+ "loss": 1.3582,
2054
+ "step": 3410
2055
+ },
2056
+ {
2057
+ "epoch": 0.39,
2058
+ "learning_rate": 0.0002,
2059
+ "loss": 1.3727,
2060
+ "step": 3420
2061
+ },
2062
+ {
2063
+ "epoch": 0.39,
2064
+ "learning_rate": 0.0002,
2065
+ "loss": 1.4673,
2066
+ "step": 3430
2067
+ },
2068
+ {
2069
+ "epoch": 0.39,
2070
+ "learning_rate": 0.0002,
2071
+ "loss": 1.4891,
2072
+ "step": 3440
2073
+ },
2074
+ {
2075
+ "epoch": 0.39,
2076
+ "learning_rate": 0.0002,
2077
+ "loss": 1.156,
2078
+ "step": 3450
2079
+ },
2080
+ {
2081
+ "epoch": 0.4,
2082
+ "learning_rate": 0.0002,
2083
+ "loss": 1.3427,
2084
+ "step": 3460
2085
+ },
2086
+ {
2087
+ "epoch": 0.4,
2088
+ "learning_rate": 0.0002,
2089
+ "loss": 1.3656,
2090
+ "step": 3470
2091
+ },
2092
+ {
2093
+ "epoch": 0.4,
2094
+ "learning_rate": 0.0002,
2095
+ "loss": 1.4977,
2096
+ "step": 3480
2097
+ },
2098
+ {
2099
+ "epoch": 0.4,
2100
+ "learning_rate": 0.0002,
2101
+ "loss": 1.4801,
2102
+ "step": 3490
2103
+ },
2104
+ {
2105
+ "epoch": 0.4,
2106
+ "learning_rate": 0.0002,
2107
+ "loss": 1.0395,
2108
+ "step": 3500
2109
+ },
2110
+ {
2111
+ "epoch": 0.4,
2112
+ "learning_rate": 0.0002,
2113
+ "loss": 1.3571,
2114
+ "step": 3510
2115
+ },
2116
+ {
2117
+ "epoch": 0.4,
2118
+ "learning_rate": 0.0002,
2119
+ "loss": 1.3911,
2120
+ "step": 3520
2121
+ },
2122
+ {
2123
+ "epoch": 0.4,
2124
+ "learning_rate": 0.0002,
2125
+ "loss": 1.4859,
2126
+ "step": 3530
2127
+ },
2128
+ {
2129
+ "epoch": 0.4,
2130
+ "learning_rate": 0.0002,
2131
+ "loss": 1.4886,
2132
+ "step": 3540
2133
+ },
2134
+ {
2135
+ "epoch": 0.41,
2136
+ "learning_rate": 0.0002,
2137
+ "loss": 1.1248,
2138
+ "step": 3550
2139
+ },
2140
+ {
2141
+ "epoch": 0.41,
2142
+ "learning_rate": 0.0002,
2143
+ "loss": 1.3461,
2144
+ "step": 3560
2145
+ },
2146
+ {
2147
+ "epoch": 0.41,
2148
+ "learning_rate": 0.0002,
2149
+ "loss": 1.4296,
2150
+ "step": 3570
2151
+ },
2152
+ {
2153
+ "epoch": 0.41,
2154
+ "learning_rate": 0.0002,
2155
+ "loss": 1.5224,
2156
+ "step": 3580
2157
+ },
2158
+ {
2159
+ "epoch": 0.41,
2160
+ "learning_rate": 0.0002,
2161
+ "loss": 1.4436,
2162
+ "step": 3590
2163
+ },
2164
+ {
2165
+ "epoch": 0.41,
2166
+ "learning_rate": 0.0002,
2167
+ "loss": 1.1429,
2168
+ "step": 3600
2169
+ },
2170
+ {
2171
+ "epoch": 0.41,
2172
+ "learning_rate": 0.0002,
2173
+ "loss": 1.4042,
2174
+ "step": 3610
2175
+ },
2176
+ {
2177
+ "epoch": 0.41,
2178
+ "learning_rate": 0.0002,
2179
+ "loss": 1.3643,
2180
+ "step": 3620
2181
+ },
2182
+ {
2183
+ "epoch": 0.41,
2184
+ "learning_rate": 0.0002,
2185
+ "loss": 1.4806,
2186
+ "step": 3630
2187
+ },
2188
+ {
2189
+ "epoch": 0.42,
2190
+ "learning_rate": 0.0002,
2191
+ "loss": 1.4694,
2192
+ "step": 3640
2193
+ },
2194
+ {
2195
+ "epoch": 0.42,
2196
+ "learning_rate": 0.0002,
2197
+ "loss": 1.1284,
2198
+ "step": 3650
2199
+ },
2200
+ {
2201
+ "epoch": 0.42,
2202
+ "learning_rate": 0.0002,
2203
+ "loss": 1.3763,
2204
+ "step": 3660
2205
+ },
2206
+ {
2207
+ "epoch": 0.42,
2208
+ "learning_rate": 0.0002,
2209
+ "loss": 1.3785,
2210
+ "step": 3670
2211
+ },
2212
+ {
2213
+ "epoch": 0.42,
2214
+ "learning_rate": 0.0002,
2215
+ "loss": 1.5124,
2216
+ "step": 3680
2217
+ },
2218
+ {
2219
+ "epoch": 0.42,
2220
+ "learning_rate": 0.0002,
2221
+ "loss": 1.4636,
2222
+ "step": 3690
2223
+ },
2224
+ {
2225
+ "epoch": 0.42,
2226
+ "learning_rate": 0.0002,
2227
+ "loss": 1.1329,
2228
+ "step": 3700
2229
+ },
2230
+ {
2231
+ "epoch": 0.42,
2232
+ "learning_rate": 0.0002,
2233
+ "loss": 1.3724,
2234
+ "step": 3710
2235
+ },
2236
+ {
2237
+ "epoch": 0.43,
2238
+ "learning_rate": 0.0002,
2239
+ "loss": 1.3632,
2240
+ "step": 3720
2241
+ },
2242
+ {
2243
+ "epoch": 0.43,
2244
+ "learning_rate": 0.0002,
2245
+ "loss": 1.4831,
2246
+ "step": 3730
2247
+ },
2248
+ {
2249
+ "epoch": 0.43,
2250
+ "learning_rate": 0.0002,
2251
+ "loss": 1.436,
2252
+ "step": 3740
2253
+ },
2254
+ {
2255
+ "epoch": 0.43,
2256
+ "learning_rate": 0.0002,
2257
+ "loss": 1.0665,
2258
+ "step": 3750
2259
+ },
2260
+ {
2261
+ "epoch": 0.43,
2262
+ "learning_rate": 0.0002,
2263
+ "loss": 1.3238,
2264
+ "step": 3760
2265
+ },
2266
+ {
2267
+ "epoch": 0.43,
2268
+ "learning_rate": 0.0002,
2269
+ "loss": 1.3803,
2270
+ "step": 3770
2271
+ },
2272
+ {
2273
+ "epoch": 0.43,
2274
+ "learning_rate": 0.0002,
2275
+ "loss": 1.4619,
2276
+ "step": 3780
2277
+ },
2278
+ {
2279
+ "epoch": 0.43,
2280
+ "learning_rate": 0.0002,
2281
+ "loss": 1.5251,
2282
+ "step": 3790
2283
+ },
2284
+ {
2285
+ "epoch": 0.43,
2286
+ "learning_rate": 0.0002,
2287
+ "loss": 1.1907,
2288
+ "step": 3800
2289
+ },
2290
+ {
2291
+ "epoch": 0.44,
2292
+ "learning_rate": 0.0002,
2293
+ "loss": 1.3207,
2294
+ "step": 3810
2295
+ },
2296
+ {
2297
+ "epoch": 0.44,
2298
+ "learning_rate": 0.0002,
2299
+ "loss": 1.3872,
2300
+ "step": 3820
2301
+ },
2302
+ {
2303
+ "epoch": 0.44,
2304
+ "learning_rate": 0.0002,
2305
+ "loss": 1.5255,
2306
+ "step": 3830
2307
+ },
2308
+ {
2309
+ "epoch": 0.44,
2310
+ "learning_rate": 0.0002,
2311
+ "loss": 1.4855,
2312
+ "step": 3840
2313
+ },
2314
+ {
2315
+ "epoch": 0.44,
2316
+ "learning_rate": 0.0002,
2317
+ "loss": 1.1058,
2318
+ "step": 3850
2319
+ },
2320
+ {
2321
+ "epoch": 0.44,
2322
+ "learning_rate": 0.0002,
2323
+ "loss": 1.3872,
2324
+ "step": 3860
2325
+ },
2326
+ {
2327
+ "epoch": 0.44,
2328
+ "learning_rate": 0.0002,
2329
+ "loss": 1.415,
2330
+ "step": 3870
2331
+ },
2332
+ {
2333
+ "epoch": 0.44,
2334
+ "learning_rate": 0.0002,
2335
+ "loss": 1.5502,
2336
+ "step": 3880
2337
+ },
2338
+ {
2339
+ "epoch": 0.44,
2340
+ "learning_rate": 0.0002,
2341
+ "loss": 1.4852,
2342
+ "step": 3890
2343
+ },
2344
+ {
2345
+ "epoch": 0.45,
2346
+ "learning_rate": 0.0002,
2347
+ "loss": 1.1746,
2348
+ "step": 3900
2349
+ },
2350
+ {
2351
+ "epoch": 0.45,
2352
+ "learning_rate": 0.0002,
2353
+ "loss": 1.3513,
2354
+ "step": 3910
2355
+ },
2356
+ {
2357
+ "epoch": 0.45,
2358
+ "learning_rate": 0.0002,
2359
+ "loss": 1.3604,
2360
+ "step": 3920
2361
+ },
2362
+ {
2363
+ "epoch": 0.45,
2364
+ "learning_rate": 0.0002,
2365
+ "loss": 1.4799,
2366
+ "step": 3930
2367
+ },
2368
+ {
2369
+ "epoch": 0.45,
2370
+ "learning_rate": 0.0002,
2371
+ "loss": 1.5093,
2372
+ "step": 3940
2373
+ },
2374
+ {
2375
+ "epoch": 0.45,
2376
+ "learning_rate": 0.0002,
2377
+ "loss": 1.1486,
2378
+ "step": 3950
2379
+ },
2380
+ {
2381
+ "epoch": 0.45,
2382
+ "learning_rate": 0.0002,
2383
+ "loss": 1.3901,
2384
+ "step": 3960
2385
+ },
2386
+ {
2387
+ "epoch": 0.45,
2388
+ "learning_rate": 0.0002,
2389
+ "loss": 1.3957,
2390
+ "step": 3970
2391
+ },
2392
+ {
2393
+ "epoch": 0.45,
2394
+ "learning_rate": 0.0002,
2395
+ "loss": 1.4782,
2396
+ "step": 3980
2397
+ },
2398
+ {
2399
+ "epoch": 0.46,
2400
+ "learning_rate": 0.0002,
2401
+ "loss": 1.5271,
2402
+ "step": 3990
2403
+ },
2404
+ {
2405
+ "epoch": 0.46,
2406
+ "learning_rate": 0.0002,
2407
+ "loss": 1.1536,
2408
+ "step": 4000
2409
+ },
2410
+ {
2411
+ "epoch": 0.46,
2412
+ "learning_rate": 0.0002,
2413
+ "loss": 1.4057,
2414
+ "step": 4010
2415
+ },
2416
+ {
2417
+ "epoch": 0.46,
2418
+ "learning_rate": 0.0002,
2419
+ "loss": 1.4112,
2420
+ "step": 4020
2421
+ },
2422
+ {
2423
+ "epoch": 0.46,
2424
+ "learning_rate": 0.0002,
2425
+ "loss": 1.5066,
2426
+ "step": 4030
2427
+ },
2428
+ {
2429
+ "epoch": 0.46,
2430
+ "learning_rate": 0.0002,
2431
+ "loss": 1.4654,
2432
+ "step": 4040
2433
+ },
2434
+ {
2435
+ "epoch": 0.46,
2436
+ "learning_rate": 0.0002,
2437
+ "loss": 1.1995,
2438
+ "step": 4050
2439
+ },
2440
+ {
2441
+ "epoch": 0.46,
2442
+ "learning_rate": 0.0002,
2443
+ "loss": 1.3912,
2444
+ "step": 4060
2445
+ },
2446
+ {
2447
+ "epoch": 0.47,
2448
+ "learning_rate": 0.0002,
2449
+ "loss": 1.4331,
2450
+ "step": 4070
2451
+ },
2452
+ {
2453
+ "epoch": 0.47,
2454
+ "learning_rate": 0.0002,
2455
+ "loss": 1.4904,
2456
+ "step": 4080
2457
+ },
2458
+ {
2459
+ "epoch": 0.47,
2460
+ "learning_rate": 0.0002,
2461
+ "loss": 1.4481,
2462
+ "step": 4090
2463
+ },
2464
+ {
2465
+ "epoch": 0.47,
2466
+ "learning_rate": 0.0002,
2467
+ "loss": 1.0531,
2468
+ "step": 4100
2469
+ },
2470
+ {
2471
+ "epoch": 0.47,
2472
+ "learning_rate": 0.0002,
2473
+ "loss": 1.4215,
2474
+ "step": 4110
2475
+ },
2476
+ {
2477
+ "epoch": 0.47,
2478
+ "learning_rate": 0.0002,
2479
+ "loss": 1.3689,
2480
+ "step": 4120
2481
+ },
2482
+ {
2483
+ "epoch": 0.47,
2484
+ "learning_rate": 0.0002,
2485
+ "loss": 1.5042,
2486
+ "step": 4130
2487
+ },
2488
+ {
2489
+ "epoch": 0.47,
2490
+ "learning_rate": 0.0002,
2491
+ "loss": 1.4755,
2492
+ "step": 4140
2493
+ },
2494
+ {
2495
+ "epoch": 0.47,
2496
+ "learning_rate": 0.0002,
2497
+ "loss": 1.1606,
2498
+ "step": 4150
2499
+ },
2500
+ {
2501
+ "epoch": 0.48,
2502
+ "learning_rate": 0.0002,
2503
+ "loss": 1.379,
2504
+ "step": 4160
2505
+ },
2506
+ {
2507
+ "epoch": 0.48,
2508
+ "learning_rate": 0.0002,
2509
+ "loss": 1.3928,
2510
+ "step": 4170
2511
+ },
2512
+ {
2513
+ "epoch": 0.48,
2514
+ "learning_rate": 0.0002,
2515
+ "loss": 1.4858,
2516
+ "step": 4180
2517
+ },
2518
+ {
2519
+ "epoch": 0.48,
2520
+ "learning_rate": 0.0002,
2521
+ "loss": 1.4799,
2522
+ "step": 4190
2523
+ },
2524
+ {
2525
+ "epoch": 0.48,
2526
+ "learning_rate": 0.0002,
2527
+ "loss": 1.1618,
2528
+ "step": 4200
2529
+ },
2530
+ {
2531
+ "epoch": 0.48,
2532
+ "learning_rate": 0.0002,
2533
+ "loss": 1.3297,
2534
+ "step": 4210
2535
+ },
2536
+ {
2537
+ "epoch": 0.48,
2538
+ "learning_rate": 0.0002,
2539
+ "loss": 1.3624,
2540
+ "step": 4220
2541
+ },
2542
+ {
2543
+ "epoch": 0.48,
2544
+ "learning_rate": 0.0002,
2545
+ "loss": 1.5152,
2546
+ "step": 4230
2547
+ },
2548
+ {
2549
+ "epoch": 0.48,
2550
+ "learning_rate": 0.0002,
2551
+ "loss": 1.4984,
2552
+ "step": 4240
2553
+ },
2554
+ {
2555
+ "epoch": 0.49,
2556
+ "learning_rate": 0.0002,
2557
+ "loss": 1.1028,
2558
+ "step": 4250
2559
+ },
2560
+ {
2561
+ "epoch": 0.49,
2562
+ "learning_rate": 0.0002,
2563
+ "loss": 1.3869,
2564
+ "step": 4260
2565
+ },
2566
+ {
2567
+ "epoch": 0.49,
2568
+ "learning_rate": 0.0002,
2569
+ "loss": 1.4085,
2570
+ "step": 4270
2571
+ },
2572
+ {
2573
+ "epoch": 0.49,
2574
+ "learning_rate": 0.0002,
2575
+ "loss": 1.4954,
2576
+ "step": 4280
2577
+ },
2578
+ {
2579
+ "epoch": 0.49,
2580
+ "learning_rate": 0.0002,
2581
+ "loss": 1.5095,
2582
+ "step": 4290
2583
+ },
2584
+ {
2585
+ "epoch": 0.49,
2586
+ "learning_rate": 0.0002,
2587
+ "loss": 1.1629,
2588
+ "step": 4300
2589
+ },
2590
+ {
2591
+ "epoch": 0.49,
2592
+ "learning_rate": 0.0002,
2593
+ "loss": 1.364,
2594
+ "step": 4310
2595
+ },
2596
+ {
2597
+ "epoch": 0.49,
2598
+ "learning_rate": 0.0002,
2599
+ "loss": 1.4175,
2600
+ "step": 4320
2601
+ },
2602
+ {
2603
+ "epoch": 0.49,
2604
+ "learning_rate": 0.0002,
2605
+ "loss": 1.4608,
2606
+ "step": 4330
2607
+ },
2608
+ {
2609
+ "epoch": 0.5,
2610
+ "learning_rate": 0.0002,
2611
+ "loss": 1.4981,
2612
+ "step": 4340
2613
+ },
2614
+ {
2615
+ "epoch": 0.5,
2616
+ "learning_rate": 0.0002,
2617
+ "loss": 1.139,
2618
+ "step": 4350
2619
+ },
2620
+ {
2621
+ "epoch": 0.5,
2622
+ "learning_rate": 0.0002,
2623
+ "loss": 1.3476,
2624
+ "step": 4360
2625
+ },
2626
+ {
2627
+ "epoch": 0.5,
2628
+ "learning_rate": 0.0002,
2629
+ "loss": 1.4028,
2630
+ "step": 4370
2631
+ },
2632
+ {
2633
+ "epoch": 0.5,
2634
+ "learning_rate": 0.0002,
2635
+ "loss": 1.5133,
2636
+ "step": 4380
2637
+ },
2638
+ {
2639
+ "epoch": 0.5,
2640
+ "learning_rate": 0.0002,
2641
+ "loss": 1.5031,
2642
+ "step": 4390
2643
+ },
2644
+ {
2645
+ "epoch": 0.5,
2646
+ "learning_rate": 0.0002,
2647
+ "loss": 1.1492,
2648
+ "step": 4400
2649
+ },
2650
+ {
2651
+ "epoch": 0.5,
2652
+ "learning_rate": 0.0002,
2653
+ "loss": 1.3831,
2654
+ "step": 4410
2655
+ },
2656
+ {
2657
+ "epoch": 0.51,
2658
+ "learning_rate": 0.0002,
2659
+ "loss": 1.4053,
2660
+ "step": 4420
2661
+ },
2662
+ {
2663
+ "epoch": 0.51,
2664
+ "learning_rate": 0.0002,
2665
+ "loss": 1.491,
2666
+ "step": 4430
2667
+ },
2668
+ {
2669
+ "epoch": 0.51,
2670
+ "learning_rate": 0.0002,
2671
+ "loss": 1.4898,
2672
+ "step": 4440
2673
+ },
2674
+ {
2675
+ "epoch": 0.51,
2676
+ "learning_rate": 0.0002,
2677
+ "loss": 1.1842,
2678
+ "step": 4450
2679
+ },
2680
+ {
2681
+ "epoch": 0.51,
2682
+ "learning_rate": 0.0002,
2683
+ "loss": 1.3801,
2684
+ "step": 4460
2685
+ },
2686
+ {
2687
+ "epoch": 0.51,
2688
+ "learning_rate": 0.0002,
2689
+ "loss": 1.3624,
2690
+ "step": 4470
2691
+ },
2692
+ {
2693
+ "epoch": 0.51,
2694
+ "learning_rate": 0.0002,
2695
+ "loss": 1.5033,
2696
+ "step": 4480
2697
+ },
2698
+ {
2699
+ "epoch": 0.51,
2700
+ "learning_rate": 0.0002,
2701
+ "loss": 1.4971,
2702
+ "step": 4490
2703
+ },
2704
+ {
2705
+ "epoch": 0.51,
2706
+ "learning_rate": 0.0002,
2707
+ "loss": 1.1225,
2708
+ "step": 4500
2709
+ },
2710
+ {
2711
+ "epoch": 0.52,
2712
+ "learning_rate": 0.0002,
2713
+ "loss": 1.3766,
2714
+ "step": 4510
2715
+ },
2716
+ {
2717
+ "epoch": 0.52,
2718
+ "learning_rate": 0.0002,
2719
+ "loss": 1.4046,
2720
+ "step": 4520
2721
+ },
2722
+ {
2723
+ "epoch": 0.52,
2724
+ "learning_rate": 0.0002,
2725
+ "loss": 1.516,
2726
+ "step": 4530
2727
+ },
2728
+ {
2729
+ "epoch": 0.52,
2730
+ "learning_rate": 0.0002,
2731
+ "loss": 1.5033,
2732
+ "step": 4540
2733
+ },
2734
+ {
2735
+ "epoch": 0.52,
2736
+ "learning_rate": 0.0002,
2737
+ "loss": 1.1002,
2738
+ "step": 4550
2739
+ },
2740
+ {
2741
+ "epoch": 0.52,
2742
+ "learning_rate": 0.0002,
2743
+ "loss": 1.3724,
2744
+ "step": 4560
2745
+ },
2746
+ {
2747
+ "epoch": 0.52,
2748
+ "learning_rate": 0.0002,
2749
+ "loss": 1.3755,
2750
+ "step": 4570
2751
+ },
2752
+ {
2753
+ "epoch": 0.52,
2754
+ "learning_rate": 0.0002,
2755
+ "loss": 1.5143,
2756
+ "step": 4580
2757
+ },
2758
+ {
2759
+ "epoch": 0.52,
2760
+ "learning_rate": 0.0002,
2761
+ "loss": 1.5542,
2762
+ "step": 4590
2763
+ },
2764
+ {
2765
+ "epoch": 0.53,
2766
+ "learning_rate": 0.0002,
2767
+ "loss": 1.1288,
2768
+ "step": 4600
2769
+ },
2770
+ {
2771
+ "epoch": 0.53,
2772
+ "learning_rate": 0.0002,
2773
+ "loss": 1.3631,
2774
+ "step": 4610
2775
+ },
2776
+ {
2777
+ "epoch": 0.53,
2778
+ "learning_rate": 0.0002,
2779
+ "loss": 1.4227,
2780
+ "step": 4620
2781
+ },
2782
+ {
2783
+ "epoch": 0.53,
2784
+ "learning_rate": 0.0002,
2785
+ "loss": 1.4712,
2786
+ "step": 4630
2787
+ },
2788
+ {
2789
+ "epoch": 0.53,
2790
+ "learning_rate": 0.0002,
2791
+ "loss": 1.4922,
2792
+ "step": 4640
2793
+ },
2794
+ {
2795
+ "epoch": 0.53,
2796
+ "learning_rate": 0.0002,
2797
+ "loss": 1.2433,
2798
+ "step": 4650
2799
+ },
2800
+ {
2801
+ "epoch": 0.53,
2802
+ "learning_rate": 0.0002,
2803
+ "loss": 1.3936,
2804
+ "step": 4660
2805
+ },
2806
+ {
2807
+ "epoch": 0.53,
2808
+ "learning_rate": 0.0002,
2809
+ "loss": 1.453,
2810
+ "step": 4670
2811
+ },
2812
+ {
2813
+ "epoch": 0.53,
2814
+ "learning_rate": 0.0002,
2815
+ "loss": 1.5069,
2816
+ "step": 4680
2817
+ },
2818
+ {
2819
+ "epoch": 0.54,
2820
+ "learning_rate": 0.0002,
2821
+ "loss": 1.5577,
2822
+ "step": 4690
2823
+ },
2824
+ {
2825
+ "epoch": 0.54,
2826
+ "learning_rate": 0.0002,
2827
+ "loss": 1.174,
2828
+ "step": 4700
2829
+ },
2830
+ {
2831
+ "epoch": 0.54,
2832
+ "learning_rate": 0.0002,
2833
+ "loss": 1.3459,
2834
+ "step": 4710
2835
+ },
2836
+ {
2837
+ "epoch": 0.54,
2838
+ "learning_rate": 0.0002,
2839
+ "loss": 1.3971,
2840
+ "step": 4720
2841
+ },
2842
+ {
2843
+ "epoch": 0.54,
2844
+ "learning_rate": 0.0002,
2845
+ "loss": 1.535,
2846
+ "step": 4730
2847
+ },
2848
+ {
2849
+ "epoch": 0.54,
2850
+ "learning_rate": 0.0002,
2851
+ "loss": 1.5099,
2852
+ "step": 4740
2853
+ },
2854
+ {
2855
+ "epoch": 0.54,
2856
+ "learning_rate": 0.0002,
2857
+ "loss": 1.0992,
2858
+ "step": 4750
2859
+ },
2860
+ {
2861
+ "epoch": 0.54,
2862
+ "learning_rate": 0.0002,
2863
+ "loss": 1.3479,
2864
+ "step": 4760
2865
+ },
2866
+ {
2867
+ "epoch": 0.55,
2868
+ "learning_rate": 0.0002,
2869
+ "loss": 1.3866,
2870
+ "step": 4770
2871
+ },
2872
+ {
2873
+ "epoch": 0.55,
2874
+ "learning_rate": 0.0002,
2875
+ "loss": 1.4732,
2876
+ "step": 4780
2877
+ },
2878
+ {
2879
+ "epoch": 0.55,
2880
+ "learning_rate": 0.0002,
2881
+ "loss": 1.5125,
2882
+ "step": 4790
2883
+ },
2884
+ {
2885
+ "epoch": 0.55,
2886
+ "learning_rate": 0.0002,
2887
+ "loss": 1.0968,
2888
+ "step": 4800
2889
+ },
2890
+ {
2891
+ "epoch": 0.55,
2892
+ "learning_rate": 0.0002,
2893
+ "loss": 1.3876,
2894
+ "step": 4810
2895
+ },
2896
+ {
2897
+ "epoch": 0.55,
2898
+ "learning_rate": 0.0002,
2899
+ "loss": 1.4057,
2900
+ "step": 4820
2901
+ },
2902
+ {
2903
+ "epoch": 0.55,
2904
+ "learning_rate": 0.0002,
2905
+ "loss": 1.5399,
2906
+ "step": 4830
2907
+ },
2908
+ {
2909
+ "epoch": 0.55,
2910
+ "learning_rate": 0.0002,
2911
+ "loss": 1.4995,
2912
+ "step": 4840
2913
+ },
2914
+ {
2915
+ "epoch": 0.55,
2916
+ "learning_rate": 0.0002,
2917
+ "loss": 1.1411,
2918
+ "step": 4850
2919
+ },
2920
+ {
2921
+ "epoch": 0.56,
2922
+ "learning_rate": 0.0002,
2923
+ "loss": 1.4131,
2924
+ "step": 4860
2925
+ },
2926
+ {
2927
+ "epoch": 0.56,
2928
+ "learning_rate": 0.0002,
2929
+ "loss": 1.4054,
2930
+ "step": 4870
2931
+ },
2932
+ {
2933
+ "epoch": 0.56,
2934
+ "learning_rate": 0.0002,
2935
+ "loss": 1.4959,
2936
+ "step": 4880
2937
+ },
2938
+ {
2939
+ "epoch": 0.56,
2940
+ "learning_rate": 0.0002,
2941
+ "loss": 1.5191,
2942
+ "step": 4890
2943
+ },
2944
+ {
2945
+ "epoch": 0.56,
2946
+ "learning_rate": 0.0002,
2947
+ "loss": 1.1974,
2948
+ "step": 4900
2949
+ },
2950
+ {
2951
+ "epoch": 0.56,
2952
+ "learning_rate": 0.0002,
2953
+ "loss": 1.4629,
2954
+ "step": 4910
2955
+ },
2956
+ {
2957
+ "epoch": 0.56,
2958
+ "learning_rate": 0.0002,
2959
+ "loss": 1.6002,
2960
+ "step": 4920
2961
+ },
2962
+ {
2963
+ "epoch": 0.56,
2964
+ "learning_rate": 0.0002,
2965
+ "loss": 1.6047,
2966
+ "step": 4930
2967
+ },
2968
+ {
2969
+ "epoch": 0.56,
2970
+ "learning_rate": 0.0002,
2971
+ "loss": 1.6153,
2972
+ "step": 4940
2973
+ },
2974
+ {
2975
+ "epoch": 0.57,
2976
+ "learning_rate": 0.0002,
2977
+ "loss": 1.2109,
2978
+ "step": 4950
2979
+ },
2980
+ {
2981
+ "epoch": 0.57,
2982
+ "learning_rate": 0.0002,
2983
+ "loss": 1.4794,
2984
+ "step": 4960
2985
+ },
2986
+ {
2987
+ "epoch": 0.57,
2988
+ "learning_rate": 0.0002,
2989
+ "loss": 1.5028,
2990
+ "step": 4970
2991
+ },
2992
+ {
2993
+ "epoch": 0.57,
2994
+ "learning_rate": 0.0002,
2995
+ "loss": 1.6208,
2996
+ "step": 4980
2997
+ },
2998
+ {
2999
+ "epoch": 0.57,
3000
+ "learning_rate": 0.0002,
3001
+ "loss": 1.6215,
3002
+ "step": 4990
3003
+ },
3004
+ {
3005
+ "epoch": 0.57,
3006
+ "learning_rate": 0.0002,
3007
+ "loss": 1.2066,
3008
+ "step": 5000
3009
+ },
3010
+ {
3011
+ "epoch": 0.57,
3012
+ "learning_rate": 0.0002,
3013
+ "loss": 1.592,
3014
+ "step": 5010
3015
+ },
3016
+ {
3017
+ "epoch": 0.57,
3018
+ "learning_rate": 0.0002,
3019
+ "loss": 1.5682,
3020
+ "step": 5020
3021
+ },
3022
+ {
3023
+ "epoch": 0.57,
3024
+ "learning_rate": 0.0002,
3025
+ "loss": 1.6461,
3026
+ "step": 5030
3027
+ },
3028
+ {
3029
+ "epoch": 0.58,
3030
+ "learning_rate": 0.0002,
3031
+ "loss": 1.6425,
3032
+ "step": 5040
3033
+ },
3034
+ {
3035
+ "epoch": 0.58,
3036
+ "learning_rate": 0.0002,
3037
+ "loss": 1.2823,
3038
+ "step": 5050
3039
+ },
3040
+ {
3041
+ "epoch": 0.58,
3042
+ "learning_rate": 0.0002,
3043
+ "loss": 1.5915,
3044
+ "step": 5060
3045
+ },
3046
+ {
3047
+ "epoch": 0.58,
3048
+ "learning_rate": 0.0002,
3049
+ "loss": 1.5673,
3050
+ "step": 5070
3051
+ },
3052
+ {
3053
+ "epoch": 0.58,
3054
+ "learning_rate": 0.0002,
3055
+ "loss": 1.724,
3056
+ "step": 5080
3057
+ },
3058
+ {
3059
+ "epoch": 0.58,
3060
+ "learning_rate": 0.0002,
3061
+ "loss": 1.7876,
3062
+ "step": 5090
3063
+ },
3064
+ {
3065
+ "epoch": 0.58,
3066
+ "learning_rate": 0.0002,
3067
+ "loss": 1.3042,
3068
+ "step": 5100
3069
+ },
3070
+ {
3071
+ "epoch": 0.58,
3072
+ "learning_rate": 0.0002,
3073
+ "loss": 1.8139,
3074
+ "step": 5110
3075
+ },
3076
+ {
3077
+ "epoch": 0.59,
3078
+ "learning_rate": 0.0002,
3079
+ "loss": 1.8781,
3080
+ "step": 5120
3081
+ },
3082
+ {
3083
+ "epoch": 0.59,
3084
+ "learning_rate": 0.0002,
3085
+ "loss": 2.0134,
3086
+ "step": 5130
3087
+ },
3088
+ {
3089
+ "epoch": 0.59,
3090
+ "learning_rate": 0.0002,
3091
+ "loss": 1.9675,
3092
+ "step": 5140
3093
+ },
3094
+ {
3095
+ "epoch": 0.59,
3096
+ "learning_rate": 0.0002,
3097
+ "loss": 1.3781,
3098
+ "step": 5150
3099
+ },
3100
+ {
3101
+ "epoch": 0.59,
3102
+ "learning_rate": 0.0002,
3103
+ "loss": 1.8037,
3104
+ "step": 5160
3105
+ },
3106
+ {
3107
+ "epoch": 0.59,
3108
+ "learning_rate": 0.0002,
3109
+ "loss": 1.9102,
3110
+ "step": 5170
3111
+ },
3112
+ {
3113
+ "epoch": 0.59,
3114
+ "learning_rate": 0.0002,
3115
+ "loss": 2.002,
3116
+ "step": 5180
3117
+ },
3118
+ {
3119
+ "epoch": 0.59,
3120
+ "learning_rate": 0.0002,
3121
+ "loss": 1.9065,
3122
+ "step": 5190
3123
+ },
3124
+ {
3125
+ "epoch": 0.59,
3126
+ "learning_rate": 0.0002,
3127
+ "loss": 1.4923,
3128
+ "step": 5200
3129
+ },
3130
+ {
3131
+ "epoch": 0.6,
3132
+ "learning_rate": 0.0002,
3133
+ "loss": 1.6716,
3134
+ "step": 5210
3135
+ },
3136
+ {
3137
+ "epoch": 0.6,
3138
+ "learning_rate": 0.0002,
3139
+ "loss": 1.664,
3140
+ "step": 5220
3141
+ },
3142
+ {
3143
+ "epoch": 0.6,
3144
+ "learning_rate": 0.0002,
3145
+ "loss": 1.6999,
3146
+ "step": 5230
3147
+ },
3148
+ {
3149
+ "epoch": 0.6,
3150
+ "learning_rate": 0.0002,
3151
+ "loss": 1.7727,
3152
+ "step": 5240
3153
+ },
3154
+ {
3155
+ "epoch": 0.6,
3156
+ "learning_rate": 0.0002,
3157
+ "loss": 1.4808,
3158
+ "step": 5250
3159
+ },
3160
+ {
3161
+ "epoch": 0.6,
3162
+ "learning_rate": 0.0002,
3163
+ "loss": 1.6987,
3164
+ "step": 5260
3165
+ },
3166
+ {
3167
+ "epoch": 0.6,
3168
+ "learning_rate": 0.0002,
3169
+ "loss": 1.8903,
3170
+ "step": 5270
3171
+ },
3172
+ {
3173
+ "epoch": 0.6,
3174
+ "learning_rate": 0.0002,
3175
+ "loss": 2.0523,
3176
+ "step": 5280
3177
+ },
3178
+ {
3179
+ "epoch": 0.6,
3180
+ "learning_rate": 0.0002,
3181
+ "loss": 1.9036,
3182
+ "step": 5290
3183
+ },
3184
+ {
3185
+ "epoch": 0.61,
3186
+ "learning_rate": 0.0002,
3187
+ "loss": 1.536,
3188
+ "step": 5300
3189
+ },
3190
+ {
3191
+ "epoch": 0.61,
3192
+ "learning_rate": 0.0002,
3193
+ "loss": 1.8511,
3194
+ "step": 5310
3195
+ },
3196
+ {
3197
+ "epoch": 0.61,
3198
+ "learning_rate": 0.0002,
3199
+ "loss": 1.8459,
3200
+ "step": 5320
3201
+ },
3202
+ {
3203
+ "epoch": 0.61,
3204
+ "learning_rate": 0.0002,
3205
+ "loss": 1.9724,
3206
+ "step": 5330
3207
+ },
3208
+ {
3209
+ "epoch": 0.61,
3210
+ "learning_rate": 0.0002,
3211
+ "loss": 1.9299,
3212
+ "step": 5340
3213
+ },
3214
+ {
3215
+ "epoch": 0.61,
3216
+ "learning_rate": 0.0002,
3217
+ "loss": 1.6449,
3218
+ "step": 5350
3219
+ },
3220
+ {
3221
+ "epoch": 0.61,
3222
+ "learning_rate": 0.0002,
3223
+ "loss": 1.7879,
3224
+ "step": 5360
3225
+ },
3226
+ {
3227
+ "epoch": 0.61,
3228
+ "learning_rate": 0.0002,
3229
+ "loss": 1.7839,
3230
+ "step": 5370
3231
+ },
3232
+ {
3233
+ "epoch": 0.61,
3234
+ "learning_rate": 0.0002,
3235
+ "loss": 1.8283,
3236
+ "step": 5380
3237
+ },
3238
+ {
3239
+ "epoch": 0.62,
3240
+ "learning_rate": 0.0002,
3241
+ "loss": 1.8055,
3242
+ "step": 5390
3243
+ },
3244
+ {
3245
+ "epoch": 0.62,
3246
+ "learning_rate": 0.0002,
3247
+ "loss": 1.4122,
3248
+ "step": 5400
3249
+ },
3250
+ {
3251
+ "epoch": 0.62,
3252
+ "learning_rate": 0.0002,
3253
+ "loss": 1.6398,
3254
+ "step": 5410
3255
+ },
3256
+ {
3257
+ "epoch": 0.62,
3258
+ "learning_rate": 0.0002,
3259
+ "loss": 1.6537,
3260
+ "step": 5420
3261
+ },
3262
+ {
3263
+ "epoch": 0.62,
3264
+ "learning_rate": 0.0002,
3265
+ "loss": 1.6973,
3266
+ "step": 5430
3267
+ },
3268
+ {
3269
+ "epoch": 0.62,
3270
+ "learning_rate": 0.0002,
3271
+ "loss": 1.8209,
3272
+ "step": 5440
3273
+ },
3274
+ {
3275
+ "epoch": 0.62,
3276
+ "learning_rate": 0.0002,
3277
+ "loss": 1.6343,
3278
+ "step": 5450
3279
+ },
3280
+ {
3281
+ "epoch": 0.62,
3282
+ "learning_rate": 0.0002,
3283
+ "loss": 2.0785,
3284
+ "step": 5460
3285
+ },
3286
+ {
3287
+ "epoch": 0.63,
3288
+ "learning_rate": 0.0002,
3289
+ "loss": 4.1734,
3290
+ "step": 5470
3291
+ },
3292
+ {
3293
+ "epoch": 0.63,
3294
+ "learning_rate": 0.0002,
3295
+ "loss": 6.7746,
3296
+ "step": 5480
3297
+ },
3298
+ {
3299
+ "epoch": 0.63,
3300
+ "learning_rate": 0.0002,
3301
+ "loss": 5.0767,
3302
+ "step": 5490
3303
+ },
3304
+ {
3305
+ "epoch": 0.63,
3306
+ "learning_rate": 0.0002,
3307
+ "loss": 3.3103,
3308
+ "step": 5500
3309
+ },
3310
+ {
3311
+ "epoch": 0.63,
3312
+ "learning_rate": 0.0002,
3313
+ "loss": 4.5979,
3314
+ "step": 5510
3315
+ },
3316
+ {
3317
+ "epoch": 0.63,
3318
+ "learning_rate": 0.0002,
3319
+ "loss": 4.3999,
3320
+ "step": 5520
3321
+ },
3322
+ {
3323
+ "epoch": 0.63,
3324
+ "learning_rate": 0.0002,
3325
+ "loss": 4.0496,
3326
+ "step": 5530
3327
+ },
3328
+ {
3329
+ "epoch": 0.63,
3330
+ "learning_rate": 0.0002,
3331
+ "loss": 4.2599,
3332
+ "step": 5540
3333
+ },
3334
+ {
3335
+ "epoch": 0.63,
3336
+ "learning_rate": 0.0002,
3337
+ "loss": 3.6741,
3338
+ "step": 5550
3339
+ },
3340
+ {
3341
+ "epoch": 0.64,
3342
+ "learning_rate": 0.0002,
3343
+ "loss": 4.7093,
3344
+ "step": 5560
3345
+ },
3346
+ {
3347
+ "epoch": 0.64,
3348
+ "learning_rate": 0.0002,
3349
+ "loss": 4.016,
3350
+ "step": 5570
3351
+ },
3352
+ {
3353
+ "epoch": 0.64,
3354
+ "learning_rate": 0.0002,
3355
+ "loss": 4.2352,
3356
+ "step": 5580
3357
+ },
3358
+ {
3359
+ "epoch": 0.64,
3360
+ "learning_rate": 0.0002,
3361
+ "loss": 3.8359,
3362
+ "step": 5590
3363
+ },
3364
+ {
3365
+ "epoch": 0.64,
3366
+ "learning_rate": 0.0002,
3367
+ "loss": 2.7654,
3368
+ "step": 5600
3369
+ },
3370
+ {
3371
+ "epoch": 0.64,
3372
+ "learning_rate": 0.0002,
3373
+ "loss": 4.6375,
3374
+ "step": 5610
3375
+ },
3376
+ {
3377
+ "epoch": 0.64,
3378
+ "learning_rate": 0.0002,
3379
+ "loss": 4.5832,
3380
+ "step": 5620
3381
+ },
3382
+ {
3383
+ "epoch": 0.64,
3384
+ "learning_rate": 0.0002,
3385
+ "loss": 3.7359,
3386
+ "step": 5630
3387
+ },
3388
+ {
3389
+ "epoch": 0.64,
3390
+ "learning_rate": 0.0002,
3391
+ "loss": 3.3395,
3392
+ "step": 5640
3393
+ },
3394
+ {
3395
+ "epoch": 0.65,
3396
+ "learning_rate": 0.0002,
3397
+ "loss": 2.6695,
3398
+ "step": 5650
3399
+ },
3400
+ {
3401
+ "epoch": 0.65,
3402
+ "learning_rate": 0.0002,
3403
+ "loss": 4.6645,
3404
+ "step": 5660
3405
+ },
3406
+ {
3407
+ "epoch": 0.65,
3408
+ "learning_rate": 0.0002,
3409
+ "loss": 4.7006,
3410
+ "step": 5670
3411
+ },
3412
+ {
3413
+ "epoch": 0.65,
3414
+ "learning_rate": 0.0002,
3415
+ "loss": 4.377,
3416
+ "step": 5680
3417
+ },
3418
+ {
3419
+ "epoch": 0.65,
3420
+ "learning_rate": 0.0002,
3421
+ "loss": 3.8076,
3422
+ "step": 5690
3423
+ },
3424
+ {
3425
+ "epoch": 0.65,
3426
+ "learning_rate": 0.0002,
3427
+ "loss": 2.5118,
3428
+ "step": 5700
3429
+ },
3430
+ {
3431
+ "epoch": 0.65,
3432
+ "learning_rate": 0.0002,
3433
+ "loss": 3.2971,
3434
+ "step": 5710
3435
+ },
3436
+ {
3437
+ "epoch": 0.65,
3438
+ "learning_rate": 0.0002,
3439
+ "loss": 3.4954,
3440
+ "step": 5720
3441
+ },
3442
+ {
3443
+ "epoch": 0.65,
3444
+ "learning_rate": 0.0002,
3445
+ "loss": 4.1866,
3446
+ "step": 5730
3447
+ },
3448
+ {
3449
+ "epoch": 0.66,
3450
+ "learning_rate": 0.0002,
3451
+ "loss": 4.3862,
3452
+ "step": 5740
3453
+ },
3454
+ {
3455
+ "epoch": 0.66,
3456
+ "learning_rate": 0.0002,
3457
+ "loss": 2.9591,
3458
+ "step": 5750
3459
+ },
3460
+ {
3461
+ "epoch": 0.66,
3462
+ "learning_rate": 0.0002,
3463
+ "loss": 4.2781,
3464
+ "step": 5760
3465
+ },
3466
+ {
3467
+ "epoch": 0.66,
3468
+ "learning_rate": 0.0002,
3469
+ "loss": 3.9847,
3470
+ "step": 5770
3471
+ },
3472
+ {
3473
+ "epoch": 0.66,
3474
+ "learning_rate": 0.0002,
3475
+ "loss": 3.802,
3476
+ "step": 5780
3477
+ },
3478
+ {
3479
+ "epoch": 0.66,
3480
+ "learning_rate": 0.0002,
3481
+ "loss": 2.9336,
3482
+ "step": 5790
3483
+ },
3484
+ {
3485
+ "epoch": 0.66,
3486
+ "learning_rate": 0.0002,
3487
+ "loss": 2.183,
3488
+ "step": 5800
3489
+ },
3490
+ {
3491
+ "epoch": 0.66,
3492
+ "learning_rate": 0.0002,
3493
+ "loss": 3.3075,
3494
+ "step": 5810
3495
+ },
3496
+ {
3497
+ "epoch": 0.67,
3498
+ "learning_rate": 0.0002,
3499
+ "loss": 3.453,
3500
+ "step": 5820
3501
+ },
3502
+ {
3503
+ "epoch": 0.67,
3504
+ "learning_rate": 0.0002,
3505
+ "loss": 3.3029,
3506
+ "step": 5830
3507
+ },
3508
+ {
3509
+ "epoch": 0.67,
3510
+ "learning_rate": 0.0002,
3511
+ "loss": 2.809,
3512
+ "step": 5840
3513
+ },
3514
+ {
3515
+ "epoch": 0.67,
3516
+ "learning_rate": 0.0002,
3517
+ "loss": 1.8697,
3518
+ "step": 5850
3519
+ },
3520
+ {
3521
+ "epoch": 0.67,
3522
+ "learning_rate": 0.0002,
3523
+ "loss": 2.76,
3524
+ "step": 5860
3525
+ },
3526
+ {
3527
+ "epoch": 0.67,
3528
+ "learning_rate": 0.0002,
3529
+ "loss": 2.9094,
3530
+ "step": 5870
3531
+ },
3532
+ {
3533
+ "epoch": 0.67,
3534
+ "learning_rate": 0.0002,
3535
+ "loss": 2.7784,
3536
+ "step": 5880
3537
+ },
3538
+ {
3539
+ "epoch": 0.67,
3540
+ "learning_rate": 0.0002,
3541
+ "loss": 2.5071,
3542
+ "step": 5890
3543
+ },
3544
+ {
3545
+ "epoch": 0.67,
3546
+ "learning_rate": 0.0002,
3547
+ "loss": 1.8809,
3548
+ "step": 5900
3549
+ },
3550
+ {
3551
+ "epoch": 0.68,
3552
+ "learning_rate": 0.0002,
3553
+ "loss": 2.459,
3554
+ "step": 5910
3555
+ },
3556
+ {
3557
+ "epoch": 0.68,
3558
+ "learning_rate": 0.0002,
3559
+ "loss": 2.8066,
3560
+ "step": 5920
3561
+ },
3562
+ {
3563
+ "epoch": 0.68,
3564
+ "learning_rate": 0.0002,
3565
+ "loss": 4.1141,
3566
+ "step": 5930
3567
+ },
3568
+ {
3569
+ "epoch": 0.68,
3570
+ "learning_rate": 0.0002,
3571
+ "loss": 4.0347,
3572
+ "step": 5940
3573
+ },
3574
+ {
3575
+ "epoch": 0.68,
3576
+ "learning_rate": 0.0002,
3577
+ "loss": 3.0017,
3578
+ "step": 5950
3579
+ },
3580
+ {
3581
+ "epoch": 0.68,
3582
+ "learning_rate": 0.0002,
3583
+ "loss": 3.2676,
3584
+ "step": 5960
3585
+ },
3586
+ {
3587
+ "epoch": 0.68,
3588
+ "learning_rate": 0.0002,
3589
+ "loss": 3.1042,
3590
+ "step": 5970
3591
+ },
3592
+ {
3593
+ "epoch": 0.68,
3594
+ "learning_rate": 0.0002,
3595
+ "loss": 3.2889,
3596
+ "step": 5980
3597
+ },
3598
+ {
3599
+ "epoch": 0.68,
3600
+ "learning_rate": 0.0002,
3601
+ "loss": 3.3139,
3602
+ "step": 5990
3603
+ },
3604
+ {
3605
+ "epoch": 0.69,
3606
+ "learning_rate": 0.0002,
3607
+ "loss": 2.4935,
3608
+ "step": 6000
3609
+ },
3610
+ {
3611
+ "epoch": 0.69,
3612
+ "learning_rate": 0.0002,
3613
+ "loss": 4.1095,
3614
+ "step": 6010
3615
+ },
3616
+ {
3617
+ "epoch": 0.69,
3618
+ "learning_rate": 0.0002,
3619
+ "loss": 3.6322,
3620
+ "step": 6020
3621
+ },
3622
+ {
3623
+ "epoch": 0.69,
3624
+ "learning_rate": 0.0002,
3625
+ "loss": 3.3309,
3626
+ "step": 6030
3627
+ },
3628
+ {
3629
+ "epoch": 0.69,
3630
+ "learning_rate": 0.0002,
3631
+ "loss": 3.0428,
3632
+ "step": 6040
3633
+ },
3634
+ {
3635
+ "epoch": 0.69,
3636
+ "learning_rate": 0.0002,
3637
+ "loss": 2.4984,
3638
+ "step": 6050
3639
+ },
3640
+ {
3641
+ "epoch": 0.69,
3642
+ "learning_rate": 0.0002,
3643
+ "loss": 3.5314,
3644
+ "step": 6060
3645
+ },
3646
+ {
3647
+ "epoch": 0.69,
3648
+ "learning_rate": 0.0002,
3649
+ "loss": 3.6307,
3650
+ "step": 6070
3651
+ },
3652
+ {
3653
+ "epoch": 0.69,
3654
+ "learning_rate": 0.0002,
3655
+ "loss": 3.2909,
3656
+ "step": 6080
3657
+ },
3658
+ {
3659
+ "epoch": 0.7,
3660
+ "learning_rate": 0.0002,
3661
+ "loss": 2.8702,
3662
+ "step": 6090
3663
+ },
3664
+ {
3665
+ "epoch": 0.7,
3666
+ "learning_rate": 0.0002,
3667
+ "loss": 2.2269,
3668
+ "step": 6100
3669
+ },
3670
+ {
3671
+ "epoch": 0.7,
3672
+ "learning_rate": 0.0002,
3673
+ "loss": 3.3475,
3674
+ "step": 6110
3675
+ },
3676
+ {
3677
+ "epoch": 0.7,
3678
+ "learning_rate": 0.0002,
3679
+ "loss": 4.0399,
3680
+ "step": 6120
3681
+ },
3682
+ {
3683
+ "epoch": 0.7,
3684
+ "learning_rate": 0.0002,
3685
+ "loss": 5.27,
3686
+ "step": 6130
3687
+ },
3688
+ {
3689
+ "epoch": 0.7,
3690
+ "learning_rate": 0.0002,
3691
+ "loss": 4.9144,
3692
+ "step": 6140
3693
+ },
3694
+ {
3695
+ "epoch": 0.7,
3696
+ "learning_rate": 0.0002,
3697
+ "loss": 3.0834,
3698
+ "step": 6150
3699
+ },
3700
+ {
3701
+ "epoch": 0.7,
3702
+ "learning_rate": 0.0002,
3703
+ "loss": 3.7826,
3704
+ "step": 6160
3705
+ },
3706
+ {
3707
+ "epoch": 0.71,
3708
+ "learning_rate": 0.0002,
3709
+ "loss": 3.4911,
3710
+ "step": 6170
3711
+ },
3712
+ {
3713
+ "epoch": 0.71,
3714
+ "learning_rate": 0.0002,
3715
+ "loss": 3.2003,
3716
+ "step": 6180
3717
+ },
3718
+ {
3719
+ "epoch": 0.71,
3720
+ "learning_rate": 0.0002,
3721
+ "loss": 2.9853,
3722
+ "step": 6190
3723
+ },
3724
+ {
3725
+ "epoch": 0.71,
3726
+ "learning_rate": 0.0002,
3727
+ "loss": 2.1288,
3728
+ "step": 6200
3729
+ },
3730
+ {
3731
+ "epoch": 0.71,
3732
+ "learning_rate": 0.0002,
3733
+ "loss": 3.1077,
3734
+ "step": 6210
3735
+ },
3736
+ {
3737
+ "epoch": 0.71,
3738
+ "learning_rate": 0.0002,
3739
+ "loss": 3.065,
3740
+ "step": 6220
3741
+ },
3742
+ {
3743
+ "epoch": 0.71,
3744
+ "learning_rate": 0.0002,
3745
+ "loss": 2.753,
3746
+ "step": 6230
3747
+ },
3748
+ {
3749
+ "epoch": 0.71,
3750
+ "learning_rate": 0.0002,
3751
+ "loss": 2.6375,
3752
+ "step": 6240
3753
+ },
3754
+ {
3755
+ "epoch": 0.71,
3756
+ "learning_rate": 0.0002,
3757
+ "loss": 2.1715,
3758
+ "step": 6250
3759
+ },
3760
+ {
3761
+ "epoch": 0.72,
3762
+ "learning_rate": 0.0002,
3763
+ "loss": 3.2288,
3764
+ "step": 6260
3765
+ },
3766
+ {
3767
+ "epoch": 0.72,
3768
+ "learning_rate": 0.0002,
3769
+ "loss": 3.2531,
3770
+ "step": 6270
3771
+ },
3772
+ {
3773
+ "epoch": 0.72,
3774
+ "learning_rate": 0.0002,
3775
+ "loss": 3.1586,
3776
+ "step": 6280
3777
+ },
3778
+ {
3779
+ "epoch": 0.72,
3780
+ "learning_rate": 0.0002,
3781
+ "loss": 3.3152,
3782
+ "step": 6290
3783
+ },
3784
+ {
3785
+ "epoch": 0.72,
3786
+ "learning_rate": 0.0002,
3787
+ "loss": 2.4145,
3788
+ "step": 6300
3789
+ },
3790
+ {
3791
+ "epoch": 0.72,
3792
+ "learning_rate": 0.0002,
3793
+ "loss": 4.0466,
3794
+ "step": 6310
3795
+ },
3796
+ {
3797
+ "epoch": 0.72,
3798
+ "learning_rate": 0.0002,
3799
+ "loss": 4.0243,
3800
+ "step": 6320
3801
+ },
3802
+ {
3803
+ "epoch": 0.72,
3804
+ "learning_rate": 0.0002,
3805
+ "loss": 3.522,
3806
+ "step": 6330
3807
+ },
3808
+ {
3809
+ "epoch": 0.72,
3810
+ "learning_rate": 0.0002,
3811
+ "loss": 3.106,
3812
+ "step": 6340
3813
+ },
3814
+ {
3815
+ "epoch": 0.73,
3816
+ "learning_rate": 0.0002,
3817
+ "loss": 2.358,
3818
+ "step": 6350
3819
+ },
3820
+ {
3821
+ "epoch": 0.73,
3822
+ "learning_rate": 0.0002,
3823
+ "loss": 4.1717,
3824
+ "step": 6360
3825
+ },
3826
+ {
3827
+ "epoch": 0.73,
3828
+ "learning_rate": 0.0002,
3829
+ "loss": 4.3043,
3830
+ "step": 6370
3831
+ },
3832
+ {
3833
+ "epoch": 0.73,
3834
+ "learning_rate": 0.0002,
3835
+ "loss": 4.1694,
3836
+ "step": 6380
3837
+ },
3838
+ {
3839
+ "epoch": 0.73,
3840
+ "learning_rate": 0.0002,
3841
+ "loss": 3.9152,
3842
+ "step": 6390
3843
+ },
3844
+ {
3845
+ "epoch": 0.73,
3846
+ "learning_rate": 0.0002,
3847
+ "loss": 2.806,
3848
+ "step": 6400
3849
+ },
3850
+ {
3851
+ "epoch": 0.73,
3852
+ "learning_rate": 0.0002,
3853
+ "loss": 5.7838,
3854
+ "step": 6410
3855
+ },
3856
+ {
3857
+ "epoch": 0.73,
3858
+ "learning_rate": 0.0002,
3859
+ "loss": 5.6406,
3860
+ "step": 6420
3861
+ },
3862
+ {
3863
+ "epoch": 0.73,
3864
+ "learning_rate": 0.0002,
3865
+ "loss": 5.0501,
3866
+ "step": 6430
3867
+ },
3868
+ {
3869
+ "epoch": 0.74,
3870
+ "learning_rate": 0.0002,
3871
+ "loss": 4.2304,
3872
+ "step": 6440
3873
+ },
3874
+ {
3875
+ "epoch": 0.74,
3876
+ "learning_rate": 0.0002,
3877
+ "loss": 2.7792,
3878
+ "step": 6450
3879
+ },
3880
+ {
3881
+ "epoch": 0.74,
3882
+ "learning_rate": 0.0002,
3883
+ "loss": 4.368,
3884
+ "step": 6460
3885
+ },
3886
+ {
3887
+ "epoch": 0.74,
3888
+ "learning_rate": 0.0002,
3889
+ "loss": 3.9682,
3890
+ "step": 6470
3891
+ },
3892
+ {
3893
+ "epoch": 0.74,
3894
+ "learning_rate": 0.0002,
3895
+ "loss": 3.7431,
3896
+ "step": 6480
3897
+ },
3898
+ {
3899
+ "epoch": 0.74,
3900
+ "learning_rate": 0.0002,
3901
+ "loss": 3.6235,
3902
+ "step": 6490
3903
+ },
3904
+ {
3905
+ "epoch": 0.74,
3906
+ "learning_rate": 0.0002,
3907
+ "loss": 2.5896,
3908
+ "step": 6500
3909
+ },
3910
+ {
3911
+ "epoch": 0.74,
3912
+ "learning_rate": 0.0002,
3913
+ "loss": 3.8402,
3914
+ "step": 6510
3915
+ },
3916
+ {
3917
+ "epoch": 0.75,
3918
+ "learning_rate": 0.0002,
3919
+ "loss": 3.4356,
3920
+ "step": 6520
3921
+ },
3922
+ {
3923
+ "epoch": 0.75,
3924
+ "learning_rate": 0.0002,
3925
+ "loss": 3.0214,
3926
+ "step": 6530
3927
+ },
3928
+ {
3929
+ "epoch": 0.75,
3930
+ "learning_rate": 0.0002,
3931
+ "loss": 2.7369,
3932
+ "step": 6540
3933
+ },
3934
+ {
3935
+ "epoch": 0.75,
3936
+ "learning_rate": 0.0002,
3937
+ "loss": 2.0243,
3938
+ "step": 6550
3939
+ },
3940
+ {
3941
+ "epoch": 0.75,
3942
+ "learning_rate": 0.0002,
3943
+ "loss": 3.0303,
3944
+ "step": 6560
3945
+ },
3946
+ {
3947
+ "epoch": 0.75,
3948
+ "learning_rate": 0.0002,
3949
+ "loss": 2.9176,
3950
+ "step": 6570
3951
+ },
3952
+ {
3953
+ "epoch": 0.75,
3954
+ "learning_rate": 0.0002,
3955
+ "loss": 2.8165,
3956
+ "step": 6580
3957
+ },
3958
+ {
3959
+ "epoch": 0.75,
3960
+ "learning_rate": 0.0002,
3961
+ "loss": 2.711,
3962
+ "step": 6590
3963
+ },
3964
+ {
3965
+ "epoch": 0.75,
3966
+ "learning_rate": 0.0002,
3967
+ "loss": 2.1248,
3968
+ "step": 6600
3969
+ },
3970
+ {
3971
+ "epoch": 0.76,
3972
+ "learning_rate": 0.0002,
3973
+ "loss": 3.3661,
3974
+ "step": 6610
3975
+ },
3976
+ {
3977
+ "epoch": 0.76,
3978
+ "learning_rate": 0.0002,
3979
+ "loss": 3.2783,
3980
+ "step": 6620
3981
+ },
3982
+ {
3983
+ "epoch": 0.76,
3984
+ "learning_rate": 0.0002,
3985
+ "loss": 3.0595,
3986
+ "step": 6630
3987
+ },
3988
+ {
3989
+ "epoch": 0.76,
3990
+ "learning_rate": 0.0002,
3991
+ "loss": 2.9547,
3992
+ "step": 6640
3993
+ },
3994
+ {
3995
+ "epoch": 0.76,
3996
+ "learning_rate": 0.0002,
3997
+ "loss": 2.3153,
3998
+ "step": 6650
3999
+ },
4000
+ {
4001
+ "epoch": 0.76,
4002
+ "learning_rate": 0.0002,
4003
+ "loss": 3.3942,
4004
+ "step": 6660
4005
+ },
4006
+ {
4007
+ "epoch": 0.76,
4008
+ "learning_rate": 0.0002,
4009
+ "loss": 3.2098,
4010
+ "step": 6670
4011
+ },
4012
+ {
4013
+ "epoch": 0.76,
4014
+ "learning_rate": 0.0002,
4015
+ "loss": 3.134,
4016
+ "step": 6680
4017
+ },
4018
+ {
4019
+ "epoch": 0.76,
4020
+ "learning_rate": 0.0002,
4021
+ "loss": 3.0744,
4022
+ "step": 6690
4023
+ },
4024
+ {
4025
+ "epoch": 0.77,
4026
+ "learning_rate": 0.0002,
4027
+ "loss": 2.702,
4028
+ "step": 6700
4029
+ },
4030
+ {
4031
+ "epoch": 0.77,
4032
+ "learning_rate": 0.0002,
4033
+ "loss": 3.6413,
4034
+ "step": 6710
4035
+ },
4036
+ {
4037
+ "epoch": 0.77,
4038
+ "learning_rate": 0.0002,
4039
+ "loss": 3.7017,
4040
+ "step": 6720
4041
+ },
4042
+ {
4043
+ "epoch": 0.77,
4044
+ "learning_rate": 0.0002,
4045
+ "loss": 3.601,
4046
+ "step": 6730
4047
+ },
4048
+ {
4049
+ "epoch": 0.77,
4050
+ "learning_rate": 0.0002,
4051
+ "loss": 3.2477,
4052
+ "step": 6740
4053
+ },
4054
+ {
4055
+ "epoch": 0.77,
4056
+ "learning_rate": 0.0002,
4057
+ "loss": 2.5295,
4058
+ "step": 6750
4059
+ },
4060
+ {
4061
+ "epoch": 0.77,
4062
+ "learning_rate": 0.0002,
4063
+ "loss": 3.6857,
4064
+ "step": 6760
4065
+ },
4066
+ {
4067
+ "epoch": 0.77,
4068
+ "learning_rate": 0.0002,
4069
+ "loss": 3.6193,
4070
+ "step": 6770
4071
+ },
4072
+ {
4073
+ "epoch": 0.77,
4074
+ "learning_rate": 0.0002,
4075
+ "loss": 3.4057,
4076
+ "step": 6780
4077
+ },
4078
+ {
4079
+ "epoch": 0.78,
4080
+ "learning_rate": 0.0002,
4081
+ "loss": 3.1279,
4082
+ "step": 6790
4083
+ },
4084
+ {
4085
+ "epoch": 0.78,
4086
+ "learning_rate": 0.0002,
4087
+ "loss": 2.3629,
4088
+ "step": 6800
4089
+ },
4090
+ {
4091
+ "epoch": 0.78,
4092
+ "learning_rate": 0.0002,
4093
+ "loss": 3.3246,
4094
+ "step": 6810
4095
+ },
4096
+ {
4097
+ "epoch": 0.78,
4098
+ "learning_rate": 0.0002,
4099
+ "loss": 3.1549,
4100
+ "step": 6820
4101
+ },
4102
+ {
4103
+ "epoch": 0.78,
4104
+ "learning_rate": 0.0002,
4105
+ "loss": 2.9732,
4106
+ "step": 6830
4107
+ },
4108
+ {
4109
+ "epoch": 0.78,
4110
+ "learning_rate": 0.0002,
4111
+ "loss": 2.7179,
4112
+ "step": 6840
4113
+ },
4114
+ {
4115
+ "epoch": 0.78,
4116
+ "learning_rate": 0.0002,
4117
+ "loss": 2.1229,
4118
+ "step": 6850
4119
+ },
4120
+ {
4121
+ "epoch": 0.78,
4122
+ "learning_rate": 0.0002,
4123
+ "loss": 3.1299,
4124
+ "step": 6860
4125
+ },
4126
+ {
4127
+ "epoch": 0.79,
4128
+ "learning_rate": 0.0002,
4129
+ "loss": 3.1563,
4130
+ "step": 6870
4131
+ },
4132
+ {
4133
+ "epoch": 0.79,
4134
+ "learning_rate": 0.0002,
4135
+ "loss": 2.8985,
4136
+ "step": 6880
4137
+ },
4138
+ {
4139
+ "epoch": 0.79,
4140
+ "learning_rate": 0.0002,
4141
+ "loss": 2.6909,
4142
+ "step": 6890
4143
+ },
4144
+ {
4145
+ "epoch": 0.79,
4146
+ "learning_rate": 0.0002,
4147
+ "loss": 2.005,
4148
+ "step": 6900
4149
+ },
4150
+ {
4151
+ "epoch": 0.79,
4152
+ "learning_rate": 0.0002,
4153
+ "loss": 3.2559,
4154
+ "step": 6910
4155
+ },
4156
+ {
4157
+ "epoch": 0.79,
4158
+ "learning_rate": 0.0002,
4159
+ "loss": 3.2412,
4160
+ "step": 6920
4161
+ },
4162
+ {
4163
+ "epoch": 0.79,
4164
+ "learning_rate": 0.0002,
4165
+ "loss": 2.8569,
4166
+ "step": 6930
4167
+ },
4168
+ {
4169
+ "epoch": 0.79,
4170
+ "learning_rate": 0.0002,
4171
+ "loss": 2.5241,
4172
+ "step": 6940
4173
+ },
4174
+ {
4175
+ "epoch": 0.79,
4176
+ "learning_rate": 0.0002,
4177
+ "loss": 1.837,
4178
+ "step": 6950
4179
+ },
4180
+ {
4181
+ "epoch": 0.8,
4182
+ "learning_rate": 0.0002,
4183
+ "loss": 2.5914,
4184
+ "step": 6960
4185
+ },
4186
+ {
4187
+ "epoch": 0.8,
4188
+ "learning_rate": 0.0002,
4189
+ "loss": 2.4782,
4190
+ "step": 6970
4191
+ },
4192
+ {
4193
+ "epoch": 0.8,
4194
+ "learning_rate": 0.0002,
4195
+ "loss": 2.4614,
4196
+ "step": 6980
4197
+ },
4198
+ {
4199
+ "epoch": 0.8,
4200
+ "learning_rate": 0.0002,
4201
+ "loss": 2.4431,
4202
+ "step": 6990
4203
+ },
4204
+ {
4205
+ "epoch": 0.8,
4206
+ "learning_rate": 0.0002,
4207
+ "loss": 1.8716,
4208
+ "step": 7000
4209
+ },
4210
+ {
4211
+ "epoch": 0.8,
4212
+ "learning_rate": 0.0002,
4213
+ "loss": 2.745,
4214
+ "step": 7010
4215
+ },
4216
+ {
4217
+ "epoch": 0.8,
4218
+ "learning_rate": 0.0002,
4219
+ "loss": 2.8138,
4220
+ "step": 7020
4221
+ },
4222
+ {
4223
+ "epoch": 0.8,
4224
+ "learning_rate": 0.0002,
4225
+ "loss": 2.6609,
4226
+ "step": 7030
4227
+ },
4228
+ {
4229
+ "epoch": 0.8,
4230
+ "learning_rate": 0.0002,
4231
+ "loss": 2.5277,
4232
+ "step": 7040
4233
+ },
4234
+ {
4235
+ "epoch": 0.81,
4236
+ "learning_rate": 0.0002,
4237
+ "loss": 2.0266,
4238
+ "step": 7050
4239
+ },
4240
+ {
4241
+ "epoch": 0.81,
4242
+ "learning_rate": 0.0002,
4243
+ "loss": 2.8222,
4244
+ "step": 7060
4245
+ },
4246
+ {
4247
+ "epoch": 0.81,
4248
+ "learning_rate": 0.0002,
4249
+ "loss": 2.7629,
4250
+ "step": 7070
4251
+ },
4252
+ {
4253
+ "epoch": 0.81,
4254
+ "learning_rate": 0.0002,
4255
+ "loss": 2.6332,
4256
+ "step": 7080
4257
+ },
4258
+ {
4259
+ "epoch": 0.81,
4260
+ "learning_rate": 0.0002,
4261
+ "loss": 2.5151,
4262
+ "step": 7090
4263
+ },
4264
+ {
4265
+ "epoch": 0.81,
4266
+ "learning_rate": 0.0002,
4267
+ "loss": 1.9864,
4268
+ "step": 7100
4269
+ },
4270
+ {
4271
+ "epoch": 0.81,
4272
+ "learning_rate": 0.0002,
4273
+ "loss": 2.6369,
4274
+ "step": 7110
4275
+ },
4276
+ {
4277
+ "epoch": 0.81,
4278
+ "learning_rate": 0.0002,
4279
+ "loss": 2.5675,
4280
+ "step": 7120
4281
+ },
4282
+ {
4283
+ "epoch": 0.81,
4284
+ "learning_rate": 0.0002,
4285
+ "loss": 2.4732,
4286
+ "step": 7130
4287
+ },
4288
+ {
4289
+ "epoch": 0.82,
4290
+ "learning_rate": 0.0002,
4291
+ "loss": 2.426,
4292
+ "step": 7140
4293
+ },
4294
+ {
4295
+ "epoch": 0.82,
4296
+ "learning_rate": 0.0002,
4297
+ "loss": 1.9873,
4298
+ "step": 7150
4299
+ },
4300
+ {
4301
+ "epoch": 0.82,
4302
+ "learning_rate": 0.0002,
4303
+ "loss": 2.6666,
4304
+ "step": 7160
4305
+ },
4306
+ {
4307
+ "epoch": 0.82,
4308
+ "learning_rate": 0.0002,
4309
+ "loss": 2.5453,
4310
+ "step": 7170
4311
+ },
4312
+ {
4313
+ "epoch": 0.82,
4314
+ "learning_rate": 0.0002,
4315
+ "loss": 2.4866,
4316
+ "step": 7180
4317
+ },
4318
+ {
4319
+ "epoch": 0.82,
4320
+ "learning_rate": 0.0002,
4321
+ "loss": 2.4204,
4322
+ "step": 7190
4323
+ },
4324
+ {
4325
+ "epoch": 0.82,
4326
+ "learning_rate": 0.0002,
4327
+ "loss": 1.9865,
4328
+ "step": 7200
4329
+ },
4330
+ {
4331
+ "epoch": 0.82,
4332
+ "learning_rate": 0.0002,
4333
+ "loss": 2.5969,
4334
+ "step": 7210
4335
+ },
4336
+ {
4337
+ "epoch": 0.83,
4338
+ "learning_rate": 0.0002,
4339
+ "loss": 2.5248,
4340
+ "step": 7220
4341
+ },
4342
+ {
4343
+ "epoch": 0.83,
4344
+ "learning_rate": 0.0002,
4345
+ "loss": 2.4691,
4346
+ "step": 7230
4347
+ },
4348
+ {
4349
+ "epoch": 0.83,
4350
+ "learning_rate": 0.0002,
4351
+ "loss": 2.4653,
4352
+ "step": 7240
4353
+ },
4354
+ {
4355
+ "epoch": 0.83,
4356
+ "learning_rate": 0.0002,
4357
+ "loss": 1.9409,
4358
+ "step": 7250
4359
+ },
4360
+ {
4361
+ "epoch": 0.83,
4362
+ "learning_rate": 0.0002,
4363
+ "loss": 2.4511,
4364
+ "step": 7260
4365
+ },
4366
+ {
4367
+ "epoch": 0.83,
4368
+ "learning_rate": 0.0002,
4369
+ "loss": 2.5139,
4370
+ "step": 7270
4371
+ },
4372
+ {
4373
+ "epoch": 0.83,
4374
+ "learning_rate": 0.0002,
4375
+ "loss": 2.4236,
4376
+ "step": 7280
4377
+ },
4378
+ {
4379
+ "epoch": 0.83,
4380
+ "learning_rate": 0.0002,
4381
+ "loss": 2.3505,
4382
+ "step": 7290
4383
+ },
4384
+ {
4385
+ "epoch": 0.83,
4386
+ "learning_rate": 0.0002,
4387
+ "loss": 1.9038,
4388
+ "step": 7300
4389
+ },
4390
+ {
4391
+ "epoch": 0.84,
4392
+ "learning_rate": 0.0002,
4393
+ "loss": 2.4344,
4394
+ "step": 7310
4395
+ },
4396
+ {
4397
+ "epoch": 0.84,
4398
+ "learning_rate": 0.0002,
4399
+ "loss": 2.3629,
4400
+ "step": 7320
4401
+ },
4402
+ {
4403
+ "epoch": 0.84,
4404
+ "learning_rate": 0.0002,
4405
+ "loss": 2.2656,
4406
+ "step": 7330
4407
+ },
4408
+ {
4409
+ "epoch": 0.84,
4410
+ "learning_rate": 0.0002,
4411
+ "loss": 2.1788,
4412
+ "step": 7340
4413
+ },
4414
+ {
4415
+ "epoch": 0.84,
4416
+ "learning_rate": 0.0002,
4417
+ "loss": 1.54,
4418
+ "step": 7350
4419
+ },
4420
+ {
4421
+ "epoch": 0.84,
4422
+ "learning_rate": 0.0002,
4423
+ "loss": 2.1208,
4424
+ "step": 7360
4425
+ },
4426
+ {
4427
+ "epoch": 0.84,
4428
+ "learning_rate": 0.0002,
4429
+ "loss": 2.1735,
4430
+ "step": 7370
4431
+ },
4432
+ {
4433
+ "epoch": 0.84,
4434
+ "learning_rate": 0.0002,
4435
+ "loss": 2.2392,
4436
+ "step": 7380
4437
+ },
4438
+ {
4439
+ "epoch": 0.84,
4440
+ "learning_rate": 0.0002,
4441
+ "loss": 2.2141,
4442
+ "step": 7390
4443
+ },
4444
+ {
4445
+ "epoch": 0.85,
4446
+ "learning_rate": 0.0002,
4447
+ "loss": 1.7179,
4448
+ "step": 7400
4449
+ },
4450
+ {
4451
+ "epoch": 0.85,
4452
+ "learning_rate": 0.0002,
4453
+ "loss": 2.2849,
4454
+ "step": 7410
4455
+ },
4456
+ {
4457
+ "epoch": 0.85,
4458
+ "learning_rate": 0.0002,
4459
+ "loss": 2.277,
4460
+ "step": 7420
4461
+ },
4462
+ {
4463
+ "epoch": 0.85,
4464
+ "learning_rate": 0.0002,
4465
+ "loss": 2.27,
4466
+ "step": 7430
4467
+ },
4468
+ {
4469
+ "epoch": 0.85,
4470
+ "learning_rate": 0.0002,
4471
+ "loss": 2.2229,
4472
+ "step": 7440
4473
+ },
4474
+ {
4475
+ "epoch": 0.85,
4476
+ "learning_rate": 0.0002,
4477
+ "loss": 1.712,
4478
+ "step": 7450
4479
+ },
4480
+ {
4481
+ "epoch": 0.85,
4482
+ "learning_rate": 0.0002,
4483
+ "loss": 2.1984,
4484
+ "step": 7460
4485
+ },
4486
+ {
4487
+ "epoch": 0.85,
4488
+ "learning_rate": 0.0002,
4489
+ "loss": 2.2542,
4490
+ "step": 7470
4491
+ },
4492
+ {
4493
+ "epoch": 0.85,
4494
+ "learning_rate": 0.0002,
4495
+ "loss": 2.3139,
4496
+ "step": 7480
4497
+ },
4498
+ {
4499
+ "epoch": 0.86,
4500
+ "learning_rate": 0.0002,
4501
+ "loss": 2.2685,
4502
+ "step": 7490
4503
+ },
4504
+ {
4505
+ "epoch": 0.86,
4506
+ "learning_rate": 0.0002,
4507
+ "loss": 1.6901,
4508
+ "step": 7500
4509
+ },
4510
+ {
4511
+ "epoch": 0.86,
4512
+ "learning_rate": 0.0002,
4513
+ "loss": 2.1149,
4514
+ "step": 7510
4515
+ },
4516
+ {
4517
+ "epoch": 0.86,
4518
+ "learning_rate": 0.0002,
4519
+ "loss": 2.0516,
4520
+ "step": 7520
4521
+ },
4522
+ {
4523
+ "epoch": 0.86,
4524
+ "learning_rate": 0.0002,
4525
+ "loss": 2.0119,
4526
+ "step": 7530
4527
+ },
4528
+ {
4529
+ "epoch": 0.86,
4530
+ "learning_rate": 0.0002,
4531
+ "loss": 2.0099,
4532
+ "step": 7540
4533
+ },
4534
+ {
4535
+ "epoch": 0.86,
4536
+ "learning_rate": 0.0002,
4537
+ "loss": 1.5823,
4538
+ "step": 7550
4539
+ },
4540
+ {
4541
+ "epoch": 0.86,
4542
+ "learning_rate": 0.0002,
4543
+ "loss": 1.9658,
4544
+ "step": 7560
4545
+ },
4546
+ {
4547
+ "epoch": 0.87,
4548
+ "learning_rate": 0.0002,
4549
+ "loss": 1.9999,
4550
+ "step": 7570
4551
+ },
4552
+ {
4553
+ "epoch": 0.87,
4554
+ "learning_rate": 0.0002,
4555
+ "loss": 2.0033,
4556
+ "step": 7580
4557
+ },
4558
+ {
4559
+ "epoch": 0.87,
4560
+ "learning_rate": 0.0002,
4561
+ "loss": 1.9746,
4562
+ "step": 7590
4563
+ },
4564
+ {
4565
+ "epoch": 0.87,
4566
+ "learning_rate": 0.0002,
4567
+ "loss": 1.5731,
4568
+ "step": 7600
4569
+ },
4570
+ {
4571
+ "epoch": 0.87,
4572
+ "learning_rate": 0.0002,
4573
+ "loss": 1.9657,
4574
+ "step": 7610
4575
+ },
4576
+ {
4577
+ "epoch": 0.87,
4578
+ "learning_rate": 0.0002,
4579
+ "loss": 1.9967,
4580
+ "step": 7620
4581
+ },
4582
+ {
4583
+ "epoch": 0.87,
4584
+ "learning_rate": 0.0002,
4585
+ "loss": 1.9404,
4586
+ "step": 7630
4587
+ },
4588
+ {
4589
+ "epoch": 0.87,
4590
+ "learning_rate": 0.0002,
4591
+ "loss": 1.8483,
4592
+ "step": 7640
4593
+ },
4594
+ {
4595
+ "epoch": 0.87,
4596
+ "learning_rate": 0.0002,
4597
+ "loss": 1.4618,
4598
+ "step": 7650
4599
+ },
4600
+ {
4601
+ "epoch": 0.88,
4602
+ "learning_rate": 0.0002,
4603
+ "loss": 1.7986,
4604
+ "step": 7660
4605
+ },
4606
+ {
4607
+ "epoch": 0.88,
4608
+ "learning_rate": 0.0002,
4609
+ "loss": 1.8033,
4610
+ "step": 7670
4611
+ },
4612
+ {
4613
+ "epoch": 0.88,
4614
+ "learning_rate": 0.0002,
4615
+ "loss": 1.8471,
4616
+ "step": 7680
4617
+ },
4618
+ {
4619
+ "epoch": 0.88,
4620
+ "learning_rate": 0.0002,
4621
+ "loss": 1.8007,
4622
+ "step": 7690
4623
+ },
4624
+ {
4625
+ "epoch": 0.88,
4626
+ "learning_rate": 0.0002,
4627
+ "loss": 1.3605,
4628
+ "step": 7700
4629
+ },
4630
+ {
4631
+ "epoch": 0.88,
4632
+ "learning_rate": 0.0002,
4633
+ "loss": 1.6258,
4634
+ "step": 7710
4635
+ },
4636
+ {
4637
+ "epoch": 0.88,
4638
+ "learning_rate": 0.0002,
4639
+ "loss": 1.733,
4640
+ "step": 7720
4641
+ },
4642
+ {
4643
+ "epoch": 0.88,
4644
+ "learning_rate": 0.0002,
4645
+ "loss": 1.812,
4646
+ "step": 7730
4647
+ },
4648
+ {
4649
+ "epoch": 0.88,
4650
+ "learning_rate": 0.0002,
4651
+ "loss": 1.7575,
4652
+ "step": 7740
4653
+ },
4654
+ {
4655
+ "epoch": 0.89,
4656
+ "learning_rate": 0.0002,
4657
+ "loss": 1.3581,
4658
+ "step": 7750
4659
+ },
4660
+ {
4661
+ "epoch": 0.89,
4662
+ "learning_rate": 0.0002,
4663
+ "loss": 1.6633,
4664
+ "step": 7760
4665
+ },
4666
+ {
4667
+ "epoch": 0.89,
4668
+ "learning_rate": 0.0002,
4669
+ "loss": 1.6169,
4670
+ "step": 7770
4671
+ },
4672
+ {
4673
+ "epoch": 0.89,
4674
+ "learning_rate": 0.0002,
4675
+ "loss": 1.7135,
4676
+ "step": 7780
4677
+ },
4678
+ {
4679
+ "epoch": 0.89,
4680
+ "learning_rate": 0.0002,
4681
+ "loss": 1.7109,
4682
+ "step": 7790
4683
+ },
4684
+ {
4685
+ "epoch": 0.89,
4686
+ "learning_rate": 0.0002,
4687
+ "loss": 1.407,
4688
+ "step": 7800
4689
+ },
4690
+ {
4691
+ "epoch": 0.89,
4692
+ "learning_rate": 0.0002,
4693
+ "loss": 1.641,
4694
+ "step": 7810
4695
+ },
4696
+ {
4697
+ "epoch": 0.89,
4698
+ "learning_rate": 0.0002,
4699
+ "loss": 1.6593,
4700
+ "step": 7820
4701
+ },
4702
+ {
4703
+ "epoch": 0.89,
4704
+ "learning_rate": 0.0002,
4705
+ "loss": 1.7338,
4706
+ "step": 7830
4707
+ },
4708
+ {
4709
+ "epoch": 0.9,
4710
+ "learning_rate": 0.0002,
4711
+ "loss": 1.7325,
4712
+ "step": 7840
4713
+ },
4714
+ {
4715
+ "epoch": 0.9,
4716
+ "learning_rate": 0.0002,
4717
+ "loss": 1.3272,
4718
+ "step": 7850
4719
+ },
4720
+ {
4721
+ "epoch": 0.9,
4722
+ "learning_rate": 0.0002,
4723
+ "loss": 1.6207,
4724
+ "step": 7860
4725
+ },
4726
+ {
4727
+ "epoch": 0.9,
4728
+ "learning_rate": 0.0002,
4729
+ "loss": 1.6328,
4730
+ "step": 7870
4731
+ },
4732
+ {
4733
+ "epoch": 0.9,
4734
+ "learning_rate": 0.0002,
4735
+ "loss": 1.6865,
4736
+ "step": 7880
4737
+ },
4738
+ {
4739
+ "epoch": 0.9,
4740
+ "learning_rate": 0.0002,
4741
+ "loss": 1.6119,
4742
+ "step": 7890
4743
+ },
4744
+ {
4745
+ "epoch": 0.9,
4746
+ "learning_rate": 0.0002,
4747
+ "loss": 1.3014,
4748
+ "step": 7900
4749
+ },
4750
+ {
4751
+ "epoch": 0.9,
4752
+ "learning_rate": 0.0002,
4753
+ "loss": 1.5949,
4754
+ "step": 7910
4755
+ },
4756
+ {
4757
+ "epoch": 0.91,
4758
+ "learning_rate": 0.0002,
4759
+ "loss": 1.5938,
4760
+ "step": 7920
4761
+ },
4762
+ {
4763
+ "epoch": 0.91,
4764
+ "learning_rate": 0.0002,
4765
+ "loss": 1.6891,
4766
+ "step": 7930
4767
+ },
4768
+ {
4769
+ "epoch": 0.91,
4770
+ "learning_rate": 0.0002,
4771
+ "loss": 1.6699,
4772
+ "step": 7940
4773
+ },
4774
+ {
4775
+ "epoch": 0.91,
4776
+ "learning_rate": 0.0002,
4777
+ "loss": 1.3358,
4778
+ "step": 7950
4779
+ },
4780
+ {
4781
+ "epoch": 0.91,
4782
+ "learning_rate": 0.0002,
4783
+ "loss": 1.5619,
4784
+ "step": 7960
4785
+ },
4786
+ {
4787
+ "epoch": 0.91,
4788
+ "learning_rate": 0.0002,
4789
+ "loss": 1.5884,
4790
+ "step": 7970
4791
+ },
4792
+ {
4793
+ "epoch": 0.91,
4794
+ "learning_rate": 0.0002,
4795
+ "loss": 1.634,
4796
+ "step": 7980
4797
+ },
4798
+ {
4799
+ "epoch": 0.91,
4800
+ "learning_rate": 0.0002,
4801
+ "loss": 1.6221,
4802
+ "step": 7990
4803
+ },
4804
+ {
4805
+ "epoch": 0.91,
4806
+ "learning_rate": 0.0002,
4807
+ "loss": 1.3082,
4808
+ "step": 8000
4809
+ },
4810
+ {
4811
+ "epoch": 0.92,
4812
+ "learning_rate": 0.0002,
4813
+ "loss": 1.5455,
4814
+ "step": 8010
4815
+ },
4816
+ {
4817
+ "epoch": 0.92,
4818
+ "learning_rate": 0.0002,
4819
+ "loss": 1.5425,
4820
+ "step": 8020
4821
+ },
4822
+ {
4823
+ "epoch": 0.92,
4824
+ "learning_rate": 0.0002,
4825
+ "loss": 1.6359,
4826
+ "step": 8030
4827
+ },
4828
+ {
4829
+ "epoch": 0.92,
4830
+ "learning_rate": 0.0002,
4831
+ "loss": 1.6494,
4832
+ "step": 8040
4833
+ },
4834
+ {
4835
+ "epoch": 0.92,
4836
+ "learning_rate": 0.0002,
4837
+ "loss": 1.2916,
4838
+ "step": 8050
4839
+ },
4840
+ {
4841
+ "epoch": 0.92,
4842
+ "learning_rate": 0.0002,
4843
+ "loss": 1.4993,
4844
+ "step": 8060
4845
+ },
4846
+ {
4847
+ "epoch": 0.92,
4848
+ "learning_rate": 0.0002,
4849
+ "loss": 1.5583,
4850
+ "step": 8070
4851
+ },
4852
+ {
4853
+ "epoch": 0.92,
4854
+ "learning_rate": 0.0002,
4855
+ "loss": 1.5923,
4856
+ "step": 8080
4857
+ },
4858
+ {
4859
+ "epoch": 0.92,
4860
+ "learning_rate": 0.0002,
4861
+ "loss": 1.6308,
4862
+ "step": 8090
4863
+ },
4864
+ {
4865
+ "epoch": 0.93,
4866
+ "learning_rate": 0.0002,
4867
+ "loss": 1.3073,
4868
+ "step": 8100
4869
+ },
4870
+ {
4871
+ "epoch": 0.93,
4872
+ "learning_rate": 0.0002,
4873
+ "loss": 1.5363,
4874
+ "step": 8110
4875
+ },
4876
+ {
4877
+ "epoch": 0.93,
4878
+ "learning_rate": 0.0002,
4879
+ "loss": 1.6257,
4880
+ "step": 8120
4881
+ },
4882
+ {
4883
+ "epoch": 0.93,
4884
+ "learning_rate": 0.0002,
4885
+ "loss": 1.6471,
4886
+ "step": 8130
4887
+ },
4888
+ {
4889
+ "epoch": 0.93,
4890
+ "learning_rate": 0.0002,
4891
+ "loss": 1.6669,
4892
+ "step": 8140
4893
+ },
4894
+ {
4895
+ "epoch": 0.93,
4896
+ "learning_rate": 0.0002,
4897
+ "loss": 1.3195,
4898
+ "step": 8150
4899
+ },
4900
+ {
4901
+ "epoch": 0.93,
4902
+ "learning_rate": 0.0002,
4903
+ "loss": 1.5359,
4904
+ "step": 8160
4905
+ },
4906
+ {
4907
+ "epoch": 0.93,
4908
+ "learning_rate": 0.0002,
4909
+ "loss": 1.5833,
4910
+ "step": 8170
4911
+ },
4912
+ {
4913
+ "epoch": 0.93,
4914
+ "learning_rate": 0.0002,
4915
+ "loss": 1.6571,
4916
+ "step": 8180
4917
+ },
4918
+ {
4919
+ "epoch": 0.94,
4920
+ "learning_rate": 0.0002,
4921
+ "loss": 1.6456,
4922
+ "step": 8190
4923
+ },
4924
+ {
4925
+ "epoch": 0.94,
4926
+ "learning_rate": 0.0002,
4927
+ "loss": 1.3148,
4928
+ "step": 8200
4929
+ },
4930
+ {
4931
+ "epoch": 0.94,
4932
+ "learning_rate": 0.0002,
4933
+ "loss": 1.5396,
4934
+ "step": 8210
4935
+ },
4936
+ {
4937
+ "epoch": 0.94,
4938
+ "learning_rate": 0.0002,
4939
+ "loss": 1.5219,
4940
+ "step": 8220
4941
+ },
4942
+ {
4943
+ "epoch": 0.94,
4944
+ "learning_rate": 0.0002,
4945
+ "loss": 1.6478,
4946
+ "step": 8230
4947
+ },
4948
+ {
4949
+ "epoch": 0.94,
4950
+ "learning_rate": 0.0002,
4951
+ "loss": 1.6146,
4952
+ "step": 8240
4953
+ },
4954
+ {
4955
+ "epoch": 0.94,
4956
+ "learning_rate": 0.0002,
4957
+ "loss": 1.284,
4958
+ "step": 8250
4959
+ },
4960
+ {
4961
+ "epoch": 0.94,
4962
+ "learning_rate": 0.0002,
4963
+ "loss": 1.5497,
4964
+ "step": 8260
4965
+ },
4966
+ {
4967
+ "epoch": 0.95,
4968
+ "learning_rate": 0.0002,
4969
+ "loss": 1.5858,
4970
+ "step": 8270
4971
+ },
4972
+ {
4973
+ "epoch": 0.95,
4974
+ "learning_rate": 0.0002,
4975
+ "loss": 1.6964,
4976
+ "step": 8280
4977
+ },
4978
+ {
4979
+ "epoch": 0.95,
4980
+ "learning_rate": 0.0002,
4981
+ "loss": 1.6518,
4982
+ "step": 8290
4983
+ },
4984
+ {
4985
+ "epoch": 0.95,
4986
+ "learning_rate": 0.0002,
4987
+ "loss": 1.4095,
4988
+ "step": 8300
4989
+ },
4990
+ {
4991
+ "epoch": 0.95,
4992
+ "learning_rate": 0.0002,
4993
+ "loss": 1.6811,
4994
+ "step": 8310
4995
+ },
4996
+ {
4997
+ "epoch": 0.95,
4998
+ "learning_rate": 0.0002,
4999
+ "loss": 1.7725,
5000
+ "step": 8320
5001
+ },
5002
+ {
5003
+ "epoch": 0.95,
5004
+ "learning_rate": 0.0002,
5005
+ "loss": 2.9559,
5006
+ "step": 8330
5007
+ },
5008
+ {
5009
+ "epoch": 0.95,
5010
+ "learning_rate": 0.0002,
5011
+ "loss": 4.9749,
5012
+ "step": 8340
5013
+ },
5014
+ {
5015
+ "epoch": 0.95,
5016
+ "learning_rate": 0.0002,
5017
+ "loss": 4.138,
5018
+ "step": 8350
5019
+ },
5020
+ {
5021
+ "epoch": 0.96,
5022
+ "learning_rate": 0.0002,
5023
+ "loss": 6.1766,
5024
+ "step": 8360
5025
+ },
5026
+ {
5027
+ "epoch": 0.96,
5028
+ "learning_rate": 0.0002,
5029
+ "loss": 5.5577,
5030
+ "step": 8370
5031
+ },
5032
+ {
5033
+ "epoch": 0.96,
5034
+ "learning_rate": 0.0002,
5035
+ "loss": 4.946,
5036
+ "step": 8380
5037
+ },
5038
+ {
5039
+ "epoch": 0.96,
5040
+ "learning_rate": 0.0002,
5041
+ "loss": 4.2825,
5042
+ "step": 8390
5043
+ },
5044
+ {
5045
+ "epoch": 0.96,
5046
+ "learning_rate": 0.0002,
5047
+ "loss": 3.2138,
5048
+ "step": 8400
5049
+ },
5050
+ {
5051
+ "epoch": 0.96,
5052
+ "learning_rate": 0.0002,
5053
+ "loss": 4.8699,
5054
+ "step": 8410
5055
+ },
5056
+ {
5057
+ "epoch": 0.96,
5058
+ "learning_rate": 0.0002,
5059
+ "loss": 4.3577,
5060
+ "step": 8420
5061
+ },
5062
+ {
5063
+ "epoch": 0.96,
5064
+ "learning_rate": 0.0002,
5065
+ "loss": 4.5203,
5066
+ "step": 8430
5067
+ },
5068
+ {
5069
+ "epoch": 0.96,
5070
+ "learning_rate": 0.0002,
5071
+ "loss": 4.2161,
5072
+ "step": 8440
5073
+ },
5074
+ {
5075
+ "epoch": 0.97,
5076
+ "learning_rate": 0.0002,
5077
+ "loss": 2.6815,
5078
+ "step": 8450
5079
+ },
5080
+ {
5081
+ "epoch": 0.97,
5082
+ "learning_rate": 0.0002,
5083
+ "loss": 3.572,
5084
+ "step": 8460
5085
+ },
5086
+ {
5087
+ "epoch": 0.97,
5088
+ "learning_rate": 0.0002,
5089
+ "loss": 3.7959,
5090
+ "step": 8470
5091
+ },
5092
+ {
5093
+ "epoch": 0.97,
5094
+ "learning_rate": 0.0002,
5095
+ "loss": 4.3023,
5096
+ "step": 8480
5097
+ },
5098
+ {
5099
+ "epoch": 0.97,
5100
+ "learning_rate": 0.0002,
5101
+ "loss": 4.6155,
5102
+ "step": 8490
5103
+ },
5104
+ {
5105
+ "epoch": 0.97,
5106
+ "learning_rate": 0.0002,
5107
+ "loss": 3.4514,
5108
+ "step": 8500
5109
+ },
5110
+ {
5111
+ "epoch": 0.97,
5112
+ "learning_rate": 0.0002,
5113
+ "loss": 4.8087,
5114
+ "step": 8510
5115
+ },
5116
+ {
5117
+ "epoch": 0.97,
5118
+ "learning_rate": 0.0002,
5119
+ "loss": 4.5272,
5120
+ "step": 8520
5121
+ },
5122
+ {
5123
+ "epoch": 0.97,
5124
+ "learning_rate": 0.0002,
5125
+ "loss": 4.1084,
5126
+ "step": 8530
5127
+ },
5128
+ {
5129
+ "epoch": 0.98,
5130
+ "learning_rate": 0.0002,
5131
+ "loss": 3.6305,
5132
+ "step": 8540
5133
+ },
5134
+ {
5135
+ "epoch": 0.98,
5136
+ "learning_rate": 0.0002,
5137
+ "loss": 2.5536,
5138
+ "step": 8550
5139
+ },
5140
+ {
5141
+ "epoch": 0.98,
5142
+ "learning_rate": 0.0002,
5143
+ "loss": 4.4515,
5144
+ "step": 8560
5145
+ },
5146
+ {
5147
+ "epoch": 0.98,
5148
+ "learning_rate": 0.0002,
5149
+ "loss": 3.4525,
5150
+ "step": 8570
5151
+ },
5152
+ {
5153
+ "epoch": 0.98,
5154
+ "learning_rate": 0.0002,
5155
+ "loss": 3.5976,
5156
+ "step": 8580
5157
+ },
5158
+ {
5159
+ "epoch": 0.98,
5160
+ "learning_rate": 0.0002,
5161
+ "loss": 3.4597,
5162
+ "step": 8590
5163
+ },
5164
+ {
5165
+ "epoch": 0.98,
5166
+ "learning_rate": 0.0002,
5167
+ "loss": 3.1653,
5168
+ "step": 8600
5169
+ },
5170
+ {
5171
+ "epoch": 0.98,
5172
+ "learning_rate": 0.0002,
5173
+ "loss": 4.5532,
5174
+ "step": 8610
5175
+ },
5176
+ {
5177
+ "epoch": 0.99,
5178
+ "learning_rate": 0.0002,
5179
+ "loss": 4.1625,
5180
+ "step": 8620
5181
+ },
5182
+ {
5183
+ "epoch": 0.99,
5184
+ "learning_rate": 0.0002,
5185
+ "loss": 4.3235,
5186
+ "step": 8630
5187
+ },
5188
+ {
5189
+ "epoch": 0.99,
5190
+ "learning_rate": 0.0002,
5191
+ "loss": 4.1094,
5192
+ "step": 8640
5193
+ },
5194
+ {
5195
+ "epoch": 0.99,
5196
+ "learning_rate": 0.0002,
5197
+ "loss": 3.1124,
5198
+ "step": 8650
5199
+ },
5200
+ {
5201
+ "epoch": 0.99,
5202
+ "learning_rate": 0.0002,
5203
+ "loss": 4.5213,
5204
+ "step": 8660
5205
+ },
5206
+ {
5207
+ "epoch": 0.99,
5208
+ "learning_rate": 0.0002,
5209
+ "loss": 4.492,
5210
+ "step": 8670
5211
+ },
5212
+ {
5213
+ "epoch": 0.99,
5214
+ "learning_rate": 0.0002,
5215
+ "loss": 4.3158,
5216
+ "step": 8680
5217
+ },
5218
+ {
5219
+ "epoch": 0.99,
5220
+ "learning_rate": 0.0002,
5221
+ "loss": 3.9548,
5222
+ "step": 8690
5223
+ },
5224
+ {
5225
+ "epoch": 0.99,
5226
+ "learning_rate": 0.0002,
5227
+ "loss": 3.7034,
5228
+ "step": 8700
5229
+ },
5230
+ {
5231
+ "epoch": 1.0,
5232
+ "learning_rate": 0.0002,
5233
+ "loss": 3.9088,
5234
+ "step": 8710
5235
+ },
5236
+ {
5237
+ "epoch": 1.0,
5238
+ "learning_rate": 0.0002,
5239
+ "loss": 3.9799,
5240
+ "step": 8720
5241
+ },
5242
+ {
5243
+ "epoch": 1.0,
5244
+ "learning_rate": 0.0002,
5245
+ "loss": 4.1253,
5246
+ "step": 8730
5247
+ },
5248
+ {
5249
+ "epoch": 1.0,
5250
+ "learning_rate": 0.0002,
5251
+ "loss": 3.9726,
5252
+ "step": 8740
5253
+ },
5254
+ {
5255
+ "epoch": 1.0,
5256
+ "learning_rate": 0.0002,
5257
+ "loss": 3.958,
5258
+ "step": 8750
5259
+ },
5260
+ {
5261
+ "epoch": 1.0,
5262
+ "learning_rate": 0.0002,
5263
+ "loss": 4.1978,
5264
+ "step": 8760
5265
+ },
5266
+ {
5267
+ "epoch": 1.0,
5268
+ "learning_rate": 0.0002,
5269
+ "loss": 4.9392,
5270
+ "step": 8770
5271
+ },
5272
+ {
5273
+ "epoch": 1.0,
5274
+ "learning_rate": 0.0002,
5275
+ "loss": 5.4691,
5276
+ "step": 8780
5277
+ },
5278
+ {
5279
+ "epoch": 1.0,
5280
+ "learning_rate": 0.0002,
5281
+ "loss": 4.2608,
5282
+ "step": 8790
5283
+ },
5284
+ {
5285
+ "epoch": 1.01,
5286
+ "learning_rate": 0.0002,
5287
+ "loss": 3.4531,
5288
+ "step": 8800
5289
+ }
5290
+ ],
5291
+ "max_steps": 10000,
5292
+ "num_train_epochs": 2,
5293
+ "total_flos": 1.1443961596526592e+18,
5294
+ "trial_name": null,
5295
+ "trial_params": null
5296
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f2a0fea313a7e98e4f6902e2266f6ac6346b0d8997a4f6744ef82af6acccc37
3
+ size 3963