Hugging Face's logo

cmu-lti
/
sotopia-pi-mistral-7b-BC_SR

PEFT
Model card Files Community
1
Pamela153 commited on
Commit
4dbd526
•
1 Parent(s): f9223d4

Upload folder using huggingface_hub

Browse files
Files changed (22) hide show
  1. README.md +19 -1
  2. adapter_config.json +21 -0
  3. adapter_model.bin +3 -0
  4. added_tokens.json +5 -0
  5. global_step730/mp_rank_00_model_states.pt +3 -0
  6. global_step730/zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  7. global_step730/zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  8. global_step730/zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  9. global_step730/zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  10. latest +1 -0
  11. rng_state_0.pth +3 -0
  12. rng_state_1.pth +3 -0
  13. rng_state_2.pth +3 -0
  14. rng_state_3.pth +3 -0
  15. scheduler.pt +3 -0
  16. special_tokens_map.json +11 -0
  17. tokenizer.json +0 -0
  18. tokenizer.model +3 -0
  19. tokenizer_config.json +47 -0
  20. trainer_state.json +4399 -0
  21. training_args.bin +3 -0
  22. zero_to_fp32.py +587 -0
README.md CHANGED
@@ -1,3 +1,21 @@
1
  ---
2
- license: mit
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ library_name: peft
3
  ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: QuantizationMethod.BITS_AND_BYTES
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: float16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.5.0
adapter_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "/data/tir/projects/tir6/bisk/ruiyiwan/selftrain/sft-round-2/Mistral-7B-Instruct-v0.1",
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": 32.0,
11
+ "lora_dropout": 0.1,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 8,
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 sha256:9114467ff4fc9eb6db12d2d19d0542044944a841097cb53b55ad0703561f746e
3
+ size 6859725
added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "</s>": 2,
3
+ "<s>": 1,
4
+ "<unk>": 0
5
+ }
global_step730/mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f1ed9d6a737505fccd50e65ef7faf8a937b0524426ed71f5f7f91b74d7ce8d6
3
+ size 8036113727
global_step730/zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87648f8081e02fc2a33999418f0e6982abbf46f63380ad5d5b12b17011ac7eb5
3
+ size 10230361
global_step730/zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:62dcc6e187a425b647d0475801f0db40f2d803075d31ffac038a516be02667fa
3
+ size 10230425
global_step730/zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:571c8afc381db61563b206b5db05a2ed3006f9fd8425fe01770d3dc99873fc9e
3
+ size 10230425
global_step730/zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea080266cd12f74467510691ce1890688df476832be03f60a4b53fd9a9c57cdf
3
+ size 10230425
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step730
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49190c40adc9ffe0db97aa9769bbcd6e1fe729de00be20564e19c0ab6c327338
3
+ size 17655
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f723f4cc8eebd1a91b89ccc5135a2a2d455b82ac54bdf5c2ed831dad7cd683d
3
+ size 17655
rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eca31ee0f1d72d21a104282bb6c7dcd269e65c62fcf052ad7b08deb8a3ac849f
3
+ size 17655
rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15cd3d58ca65afff184e6a2d55b92c1a7a43918b489e3dc7e6ee195062998309
3
+ size 17655
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6054d2b4a50cb3770a637d0f74920957f83535d8f29cad1fdf6572a793f4c400
3
+ size 627
special_tokens_map.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<unk>",
4
+ "<s>",
5
+ "</s>"
6
+ ],
7
+ "bos_token": "<s>",
8
+ "eos_token": "</s>",
9
+ "pad_token": "<unk>",
10
+ "unk_token": "<unk>"
11
+ }
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:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
tokenizer_config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<unk>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "1": {
12
+ "content": "<s>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "2": {
20
+ "content": "</s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ }
27
+ },
28
+ "additional_special_tokens": [
29
+ "<unk>",
30
+ "<s>",
31
+ "</s>"
32
+ ],
33
+ "bos_token": "<s>",
34
+ "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token + ' ' }}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
35
+ "clean_up_tokenization_spaces": false,
36
+ "eos_token": "</s>",
37
+ "legacy": true,
38
+ "model_max_length": 1000000000000000019884624838656,
39
+ "pad_token": "<unk>",
40
+ "padding_side": "right",
41
+ "sp_model_kwargs": {},
42
+ "spaces_between_special_tokens": false,
43
+ "split_special_tokens": false,
44
+ "tokenizer_class": "LlamaTokenizer",
45
+ "unk_token": "<unk>",
46
+ "use_default_system_prompt": true
47
+ }
trainer_state.json ADDED
@@ -0,0 +1,4399 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 5.0,
5
+ "eval_steps": 500,
6
+ "global_step": 730,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 5.681818181818182e-07,
14
+ "loss": 1.6669,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 1.1363636363636364e-06,
20
+ "loss": 1.4274,
21
+ "step": 2
22
+ },
23
+ {
24
+ "epoch": 0.02,
25
+ "learning_rate": 1.7045454545454546e-06,
26
+ "loss": 1.5736,
27
+ "step": 3
28
+ },
29
+ {
30
+ "epoch": 0.03,
31
+ "learning_rate": 2.2727272727272728e-06,
32
+ "loss": 1.6031,
33
+ "step": 4
34
+ },
35
+ {
36
+ "epoch": 0.03,
37
+ "learning_rate": 2.840909090909091e-06,
38
+ "loss": 1.4013,
39
+ "step": 5
40
+ },
41
+ {
42
+ "epoch": 0.04,
43
+ "learning_rate": 3.409090909090909e-06,
44
+ "loss": 1.4383,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 0.05,
49
+ "learning_rate": 3.9772727272727275e-06,
50
+ "loss": 1.4783,
51
+ "step": 7
52
+ },
53
+ {
54
+ "epoch": 0.05,
55
+ "learning_rate": 4.5454545454545455e-06,
56
+ "loss": 1.4839,
57
+ "step": 8
58
+ },
59
+ {
60
+ "epoch": 0.06,
61
+ "learning_rate": 5.113636363636364e-06,
62
+ "loss": 1.2941,
63
+ "step": 9
64
+ },
65
+ {
66
+ "epoch": 0.07,
67
+ "learning_rate": 5.681818181818182e-06,
68
+ "loss": 1.5711,
69
+ "step": 10
70
+ },
71
+ {
72
+ "epoch": 0.08,
73
+ "learning_rate": 6.25e-06,
74
+ "loss": 1.5195,
75
+ "step": 11
76
+ },
77
+ {
78
+ "epoch": 0.08,
79
+ "learning_rate": 6.818181818181818e-06,
80
+ "loss": 1.535,
81
+ "step": 12
82
+ },
83
+ {
84
+ "epoch": 0.09,
85
+ "learning_rate": 7.386363636363637e-06,
86
+ "loss": 1.3355,
87
+ "step": 13
88
+ },
89
+ {
90
+ "epoch": 0.1,
91
+ "learning_rate": 7.954545454545455e-06,
92
+ "loss": 1.5082,
93
+ "step": 14
94
+ },
95
+ {
96
+ "epoch": 0.1,
97
+ "learning_rate": 8.522727272727273e-06,
98
+ "loss": 1.4538,
99
+ "step": 15
100
+ },
101
+ {
102
+ "epoch": 0.11,
103
+ "learning_rate": 9.090909090909091e-06,
104
+ "loss": 1.3486,
105
+ "step": 16
106
+ },
107
+ {
108
+ "epoch": 0.12,
109
+ "learning_rate": 9.659090909090909e-06,
110
+ "loss": 1.4842,
111
+ "step": 17
112
+ },
113
+ {
114
+ "epoch": 0.12,
115
+ "learning_rate": 1.0227272727272729e-05,
116
+ "loss": 1.4197,
117
+ "step": 18
118
+ },
119
+ {
120
+ "epoch": 0.13,
121
+ "learning_rate": 1.0795454545454547e-05,
122
+ "loss": 1.4488,
123
+ "step": 19
124
+ },
125
+ {
126
+ "epoch": 0.14,
127
+ "learning_rate": 1.1363636363636365e-05,
128
+ "loss": 1.4491,
129
+ "step": 20
130
+ },
131
+ {
132
+ "epoch": 0.14,
133
+ "learning_rate": 1.1931818181818183e-05,
134
+ "loss": 1.3609,
135
+ "step": 21
136
+ },
137
+ {
138
+ "epoch": 0.15,
139
+ "learning_rate": 1.25e-05,
140
+ "loss": 1.4129,
141
+ "step": 22
142
+ },
143
+ {
144
+ "epoch": 0.16,
145
+ "learning_rate": 1.3068181818181819e-05,
146
+ "loss": 1.1582,
147
+ "step": 23
148
+ },
149
+ {
150
+ "epoch": 0.16,
151
+ "learning_rate": 1.3636363636363637e-05,
152
+ "loss": 1.4224,
153
+ "step": 24
154
+ },
155
+ {
156
+ "epoch": 0.17,
157
+ "learning_rate": 1.4204545454545456e-05,
158
+ "loss": 1.4216,
159
+ "step": 25
160
+ },
161
+ {
162
+ "epoch": 0.18,
163
+ "learning_rate": 1.4772727272727274e-05,
164
+ "loss": 1.4054,
165
+ "step": 26
166
+ },
167
+ {
168
+ "epoch": 0.18,
169
+ "learning_rate": 1.534090909090909e-05,
170
+ "loss": 1.3476,
171
+ "step": 27
172
+ },
173
+ {
174
+ "epoch": 0.19,
175
+ "learning_rate": 1.590909090909091e-05,
176
+ "loss": 1.4123,
177
+ "step": 28
178
+ },
179
+ {
180
+ "epoch": 0.2,
181
+ "learning_rate": 1.6477272727272726e-05,
182
+ "loss": 1.2541,
183
+ "step": 29
184
+ },
185
+ {
186
+ "epoch": 0.21,
187
+ "learning_rate": 1.7045454545454546e-05,
188
+ "loss": 1.3971,
189
+ "step": 30
190
+ },
191
+ {
192
+ "epoch": 0.21,
193
+ "learning_rate": 1.7613636363636366e-05,
194
+ "loss": 1.4469,
195
+ "step": 31
196
+ },
197
+ {
198
+ "epoch": 0.22,
199
+ "learning_rate": 1.8181818181818182e-05,
200
+ "loss": 1.3165,
201
+ "step": 32
202
+ },
203
+ {
204
+ "epoch": 0.23,
205
+ "learning_rate": 1.8750000000000002e-05,
206
+ "loss": 1.3095,
207
+ "step": 33
208
+ },
209
+ {
210
+ "epoch": 0.23,
211
+ "learning_rate": 1.9318181818181818e-05,
212
+ "loss": 1.2892,
213
+ "step": 34
214
+ },
215
+ {
216
+ "epoch": 0.24,
217
+ "learning_rate": 1.9886363636363638e-05,
218
+ "loss": 1.2706,
219
+ "step": 35
220
+ },
221
+ {
222
+ "epoch": 0.25,
223
+ "learning_rate": 2.0454545454545457e-05,
224
+ "loss": 1.3491,
225
+ "step": 36
226
+ },
227
+ {
228
+ "epoch": 0.25,
229
+ "learning_rate": 2.1022727272727274e-05,
230
+ "loss": 1.3282,
231
+ "step": 37
232
+ },
233
+ {
234
+ "epoch": 0.26,
235
+ "learning_rate": 2.1590909090909093e-05,
236
+ "loss": 1.1812,
237
+ "step": 38
238
+ },
239
+ {
240
+ "epoch": 0.27,
241
+ "learning_rate": 2.215909090909091e-05,
242
+ "loss": 1.3438,
243
+ "step": 39
244
+ },
245
+ {
246
+ "epoch": 0.27,
247
+ "learning_rate": 2.272727272727273e-05,
248
+ "loss": 1.3827,
249
+ "step": 40
250
+ },
251
+ {
252
+ "epoch": 0.28,
253
+ "learning_rate": 2.3295454545454546e-05,
254
+ "loss": 1.2663,
255
+ "step": 41
256
+ },
257
+ {
258
+ "epoch": 0.29,
259
+ "learning_rate": 2.3863636363636365e-05,
260
+ "loss": 1.3777,
261
+ "step": 42
262
+ },
263
+ {
264
+ "epoch": 0.29,
265
+ "learning_rate": 2.4431818181818185e-05,
266
+ "loss": 1.2316,
267
+ "step": 43
268
+ },
269
+ {
270
+ "epoch": 0.3,
271
+ "learning_rate": 2.5e-05,
272
+ "loss": 1.3528,
273
+ "step": 44
274
+ },
275
+ {
276
+ "epoch": 0.31,
277
+ "learning_rate": 2.5568181818181817e-05,
278
+ "loss": 1.379,
279
+ "step": 45
280
+ },
281
+ {
282
+ "epoch": 0.32,
283
+ "learning_rate": 2.6136363636363637e-05,
284
+ "loss": 1.3162,
285
+ "step": 46
286
+ },
287
+ {
288
+ "epoch": 0.32,
289
+ "learning_rate": 2.6704545454545453e-05,
290
+ "loss": 1.2006,
291
+ "step": 47
292
+ },
293
+ {
294
+ "epoch": 0.33,
295
+ "learning_rate": 2.7272727272727273e-05,
296
+ "loss": 1.4458,
297
+ "step": 48
298
+ },
299
+ {
300
+ "epoch": 0.34,
301
+ "learning_rate": 2.784090909090909e-05,
302
+ "loss": 1.2852,
303
+ "step": 49
304
+ },
305
+ {
306
+ "epoch": 0.34,
307
+ "learning_rate": 2.8409090909090912e-05,
308
+ "loss": 1.2553,
309
+ "step": 50
310
+ },
311
+ {
312
+ "epoch": 0.35,
313
+ "learning_rate": 2.8977272727272732e-05,
314
+ "loss": 1.4476,
315
+ "step": 51
316
+ },
317
+ {
318
+ "epoch": 0.36,
319
+ "learning_rate": 2.954545454545455e-05,
320
+ "loss": 1.2309,
321
+ "step": 52
322
+ },
323
+ {
324
+ "epoch": 0.36,
325
+ "learning_rate": 3.0113636363636365e-05,
326
+ "loss": 1.3228,
327
+ "step": 53
328
+ },
329
+ {
330
+ "epoch": 0.37,
331
+ "learning_rate": 3.068181818181818e-05,
332
+ "loss": 1.3367,
333
+ "step": 54
334
+ },
335
+ {
336
+ "epoch": 0.38,
337
+ "learning_rate": 3.125e-05,
338
+ "loss": 1.2866,
339
+ "step": 55
340
+ },
341
+ {
342
+ "epoch": 0.38,
343
+ "learning_rate": 3.181818181818182e-05,
344
+ "loss": 1.3093,
345
+ "step": 56
346
+ },
347
+ {
348
+ "epoch": 0.39,
349
+ "learning_rate": 3.238636363636364e-05,
350
+ "loss": 1.3932,
351
+ "step": 57
352
+ },
353
+ {
354
+ "epoch": 0.4,
355
+ "learning_rate": 3.295454545454545e-05,
356
+ "loss": 1.1568,
357
+ "step": 58
358
+ },
359
+ {
360
+ "epoch": 0.4,
361
+ "learning_rate": 3.352272727272727e-05,
362
+ "loss": 1.2535,
363
+ "step": 59
364
+ },
365
+ {
366
+ "epoch": 0.41,
367
+ "learning_rate": 3.409090909090909e-05,
368
+ "loss": 1.2189,
369
+ "step": 60
370
+ },
371
+ {
372
+ "epoch": 0.42,
373
+ "learning_rate": 3.465909090909091e-05,
374
+ "loss": 1.149,
375
+ "step": 61
376
+ },
377
+ {
378
+ "epoch": 0.42,
379
+ "learning_rate": 3.522727272727273e-05,
380
+ "loss": 1.3026,
381
+ "step": 62
382
+ },
383
+ {
384
+ "epoch": 0.43,
385
+ "learning_rate": 3.579545454545455e-05,
386
+ "loss": 1.2468,
387
+ "step": 63
388
+ },
389
+ {
390
+ "epoch": 0.44,
391
+ "learning_rate": 3.6363636363636364e-05,
392
+ "loss": 1.4326,
393
+ "step": 64
394
+ },
395
+ {
396
+ "epoch": 0.45,
397
+ "learning_rate": 3.6931818181818184e-05,
398
+ "loss": 1.2512,
399
+ "step": 65
400
+ },
401
+ {
402
+ "epoch": 0.45,
403
+ "learning_rate": 3.7500000000000003e-05,
404
+ "loss": 1.3661,
405
+ "step": 66
406
+ },
407
+ {
408
+ "epoch": 0.46,
409
+ "learning_rate": 3.8068181818181816e-05,
410
+ "loss": 1.2677,
411
+ "step": 67
412
+ },
413
+ {
414
+ "epoch": 0.47,
415
+ "learning_rate": 3.8636363636363636e-05,
416
+ "loss": 1.354,
417
+ "step": 68
418
+ },
419
+ {
420
+ "epoch": 0.47,
421
+ "learning_rate": 3.9204545454545456e-05,
422
+ "loss": 1.2258,
423
+ "step": 69
424
+ },
425
+ {
426
+ "epoch": 0.48,
427
+ "learning_rate": 3.9772727272727275e-05,
428
+ "loss": 1.1839,
429
+ "step": 70
430
+ },
431
+ {
432
+ "epoch": 0.49,
433
+ "learning_rate": 4.034090909090909e-05,
434
+ "loss": 1.4097,
435
+ "step": 71
436
+ },
437
+ {
438
+ "epoch": 0.49,
439
+ "learning_rate": 4.0909090909090915e-05,
440
+ "loss": 1.2499,
441
+ "step": 72
442
+ },
443
+ {
444
+ "epoch": 0.5,
445
+ "learning_rate": 4.1477272727272734e-05,
446
+ "loss": 1.3991,
447
+ "step": 73
448
+ },
449
+ {
450
+ "epoch": 0.51,
451
+ "learning_rate": 4.204545454545455e-05,
452
+ "loss": 1.2102,
453
+ "step": 74
454
+ },
455
+ {
456
+ "epoch": 0.51,
457
+ "learning_rate": 4.261363636363637e-05,
458
+ "loss": 1.3301,
459
+ "step": 75
460
+ },
461
+ {
462
+ "epoch": 0.52,
463
+ "learning_rate": 4.318181818181819e-05,
464
+ "loss": 1.2872,
465
+ "step": 76
466
+ },
467
+ {
468
+ "epoch": 0.53,
469
+ "learning_rate": 4.375e-05,
470
+ "loss": 1.2391,
471
+ "step": 77
472
+ },
473
+ {
474
+ "epoch": 0.53,
475
+ "learning_rate": 4.431818181818182e-05,
476
+ "loss": 1.2094,
477
+ "step": 78
478
+ },
479
+ {
480
+ "epoch": 0.54,
481
+ "learning_rate": 4.488636363636364e-05,
482
+ "loss": 1.1217,
483
+ "step": 79
484
+ },
485
+ {
486
+ "epoch": 0.55,
487
+ "learning_rate": 4.545454545454546e-05,
488
+ "loss": 1.3957,
489
+ "step": 80
490
+ },
491
+ {
492
+ "epoch": 0.55,
493
+ "learning_rate": 4.602272727272727e-05,
494
+ "loss": 1.1267,
495
+ "step": 81
496
+ },
497
+ {
498
+ "epoch": 0.56,
499
+ "learning_rate": 4.659090909090909e-05,
500
+ "loss": 1.0954,
501
+ "step": 82
502
+ },
503
+ {
504
+ "epoch": 0.57,
505
+ "learning_rate": 4.715909090909091e-05,
506
+ "loss": 1.412,
507
+ "step": 83
508
+ },
509
+ {
510
+ "epoch": 0.58,
511
+ "learning_rate": 4.772727272727273e-05,
512
+ "loss": 1.076,
513
+ "step": 84
514
+ },
515
+ {
516
+ "epoch": 0.58,
517
+ "learning_rate": 4.829545454545455e-05,
518
+ "loss": 1.2825,
519
+ "step": 85
520
+ },
521
+ {
522
+ "epoch": 0.59,
523
+ "learning_rate": 4.886363636363637e-05,
524
+ "loss": 1.1066,
525
+ "step": 86
526
+ },
527
+ {
528
+ "epoch": 0.6,
529
+ "learning_rate": 4.943181818181818e-05,
530
+ "loss": 1.2088,
531
+ "step": 87
532
+ },
533
+ {
534
+ "epoch": 0.6,
535
+ "learning_rate": 5e-05,
536
+ "loss": 1.1819,
537
+ "step": 88
538
+ },
539
+ {
540
+ "epoch": 0.61,
541
+ "learning_rate": 4.999998461763134e-05,
542
+ "loss": 1.2009,
543
+ "step": 89
544
+ },
545
+ {
546
+ "epoch": 0.62,
547
+ "learning_rate": 4.9999938470544264e-05,
548
+ "loss": 1.1569,
549
+ "step": 90
550
+ },
551
+ {
552
+ "epoch": 0.62,
553
+ "learning_rate": 4.999986155879557e-05,
554
+ "loss": 0.9333,
555
+ "step": 91
556
+ },
557
+ {
558
+ "epoch": 0.63,
559
+ "learning_rate": 4.9999753882479914e-05,
560
+ "loss": 1.2257,
561
+ "step": 92
562
+ },
563
+ {
564
+ "epoch": 0.64,
565
+ "learning_rate": 4.9999615441729785e-05,
566
+ "loss": 1.26,
567
+ "step": 93
568
+ },
569
+ {
570
+ "epoch": 0.64,
571
+ "learning_rate": 4.999944623671556e-05,
572
+ "loss": 1.185,
573
+ "step": 94
574
+ },
575
+ {
576
+ "epoch": 0.65,
577
+ "learning_rate": 4.9999246267645464e-05,
578
+ "loss": 1.2355,
579
+ "step": 95
580
+ },
581
+ {
582
+ "epoch": 0.66,
583
+ "learning_rate": 4.999901553476555e-05,
584
+ "loss": 1.323,
585
+ "step": 96
586
+ },
587
+ {
588
+ "epoch": 0.66,
589
+ "learning_rate": 4.9998754038359785e-05,
590
+ "loss": 1.2167,
591
+ "step": 97
592
+ },
593
+ {
594
+ "epoch": 0.67,
595
+ "learning_rate": 4.999846177874995e-05,
596
+ "loss": 1.2381,
597
+ "step": 98
598
+ },
599
+ {
600
+ "epoch": 0.68,
601
+ "learning_rate": 4.9998138756295704e-05,
602
+ "loss": 1.1787,
603
+ "step": 99
604
+ },
605
+ {
606
+ "epoch": 0.68,
607
+ "learning_rate": 4.999778497139455e-05,
608
+ "loss": 1.2369,
609
+ "step": 100
610
+ },
611
+ {
612
+ "epoch": 0.69,
613
+ "learning_rate": 4.9997400424481844e-05,
614
+ "loss": 1.265,
615
+ "step": 101
616
+ },
617
+ {
618
+ "epoch": 0.7,
619
+ "learning_rate": 4.999698511603082e-05,
620
+ "loss": 1.1829,
621
+ "step": 102
622
+ },
623
+ {
624
+ "epoch": 0.71,
625
+ "learning_rate": 4.9996539046552546e-05,
626
+ "loss": 1.2228,
627
+ "step": 103
628
+ },
629
+ {
630
+ "epoch": 0.71,
631
+ "learning_rate": 4.999606221659595e-05,
632
+ "loss": 1.4031,
633
+ "step": 104
634
+ },
635
+ {
636
+ "epoch": 0.72,
637
+ "learning_rate": 4.999555462674781e-05,
638
+ "loss": 1.2798,
639
+ "step": 105
640
+ },
641
+ {
642
+ "epoch": 0.73,
643
+ "learning_rate": 4.999501627763277e-05,
644
+ "loss": 1.1808,
645
+ "step": 106
646
+ },
647
+ {
648
+ "epoch": 0.73,
649
+ "learning_rate": 4.999444716991332e-05,
650
+ "loss": 1.2457,
651
+ "step": 107
652
+ },
653
+ {
654
+ "epoch": 0.74,
655
+ "learning_rate": 4.9993847304289774e-05,
656
+ "loss": 1.3655,
657
+ "step": 108
658
+ },
659
+ {
660
+ "epoch": 0.75,
661
+ "learning_rate": 4.9993216681500346e-05,
662
+ "loss": 1.187,
663
+ "step": 109
664
+ },
665
+ {
666
+ "epoch": 0.75,
667
+ "learning_rate": 4.999255530232105e-05,
668
+ "loss": 1.1713,
669
+ "step": 110
670
+ },
671
+ {
672
+ "epoch": 0.76,
673
+ "learning_rate": 4.99918631675658e-05,
674
+ "loss": 1.1947,
675
+ "step": 111
676
+ },
677
+ {
678
+ "epoch": 0.77,
679
+ "learning_rate": 4.9991140278086316e-05,
680
+ "loss": 1.1169,
681
+ "step": 112
682
+ },
683
+ {
684
+ "epoch": 0.77,
685
+ "learning_rate": 4.999038663477218e-05,
686
+ "loss": 1.2369,
687
+ "step": 113
688
+ },
689
+ {
690
+ "epoch": 0.78,
691
+ "learning_rate": 4.99896022385508e-05,
692
+ "loss": 1.2075,
693
+ "step": 114
694
+ },
695
+ {
696
+ "epoch": 0.79,
697
+ "learning_rate": 4.998878709038748e-05,
698
+ "loss": 1.2208,
699
+ "step": 115
700
+ },
701
+ {
702
+ "epoch": 0.79,
703
+ "learning_rate": 4.99879411912853e-05,
704
+ "loss": 1.1711,
705
+ "step": 116
706
+ },
707
+ {
708
+ "epoch": 0.8,
709
+ "learning_rate": 4.998706454228524e-05,
710
+ "loss": 1.2532,
711
+ "step": 117
712
+ },
713
+ {
714
+ "epoch": 0.81,
715
+ "learning_rate": 4.998615714446608e-05,
716
+ "loss": 1.1868,
717
+ "step": 118
718
+ },
719
+ {
720
+ "epoch": 0.82,
721
+ "learning_rate": 4.998521899894446e-05,
722
+ "loss": 1.2798,
723
+ "step": 119
724
+ },
725
+ {
726
+ "epoch": 0.82,
727
+ "learning_rate": 4.998425010687484e-05,
728
+ "loss": 1.1243,
729
+ "step": 120
730
+ },
731
+ {
732
+ "epoch": 0.83,
733
+ "learning_rate": 4.998325046944955e-05,
734
+ "loss": 1.3059,
735
+ "step": 121
736
+ },
737
+ {
738
+ "epoch": 0.84,
739
+ "learning_rate": 4.9982220087898715e-05,
740
+ "loss": 1.2788,
741
+ "step": 122
742
+ },
743
+ {
744
+ "epoch": 0.84,
745
+ "learning_rate": 4.998115896349032e-05,
746
+ "loss": 1.1768,
747
+ "step": 123
748
+ },
749
+ {
750
+ "epoch": 0.85,
751
+ "learning_rate": 4.998006709753017e-05,
752
+ "loss": 1.2633,
753
+ "step": 124
754
+ },
755
+ {
756
+ "epoch": 0.86,
757
+ "learning_rate": 4.997894449136191e-05,
758
+ "loss": 1.0365,
759
+ "step": 125
760
+ },
761
+ {
762
+ "epoch": 0.86,
763
+ "learning_rate": 4.9977791146367005e-05,
764
+ "loss": 1.2884,
765
+ "step": 126
766
+ },
767
+ {
768
+ "epoch": 0.87,
769
+ "learning_rate": 4.997660706396474e-05,
770
+ "loss": 1.2045,
771
+ "step": 127
772
+ },
773
+ {
774
+ "epoch": 0.88,
775
+ "learning_rate": 4.9975392245612254e-05,
776
+ "loss": 1.2226,
777
+ "step": 128
778
+ },
779
+ {
780
+ "epoch": 0.88,
781
+ "learning_rate": 4.997414669280446e-05,
782
+ "loss": 1.237,
783
+ "step": 129
784
+ },
785
+ {
786
+ "epoch": 0.89,
787
+ "learning_rate": 4.997287040707415e-05,
788
+ "loss": 1.1576,
789
+ "step": 130
790
+ },
791
+ {
792
+ "epoch": 0.9,
793
+ "learning_rate": 4.99715633899919e-05,
794
+ "loss": 1.2579,
795
+ "step": 131
796
+ },
797
+ {
798
+ "epoch": 0.9,
799
+ "learning_rate": 4.99702256431661e-05,
800
+ "loss": 1.3043,
801
+ "step": 132
802
+ },
803
+ {
804
+ "epoch": 0.91,
805
+ "learning_rate": 4.996885716824299e-05,
806
+ "loss": 1.2905,
807
+ "step": 133
808
+ },
809
+ {
810
+ "epoch": 0.92,
811
+ "learning_rate": 4.996745796690657e-05,
812
+ "loss": 1.1184,
813
+ "step": 134
814
+ },
815
+ {
816
+ "epoch": 0.92,
817
+ "learning_rate": 4.996602804087871e-05,
818
+ "loss": 1.0862,
819
+ "step": 135
820
+ },
821
+ {
822
+ "epoch": 0.93,
823
+ "learning_rate": 4.996456739191905e-05,
824
+ "loss": 1.2715,
825
+ "step": 136
826
+ },
827
+ {
828
+ "epoch": 0.94,
829
+ "learning_rate": 4.996307602182505e-05,
830
+ "loss": 1.0201,
831
+ "step": 137
832
+ },
833
+ {
834
+ "epoch": 0.95,
835
+ "learning_rate": 4.996155393243198e-05,
836
+ "loss": 1.3834,
837
+ "step": 138
838
+ },
839
+ {
840
+ "epoch": 0.95,
841
+ "learning_rate": 4.996000112561289e-05,
842
+ "loss": 1.1772,
843
+ "step": 139
844
+ },
845
+ {
846
+ "epoch": 0.96,
847
+ "learning_rate": 4.995841760327867e-05,
848
+ "loss": 1.14,
849
+ "step": 140
850
+ },
851
+ {
852
+ "epoch": 0.97,
853
+ "learning_rate": 4.995680336737797e-05,
854
+ "loss": 1.3104,
855
+ "step": 141
856
+ },
857
+ {
858
+ "epoch": 0.97,
859
+ "learning_rate": 4.995515841989726e-05,
860
+ "loss": 1.1304,
861
+ "step": 142
862
+ },
863
+ {
864
+ "epoch": 0.98,
865
+ "learning_rate": 4.9953482762860796e-05,
866
+ "loss": 1.0273,
867
+ "step": 143
868
+ },
869
+ {
870
+ "epoch": 0.99,
871
+ "learning_rate": 4.995177639833062e-05,
872
+ "loss": 1.2024,
873
+ "step": 144
874
+ },
875
+ {
876
+ "epoch": 0.99,
877
+ "learning_rate": 4.995003932840657e-05,
878
+ "loss": 1.2857,
879
+ "step": 145
880
+ },
881
+ {
882
+ "epoch": 1.0,
883
+ "learning_rate": 4.994827155522625e-05,
884
+ "loss": 1.1091,
885
+ "step": 146
886
+ },
887
+ {
888
+ "epoch": 1.01,
889
+ "learning_rate": 4.994647308096509e-05,
890
+ "loss": 1.3138,
891
+ "step": 147
892
+ },
893
+ {
894
+ "epoch": 1.01,
895
+ "learning_rate": 4.994464390783625e-05,
896
+ "loss": 1.1692,
897
+ "step": 148
898
+ },
899
+ {
900
+ "epoch": 1.02,
901
+ "learning_rate": 4.994278403809071e-05,
902
+ "loss": 1.2849,
903
+ "step": 149
904
+ },
905
+ {
906
+ "epoch": 1.03,
907
+ "learning_rate": 4.994089347401719e-05,
908
+ "loss": 1.3361,
909
+ "step": 150
910
+ },
911
+ {
912
+ "epoch": 1.03,
913
+ "learning_rate": 4.993897221794221e-05,
914
+ "loss": 1.1463,
915
+ "step": 151
916
+ },
917
+ {
918
+ "epoch": 1.04,
919
+ "learning_rate": 4.993702027223004e-05,
920
+ "loss": 1.1808,
921
+ "step": 152
922
+ },
923
+ {
924
+ "epoch": 1.05,
925
+ "learning_rate": 4.9935037639282725e-05,
926
+ "loss": 1.1949,
927
+ "step": 153
928
+ },
929
+ {
930
+ "epoch": 1.05,
931
+ "learning_rate": 4.993302432154008e-05,
932
+ "loss": 1.2023,
933
+ "step": 154
934
+ },
935
+ {
936
+ "epoch": 1.06,
937
+ "learning_rate": 4.993098032147966e-05,
938
+ "loss": 0.9441,
939
+ "step": 155
940
+ },
941
+ {
942
+ "epoch": 1.07,
943
+ "learning_rate": 4.9928905641616794e-05,
944
+ "loss": 1.26,
945
+ "step": 156
946
+ },
947
+ {
948
+ "epoch": 1.08,
949
+ "learning_rate": 4.992680028450457e-05,
950
+ "loss": 1.2594,
951
+ "step": 157
952
+ },
953
+ {
954
+ "epoch": 1.08,
955
+ "learning_rate": 4.9924664252733816e-05,
956
+ "loss": 1.2523,
957
+ "step": 158
958
+ },
959
+ {
960
+ "epoch": 1.09,
961
+ "learning_rate": 4.99224975489331e-05,
962
+ "loss": 1.0497,
963
+ "step": 159
964
+ },
965
+ {
966
+ "epoch": 1.1,
967
+ "learning_rate": 4.992030017576875e-05,
968
+ "loss": 1.2882,
969
+ "step": 160
970
+ },
971
+ {
972
+ "epoch": 1.1,
973
+ "learning_rate": 4.991807213594484e-05,
974
+ "loss": 1.1562,
975
+ "step": 161
976
+ },
977
+ {
978
+ "epoch": 1.11,
979
+ "learning_rate": 4.9915813432203165e-05,
980
+ "loss": 1.1235,
981
+ "step": 162
982
+ },
983
+ {
984
+ "epoch": 1.12,
985
+ "learning_rate": 4.991352406732325e-05,
986
+ "loss": 1.2257,
987
+ "step": 163
988
+ },
989
+ {
990
+ "epoch": 1.12,
991
+ "learning_rate": 4.991120404412238e-05,
992
+ "loss": 1.2162,
993
+ "step": 164
994
+ },
995
+ {
996
+ "epoch": 1.13,
997
+ "learning_rate": 4.990885336545555e-05,
998
+ "loss": 1.2173,
999
+ "step": 165
1000
+ },
1001
+ {
1002
+ "epoch": 1.14,
1003
+ "learning_rate": 4.9906472034215466e-05,
1004
+ "loss": 1.2527,
1005
+ "step": 166
1006
+ },
1007
+ {
1008
+ "epoch": 1.14,
1009
+ "learning_rate": 4.990406005333259e-05,
1010
+ "loss": 1.1254,
1011
+ "step": 167
1012
+ },
1013
+ {
1014
+ "epoch": 1.15,
1015
+ "learning_rate": 4.9901617425775067e-05,
1016
+ "loss": 1.2175,
1017
+ "step": 168
1018
+ },
1019
+ {
1020
+ "epoch": 1.16,
1021
+ "learning_rate": 4.989914415454877e-05,
1022
+ "loss": 0.9104,
1023
+ "step": 169
1024
+ },
1025
+ {
1026
+ "epoch": 1.16,
1027
+ "learning_rate": 4.9896640242697276e-05,
1028
+ "loss": 1.2549,
1029
+ "step": 170
1030
+ },
1031
+ {
1032
+ "epoch": 1.17,
1033
+ "learning_rate": 4.9894105693301896e-05,
1034
+ "loss": 1.2029,
1035
+ "step": 171
1036
+ },
1037
+ {
1038
+ "epoch": 1.18,
1039
+ "learning_rate": 4.989154050948159e-05,
1040
+ "loss": 1.1945,
1041
+ "step": 172
1042
+ },
1043
+ {
1044
+ "epoch": 1.18,
1045
+ "learning_rate": 4.988894469439306e-05,
1046
+ "loss": 1.1285,
1047
+ "step": 173
1048
+ },
1049
+ {
1050
+ "epoch": 1.19,
1051
+ "learning_rate": 4.988631825123069e-05,
1052
+ "loss": 1.2458,
1053
+ "step": 174
1054
+ },
1055
+ {
1056
+ "epoch": 1.2,
1057
+ "learning_rate": 4.988366118322655e-05,
1058
+ "loss": 1.0894,
1059
+ "step": 175
1060
+ },
1061
+ {
1062
+ "epoch": 1.21,
1063
+ "learning_rate": 4.9880973493650394e-05,
1064
+ "loss": 1.2417,
1065
+ "step": 176
1066
+ },
1067
+ {
1068
+ "epoch": 1.21,
1069
+ "learning_rate": 4.9878255185809684e-05,
1070
+ "loss": 1.2962,
1071
+ "step": 177
1072
+ },
1073
+ {
1074
+ "epoch": 1.22,
1075
+ "learning_rate": 4.987550626304952e-05,
1076
+ "loss": 1.1519,
1077
+ "step": 178
1078
+ },
1079
+ {
1080
+ "epoch": 1.23,
1081
+ "learning_rate": 4.987272672875271e-05,
1082
+ "loss": 1.116,
1083
+ "step": 179
1084
+ },
1085
+ {
1086
+ "epoch": 1.23,
1087
+ "learning_rate": 4.986991658633972e-05,
1088
+ "loss": 1.0908,
1089
+ "step": 180
1090
+ },
1091
+ {
1092
+ "epoch": 1.24,
1093
+ "learning_rate": 4.986707583926867e-05,
1094
+ "loss": 1.0814,
1095
+ "step": 181
1096
+ },
1097
+ {
1098
+ "epoch": 1.25,
1099
+ "learning_rate": 4.986420449103536e-05,
1100
+ "loss": 1.182,
1101
+ "step": 182
1102
+ },
1103
+ {
1104
+ "epoch": 1.25,
1105
+ "learning_rate": 4.986130254517325e-05,
1106
+ "loss": 1.1839,
1107
+ "step": 183
1108
+ },
1109
+ {
1110
+ "epoch": 1.26,
1111
+ "learning_rate": 4.985837000525343e-05,
1112
+ "loss": 1.0252,
1113
+ "step": 184
1114
+ },
1115
+ {
1116
+ "epoch": 1.27,
1117
+ "learning_rate": 4.985540687488466e-05,
1118
+ "loss": 1.219,
1119
+ "step": 185
1120
+ },
1121
+ {
1122
+ "epoch": 1.27,
1123
+ "learning_rate": 4.985241315771334e-05,
1124
+ "loss": 1.2573,
1125
+ "step": 186
1126
+ },
1127
+ {
1128
+ "epoch": 1.28,
1129
+ "learning_rate": 4.98493888574235e-05,
1130
+ "loss": 1.1408,
1131
+ "step": 187
1132
+ },
1133
+ {
1134
+ "epoch": 1.29,
1135
+ "learning_rate": 4.9846333977736813e-05,
1136
+ "loss": 1.2265,
1137
+ "step": 188
1138
+ },
1139
+ {
1140
+ "epoch": 1.29,
1141
+ "learning_rate": 4.984324852241259e-05,
1142
+ "loss": 1.0908,
1143
+ "step": 189
1144
+ },
1145
+ {
1146
+ "epoch": 1.3,
1147
+ "learning_rate": 4.984013249524775e-05,
1148
+ "loss": 1.2468,
1149
+ "step": 190
1150
+ },
1151
+ {
1152
+ "epoch": 1.31,
1153
+ "learning_rate": 4.9836985900076844e-05,
1154
+ "loss": 1.2172,
1155
+ "step": 191
1156
+ },
1157
+ {
1158
+ "epoch": 1.32,
1159
+ "learning_rate": 4.983380874077204e-05,
1160
+ "loss": 1.2108,
1161
+ "step": 192
1162
+ },
1163
+ {
1164
+ "epoch": 1.32,
1165
+ "learning_rate": 4.9830601021243125e-05,
1166
+ "loss": 1.0567,
1167
+ "step": 193
1168
+ },
1169
+ {
1170
+ "epoch": 1.33,
1171
+ "learning_rate": 4.982736274543748e-05,
1172
+ "loss": 1.3061,
1173
+ "step": 194
1174
+ },
1175
+ {
1176
+ "epoch": 1.34,
1177
+ "learning_rate": 4.982409391734009e-05,
1178
+ "loss": 1.1814,
1179
+ "step": 195
1180
+ },
1181
+ {
1182
+ "epoch": 1.34,
1183
+ "learning_rate": 4.982079454097354e-05,
1184
+ "loss": 1.1392,
1185
+ "step": 196
1186
+ },
1187
+ {
1188
+ "epoch": 1.35,
1189
+ "learning_rate": 4.981746462039801e-05,
1190
+ "loss": 1.3159,
1191
+ "step": 197
1192
+ },
1193
+ {
1194
+ "epoch": 1.36,
1195
+ "learning_rate": 4.981410415971127e-05,
1196
+ "loss": 1.1255,
1197
+ "step": 198
1198
+ },
1199
+ {
1200
+ "epoch": 1.36,
1201
+ "learning_rate": 4.981071316304867e-05,
1202
+ "loss": 1.2056,
1203
+ "step": 199
1204
+ },
1205
+ {
1206
+ "epoch": 1.37,
1207
+ "learning_rate": 4.980729163458312e-05,
1208
+ "loss": 1.2303,
1209
+ "step": 200
1210
+ },
1211
+ {
1212
+ "epoch": 1.38,
1213
+ "learning_rate": 4.980383957852512e-05,
1214
+ "loss": 1.1711,
1215
+ "step": 201
1216
+ },
1217
+ {
1218
+ "epoch": 1.38,
1219
+ "learning_rate": 4.9800356999122746e-05,
1220
+ "loss": 1.203,
1221
+ "step": 202
1222
+ },
1223
+ {
1224
+ "epoch": 1.39,
1225
+ "learning_rate": 4.9796843900661613e-05,
1226
+ "loss": 1.2845,
1227
+ "step": 203
1228
+ },
1229
+ {
1230
+ "epoch": 1.4,
1231
+ "learning_rate": 4.979330028746491e-05,
1232
+ "loss": 1.0433,
1233
+ "step": 204
1234
+ },
1235
+ {
1236
+ "epoch": 1.4,
1237
+ "learning_rate": 4.978972616389337e-05,
1238
+ "loss": 1.1515,
1239
+ "step": 205
1240
+ },
1241
+ {
1242
+ "epoch": 1.41,
1243
+ "learning_rate": 4.9786121534345265e-05,
1244
+ "loss": 1.1155,
1245
+ "step": 206
1246
+ },
1247
+ {
1248
+ "epoch": 1.42,
1249
+ "learning_rate": 4.978248640325641e-05,
1250
+ "loss": 1.0288,
1251
+ "step": 207
1252
+ },
1253
+ {
1254
+ "epoch": 1.42,
1255
+ "learning_rate": 4.9778820775100185e-05,
1256
+ "loss": 1.1976,
1257
+ "step": 208
1258
+ },
1259
+ {
1260
+ "epoch": 1.43,
1261
+ "learning_rate": 4.977512465438744e-05,
1262
+ "loss": 1.1593,
1263
+ "step": 209
1264
+ },
1265
+ {
1266
+ "epoch": 1.44,
1267
+ "learning_rate": 4.9771398045666606e-05,
1268
+ "loss": 1.3097,
1269
+ "step": 210
1270
+ },
1271
+ {
1272
+ "epoch": 1.45,
1273
+ "learning_rate": 4.97676409535236e-05,
1274
+ "loss": 1.1394,
1275
+ "step": 211
1276
+ },
1277
+ {
1278
+ "epoch": 1.45,
1279
+ "learning_rate": 4.976385338258186e-05,
1280
+ "loss": 1.2662,
1281
+ "step": 212
1282
+ },
1283
+ {
1284
+ "epoch": 1.46,
1285
+ "learning_rate": 4.976003533750233e-05,
1286
+ "loss": 1.1382,
1287
+ "step": 213
1288
+ },
1289
+ {
1290
+ "epoch": 1.47,
1291
+ "learning_rate": 4.9756186822983464e-05,
1292
+ "loss": 1.248,
1293
+ "step": 214
1294
+ },
1295
+ {
1296
+ "epoch": 1.47,
1297
+ "learning_rate": 4.975230784376119e-05,
1298
+ "loss": 1.1389,
1299
+ "step": 215
1300
+ },
1301
+ {
1302
+ "epoch": 1.48,
1303
+ "learning_rate": 4.974839840460895e-05,
1304
+ "loss": 1.0841,
1305
+ "step": 216
1306
+ },
1307
+ {
1308
+ "epoch": 1.49,
1309
+ "learning_rate": 4.974445851033765e-05,
1310
+ "loss": 1.317,
1311
+ "step": 217
1312
+ },
1313
+ {
1314
+ "epoch": 1.49,
1315
+ "learning_rate": 4.97404881657957e-05,
1316
+ "loss": 1.1366,
1317
+ "step": 218
1318
+ },
1319
+ {
1320
+ "epoch": 1.5,
1321
+ "learning_rate": 4.973648737586894e-05,
1322
+ "loss": 1.2806,
1323
+ "step": 219
1324
+ },
1325
+ {
1326
+ "epoch": 1.51,
1327
+ "learning_rate": 4.973245614548072e-05,
1328
+ "loss": 1.1281,
1329
+ "step": 220
1330
+ },
1331
+ {
1332
+ "epoch": 1.51,
1333
+ "learning_rate": 4.972839447959181e-05,
1334
+ "loss": 1.2346,
1335
+ "step": 221
1336
+ },
1337
+ {
1338
+ "epoch": 1.52,
1339
+ "learning_rate": 4.9724302383200477e-05,
1340
+ "loss": 1.2117,
1341
+ "step": 222
1342
+ },
1343
+ {
1344
+ "epoch": 1.53,
1345
+ "learning_rate": 4.97201798613424e-05,
1346
+ "loss": 1.1455,
1347
+ "step": 223
1348
+ },
1349
+ {
1350
+ "epoch": 1.53,
1351
+ "learning_rate": 4.9716026919090705e-05,
1352
+ "loss": 1.0981,
1353
+ "step": 224
1354
+ },
1355
+ {
1356
+ "epoch": 1.54,
1357
+ "learning_rate": 4.971184356155597e-05,
1358
+ "loss": 1.023,
1359
+ "step": 225
1360
+ },
1361
+ {
1362
+ "epoch": 1.55,
1363
+ "learning_rate": 4.970762979388618e-05,
1364
+ "loss": 1.2858,
1365
+ "step": 226
1366
+ },
1367
+ {
1368
+ "epoch": 1.55,
1369
+ "learning_rate": 4.9703385621266766e-05,
1370
+ "loss": 1.0097,
1371
+ "step": 227
1372
+ },
1373
+ {
1374
+ "epoch": 1.56,
1375
+ "learning_rate": 4.9699111048920554e-05,
1376
+ "loss": 1.0166,
1377
+ "step": 228
1378
+ },
1379
+ {
1380
+ "epoch": 1.57,
1381
+ "learning_rate": 4.969480608210779e-05,
1382
+ "loss": 1.3232,
1383
+ "step": 229
1384
+ },
1385
+ {
1386
+ "epoch": 1.58,
1387
+ "learning_rate": 4.9690470726126115e-05,
1388
+ "loss": 0.9815,
1389
+ "step": 230
1390
+ },
1391
+ {
1392
+ "epoch": 1.58,
1393
+ "learning_rate": 4.968610498631058e-05,
1394
+ "loss": 1.175,
1395
+ "step": 231
1396
+ },
1397
+ {
1398
+ "epoch": 1.59,
1399
+ "learning_rate": 4.9681708868033616e-05,
1400
+ "loss": 1.0065,
1401
+ "step": 232
1402
+ },
1403
+ {
1404
+ "epoch": 1.6,
1405
+ "learning_rate": 4.967728237670504e-05,
1406
+ "loss": 1.0978,
1407
+ "step": 233
1408
+ },
1409
+ {
1410
+ "epoch": 1.6,
1411
+ "learning_rate": 4.967282551777205e-05,
1412
+ "loss": 1.0995,
1413
+ "step": 234
1414
+ },
1415
+ {
1416
+ "epoch": 1.61,
1417
+ "learning_rate": 4.96683382967192e-05,
1418
+ "loss": 1.1151,
1419
+ "step": 235
1420
+ },
1421
+ {
1422
+ "epoch": 1.62,
1423
+ "learning_rate": 4.966382071906843e-05,
1424
+ "loss": 1.0575,
1425
+ "step": 236
1426
+ },
1427
+ {
1428
+ "epoch": 1.62,
1429
+ "learning_rate": 4.965927279037901e-05,
1430
+ "loss": 0.8452,
1431
+ "step": 237
1432
+ },
1433
+ {
1434
+ "epoch": 1.63,
1435
+ "learning_rate": 4.965469451624759e-05,
1436
+ "loss": 1.1209,
1437
+ "step": 238
1438
+ },
1439
+ {
1440
+ "epoch": 1.64,
1441
+ "learning_rate": 4.965008590230812e-05,
1442
+ "loss": 1.171,
1443
+ "step": 239
1444
+ },
1445
+ {
1446
+ "epoch": 1.64,
1447
+ "learning_rate": 4.9645446954231936e-05,
1448
+ "loss": 1.0801,
1449
+ "step": 240
1450
+ },
1451
+ {
1452
+ "epoch": 1.65,
1453
+ "learning_rate": 4.9640777677727674e-05,
1454
+ "loss": 1.1266,
1455
+ "step": 241
1456
+ },
1457
+ {
1458
+ "epoch": 1.66,
1459
+ "learning_rate": 4.963607807854129e-05,
1460
+ "loss": 1.2221,
1461
+ "step": 242
1462
+ },
1463
+ {
1464
+ "epoch": 1.66,
1465
+ "learning_rate": 4.963134816245606e-05,
1466
+ "loss": 1.1198,
1467
+ "step": 243
1468
+ },
1469
+ {
1470
+ "epoch": 1.67,
1471
+ "learning_rate": 4.962658793529258e-05,
1472
+ "loss": 1.1369,
1473
+ "step": 244
1474
+ },
1475
+ {
1476
+ "epoch": 1.68,
1477
+ "learning_rate": 4.962179740290873e-05,
1478
+ "loss": 1.0799,
1479
+ "step": 245
1480
+ },
1481
+ {
1482
+ "epoch": 1.68,
1483
+ "learning_rate": 4.961697657119968e-05,
1484
+ "loss": 1.1408,
1485
+ "step": 246
1486
+ },
1487
+ {
1488
+ "epoch": 1.69,
1489
+ "learning_rate": 4.9612125446097905e-05,
1490
+ "loss": 1.1629,
1491
+ "step": 247
1492
+ },
1493
+ {
1494
+ "epoch": 1.7,
1495
+ "learning_rate": 4.9607244033573156e-05,
1496
+ "loss": 1.0933,
1497
+ "step": 248
1498
+ },
1499
+ {
1500
+ "epoch": 1.71,
1501
+ "learning_rate": 4.960233233963242e-05,
1502
+ "loss": 1.1474,
1503
+ "step": 249
1504
+ },
1505
+ {
1506
+ "epoch": 1.71,
1507
+ "learning_rate": 4.9597390370320006e-05,
1508
+ "loss": 1.3268,
1509
+ "step": 250
1510
+ },
1511
+ {
1512
+ "epoch": 1.72,
1513
+ "learning_rate": 4.959241813171743e-05,
1514
+ "loss": 1.1895,
1515
+ "step": 251
1516
+ },
1517
+ {
1518
+ "epoch": 1.73,
1519
+ "learning_rate": 4.95874156299435e-05,
1520
+ "loss": 1.0743,
1521
+ "step": 252
1522
+ },
1523
+ {
1524
+ "epoch": 1.73,
1525
+ "learning_rate": 4.958238287115421e-05,
1526
+ "loss": 1.16,
1527
+ "step": 253
1528
+ },
1529
+ {
1530
+ "epoch": 1.74,
1531
+ "learning_rate": 4.957731986154285e-05,
1532
+ "loss": 1.2787,
1533
+ "step": 254
1534
+ },
1535
+ {
1536
+ "epoch": 1.75,
1537
+ "learning_rate": 4.9572226607339886e-05,
1538
+ "loss": 1.0935,
1539
+ "step": 255
1540
+ },
1541
+ {
1542
+ "epoch": 1.75,
1543
+ "learning_rate": 4.956710311481303e-05,
1544
+ "loss": 1.0916,
1545
+ "step": 256
1546
+ },
1547
+ {
1548
+ "epoch": 1.76,
1549
+ "learning_rate": 4.9561949390267196e-05,
1550
+ "loss": 1.1169,
1551
+ "step": 257
1552
+ },
1553
+ {
1554
+ "epoch": 1.77,
1555
+ "learning_rate": 4.95567654400445e-05,
1556
+ "loss": 1.0298,
1557
+ "step": 258
1558
+ },
1559
+ {
1560
+ "epoch": 1.77,
1561
+ "learning_rate": 4.955155127052428e-05,
1562
+ "loss": 1.1667,
1563
+ "step": 259
1564
+ },
1565
+ {
1566
+ "epoch": 1.78,
1567
+ "learning_rate": 4.9546306888123003e-05,
1568
+ "loss": 1.1274,
1569
+ "step": 260
1570
+ },
1571
+ {
1572
+ "epoch": 1.79,
1573
+ "learning_rate": 4.9541032299294375e-05,
1574
+ "loss": 1.1245,
1575
+ "step": 261
1576
+ },
1577
+ {
1578
+ "epoch": 1.79,
1579
+ "learning_rate": 4.953572751052924e-05,
1580
+ "loss": 1.0894,
1581
+ "step": 262
1582
+ },
1583
+ {
1584
+ "epoch": 1.8,
1585
+ "learning_rate": 4.9530392528355626e-05,
1586
+ "loss": 1.1729,
1587
+ "step": 263
1588
+ },
1589
+ {
1590
+ "epoch": 1.81,
1591
+ "learning_rate": 4.9525027359338696e-05,
1592
+ "loss": 1.096,
1593
+ "step": 264
1594
+ },
1595
+ {
1596
+ "epoch": 1.82,
1597
+ "learning_rate": 4.951963201008076e-05,
1598
+ "loss": 1.1825,
1599
+ "step": 265
1600
+ },
1601
+ {
1602
+ "epoch": 1.82,
1603
+ "learning_rate": 4.9514206487221304e-05,
1604
+ "loss": 1.0211,
1605
+ "step": 266
1606
+ },
1607
+ {
1608
+ "epoch": 1.83,
1609
+ "learning_rate": 4.95087507974369e-05,
1610
+ "loss": 1.2285,
1611
+ "step": 267
1612
+ },
1613
+ {
1614
+ "epoch": 1.84,
1615
+ "learning_rate": 4.9503264947441275e-05,
1616
+ "loss": 1.2042,
1617
+ "step": 268
1618
+ },
1619
+ {
1620
+ "epoch": 1.84,
1621
+ "learning_rate": 4.949774894398524e-05,
1622
+ "loss": 1.088,
1623
+ "step": 269
1624
+ },
1625
+ {
1626
+ "epoch": 1.85,
1627
+ "learning_rate": 4.949220279385676e-05,
1628
+ "loss": 1.1779,
1629
+ "step": 270
1630
+ },
1631
+ {
1632
+ "epoch": 1.86,
1633
+ "learning_rate": 4.948662650388084e-05,
1634
+ "loss": 0.9534,
1635
+ "step": 271
1636
+ },
1637
+ {
1638
+ "epoch": 1.86,
1639
+ "learning_rate": 4.948102008091962e-05,
1640
+ "loss": 1.1841,
1641
+ "step": 272
1642
+ },
1643
+ {
1644
+ "epoch": 1.87,
1645
+ "learning_rate": 4.94753835318723e-05,
1646
+ "loss": 1.1233,
1647
+ "step": 273
1648
+ },
1649
+ {
1650
+ "epoch": 1.88,
1651
+ "learning_rate": 4.946971686367516e-05,
1652
+ "loss": 1.1338,
1653
+ "step": 274
1654
+ },
1655
+ {
1656
+ "epoch": 1.88,
1657
+ "learning_rate": 4.9464020083301544e-05,
1658
+ "loss": 1.1575,
1659
+ "step": 275
1660
+ },
1661
+ {
1662
+ "epoch": 1.89,
1663
+ "learning_rate": 4.945829319776184e-05,
1664
+ "loss": 1.0758,
1665
+ "step": 276
1666
+ },
1667
+ {
1668
+ "epoch": 1.9,
1669
+ "learning_rate": 4.945253621410351e-05,
1670
+ "loss": 1.172,
1671
+ "step": 277
1672
+ },
1673
+ {
1674
+ "epoch": 1.9,
1675
+ "learning_rate": 4.944674913941102e-05,
1676
+ "loss": 1.2216,
1677
+ "step": 278
1678
+ },
1679
+ {
1680
+ "epoch": 1.91,
1681
+ "learning_rate": 4.944093198080589e-05,
1682
+ "loss": 1.2167,
1683
+ "step": 279
1684
+ },
1685
+ {
1686
+ "epoch": 1.92,
1687
+ "learning_rate": 4.9435084745446666e-05,
1688
+ "loss": 1.0337,
1689
+ "step": 280
1690
+ },
1691
+ {
1692
+ "epoch": 1.92,
1693
+ "learning_rate": 4.9429207440528876e-05,
1694
+ "loss": 1.0044,
1695
+ "step": 281
1696
+ },
1697
+ {
1698
+ "epoch": 1.93,
1699
+ "learning_rate": 4.942330007328509e-05,
1700
+ "loss": 1.1916,
1701
+ "step": 282
1702
+ },
1703
+ {
1704
+ "epoch": 1.94,
1705
+ "learning_rate": 4.9417362650984835e-05,
1706
+ "loss": 0.9445,
1707
+ "step": 283
1708
+ },
1709
+ {
1710
+ "epoch": 1.95,
1711
+ "learning_rate": 4.941139518093464e-05,
1712
+ "loss": 1.3121,
1713
+ "step": 284
1714
+ },
1715
+ {
1716
+ "epoch": 1.95,
1717
+ "learning_rate": 4.9405397670478014e-05,
1718
+ "loss": 1.1049,
1719
+ "step": 285
1720
+ },
1721
+ {
1722
+ "epoch": 1.96,
1723
+ "learning_rate": 4.9399370126995444e-05,
1724
+ "loss": 1.062,
1725
+ "step": 286
1726
+ },
1727
+ {
1728
+ "epoch": 1.97,
1729
+ "learning_rate": 4.939331255790434e-05,
1730
+ "loss": 1.2379,
1731
+ "step": 287
1732
+ },
1733
+ {
1734
+ "epoch": 1.97,
1735
+ "learning_rate": 4.93872249706591e-05,
1736
+ "loss": 1.0523,
1737
+ "step": 288
1738
+ },
1739
+ {
1740
+ "epoch": 1.98,
1741
+ "learning_rate": 4.938110737275104e-05,
1742
+ "loss": 0.9423,
1743
+ "step": 289
1744
+ },
1745
+ {
1746
+ "epoch": 1.99,
1747
+ "learning_rate": 4.937495977170841e-05,
1748
+ "loss": 1.1224,
1749
+ "step": 290
1750
+ },
1751
+ {
1752
+ "epoch": 1.99,
1753
+ "learning_rate": 4.936878217509637e-05,
1754
+ "loss": 1.2134,
1755
+ "step": 291
1756
+ },
1757
+ {
1758
+ "epoch": 2.0,
1759
+ "learning_rate": 4.936257459051703e-05,
1760
+ "loss": 0.9984,
1761
+ "step": 292
1762
+ },
1763
+ {
1764
+ "epoch": 2.01,
1765
+ "learning_rate": 4.9356337025609365e-05,
1766
+ "loss": 1.1952,
1767
+ "step": 293
1768
+ },
1769
+ {
1770
+ "epoch": 2.01,
1771
+ "learning_rate": 4.935006948804927e-05,
1772
+ "loss": 1.0955,
1773
+ "step": 294
1774
+ },
1775
+ {
1776
+ "epoch": 2.02,
1777
+ "learning_rate": 4.934377198554948e-05,
1778
+ "loss": 1.1969,
1779
+ "step": 295
1780
+ },
1781
+ {
1782
+ "epoch": 2.03,
1783
+ "learning_rate": 4.933744452585966e-05,
1784
+ "loss": 1.2647,
1785
+ "step": 296
1786
+ },
1787
+ {
1788
+ "epoch": 2.03,
1789
+ "learning_rate": 4.933108711676632e-05,
1790
+ "loss": 1.0782,
1791
+ "step": 297
1792
+ },
1793
+ {
1794
+ "epoch": 2.04,
1795
+ "learning_rate": 4.9324699766092806e-05,
1796
+ "loss": 1.0923,
1797
+ "step": 298
1798
+ },
1799
+ {
1800
+ "epoch": 2.05,
1801
+ "learning_rate": 4.931828248169933e-05,
1802
+ "loss": 1.1035,
1803
+ "step": 299
1804
+ },
1805
+ {
1806
+ "epoch": 2.05,
1807
+ "learning_rate": 4.9311835271482943e-05,
1808
+ "loss": 1.1281,
1809
+ "step": 300
1810
+ },
1811
+ {
1812
+ "epoch": 2.06,
1813
+ "learning_rate": 4.93053581433775e-05,
1814
+ "loss": 0.8405,
1815
+ "step": 301
1816
+ },
1817
+ {
1818
+ "epoch": 2.07,
1819
+ "learning_rate": 4.9298851105353696e-05,
1820
+ "loss": 1.1676,
1821
+ "step": 302
1822
+ },
1823
+ {
1824
+ "epoch": 2.08,
1825
+ "learning_rate": 4.929231416541901e-05,
1826
+ "loss": 1.192,
1827
+ "step": 303
1828
+ },
1829
+ {
1830
+ "epoch": 2.08,
1831
+ "learning_rate": 4.9285747331617746e-05,
1832
+ "loss": 1.183,
1833
+ "step": 304
1834
+ },
1835
+ {
1836
+ "epoch": 2.09,
1837
+ "learning_rate": 4.927915061203099e-05,
1838
+ "loss": 0.9712,
1839
+ "step": 305
1840
+ },
1841
+ {
1842
+ "epoch": 2.1,
1843
+ "learning_rate": 4.927252401477657e-05,
1844
+ "loss": 1.1965,
1845
+ "step": 306
1846
+ },
1847
+ {
1848
+ "epoch": 2.1,
1849
+ "learning_rate": 4.926586754800912e-05,
1850
+ "loss": 1.0713,
1851
+ "step": 307
1852
+ },
1853
+ {
1854
+ "epoch": 2.11,
1855
+ "learning_rate": 4.925918121992002e-05,
1856
+ "loss": 1.0439,
1857
+ "step": 308
1858
+ },
1859
+ {
1860
+ "epoch": 2.12,
1861
+ "learning_rate": 4.925246503873739e-05,
1862
+ "loss": 1.1367,
1863
+ "step": 309
1864
+ },
1865
+ {
1866
+ "epoch": 2.12,
1867
+ "learning_rate": 4.92457190127261e-05,
1868
+ "loss": 1.1343,
1869
+ "step": 310
1870
+ },
1871
+ {
1872
+ "epoch": 2.13,
1873
+ "learning_rate": 4.923894315018773e-05,
1874
+ "loss": 1.1216,
1875
+ "step": 311
1876
+ },
1877
+ {
1878
+ "epoch": 2.14,
1879
+ "learning_rate": 4.923213745946059e-05,
1880
+ "loss": 1.1822,
1881
+ "step": 312
1882
+ },
1883
+ {
1884
+ "epoch": 2.14,
1885
+ "learning_rate": 4.922530194891969e-05,
1886
+ "loss": 1.0442,
1887
+ "step": 313
1888
+ },
1889
+ {
1890
+ "epoch": 2.15,
1891
+ "learning_rate": 4.921843662697673e-05,
1892
+ "loss": 1.139,
1893
+ "step": 314
1894
+ },
1895
+ {
1896
+ "epoch": 2.16,
1897
+ "learning_rate": 4.921154150208012e-05,
1898
+ "loss": 0.8287,
1899
+ "step": 315
1900
+ },
1901
+ {
1902
+ "epoch": 2.16,
1903
+ "learning_rate": 4.920461658271492e-05,
1904
+ "loss": 1.1756,
1905
+ "step": 316
1906
+ },
1907
+ {
1908
+ "epoch": 2.17,
1909
+ "learning_rate": 4.919766187740286e-05,
1910
+ "loss": 1.1312,
1911
+ "step": 317
1912
+ },
1913
+ {
1914
+ "epoch": 2.18,
1915
+ "learning_rate": 4.9190677394702324e-05,
1916
+ "loss": 1.122,
1917
+ "step": 318
1918
+ },
1919
+ {
1920
+ "epoch": 2.18,
1921
+ "learning_rate": 4.9183663143208355e-05,
1922
+ "loss": 1.0514,
1923
+ "step": 319
1924
+ },
1925
+ {
1926
+ "epoch": 2.19,
1927
+ "learning_rate": 4.9176619131552604e-05,
1928
+ "loss": 1.1719,
1929
+ "step": 320
1930
+ },
1931
+ {
1932
+ "epoch": 2.2,
1933
+ "learning_rate": 4.916954536840337e-05,
1934
+ "loss": 1.0202,
1935
+ "step": 321
1936
+ },
1937
+ {
1938
+ "epoch": 2.21,
1939
+ "learning_rate": 4.916244186246555e-05,
1940
+ "loss": 1.1547,
1941
+ "step": 322
1942
+ },
1943
+ {
1944
+ "epoch": 2.21,
1945
+ "learning_rate": 4.9155308622480625e-05,
1946
+ "loss": 1.2142,
1947
+ "step": 323
1948
+ },
1949
+ {
1950
+ "epoch": 2.22,
1951
+ "learning_rate": 4.914814565722671e-05,
1952
+ "loss": 1.0676,
1953
+ "step": 324
1954
+ },
1955
+ {
1956
+ "epoch": 2.23,
1957
+ "learning_rate": 4.9140952975518465e-05,
1958
+ "loss": 1.0258,
1959
+ "step": 325
1960
+ },
1961
+ {
1962
+ "epoch": 2.23,
1963
+ "learning_rate": 4.9133730586207124e-05,
1964
+ "loss": 1.0105,
1965
+ "step": 326
1966
+ },
1967
+ {
1968
+ "epoch": 2.24,
1969
+ "learning_rate": 4.9133730586207124e-05,
1970
+ "loss": 0.9735,
1971
+ "step": 327
1972
+ },
1973
+ {
1974
+ "epoch": 2.25,
1975
+ "learning_rate": 4.912647849818049e-05,
1976
+ "loss": 1.1046,
1977
+ "step": 328
1978
+ },
1979
+ {
1980
+ "epoch": 2.25,
1981
+ "learning_rate": 4.91191967203629e-05,
1982
+ "loss": 1.1104,
1983
+ "step": 329
1984
+ },
1985
+ {
1986
+ "epoch": 2.26,
1987
+ "learning_rate": 4.911188526171524e-05,
1988
+ "loss": 0.9328,
1989
+ "step": 330
1990
+ },
1991
+ {
1992
+ "epoch": 2.27,
1993
+ "learning_rate": 4.910454413123491e-05,
1994
+ "loss": 1.1438,
1995
+ "step": 331
1996
+ },
1997
+ {
1998
+ "epoch": 2.27,
1999
+ "learning_rate": 4.909717333795584e-05,
2000
+ "loss": 1.1995,
2001
+ "step": 332
2002
+ },
2003
+ {
2004
+ "epoch": 2.28,
2005
+ "learning_rate": 4.908977289094843e-05,
2006
+ "loss": 1.057,
2007
+ "step": 333
2008
+ },
2009
+ {
2010
+ "epoch": 2.29,
2011
+ "learning_rate": 4.90823427993196e-05,
2012
+ "loss": 1.1436,
2013
+ "step": 334
2014
+ },
2015
+ {
2016
+ "epoch": 2.29,
2017
+ "learning_rate": 4.9074883072212744e-05,
2018
+ "loss": 1.0117,
2019
+ "step": 335
2020
+ },
2021
+ {
2022
+ "epoch": 2.3,
2023
+ "learning_rate": 4.906739371880773e-05,
2024
+ "loss": 1.1685,
2025
+ "step": 336
2026
+ },
2027
+ {
2028
+ "epoch": 2.31,
2029
+ "learning_rate": 4.9059874748320876e-05,
2030
+ "loss": 1.1363,
2031
+ "step": 337
2032
+ },
2033
+ {
2034
+ "epoch": 2.32,
2035
+ "learning_rate": 4.9052326170004936e-05,
2036
+ "loss": 1.1357,
2037
+ "step": 338
2038
+ },
2039
+ {
2040
+ "epoch": 2.32,
2041
+ "learning_rate": 4.904474799314913e-05,
2042
+ "loss": 0.974,
2043
+ "step": 339
2044
+ },
2045
+ {
2046
+ "epoch": 2.33,
2047
+ "learning_rate": 4.9037140227079065e-05,
2048
+ "loss": 1.2257,
2049
+ "step": 340
2050
+ },
2051
+ {
2052
+ "epoch": 2.34,
2053
+ "learning_rate": 4.902950288115679e-05,
2054
+ "loss": 1.1061,
2055
+ "step": 341
2056
+ },
2057
+ {
2058
+ "epoch": 2.34,
2059
+ "learning_rate": 4.902183596478073e-05,
2060
+ "loss": 1.0701,
2061
+ "step": 342
2062
+ },
2063
+ {
2064
+ "epoch": 2.35,
2065
+ "learning_rate": 4.901413948738572e-05,
2066
+ "loss": 1.2365,
2067
+ "step": 343
2068
+ },
2069
+ {
2070
+ "epoch": 2.36,
2071
+ "learning_rate": 4.900641345844298e-05,
2072
+ "loss": 1.0575,
2073
+ "step": 344
2074
+ },
2075
+ {
2076
+ "epoch": 2.36,
2077
+ "learning_rate": 4.899865788746005e-05,
2078
+ "loss": 1.1325,
2079
+ "step": 345
2080
+ },
2081
+ {
2082
+ "epoch": 2.37,
2083
+ "learning_rate": 4.899087278398088e-05,
2084
+ "loss": 1.1504,
2085
+ "step": 346
2086
+ },
2087
+ {
2088
+ "epoch": 2.38,
2089
+ "learning_rate": 4.8983058157585717e-05,
2090
+ "loss": 1.0893,
2091
+ "step": 347
2092
+ },
2093
+ {
2094
+ "epoch": 2.38,
2095
+ "learning_rate": 4.897521401789116e-05,
2096
+ "loss": 1.1198,
2097
+ "step": 348
2098
+ },
2099
+ {
2100
+ "epoch": 2.39,
2101
+ "learning_rate": 4.896734037455014e-05,
2102
+ "loss": 1.1958,
2103
+ "step": 349
2104
+ },
2105
+ {
2106
+ "epoch": 2.4,
2107
+ "learning_rate": 4.895943723725187e-05,
2108
+ "loss": 0.958,
2109
+ "step": 350
2110
+ },
2111
+ {
2112
+ "epoch": 2.4,
2113
+ "learning_rate": 4.895150461572187e-05,
2114
+ "loss": 1.0629,
2115
+ "step": 351
2116
+ },
2117
+ {
2118
+ "epoch": 2.41,
2119
+ "learning_rate": 4.894354251972193e-05,
2120
+ "loss": 1.0446,
2121
+ "step": 352
2122
+ },
2123
+ {
2124
+ "epoch": 2.42,
2125
+ "learning_rate": 4.893555095905014e-05,
2126
+ "loss": 0.9373,
2127
+ "step": 353
2128
+ },
2129
+ {
2130
+ "epoch": 2.42,
2131
+ "learning_rate": 4.892752994354082e-05,
2132
+ "loss": 1.11,
2133
+ "step": 354
2134
+ },
2135
+ {
2136
+ "epoch": 2.43,
2137
+ "learning_rate": 4.8919479483064544e-05,
2138
+ "loss": 1.0709,
2139
+ "step": 355
2140
+ },
2141
+ {
2142
+ "epoch": 2.44,
2143
+ "learning_rate": 4.891139958752813e-05,
2144
+ "loss": 1.2324,
2145
+ "step": 356
2146
+ },
2147
+ {
2148
+ "epoch": 2.45,
2149
+ "learning_rate": 4.890329026687462e-05,
2150
+ "loss": 1.0608,
2151
+ "step": 357
2152
+ },
2153
+ {
2154
+ "epoch": 2.45,
2155
+ "learning_rate": 4.889515153108324e-05,
2156
+ "loss": 1.1808,
2157
+ "step": 358
2158
+ },
2159
+ {
2160
+ "epoch": 2.46,
2161
+ "learning_rate": 4.888698339016945e-05,
2162
+ "loss": 1.0626,
2163
+ "step": 359
2164
+ },
2165
+ {
2166
+ "epoch": 2.47,
2167
+ "learning_rate": 4.887878585418487e-05,
2168
+ "loss": 1.1625,
2169
+ "step": 360
2170
+ },
2171
+ {
2172
+ "epoch": 2.47,
2173
+ "learning_rate": 4.88705589332173e-05,
2174
+ "loss": 1.0463,
2175
+ "step": 361
2176
+ },
2177
+ {
2178
+ "epoch": 2.48,
2179
+ "learning_rate": 4.8862302637390714e-05,
2180
+ "loss": 0.9999,
2181
+ "step": 362
2182
+ },
2183
+ {
2184
+ "epoch": 2.49,
2185
+ "learning_rate": 4.88540169768652e-05,
2186
+ "loss": 1.2328,
2187
+ "step": 363
2188
+ },
2189
+ {
2190
+ "epoch": 2.49,
2191
+ "learning_rate": 4.884570196183703e-05,
2192
+ "loss": 1.063,
2193
+ "step": 364
2194
+ },
2195
+ {
2196
+ "epoch": 2.5,
2197
+ "learning_rate": 4.883735760253856e-05,
2198
+ "loss": 1.1949,
2199
+ "step": 365
2200
+ },
2201
+ {
2202
+ "epoch": 2.51,
2203
+ "learning_rate": 4.882898390923828e-05,
2204
+ "loss": 1.0521,
2205
+ "step": 366
2206
+ },
2207
+ {
2208
+ "epoch": 2.51,
2209
+ "learning_rate": 4.882058089224075e-05,
2210
+ "loss": 1.1375,
2211
+ "step": 367
2212
+ },
2213
+ {
2214
+ "epoch": 2.52,
2215
+ "learning_rate": 4.881214856188666e-05,
2216
+ "loss": 1.1363,
2217
+ "step": 368
2218
+ },
2219
+ {
2220
+ "epoch": 2.53,
2221
+ "learning_rate": 4.8803686928552736e-05,
2222
+ "loss": 1.0669,
2223
+ "step": 369
2224
+ },
2225
+ {
2226
+ "epoch": 2.53,
2227
+ "learning_rate": 4.879519600265177e-05,
2228
+ "loss": 1.0104,
2229
+ "step": 370
2230
+ },
2231
+ {
2232
+ "epoch": 2.54,
2233
+ "learning_rate": 4.8786675794632606e-05,
2234
+ "loss": 0.943,
2235
+ "step": 371
2236
+ },
2237
+ {
2238
+ "epoch": 2.55,
2239
+ "learning_rate": 4.8778126314980136e-05,
2240
+ "loss": 1.2132,
2241
+ "step": 372
2242
+ },
2243
+ {
2244
+ "epoch": 2.55,
2245
+ "learning_rate": 4.876954757421523e-05,
2246
+ "loss": 0.9231,
2247
+ "step": 373
2248
+ },
2249
+ {
2250
+ "epoch": 2.56,
2251
+ "learning_rate": 4.876093958289484e-05,
2252
+ "loss": 0.9422,
2253
+ "step": 374
2254
+ },
2255
+ {
2256
+ "epoch": 2.57,
2257
+ "learning_rate": 4.8752302351611836e-05,
2258
+ "loss": 1.2366,
2259
+ "step": 375
2260
+ },
2261
+ {
2262
+ "epoch": 2.58,
2263
+ "learning_rate": 4.8743635890995124e-05,
2264
+ "loss": 0.904,
2265
+ "step": 376
2266
+ },
2267
+ {
2268
+ "epoch": 2.58,
2269
+ "learning_rate": 4.873494021170953e-05,
2270
+ "loss": 1.0879,
2271
+ "step": 377
2272
+ },
2273
+ {
2274
+ "epoch": 2.59,
2275
+ "learning_rate": 4.8726215324455905e-05,
2276
+ "loss": 0.9262,
2277
+ "step": 378
2278
+ },
2279
+ {
2280
+ "epoch": 2.6,
2281
+ "learning_rate": 4.8717461239970975e-05,
2282
+ "loss": 1.0145,
2283
+ "step": 379
2284
+ },
2285
+ {
2286
+ "epoch": 2.6,
2287
+ "learning_rate": 4.870867796902743e-05,
2288
+ "loss": 1.0186,
2289
+ "step": 380
2290
+ },
2291
+ {
2292
+ "epoch": 2.61,
2293
+ "learning_rate": 4.8699865522433884e-05,
2294
+ "loss": 1.0292,
2295
+ "step": 381
2296
+ },
2297
+ {
2298
+ "epoch": 2.62,
2299
+ "learning_rate": 4.8691023911034826e-05,
2300
+ "loss": 0.9748,
2301
+ "step": 382
2302
+ },
2303
+ {
2304
+ "epoch": 2.62,
2305
+ "learning_rate": 4.868215314571065e-05,
2306
+ "loss": 0.7651,
2307
+ "step": 383
2308
+ },
2309
+ {
2310
+ "epoch": 2.63,
2311
+ "learning_rate": 4.867325323737765e-05,
2312
+ "loss": 1.0372,
2313
+ "step": 384
2314
+ },
2315
+ {
2316
+ "epoch": 2.64,
2317
+ "learning_rate": 4.866432419698792e-05,
2318
+ "loss": 1.0903,
2319
+ "step": 385
2320
+ },
2321
+ {
2322
+ "epoch": 2.64,
2323
+ "learning_rate": 4.8655366035529483e-05,
2324
+ "loss": 1.0043,
2325
+ "step": 386
2326
+ },
2327
+ {
2328
+ "epoch": 2.65,
2329
+ "learning_rate": 4.8646378764026126e-05,
2330
+ "loss": 1.0399,
2331
+ "step": 387
2332
+ },
2333
+ {
2334
+ "epoch": 2.66,
2335
+ "learning_rate": 4.863736239353752e-05,
2336
+ "loss": 1.1461,
2337
+ "step": 388
2338
+ },
2339
+ {
2340
+ "epoch": 2.66,
2341
+ "learning_rate": 4.862831693515909e-05,
2342
+ "loss": 1.0312,
2343
+ "step": 389
2344
+ },
2345
+ {
2346
+ "epoch": 2.67,
2347
+ "learning_rate": 4.8619242400022096e-05,
2348
+ "loss": 1.0527,
2349
+ "step": 390
2350
+ },
2351
+ {
2352
+ "epoch": 2.68,
2353
+ "learning_rate": 4.861013879929357e-05,
2354
+ "loss": 0.9872,
2355
+ "step": 391
2356
+ },
2357
+ {
2358
+ "epoch": 2.68,
2359
+ "learning_rate": 4.8601006144176284e-05,
2360
+ "loss": 1.0633,
2361
+ "step": 392
2362
+ },
2363
+ {
2364
+ "epoch": 2.69,
2365
+ "learning_rate": 4.859184444590882e-05,
2366
+ "loss": 1.0893,
2367
+ "step": 393
2368
+ },
2369
+ {
2370
+ "epoch": 2.7,
2371
+ "learning_rate": 4.859184444590882e-05,
2372
+ "loss": 1.0137,
2373
+ "step": 394
2374
+ },
2375
+ {
2376
+ "epoch": 2.71,
2377
+ "learning_rate": 4.858265371576545e-05,
2378
+ "loss": 1.0704,
2379
+ "step": 395
2380
+ },
2381
+ {
2382
+ "epoch": 2.71,
2383
+ "learning_rate": 4.857343396505618e-05,
2384
+ "loss": 1.2629,
2385
+ "step": 396
2386
+ },
2387
+ {
2388
+ "epoch": 2.72,
2389
+ "learning_rate": 4.856418520512676e-05,
2390
+ "loss": 1.1081,
2391
+ "step": 397
2392
+ },
2393
+ {
2394
+ "epoch": 2.73,
2395
+ "learning_rate": 4.85549074473586e-05,
2396
+ "loss": 0.9953,
2397
+ "step": 398
2398
+ },
2399
+ {
2400
+ "epoch": 2.73,
2401
+ "learning_rate": 4.8545600703168824e-05,
2402
+ "loss": 1.0844,
2403
+ "step": 399
2404
+ },
2405
+ {
2406
+ "epoch": 2.74,
2407
+ "learning_rate": 4.85362649840102e-05,
2408
+ "loss": 1.1951,
2409
+ "step": 400
2410
+ },
2411
+ {
2412
+ "epoch": 2.75,
2413
+ "learning_rate": 4.852690030137118e-05,
2414
+ "loss": 1.018,
2415
+ "step": 401
2416
+ },
2417
+ {
2418
+ "epoch": 2.75,
2419
+ "learning_rate": 4.851750666677584e-05,
2420
+ "loss": 1.0113,
2421
+ "step": 402
2422
+ },
2423
+ {
2424
+ "epoch": 2.76,
2425
+ "learning_rate": 4.8508084091783874e-05,
2426
+ "loss": 1.039,
2427
+ "step": 403
2428
+ },
2429
+ {
2430
+ "epoch": 2.77,
2431
+ "learning_rate": 4.8498632587990625e-05,
2432
+ "loss": 0.953,
2433
+ "step": 404
2434
+ },
2435
+ {
2436
+ "epoch": 2.77,
2437
+ "learning_rate": 4.8489152167026995e-05,
2438
+ "loss": 1.1071,
2439
+ "step": 405
2440
+ },
2441
+ {
2442
+ "epoch": 2.78,
2443
+ "learning_rate": 4.8479642840559505e-05,
2444
+ "loss": 1.048,
2445
+ "step": 406
2446
+ },
2447
+ {
2448
+ "epoch": 2.79,
2449
+ "learning_rate": 4.847010462029022e-05,
2450
+ "loss": 1.0457,
2451
+ "step": 407
2452
+ },
2453
+ {
2454
+ "epoch": 2.79,
2455
+ "learning_rate": 4.8460537517956794e-05,
2456
+ "loss": 1.0223,
2457
+ "step": 408
2458
+ },
2459
+ {
2460
+ "epoch": 2.8,
2461
+ "learning_rate": 4.845094154533239e-05,
2462
+ "loss": 1.1077,
2463
+ "step": 409
2464
+ },
2465
+ {
2466
+ "epoch": 2.81,
2467
+ "learning_rate": 4.84413167142257e-05,
2468
+ "loss": 1.0166,
2469
+ "step": 410
2470
+ },
2471
+ {
2472
+ "epoch": 2.82,
2473
+ "learning_rate": 4.8431663036480955e-05,
2474
+ "loss": 1.0992,
2475
+ "step": 411
2476
+ },
2477
+ {
2478
+ "epoch": 2.82,
2479
+ "learning_rate": 4.8421980523977864e-05,
2480
+ "loss": 0.9356,
2481
+ "step": 412
2482
+ },
2483
+ {
2484
+ "epoch": 2.83,
2485
+ "learning_rate": 4.841226918863162e-05,
2486
+ "loss": 1.1566,
2487
+ "step": 413
2488
+ },
2489
+ {
2490
+ "epoch": 2.84,
2491
+ "learning_rate": 4.840252904239291e-05,
2492
+ "loss": 1.1303,
2493
+ "step": 414
2494
+ },
2495
+ {
2496
+ "epoch": 2.84,
2497
+ "learning_rate": 4.839276009724783e-05,
2498
+ "loss": 1.0215,
2499
+ "step": 415
2500
+ },
2501
+ {
2502
+ "epoch": 2.85,
2503
+ "learning_rate": 4.838296236521796e-05,
2504
+ "loss": 1.087,
2505
+ "step": 416
2506
+ },
2507
+ {
2508
+ "epoch": 2.86,
2509
+ "learning_rate": 4.837313585836027e-05,
2510
+ "loss": 0.8805,
2511
+ "step": 417
2512
+ },
2513
+ {
2514
+ "epoch": 2.86,
2515
+ "learning_rate": 4.836328058876717e-05,
2516
+ "loss": 1.0968,
2517
+ "step": 418
2518
+ },
2519
+ {
2520
+ "epoch": 2.87,
2521
+ "learning_rate": 4.8353396568566454e-05,
2522
+ "loss": 1.0483,
2523
+ "step": 419
2524
+ },
2525
+ {
2526
+ "epoch": 2.88,
2527
+ "learning_rate": 4.834348380992127e-05,
2528
+ "loss": 1.0528,
2529
+ "step": 420
2530
+ },
2531
+ {
2532
+ "epoch": 2.88,
2533
+ "learning_rate": 4.833354232503019e-05,
2534
+ "loss": 1.0813,
2535
+ "step": 421
2536
+ },
2537
+ {
2538
+ "epoch": 2.89,
2539
+ "learning_rate": 4.832357212612707e-05,
2540
+ "loss": 1.0018,
2541
+ "step": 422
2542
+ },
2543
+ {
2544
+ "epoch": 2.9,
2545
+ "learning_rate": 4.8313573225481143e-05,
2546
+ "loss": 1.088,
2547
+ "step": 423
2548
+ },
2549
+ {
2550
+ "epoch": 2.9,
2551
+ "learning_rate": 4.830354563539696e-05,
2552
+ "loss": 1.1499,
2553
+ "step": 424
2554
+ },
2555
+ {
2556
+ "epoch": 2.91,
2557
+ "learning_rate": 4.8293489368214354e-05,
2558
+ "loss": 1.13,
2559
+ "step": 425
2560
+ },
2561
+ {
2562
+ "epoch": 2.92,
2563
+ "learning_rate": 4.8283404436308464e-05,
2564
+ "loss": 0.952,
2565
+ "step": 426
2566
+ },
2567
+ {
2568
+ "epoch": 2.92,
2569
+ "learning_rate": 4.8273290852089704e-05,
2570
+ "loss": 0.9332,
2571
+ "step": 427
2572
+ },
2573
+ {
2574
+ "epoch": 2.93,
2575
+ "learning_rate": 4.826314862800375e-05,
2576
+ "loss": 1.1237,
2577
+ "step": 428
2578
+ },
2579
+ {
2580
+ "epoch": 2.94,
2581
+ "learning_rate": 4.825297777653151e-05,
2582
+ "loss": 0.874,
2583
+ "step": 429
2584
+ },
2585
+ {
2586
+ "epoch": 2.95,
2587
+ "learning_rate": 4.824277831018913e-05,
2588
+ "loss": 1.2359,
2589
+ "step": 430
2590
+ },
2591
+ {
2592
+ "epoch": 2.95,
2593
+ "learning_rate": 4.823255024152796e-05,
2594
+ "loss": 1.0285,
2595
+ "step": 431
2596
+ },
2597
+ {
2598
+ "epoch": 2.96,
2599
+ "learning_rate": 4.822229358313456e-05,
2600
+ "loss": 0.9887,
2601
+ "step": 432
2602
+ },
2603
+ {
2604
+ "epoch": 2.97,
2605
+ "learning_rate": 4.8212008347630664e-05,
2606
+ "loss": 1.1521,
2607
+ "step": 433
2608
+ },
2609
+ {
2610
+ "epoch": 2.97,
2611
+ "learning_rate": 4.820169454767318e-05,
2612
+ "loss": 0.9784,
2613
+ "step": 434
2614
+ },
2615
+ {
2616
+ "epoch": 2.98,
2617
+ "learning_rate": 4.819135219595416e-05,
2618
+ "loss": 0.8608,
2619
+ "step": 435
2620
+ },
2621
+ {
2622
+ "epoch": 2.99,
2623
+ "learning_rate": 4.818098130520078e-05,
2624
+ "loss": 1.0483,
2625
+ "step": 436
2626
+ },
2627
+ {
2628
+ "epoch": 2.99,
2629
+ "learning_rate": 4.817058188817536e-05,
2630
+ "loss": 1.1391,
2631
+ "step": 437
2632
+ },
2633
+ {
2634
+ "epoch": 3.0,
2635
+ "learning_rate": 4.8160153957675316e-05,
2636
+ "loss": 0.8631,
2637
+ "step": 438
2638
+ },
2639
+ {
2640
+ "epoch": 3.01,
2641
+ "learning_rate": 4.8149697526533145e-05,
2642
+ "loss": 1.0571,
2643
+ "step": 439
2644
+ },
2645
+ {
2646
+ "epoch": 3.01,
2647
+ "learning_rate": 4.813921260761642e-05,
2648
+ "loss": 1.022,
2649
+ "step": 440
2650
+ },
2651
+ {
2652
+ "epoch": 3.02,
2653
+ "learning_rate": 4.8128699213827785e-05,
2654
+ "loss": 1.1186,
2655
+ "step": 441
2656
+ },
2657
+ {
2658
+ "epoch": 3.03,
2659
+ "learning_rate": 4.81181573581049e-05,
2660
+ "loss": 1.1906,
2661
+ "step": 442
2662
+ },
2663
+ {
2664
+ "epoch": 3.03,
2665
+ "learning_rate": 4.8107587053420455e-05,
2666
+ "loss": 1.0026,
2667
+ "step": 443
2668
+ },
2669
+ {
2670
+ "epoch": 3.04,
2671
+ "learning_rate": 4.8096988312782174e-05,
2672
+ "loss": 1.011,
2673
+ "step": 444
2674
+ },
2675
+ {
2676
+ "epoch": 3.05,
2677
+ "learning_rate": 4.808636114923274e-05,
2678
+ "loss": 1.0215,
2679
+ "step": 445
2680
+ },
2681
+ {
2682
+ "epoch": 3.05,
2683
+ "learning_rate": 4.807570557584984e-05,
2684
+ "loss": 1.0448,
2685
+ "step": 446
2686
+ },
2687
+ {
2688
+ "epoch": 3.06,
2689
+ "learning_rate": 4.8065021605746104e-05,
2690
+ "loss": 0.7449,
2691
+ "step": 447
2692
+ },
2693
+ {
2694
+ "epoch": 3.07,
2695
+ "learning_rate": 4.8054309252069114e-05,
2696
+ "loss": 1.0893,
2697
+ "step": 448
2698
+ },
2699
+ {
2700
+ "epoch": 3.08,
2701
+ "learning_rate": 4.8043568528001384e-05,
2702
+ "loss": 1.1073,
2703
+ "step": 449
2704
+ },
2705
+ {
2706
+ "epoch": 3.08,
2707
+ "learning_rate": 4.803279944676032e-05,
2708
+ "loss": 1.1046,
2709
+ "step": 450
2710
+ },
2711
+ {
2712
+ "epoch": 3.09,
2713
+ "learning_rate": 4.802200202159827e-05,
2714
+ "loss": 0.8914,
2715
+ "step": 451
2716
+ },
2717
+ {
2718
+ "epoch": 3.1,
2719
+ "learning_rate": 4.8011176265802415e-05,
2720
+ "loss": 1.1037,
2721
+ "step": 452
2722
+ },
2723
+ {
2724
+ "epoch": 3.1,
2725
+ "learning_rate": 4.8000322192694814e-05,
2726
+ "loss": 0.9823,
2727
+ "step": 453
2728
+ },
2729
+ {
2730
+ "epoch": 3.11,
2731
+ "learning_rate": 4.798943981563238e-05,
2732
+ "loss": 0.9635,
2733
+ "step": 454
2734
+ },
2735
+ {
2736
+ "epoch": 3.12,
2737
+ "learning_rate": 4.797852914800684e-05,
2738
+ "loss": 1.0584,
2739
+ "step": 455
2740
+ },
2741
+ {
2742
+ "epoch": 3.12,
2743
+ "learning_rate": 4.796759020324477e-05,
2744
+ "loss": 1.0476,
2745
+ "step": 456
2746
+ },
2747
+ {
2748
+ "epoch": 3.13,
2749
+ "learning_rate": 4.7956622994807495e-05,
2750
+ "loss": 1.0305,
2751
+ "step": 457
2752
+ },
2753
+ {
2754
+ "epoch": 3.14,
2755
+ "learning_rate": 4.7945627536191166e-05,
2756
+ "loss": 1.1159,
2757
+ "step": 458
2758
+ },
2759
+ {
2760
+ "epoch": 3.14,
2761
+ "learning_rate": 4.7934603840926675e-05,
2762
+ "loss": 0.9567,
2763
+ "step": 459
2764
+ },
2765
+ {
2766
+ "epoch": 3.15,
2767
+ "learning_rate": 4.7923551922579656e-05,
2768
+ "loss": 1.0536,
2769
+ "step": 460
2770
+ },
2771
+ {
2772
+ "epoch": 3.16,
2773
+ "learning_rate": 4.7912471794750496e-05,
2774
+ "loss": 0.7428,
2775
+ "step": 461
2776
+ },
2777
+ {
2778
+ "epoch": 3.16,
2779
+ "learning_rate": 4.790136347107427e-05,
2780
+ "loss": 1.0945,
2781
+ "step": 462
2782
+ },
2783
+ {
2784
+ "epoch": 3.17,
2785
+ "learning_rate": 4.7890226965220785e-05,
2786
+ "loss": 1.0525,
2787
+ "step": 463
2788
+ },
2789
+ {
2790
+ "epoch": 3.18,
2791
+ "learning_rate": 4.787906229089448e-05,
2792
+ "loss": 1.0444,
2793
+ "step": 464
2794
+ },
2795
+ {
2796
+ "epoch": 3.18,
2797
+ "learning_rate": 4.7867869461834514e-05,
2798
+ "loss": 0.9727,
2799
+ "step": 465
2800
+ },
2801
+ {
2802
+ "epoch": 3.19,
2803
+ "learning_rate": 4.785664849181465e-05,
2804
+ "loss": 1.097,
2805
+ "step": 466
2806
+ },
2807
+ {
2808
+ "epoch": 3.2,
2809
+ "learning_rate": 4.78453993946433e-05,
2810
+ "loss": 0.953,
2811
+ "step": 467
2812
+ },
2813
+ {
2814
+ "epoch": 3.21,
2815
+ "learning_rate": 4.783412218416348e-05,
2816
+ "loss": 1.0736,
2817
+ "step": 468
2818
+ },
2819
+ {
2820
+ "epoch": 3.21,
2821
+ "learning_rate": 4.78228168742528e-05,
2822
+ "loss": 1.1363,
2823
+ "step": 469
2824
+ },
2825
+ {
2826
+ "epoch": 3.22,
2827
+ "learning_rate": 4.781148347882347e-05,
2828
+ "loss": 0.9819,
2829
+ "step": 470
2830
+ },
2831
+ {
2832
+ "epoch": 3.23,
2833
+ "learning_rate": 4.780012201182225e-05,
2834
+ "loss": 0.9414,
2835
+ "step": 471
2836
+ },
2837
+ {
2838
+ "epoch": 3.23,
2839
+ "learning_rate": 4.7788732487230434e-05,
2840
+ "loss": 0.9277,
2841
+ "step": 472
2842
+ },
2843
+ {
2844
+ "epoch": 3.24,
2845
+ "learning_rate": 4.777731491906384e-05,
2846
+ "loss": 0.9638,
2847
+ "step": 473
2848
+ },
2849
+ {
2850
+ "epoch": 3.25,
2851
+ "learning_rate": 4.7765869321372836e-05,
2852
+ "loss": 1.0214,
2853
+ "step": 474
2854
+ },
2855
+ {
2856
+ "epoch": 3.25,
2857
+ "learning_rate": 4.7754395708242226e-05,
2858
+ "loss": 1.0234,
2859
+ "step": 475
2860
+ },
2861
+ {
2862
+ "epoch": 3.26,
2863
+ "learning_rate": 4.774289409379133e-05,
2864
+ "loss": 0.8438,
2865
+ "step": 476
2866
+ },
2867
+ {
2868
+ "epoch": 3.27,
2869
+ "learning_rate": 4.773136449217391e-05,
2870
+ "loss": 1.0616,
2871
+ "step": 477
2872
+ },
2873
+ {
2874
+ "epoch": 3.27,
2875
+ "learning_rate": 4.771980691757819e-05,
2876
+ "loss": 1.1344,
2877
+ "step": 478
2878
+ },
2879
+ {
2880
+ "epoch": 3.28,
2881
+ "learning_rate": 4.770822138422677e-05,
2882
+ "loss": 0.9735,
2883
+ "step": 479
2884
+ },
2885
+ {
2886
+ "epoch": 3.29,
2887
+ "learning_rate": 4.769660790637671e-05,
2888
+ "loss": 1.061,
2889
+ "step": 480
2890
+ },
2891
+ {
2892
+ "epoch": 3.29,
2893
+ "learning_rate": 4.768496649831942e-05,
2894
+ "loss": 0.9322,
2895
+ "step": 481
2896
+ },
2897
+ {
2898
+ "epoch": 3.3,
2899
+ "learning_rate": 4.767329717438071e-05,
2900
+ "loss": 1.0904,
2901
+ "step": 482
2902
+ },
2903
+ {
2904
+ "epoch": 3.31,
2905
+ "learning_rate": 4.76615999489207e-05,
2906
+ "loss": 1.0558,
2907
+ "step": 483
2908
+ },
2909
+ {
2910
+ "epoch": 3.32,
2911
+ "learning_rate": 4.76498748363339e-05,
2912
+ "loss": 1.0602,
2913
+ "step": 484
2914
+ },
2915
+ {
2916
+ "epoch": 3.32,
2917
+ "learning_rate": 4.76381218510491e-05,
2918
+ "loss": 0.9079,
2919
+ "step": 485
2920
+ },
2921
+ {
2922
+ "epoch": 3.33,
2923
+ "learning_rate": 4.76263410075294e-05,
2924
+ "loss": 1.1412,
2925
+ "step": 486
2926
+ },
2927
+ {
2928
+ "epoch": 3.34,
2929
+ "learning_rate": 4.7614532320272174e-05,
2930
+ "loss": 1.0225,
2931
+ "step": 487
2932
+ },
2933
+ {
2934
+ "epoch": 3.34,
2935
+ "learning_rate": 4.760269580380909e-05,
2936
+ "loss": 1.0012,
2937
+ "step": 488
2938
+ },
2939
+ {
2940
+ "epoch": 3.35,
2941
+ "learning_rate": 4.759083147270602e-05,
2942
+ "loss": 1.1793,
2943
+ "step": 489
2944
+ },
2945
+ {
2946
+ "epoch": 3.36,
2947
+ "learning_rate": 4.7578939341563095e-05,
2948
+ "loss": 0.9858,
2949
+ "step": 490
2950
+ },
2951
+ {
2952
+ "epoch": 3.36,
2953
+ "learning_rate": 4.7567019425014644e-05,
2954
+ "loss": 1.059,
2955
+ "step": 491
2956
+ },
2957
+ {
2958
+ "epoch": 3.37,
2959
+ "learning_rate": 4.755507173772919e-05,
2960
+ "loss": 1.0591,
2961
+ "step": 492
2962
+ },
2963
+ {
2964
+ "epoch": 3.38,
2965
+ "learning_rate": 4.754309629440943e-05,
2966
+ "loss": 1.0188,
2967
+ "step": 493
2968
+ },
2969
+ {
2970
+ "epoch": 3.38,
2971
+ "learning_rate": 4.753109310979224e-05,
2972
+ "loss": 1.0426,
2973
+ "step": 494
2974
+ },
2975
+ {
2976
+ "epoch": 3.39,
2977
+ "learning_rate": 4.751906219864857e-05,
2978
+ "loss": 1.1116,
2979
+ "step": 495
2980
+ },
2981
+ {
2982
+ "epoch": 3.4,
2983
+ "learning_rate": 4.750700357578357e-05,
2984
+ "loss": 0.8753,
2985
+ "step": 496
2986
+ },
2987
+ {
2988
+ "epoch": 3.4,
2989
+ "learning_rate": 4.749491725603644e-05,
2990
+ "loss": 0.9712,
2991
+ "step": 497
2992
+ },
2993
+ {
2994
+ "epoch": 3.41,
2995
+ "learning_rate": 4.7482803254280484e-05,
2996
+ "loss": 0.9633,
2997
+ "step": 498
2998
+ },
2999
+ {
3000
+ "epoch": 3.42,
3001
+ "learning_rate": 4.747066158542306e-05,
3002
+ "loss": 0.8521,
3003
+ "step": 499
3004
+ },
3005
+ {
3006
+ "epoch": 3.42,
3007
+ "learning_rate": 4.7458492264405574e-05,
3008
+ "loss": 1.0322,
3009
+ "step": 500
3010
+ },
3011
+ {
3012
+ "epoch": 3.43,
3013
+ "learning_rate": 4.7446295306203474e-05,
3014
+ "loss": 0.9904,
3015
+ "step": 501
3016
+ },
3017
+ {
3018
+ "epoch": 3.44,
3019
+ "learning_rate": 4.743407072582621e-05,
3020
+ "loss": 1.1605,
3021
+ "step": 502
3022
+ },
3023
+ {
3024
+ "epoch": 3.45,
3025
+ "learning_rate": 4.742181853831721e-05,
3026
+ "loss": 0.9827,
3027
+ "step": 503
3028
+ },
3029
+ {
3030
+ "epoch": 3.45,
3031
+ "learning_rate": 4.74095387587539e-05,
3032
+ "loss": 1.0952,
3033
+ "step": 504
3034
+ },
3035
+ {
3036
+ "epoch": 3.46,
3037
+ "learning_rate": 4.739723140224763e-05,
3038
+ "loss": 0.9814,
3039
+ "step": 505
3040
+ },
3041
+ {
3042
+ "epoch": 3.47,
3043
+ "learning_rate": 4.738489648394373e-05,
3044
+ "loss": 1.0741,
3045
+ "step": 506
3046
+ },
3047
+ {
3048
+ "epoch": 3.47,
3049
+ "learning_rate": 4.73725340190214e-05,
3050
+ "loss": 0.9566,
3051
+ "step": 507
3052
+ },
3053
+ {
3054
+ "epoch": 3.48,
3055
+ "learning_rate": 4.736014402269376e-05,
3056
+ "loss": 0.913,
3057
+ "step": 508
3058
+ },
3059
+ {
3060
+ "epoch": 3.49,
3061
+ "learning_rate": 4.734772651020782e-05,
3062
+ "loss": 1.151,
3063
+ "step": 509
3064
+ },
3065
+ {
3066
+ "epoch": 3.49,
3067
+ "learning_rate": 4.733528149684444e-05,
3068
+ "loss": 0.9932,
3069
+ "step": 510
3070
+ },
3071
+ {
3072
+ "epoch": 3.5,
3073
+ "learning_rate": 4.732280899791832e-05,
3074
+ "loss": 1.1216,
3075
+ "step": 511
3076
+ },
3077
+ {
3078
+ "epoch": 3.51,
3079
+ "learning_rate": 4.7310309028777976e-05,
3080
+ "loss": 0.9726,
3081
+ "step": 512
3082
+ },
3083
+ {
3084
+ "epoch": 3.51,
3085
+ "learning_rate": 4.729778160480576e-05,
3086
+ "loss": 1.0382,
3087
+ "step": 513
3088
+ },
3089
+ {
3090
+ "epoch": 3.52,
3091
+ "learning_rate": 4.728522674141776e-05,
3092
+ "loss": 1.0765,
3093
+ "step": 514
3094
+ },
3095
+ {
3096
+ "epoch": 3.53,
3097
+ "learning_rate": 4.727264445406388e-05,
3098
+ "loss": 1.0005,
3099
+ "step": 515
3100
+ },
3101
+ {
3102
+ "epoch": 3.53,
3103
+ "learning_rate": 4.726003475822775e-05,
3104
+ "loss": 0.931,
3105
+ "step": 516
3106
+ },
3107
+ {
3108
+ "epoch": 3.54,
3109
+ "learning_rate": 4.7247397669426716e-05,
3110
+ "loss": 0.8656,
3111
+ "step": 517
3112
+ },
3113
+ {
3114
+ "epoch": 3.55,
3115
+ "learning_rate": 4.723473320321186e-05,
3116
+ "loss": 1.1407,
3117
+ "step": 518
3118
+ },
3119
+ {
3120
+ "epoch": 3.55,
3121
+ "learning_rate": 4.7222041375167936e-05,
3122
+ "loss": 0.845,
3123
+ "step": 519
3124
+ },
3125
+ {
3126
+ "epoch": 3.56,
3127
+ "learning_rate": 4.720932220091337e-05,
3128
+ "loss": 0.8612,
3129
+ "step": 520
3130
+ },
3131
+ {
3132
+ "epoch": 3.57,
3133
+ "learning_rate": 4.7196575696100254e-05,
3134
+ "loss": 1.1551,
3135
+ "step": 521
3136
+ },
3137
+ {
3138
+ "epoch": 3.58,
3139
+ "learning_rate": 4.7183801876414294e-05,
3140
+ "loss": 0.8392,
3141
+ "step": 522
3142
+ },
3143
+ {
3144
+ "epoch": 3.58,
3145
+ "learning_rate": 4.717100075757482e-05,
3146
+ "loss": 1.0083,
3147
+ "step": 523
3148
+ },
3149
+ {
3150
+ "epoch": 3.59,
3151
+ "learning_rate": 4.715817235533476e-05,
3152
+ "loss": 0.8475,
3153
+ "step": 524
3154
+ },
3155
+ {
3156
+ "epoch": 3.6,
3157
+ "learning_rate": 4.71453166854806e-05,
3158
+ "loss": 0.936,
3159
+ "step": 525
3160
+ },
3161
+ {
3162
+ "epoch": 3.6,
3163
+ "learning_rate": 4.7132433763832404e-05,
3164
+ "loss": 0.9306,
3165
+ "step": 526
3166
+ },
3167
+ {
3168
+ "epoch": 3.61,
3169
+ "learning_rate": 4.711952360624376e-05,
3170
+ "loss": 0.9495,
3171
+ "step": 527
3172
+ },
3173
+ {
3174
+ "epoch": 3.62,
3175
+ "learning_rate": 4.710658622860176e-05,
3176
+ "loss": 0.8872,
3177
+ "step": 528
3178
+ },
3179
+ {
3180
+ "epoch": 3.62,
3181
+ "learning_rate": 4.709362164682702e-05,
3182
+ "loss": 0.6822,
3183
+ "step": 529
3184
+ },
3185
+ {
3186
+ "epoch": 3.63,
3187
+ "learning_rate": 4.70806298768736e-05,
3188
+ "loss": 0.9525,
3189
+ "step": 530
3190
+ },
3191
+ {
3192
+ "epoch": 3.64,
3193
+ "learning_rate": 4.706761093472906e-05,
3194
+ "loss": 1.0159,
3195
+ "step": 531
3196
+ },
3197
+ {
3198
+ "epoch": 3.64,
3199
+ "learning_rate": 4.705456483641435e-05,
3200
+ "loss": 0.9192,
3201
+ "step": 532
3202
+ },
3203
+ {
3204
+ "epoch": 3.65,
3205
+ "learning_rate": 4.704149159798387e-05,
3206
+ "loss": 0.9538,
3207
+ "step": 533
3208
+ },
3209
+ {
3210
+ "epoch": 3.66,
3211
+ "learning_rate": 4.702839123552541e-05,
3212
+ "loss": 1.0676,
3213
+ "step": 534
3214
+ },
3215
+ {
3216
+ "epoch": 3.66,
3217
+ "learning_rate": 4.7015263765160154e-05,
3218
+ "loss": 0.9409,
3219
+ "step": 535
3220
+ },
3221
+ {
3222
+ "epoch": 3.67,
3223
+ "learning_rate": 4.70021092030426e-05,
3224
+ "loss": 0.9745,
3225
+ "step": 536
3226
+ },
3227
+ {
3228
+ "epoch": 3.68,
3229
+ "learning_rate": 4.698892756536064e-05,
3230
+ "loss": 0.8958,
3231
+ "step": 537
3232
+ },
3233
+ {
3234
+ "epoch": 3.68,
3235
+ "learning_rate": 4.697571886833544e-05,
3236
+ "loss": 1.0001,
3237
+ "step": 538
3238
+ },
3239
+ {
3240
+ "epoch": 3.69,
3241
+ "learning_rate": 4.696248312822149e-05,
3242
+ "loss": 1.026,
3243
+ "step": 539
3244
+ },
3245
+ {
3246
+ "epoch": 3.7,
3247
+ "learning_rate": 4.6949220361306555e-05,
3248
+ "loss": 1.0302,
3249
+ "step": 540
3250
+ },
3251
+ {
3252
+ "epoch": 3.71,
3253
+ "learning_rate": 4.693593058391165e-05,
3254
+ "loss": 0.9987,
3255
+ "step": 541
3256
+ },
3257
+ {
3258
+ "epoch": 3.71,
3259
+ "learning_rate": 4.692261381239105e-05,
3260
+ "loss": 1.1951,
3261
+ "step": 542
3262
+ },
3263
+ {
3264
+ "epoch": 3.72,
3265
+ "learning_rate": 4.690927006313222e-05,
3266
+ "loss": 1.0323,
3267
+ "step": 543
3268
+ },
3269
+ {
3270
+ "epoch": 3.73,
3271
+ "learning_rate": 4.689589935255585e-05,
3272
+ "loss": 0.9272,
3273
+ "step": 544
3274
+ },
3275
+ {
3276
+ "epoch": 3.73,
3277
+ "learning_rate": 4.688250169711578e-05,
3278
+ "loss": 0.9996,
3279
+ "step": 545
3280
+ },
3281
+ {
3282
+ "epoch": 3.74,
3283
+ "learning_rate": 4.686907711329903e-05,
3284
+ "loss": 1.1128,
3285
+ "step": 546
3286
+ },
3287
+ {
3288
+ "epoch": 3.75,
3289
+ "learning_rate": 4.6855625617625763e-05,
3290
+ "loss": 0.9461,
3291
+ "step": 547
3292
+ },
3293
+ {
3294
+ "epoch": 3.75,
3295
+ "learning_rate": 4.684214722664924e-05,
3296
+ "loss": 0.9449,
3297
+ "step": 548
3298
+ },
3299
+ {
3300
+ "epoch": 3.76,
3301
+ "learning_rate": 4.682864195695582e-05,
3302
+ "loss": 0.9528,
3303
+ "step": 549
3304
+ },
3305
+ {
3306
+ "epoch": 3.77,
3307
+ "learning_rate": 4.681510982516496e-05,
3308
+ "loss": 0.8721,
3309
+ "step": 550
3310
+ },
3311
+ {
3312
+ "epoch": 3.77,
3313
+ "learning_rate": 4.680155084792914e-05,
3314
+ "loss": 1.0428,
3315
+ "step": 551
3316
+ },
3317
+ {
3318
+ "epoch": 3.78,
3319
+ "learning_rate": 4.678796504193392e-05,
3320
+ "loss": 1.0111,
3321
+ "step": 552
3322
+ },
3323
+ {
3324
+ "epoch": 3.79,
3325
+ "learning_rate": 4.677435242389784e-05,
3326
+ "loss": 0.9557,
3327
+ "step": 553
3328
+ },
3329
+ {
3330
+ "epoch": 3.79,
3331
+ "learning_rate": 4.676071301057243e-05,
3332
+ "loss": 0.9475,
3333
+ "step": 554
3334
+ },
3335
+ {
3336
+ "epoch": 3.8,
3337
+ "learning_rate": 4.674704681874223e-05,
3338
+ "loss": 1.039,
3339
+ "step": 555
3340
+ },
3341
+ {
3342
+ "epoch": 3.81,
3343
+ "learning_rate": 4.6733353865224694e-05,
3344
+ "loss": 0.9335,
3345
+ "step": 556
3346
+ },
3347
+ {
3348
+ "epoch": 3.82,
3349
+ "learning_rate": 4.671963416687024e-05,
3350
+ "loss": 1.0222,
3351
+ "step": 557
3352
+ },
3353
+ {
3354
+ "epoch": 3.82,
3355
+ "learning_rate": 4.670588774056218e-05,
3356
+ "loss": 0.8481,
3357
+ "step": 558
3358
+ },
3359
+ {
3360
+ "epoch": 3.83,
3361
+ "learning_rate": 4.669211460321673e-05,
3362
+ "loss": 1.0872,
3363
+ "step": 559
3364
+ },
3365
+ {
3366
+ "epoch": 3.84,
3367
+ "learning_rate": 4.667831477178295e-05,
3368
+ "loss": 1.0565,
3369
+ "step": 560
3370
+ },
3371
+ {
3372
+ "epoch": 3.84,
3373
+ "learning_rate": 4.666448826324278e-05,
3374
+ "loss": 0.9574,
3375
+ "step": 561
3376
+ },
3377
+ {
3378
+ "epoch": 3.85,
3379
+ "learning_rate": 4.665063509461097e-05,
3380
+ "loss": 1.0013,
3381
+ "step": 562
3382
+ },
3383
+ {
3384
+ "epoch": 3.86,
3385
+ "learning_rate": 4.663675528293509e-05,
3386
+ "loss": 0.807,
3387
+ "step": 563
3388
+ },
3389
+ {
3390
+ "epoch": 3.86,
3391
+ "learning_rate": 4.662284884529549e-05,
3392
+ "loss": 1.0153,
3393
+ "step": 564
3394
+ },
3395
+ {
3396
+ "epoch": 3.87,
3397
+ "learning_rate": 4.660891579880528e-05,
3398
+ "loss": 0.9717,
3399
+ "step": 565
3400
+ },
3401
+ {
3402
+ "epoch": 3.88,
3403
+ "learning_rate": 4.6594956160610325e-05,
3404
+ "loss": 0.9686,
3405
+ "step": 566
3406
+ },
3407
+ {
3408
+ "epoch": 3.88,
3409
+ "learning_rate": 4.6580969947889216e-05,
3410
+ "loss": 1.0015,
3411
+ "step": 567
3412
+ },
3413
+ {
3414
+ "epoch": 3.89,
3415
+ "learning_rate": 4.656695717785323e-05,
3416
+ "loss": 0.9288,
3417
+ "step": 568
3418
+ },
3419
+ {
3420
+ "epoch": 3.9,
3421
+ "learning_rate": 4.6552917867746324e-05,
3422
+ "loss": 1.0121,
3423
+ "step": 569
3424
+ },
3425
+ {
3426
+ "epoch": 3.9,
3427
+ "learning_rate": 4.653885203484515e-05,
3428
+ "loss": 1.0764,
3429
+ "step": 570
3430
+ },
3431
+ {
3432
+ "epoch": 3.91,
3433
+ "learning_rate": 4.652475969645896e-05,
3434
+ "loss": 1.0444,
3435
+ "step": 571
3436
+ },
3437
+ {
3438
+ "epoch": 3.92,
3439
+ "learning_rate": 4.651064086992965e-05,
3440
+ "loss": 0.8733,
3441
+ "step": 572
3442
+ },
3443
+ {
3444
+ "epoch": 3.92,
3445
+ "learning_rate": 4.6496495572631675e-05,
3446
+ "loss": 0.8512,
3447
+ "step": 573
3448
+ },
3449
+ {
3450
+ "epoch": 3.93,
3451
+ "learning_rate": 4.6482323821972105e-05,
3452
+ "loss": 1.0505,
3453
+ "step": 574
3454
+ },
3455
+ {
3456
+ "epoch": 3.94,
3457
+ "learning_rate": 4.6468125635390556e-05,
3458
+ "loss": 0.7957,
3459
+ "step": 575
3460
+ },
3461
+ {
3462
+ "epoch": 3.95,
3463
+ "learning_rate": 4.6453901030359154e-05,
3464
+ "loss": 1.1614,
3465
+ "step": 576
3466
+ },
3467
+ {
3468
+ "epoch": 3.95,
3469
+ "learning_rate": 4.6439650024382547e-05,
3470
+ "loss": 0.9545,
3471
+ "step": 577
3472
+ },
3473
+ {
3474
+ "epoch": 3.96,
3475
+ "learning_rate": 4.6439650024382547e-05,
3476
+ "loss": 0.9162,
3477
+ "step": 578
3478
+ },
3479
+ {
3480
+ "epoch": 3.97,
3481
+ "learning_rate": 4.642537263499788e-05,
3482
+ "loss": 1.0761,
3483
+ "step": 579
3484
+ },
3485
+ {
3486
+ "epoch": 3.97,
3487
+ "learning_rate": 4.6411068879774754e-05,
3488
+ "loss": 0.9022,
3489
+ "step": 580
3490
+ },
3491
+ {
3492
+ "epoch": 3.98,
3493
+ "learning_rate": 4.639673877631523e-05,
3494
+ "loss": 0.7747,
3495
+ "step": 581
3496
+ },
3497
+ {
3498
+ "epoch": 3.99,
3499
+ "learning_rate": 4.6382382342253775e-05,
3500
+ "loss": 0.9697,
3501
+ "step": 582
3502
+ },
3503
+ {
3504
+ "epoch": 3.99,
3505
+ "learning_rate": 4.636799959525726e-05,
3506
+ "loss": 1.0663,
3507
+ "step": 583
3508
+ },
3509
+ {
3510
+ "epoch": 4.0,
3511
+ "learning_rate": 4.635359055302495e-05,
3512
+ "loss": 0.7292,
3513
+ "step": 584
3514
+ },
3515
+ {
3516
+ "epoch": 4.01,
3517
+ "learning_rate": 4.633915523328846e-05,
3518
+ "loss": 0.9178,
3519
+ "step": 585
3520
+ },
3521
+ {
3522
+ "epoch": 4.01,
3523
+ "learning_rate": 4.632469365381174e-05,
3524
+ "loss": 0.9554,
3525
+ "step": 586
3526
+ },
3527
+ {
3528
+ "epoch": 4.02,
3529
+ "learning_rate": 4.631020583239107e-05,
3530
+ "loss": 1.0515,
3531
+ "step": 587
3532
+ },
3533
+ {
3534
+ "epoch": 4.03,
3535
+ "learning_rate": 4.629569178685499e-05,
3536
+ "loss": 1.1199,
3537
+ "step": 588
3538
+ },
3539
+ {
3540
+ "epoch": 4.03,
3541
+ "learning_rate": 4.628115153506435e-05,
3542
+ "loss": 0.9295,
3543
+ "step": 589
3544
+ },
3545
+ {
3546
+ "epoch": 4.04,
3547
+ "learning_rate": 4.626658509491222e-05,
3548
+ "loss": 0.9321,
3549
+ "step": 590
3550
+ },
3551
+ {
3552
+ "epoch": 4.05,
3553
+ "learning_rate": 4.625199248432391e-05,
3554
+ "loss": 0.9355,
3555
+ "step": 591
3556
+ },
3557
+ {
3558
+ "epoch": 4.05,
3559
+ "learning_rate": 4.6237373721256935e-05,
3560
+ "loss": 0.9687,
3561
+ "step": 592
3562
+ },
3563
+ {
3564
+ "epoch": 4.06,
3565
+ "learning_rate": 4.6222728823700995e-05,
3566
+ "loss": 0.6614,
3567
+ "step": 593
3568
+ },
3569
+ {
3570
+ "epoch": 4.07,
3571
+ "learning_rate": 4.620805780967794e-05,
3572
+ "loss": 1.0145,
3573
+ "step": 594
3574
+ },
3575
+ {
3576
+ "epoch": 4.08,
3577
+ "learning_rate": 4.619336069724177e-05,
3578
+ "loss": 1.0242,
3579
+ "step": 595
3580
+ },
3581
+ {
3582
+ "epoch": 4.08,
3583
+ "learning_rate": 4.617863750447861e-05,
3584
+ "loss": 1.0289,
3585
+ "step": 596
3586
+ },
3587
+ {
3588
+ "epoch": 4.09,
3589
+ "learning_rate": 4.616388824950664e-05,
3590
+ "loss": 0.8058,
3591
+ "step": 597
3592
+ },
3593
+ {
3594
+ "epoch": 4.1,
3595
+ "learning_rate": 4.614911295047615e-05,
3596
+ "loss": 1.0225,
3597
+ "step": 598
3598
+ },
3599
+ {
3600
+ "epoch": 4.1,
3601
+ "learning_rate": 4.613431162556947e-05,
3602
+ "loss": 0.8832,
3603
+ "step": 599
3604
+ },
3605
+ {
3606
+ "epoch": 4.11,
3607
+ "learning_rate": 4.6119484293000955e-05,
3608
+ "loss": 0.8861,
3609
+ "step": 600
3610
+ },
3611
+ {
3612
+ "epoch": 4.12,
3613
+ "learning_rate": 4.610463097101696e-05,
3614
+ "loss": 0.9833,
3615
+ "step": 601
3616
+ },
3617
+ {
3618
+ "epoch": 4.12,
3619
+ "learning_rate": 4.6089751677895836e-05,
3620
+ "loss": 0.9703,
3621
+ "step": 602
3622
+ },
3623
+ {
3624
+ "epoch": 4.13,
3625
+ "learning_rate": 4.607484643194788e-05,
3626
+ "loss": 0.9511,
3627
+ "step": 603
3628
+ },
3629
+ {
3630
+ "epoch": 4.14,
3631
+ "learning_rate": 4.605991525151533e-05,
3632
+ "loss": 1.0516,
3633
+ "step": 604
3634
+ },
3635
+ {
3636
+ "epoch": 4.14,
3637
+ "learning_rate": 4.604495815497234e-05,
3638
+ "loss": 0.8705,
3639
+ "step": 605
3640
+ },
3641
+ {
3642
+ "epoch": 4.15,
3643
+ "learning_rate": 4.6029975160724945e-05,
3644
+ "loss": 0.9725,
3645
+ "step": 606
3646
+ },
3647
+ {
3648
+ "epoch": 4.16,
3649
+ "learning_rate": 4.6014966287211084e-05,
3650
+ "loss": 0.6535,
3651
+ "step": 607
3652
+ },
3653
+ {
3654
+ "epoch": 4.16,
3655
+ "learning_rate": 4.59999315529005e-05,
3656
+ "loss": 1.013,
3657
+ "step": 608
3658
+ },
3659
+ {
3660
+ "epoch": 4.17,
3661
+ "learning_rate": 4.598487097629479e-05,
3662
+ "loss": 0.9731,
3663
+ "step": 609
3664
+ },
3665
+ {
3666
+ "epoch": 4.18,
3667
+ "learning_rate": 4.5969784575927324e-05,
3668
+ "loss": 0.9715,
3669
+ "step": 610
3670
+ },
3671
+ {
3672
+ "epoch": 4.18,
3673
+ "learning_rate": 4.595467237036329e-05,
3674
+ "loss": 0.898,
3675
+ "step": 611
3676
+ },
3677
+ {
3678
+ "epoch": 4.19,
3679
+ "learning_rate": 4.59395343781996e-05,
3680
+ "loss": 1.0253,
3681
+ "step": 612
3682
+ },
3683
+ {
3684
+ "epoch": 4.2,
3685
+ "learning_rate": 4.5924370618064913e-05,
3686
+ "loss": 0.8818,
3687
+ "step": 613
3688
+ },
3689
+ {
3690
+ "epoch": 4.21,
3691
+ "learning_rate": 4.590918110861958e-05,
3692
+ "loss": 0.9952,
3693
+ "step": 614
3694
+ },
3695
+ {
3696
+ "epoch": 4.21,
3697
+ "learning_rate": 4.5893965868555664e-05,
3698
+ "loss": 1.0598,
3699
+ "step": 615
3700
+ },
3701
+ {
3702
+ "epoch": 4.22,
3703
+ "learning_rate": 4.5878724916596874e-05,
3704
+ "loss": 0.9002,
3705
+ "step": 616
3706
+ },
3707
+ {
3708
+ "epoch": 4.23,
3709
+ "learning_rate": 4.586345827149856e-05,
3710
+ "loss": 0.8679,
3711
+ "step": 617
3712
+ },
3713
+ {
3714
+ "epoch": 4.23,
3715
+ "learning_rate": 4.58481659520477e-05,
3716
+ "loss": 0.8458,
3717
+ "step": 618
3718
+ },
3719
+ {
3720
+ "epoch": 4.24,
3721
+ "learning_rate": 4.5832847977062874e-05,
3722
+ "loss": 0.8581,
3723
+ "step": 619
3724
+ },
3725
+ {
3726
+ "epoch": 4.25,
3727
+ "learning_rate": 4.581750436539421e-05,
3728
+ "loss": 0.9438,
3729
+ "step": 620
3730
+ },
3731
+ {
3732
+ "epoch": 4.25,
3733
+ "learning_rate": 4.5802135135923386e-05,
3734
+ "loss": 0.9411,
3735
+ "step": 621
3736
+ },
3737
+ {
3738
+ "epoch": 4.26,
3739
+ "learning_rate": 4.5786740307563636e-05,
3740
+ "loss": 0.7669,
3741
+ "step": 622
3742
+ },
3743
+ {
3744
+ "epoch": 4.27,
3745
+ "learning_rate": 4.5771319899259656e-05,
3746
+ "loss": 0.9775,
3747
+ "step": 623
3748
+ },
3749
+ {
3750
+ "epoch": 4.27,
3751
+ "learning_rate": 4.5755873929987634e-05,
3752
+ "loss": 1.0761,
3753
+ "step": 624
3754
+ },
3755
+ {
3756
+ "epoch": 4.28,
3757
+ "learning_rate": 4.5740402418755246e-05,
3758
+ "loss": 0.8978,
3759
+ "step": 625
3760
+ },
3761
+ {
3762
+ "epoch": 4.29,
3763
+ "learning_rate": 4.572490538460154e-05,
3764
+ "loss": 0.9787,
3765
+ "step": 626
3766
+ },
3767
+ {
3768
+ "epoch": 4.29,
3769
+ "learning_rate": 4.570938284659702e-05,
3770
+ "loss": 0.8622,
3771
+ "step": 627
3772
+ },
3773
+ {
3774
+ "epoch": 4.3,
3775
+ "learning_rate": 4.5693834823843556e-05,
3776
+ "loss": 1.0193,
3777
+ "step": 628
3778
+ },
3779
+ {
3780
+ "epoch": 4.31,
3781
+ "learning_rate": 4.5678261335474384e-05,
3782
+ "loss": 0.9681,
3783
+ "step": 629
3784
+ },
3785
+ {
3786
+ "epoch": 4.32,
3787
+ "learning_rate": 4.566266240065406e-05,
3788
+ "loss": 0.9882,
3789
+ "step": 630
3790
+ },
3791
+ {
3792
+ "epoch": 4.32,
3793
+ "learning_rate": 4.564703803857849e-05,
3794
+ "loss": 0.8362,
3795
+ "step": 631
3796
+ },
3797
+ {
3798
+ "epoch": 4.33,
3799
+ "learning_rate": 4.5631388268474837e-05,
3800
+ "loss": 1.0627,
3801
+ "step": 632
3802
+ },
3803
+ {
3804
+ "epoch": 4.34,
3805
+ "learning_rate": 4.5615713109601544e-05,
3806
+ "loss": 0.9333,
3807
+ "step": 633
3808
+ },
3809
+ {
3810
+ "epoch": 4.34,
3811
+ "learning_rate": 4.56000125812483e-05,
3812
+ "loss": 0.93,
3813
+ "step": 634
3814
+ },
3815
+ {
3816
+ "epoch": 4.35,
3817
+ "learning_rate": 4.558428670273601e-05,
3818
+ "loss": 1.1188,
3819
+ "step": 635
3820
+ },
3821
+ {
3822
+ "epoch": 4.36,
3823
+ "learning_rate": 4.556853549341679e-05,
3824
+ "loss": 0.9112,
3825
+ "step": 636
3826
+ },
3827
+ {
3828
+ "epoch": 4.36,
3829
+ "learning_rate": 4.555275897267388e-05,
3830
+ "loss": 0.9868,
3831
+ "step": 637
3832
+ },
3833
+ {
3834
+ "epoch": 4.37,
3835
+ "learning_rate": 4.553695715992172e-05,
3836
+ "loss": 0.974,
3837
+ "step": 638
3838
+ },
3839
+ {
3840
+ "epoch": 4.38,
3841
+ "learning_rate": 4.552113007460586e-05,
3842
+ "loss": 0.952,
3843
+ "step": 639
3844
+ },
3845
+ {
3846
+ "epoch": 4.38,
3847
+ "learning_rate": 4.550527773620293e-05,
3848
+ "loss": 0.9677,
3849
+ "step": 640
3850
+ },
3851
+ {
3852
+ "epoch": 4.39,
3853
+ "learning_rate": 4.5489400164220666e-05,
3854
+ "loss": 1.0329,
3855
+ "step": 641
3856
+ },
3857
+ {
3858
+ "epoch": 4.4,
3859
+ "learning_rate": 4.5473497378197835e-05,
3860
+ "loss": 0.7958,
3861
+ "step": 642
3862
+ },
3863
+ {
3864
+ "epoch": 4.4,
3865
+ "learning_rate": 4.545756939770423e-05,
3866
+ "loss": 0.8831,
3867
+ "step": 643
3868
+ },
3869
+ {
3870
+ "epoch": 4.41,
3871
+ "learning_rate": 4.5441616242340665e-05,
3872
+ "loss": 0.8805,
3873
+ "step": 644
3874
+ },
3875
+ {
3876
+ "epoch": 4.42,
3877
+ "learning_rate": 4.542563793173893e-05,
3878
+ "loss": 0.7746,
3879
+ "step": 645
3880
+ },
3881
+ {
3882
+ "epoch": 4.42,
3883
+ "learning_rate": 4.540963448556176e-05,
3884
+ "loss": 0.9577,
3885
+ "step": 646
3886
+ },
3887
+ {
3888
+ "epoch": 4.43,
3889
+ "learning_rate": 4.539360592350282e-05,
3890
+ "loss": 0.9146,
3891
+ "step": 647
3892
+ },
3893
+ {
3894
+ "epoch": 4.44,
3895
+ "learning_rate": 4.537755226528671e-05,
3896
+ "loss": 1.1038,
3897
+ "step": 648
3898
+ },
3899
+ {
3900
+ "epoch": 4.45,
3901
+ "learning_rate": 4.5361473530668874e-05,
3902
+ "loss": 0.9022,
3903
+ "step": 649
3904
+ },
3905
+ {
3906
+ "epoch": 4.45,
3907
+ "learning_rate": 4.534536973943564e-05,
3908
+ "loss": 1.0197,
3909
+ "step": 650
3910
+ },
3911
+ {
3912
+ "epoch": 4.46,
3913
+ "learning_rate": 4.532924091140417e-05,
3914
+ "loss": 0.9027,
3915
+ "step": 651
3916
+ },
3917
+ {
3918
+ "epoch": 4.47,
3919
+ "learning_rate": 4.531308706642243e-05,
3920
+ "loss": 0.9918,
3921
+ "step": 652
3922
+ },
3923
+ {
3924
+ "epoch": 4.47,
3925
+ "learning_rate": 4.529690822436916e-05,
3926
+ "loss": 0.875,
3927
+ "step": 653
3928
+ },
3929
+ {
3930
+ "epoch": 4.48,
3931
+ "learning_rate": 4.528070440515388e-05,
3932
+ "loss": 0.8319,
3933
+ "step": 654
3934
+ },
3935
+ {
3936
+ "epoch": 4.49,
3937
+ "learning_rate": 4.526447562871685e-05,
3938
+ "loss": 1.0758,
3939
+ "step": 655
3940
+ },
3941
+ {
3942
+ "epoch": 4.49,
3943
+ "learning_rate": 4.5248221915029014e-05,
3944
+ "loss": 0.9287,
3945
+ "step": 656
3946
+ },
3947
+ {
3948
+ "epoch": 4.5,
3949
+ "learning_rate": 4.523194328409203e-05,
3950
+ "loss": 1.047,
3951
+ "step": 657
3952
+ },
3953
+ {
3954
+ "epoch": 4.51,
3955
+ "learning_rate": 4.5215639755938214e-05,
3956
+ "loss": 0.9021,
3957
+ "step": 658
3958
+ },
3959
+ {
3960
+ "epoch": 4.51,
3961
+ "learning_rate": 4.519931135063051e-05,
3962
+ "loss": 0.9434,
3963
+ "step": 659
3964
+ },
3965
+ {
3966
+ "epoch": 4.52,
3967
+ "learning_rate": 4.518295808826249e-05,
3968
+ "loss": 1.0068,
3969
+ "step": 660
3970
+ },
3971
+ {
3972
+ "epoch": 4.53,
3973
+ "learning_rate": 4.5166579988958296e-05,
3974
+ "loss": 0.9381,
3975
+ "step": 661
3976
+ },
3977
+ {
3978
+ "epoch": 4.53,
3979
+ "learning_rate": 4.515017707287265e-05,
3980
+ "loss": 0.8563,
3981
+ "step": 662
3982
+ },
3983
+ {
3984
+ "epoch": 4.54,
3985
+ "learning_rate": 4.5133749360190805e-05,
3986
+ "loss": 0.7951,
3987
+ "step": 663
3988
+ },
3989
+ {
3990
+ "epoch": 4.55,
3991
+ "learning_rate": 4.5117296871128546e-05,
3992
+ "loss": 1.0672,
3993
+ "step": 664
3994
+ },
3995
+ {
3996
+ "epoch": 4.55,
3997
+ "learning_rate": 4.510081962593211e-05,
3998
+ "loss": 0.7677,
3999
+ "step": 665
4000
+ },
4001
+ {
4002
+ "epoch": 4.56,
4003
+ "learning_rate": 4.508431764487824e-05,
4004
+ "loss": 0.774,
4005
+ "step": 666
4006
+ },
4007
+ {
4008
+ "epoch": 4.57,
4009
+ "learning_rate": 4.5067790948274094e-05,
4010
+ "loss": 1.0797,
4011
+ "step": 667
4012
+ },
4013
+ {
4014
+ "epoch": 4.58,
4015
+ "learning_rate": 4.5051239556457244e-05,
4016
+ "loss": 0.7631,
4017
+ "step": 668
4018
+ },
4019
+ {
4020
+ "epoch": 4.58,
4021
+ "learning_rate": 4.503466348979568e-05,
4022
+ "loss": 0.9337,
4023
+ "step": 669
4024
+ },
4025
+ {
4026
+ "epoch": 4.59,
4027
+ "learning_rate": 4.501806276868772e-05,
4028
+ "loss": 0.7778,
4029
+ "step": 670
4030
+ },
4031
+ {
4032
+ "epoch": 4.6,
4033
+ "learning_rate": 4.500143741356203e-05,
4034
+ "loss": 0.8628,
4035
+ "step": 671
4036
+ },
4037
+ {
4038
+ "epoch": 4.6,
4039
+ "learning_rate": 4.4984787444877616e-05,
4040
+ "loss": 0.8446,
4041
+ "step": 672
4042
+ },
4043
+ {
4044
+ "epoch": 4.61,
4045
+ "learning_rate": 4.4968112883123734e-05,
4046
+ "loss": 0.8722,
4047
+ "step": 673
4048
+ },
4049
+ {
4050
+ "epoch": 4.62,
4051
+ "learning_rate": 4.495141374881995e-05,
4052
+ "loss": 0.8005,
4053
+ "step": 674
4054
+ },
4055
+ {
4056
+ "epoch": 4.62,
4057
+ "learning_rate": 4.493469006251601e-05,
4058
+ "loss": 0.5906,
4059
+ "step": 675
4060
+ },
4061
+ {
4062
+ "epoch": 4.63,
4063
+ "learning_rate": 4.491794184479194e-05,
4064
+ "loss": 0.8642,
4065
+ "step": 676
4066
+ },
4067
+ {
4068
+ "epoch": 4.64,
4069
+ "learning_rate": 4.49011691162579e-05,
4070
+ "loss": 0.9482,
4071
+ "step": 677
4072
+ },
4073
+ {
4074
+ "epoch": 4.64,
4075
+ "learning_rate": 4.488437189755424e-05,
4076
+ "loss": 0.8409,
4077
+ "step": 678
4078
+ },
4079
+ {
4080
+ "epoch": 4.65,
4081
+ "learning_rate": 4.486755020935144e-05,
4082
+ "loss": 0.8732,
4083
+ "step": 679
4084
+ },
4085
+ {
4086
+ "epoch": 4.66,
4087
+ "learning_rate": 4.485070407235009e-05,
4088
+ "loss": 0.9934,
4089
+ "step": 680
4090
+ },
4091
+ {
4092
+ "epoch": 4.66,
4093
+ "learning_rate": 4.4833833507280884e-05,
4094
+ "loss": 0.858,
4095
+ "step": 681
4096
+ },
4097
+ {
4098
+ "epoch": 4.67,
4099
+ "learning_rate": 4.481693853490454e-05,
4100
+ "loss": 0.8971,
4101
+ "step": 682
4102
+ },
4103
+ {
4104
+ "epoch": 4.68,
4105
+ "learning_rate": 4.480001917601185e-05,
4106
+ "loss": 0.8115,
4107
+ "step": 683
4108
+ },
4109
+ {
4110
+ "epoch": 4.68,
4111
+ "learning_rate": 4.478307545142359e-05,
4112
+ "loss": 0.9325,
4113
+ "step": 684
4114
+ },
4115
+ {
4116
+ "epoch": 4.69,
4117
+ "learning_rate": 4.476610738199053e-05,
4118
+ "loss": 0.956,
4119
+ "step": 685
4120
+ },
4121
+ {
4122
+ "epoch": 4.7,
4123
+ "learning_rate": 4.4749114988593396e-05,
4124
+ "loss": 0.9499,
4125
+ "step": 686
4126
+ },
4127
+ {
4128
+ "epoch": 4.71,
4129
+ "learning_rate": 4.473209829214286e-05,
4130
+ "loss": 0.9287,
4131
+ "step": 687
4132
+ },
4133
+ {
4134
+ "epoch": 4.71,
4135
+ "learning_rate": 4.471505731357949e-05,
4136
+ "loss": 1.1292,
4137
+ "step": 688
4138
+ },
4139
+ {
4140
+ "epoch": 4.72,
4141
+ "learning_rate": 4.4697992073873724e-05,
4142
+ "loss": 0.9526,
4143
+ "step": 689
4144
+ },
4145
+ {
4146
+ "epoch": 4.73,
4147
+ "learning_rate": 4.468090259402587e-05,
4148
+ "loss": 0.8573,
4149
+ "step": 690
4150
+ },
4151
+ {
4152
+ "epoch": 4.73,
4153
+ "learning_rate": 4.466378889506607e-05,
4154
+ "loss": 0.9222,
4155
+ "step": 691
4156
+ },
4157
+ {
4158
+ "epoch": 4.74,
4159
+ "learning_rate": 4.464665099805424e-05,
4160
+ "loss": 1.0377,
4161
+ "step": 692
4162
+ },
4163
+ {
4164
+ "epoch": 4.75,
4165
+ "learning_rate": 4.462948892408012e-05,
4166
+ "loss": 0.8685,
4167
+ "step": 693
4168
+ },
4169
+ {
4170
+ "epoch": 4.75,
4171
+ "learning_rate": 4.4612302694263174e-05,
4172
+ "loss": 0.8759,
4173
+ "step": 694
4174
+ },
4175
+ {
4176
+ "epoch": 4.76,
4177
+ "learning_rate": 4.4595092329752583e-05,
4178
+ "loss": 0.8718,
4179
+ "step": 695
4180
+ },
4181
+ {
4182
+ "epoch": 4.77,
4183
+ "learning_rate": 4.457785785172726e-05,
4184
+ "loss": 0.7879,
4185
+ "step": 696
4186
+ },
4187
+ {
4188
+ "epoch": 4.77,
4189
+ "learning_rate": 4.4560599281395755e-05,
4190
+ "loss": 0.9711,
4191
+ "step": 697
4192
+ },
4193
+ {
4194
+ "epoch": 4.78,
4195
+ "learning_rate": 4.454331663999629e-05,
4196
+ "loss": 0.9297,
4197
+ "step": 698
4198
+ },
4199
+ {
4200
+ "epoch": 4.79,
4201
+ "learning_rate": 4.4526009948796703e-05,
4202
+ "loss": 0.8714,
4203
+ "step": 699
4204
+ },
4205
+ {
4206
+ "epoch": 4.79,
4207
+ "learning_rate": 4.4508679229094425e-05,
4208
+ "loss": 0.8773,
4209
+ "step": 700
4210
+ },
4211
+ {
4212
+ "epoch": 4.8,
4213
+ "learning_rate": 4.449132450221646e-05,
4214
+ "loss": 0.973,
4215
+ "step": 701
4216
+ },
4217
+ {
4218
+ "epoch": 4.81,
4219
+ "learning_rate": 4.447394578951935e-05,
4220
+ "loss": 0.8554,
4221
+ "step": 702
4222
+ },
4223
+ {
4224
+ "epoch": 4.82,
4225
+ "learning_rate": 4.445654311238915e-05,
4226
+ "loss": 0.955,
4227
+ "step": 703
4228
+ },
4229
+ {
4230
+ "epoch": 4.82,
4231
+ "learning_rate": 4.443911649224143e-05,
4232
+ "loss": 0.7657,
4233
+ "step": 704
4234
+ },
4235
+ {
4236
+ "epoch": 4.83,
4237
+ "learning_rate": 4.442166595052118e-05,
4238
+ "loss": 1.0131,
4239
+ "step": 705
4240
+ },
4241
+ {
4242
+ "epoch": 4.84,
4243
+ "learning_rate": 4.4404191508702875e-05,
4244
+ "loss": 0.9818,
4245
+ "step": 706
4246
+ },
4247
+ {
4248
+ "epoch": 4.84,
4249
+ "learning_rate": 4.4386693188290376e-05,
4250
+ "loss": 0.8895,
4251
+ "step": 707
4252
+ },
4253
+ {
4254
+ "epoch": 4.85,
4255
+ "learning_rate": 4.4369171010816925e-05,
4256
+ "loss": 0.9271,
4257
+ "step": 708
4258
+ },
4259
+ {
4260
+ "epoch": 4.86,
4261
+ "learning_rate": 4.435162499784513e-05,
4262
+ "loss": 0.7359,
4263
+ "step": 709
4264
+ },
4265
+ {
4266
+ "epoch": 4.86,
4267
+ "learning_rate": 4.433405517096693e-05,
4268
+ "loss": 0.9381,
4269
+ "step": 710
4270
+ },
4271
+ {
4272
+ "epoch": 4.87,
4273
+ "learning_rate": 4.431646155180358e-05,
4274
+ "loss": 0.899,
4275
+ "step": 711
4276
+ },
4277
+ {
4278
+ "epoch": 4.88,
4279
+ "learning_rate": 4.4298844162005585e-05,
4280
+ "loss": 0.8745,
4281
+ "step": 712
4282
+ },
4283
+ {
4284
+ "epoch": 4.88,
4285
+ "learning_rate": 4.4281203023252735e-05,
4286
+ "loss": 0.9215,
4287
+ "step": 713
4288
+ },
4289
+ {
4290
+ "epoch": 4.89,
4291
+ "learning_rate": 4.426353815725403e-05,
4292
+ "loss": 0.8616,
4293
+ "step": 714
4294
+ },
4295
+ {
4296
+ "epoch": 4.9,
4297
+ "learning_rate": 4.4245849585747654e-05,
4298
+ "loss": 0.9401,
4299
+ "step": 715
4300
+ },
4301
+ {
4302
+ "epoch": 4.9,
4303
+ "learning_rate": 4.422813733050099e-05,
4304
+ "loss": 1.0099,
4305
+ "step": 716
4306
+ },
4307
+ {
4308
+ "epoch": 4.91,
4309
+ "learning_rate": 4.4210401413310556e-05,
4310
+ "loss": 0.9682,
4311
+ "step": 717
4312
+ },
4313
+ {
4314
+ "epoch": 4.92,
4315
+ "learning_rate": 4.4192641856001976e-05,
4316
+ "loss": 0.7944,
4317
+ "step": 718
4318
+ },
4319
+ {
4320
+ "epoch": 4.92,
4321
+ "learning_rate": 4.417485868042998e-05,
4322
+ "loss": 0.7715,
4323
+ "step": 719
4324
+ },
4325
+ {
4326
+ "epoch": 4.93,
4327
+ "learning_rate": 4.415705190847835e-05,
4328
+ "loss": 0.9758,
4329
+ "step": 720
4330
+ },
4331
+ {
4332
+ "epoch": 4.94,
4333
+ "learning_rate": 4.413922156205992e-05,
4334
+ "loss": 0.7187,
4335
+ "step": 721
4336
+ },
4337
+ {
4338
+ "epoch": 4.95,
4339
+ "learning_rate": 4.412136766311652e-05,
4340
+ "loss": 1.0898,
4341
+ "step": 722
4342
+ },
4343
+ {
4344
+ "epoch": 4.95,
4345
+ "learning_rate": 4.410349023361898e-05,
4346
+ "loss": 0.8747,
4347
+ "step": 723
4348
+ },
4349
+ {
4350
+ "epoch": 4.96,
4351
+ "learning_rate": 4.4085589295567065e-05,
4352
+ "loss": 0.9458,
4353
+ "step": 724
4354
+ },
4355
+ {
4356
+ "epoch": 4.97,
4357
+ "learning_rate": 4.406766487098949e-05,
4358
+ "loss": 1.0059,
4359
+ "step": 725
4360
+ },
4361
+ {
4362
+ "epoch": 4.97,
4363
+ "learning_rate": 4.4049716981943866e-05,
4364
+ "loss": 0.8319,
4365
+ "step": 726
4366
+ },
4367
+ {
4368
+ "epoch": 4.98,
4369
+ "learning_rate": 4.4031745650516666e-05,
4370
+ "loss": 0.7012,
4371
+ "step": 727
4372
+ },
4373
+ {
4374
+ "epoch": 4.99,
4375
+ "learning_rate": 4.401375089882324e-05,
4376
+ "loss": 0.9013,
4377
+ "step": 728
4378
+ },
4379
+ {
4380
+ "epoch": 4.99,
4381
+ "learning_rate": 4.399573274900771e-05,
4382
+ "loss": 0.9979,
4383
+ "step": 729
4384
+ },
4385
+ {
4386
+ "epoch": 5.0,
4387
+ "learning_rate": 4.397769122324305e-05,
4388
+ "loss": 0.6219,
4389
+ "step": 730
4390
+ }
4391
+ ],
4392
+ "logging_steps": 1.0,
4393
+ "max_steps": 2920,
4394
+ "num_train_epochs": 20,
4395
+ "save_steps": 1000000000,
4396
+ "total_flos": 9.79415936794624e+17,
4397
+ "trial_name": null,
4398
+ "trial_params": null
4399
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ad79b13fda074e9480758908f5e9bd421496a609af36b2acd03460d933cd7d4
3
+ size 5563
zero_to_fp32.py ADDED
@@ -0,0 +1,587 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
252
+ param_shapes = zero_model_states[0].param_shapes
253
+
254
+ # Reconstruction protocol:
255
+ #
256
+ # XXX: document this
257
+
258
+ if debug:
259
+ for i in range(world_size):
260
+ for j in range(len(fp32_flat_groups[0])):
261
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
262
+
263
+ # XXX: memory usage doubles here (zero2)
264
+ num_param_groups = len(fp32_flat_groups[0])
265
+ merged_single_partition_of_fp32_groups = []
266
+ for i in range(num_param_groups):
267
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
268
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
269
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
270
+ avail_numel = sum(
271
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
272
+
273
+ if debug:
274
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
275
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
276
+ # not asserting if there is a mismatch due to possible padding
277
+ print(f"Have {avail_numel} numels to process.")
278
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
279
+
280
+ # params
281
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
282
+ # out-of-core computing solution
283
+ total_numel = 0
284
+ total_params = 0
285
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
286
+ offset = 0
287
+ avail_numel = full_single_fp32_vector.numel()
288
+ for name, shape in shapes.items():
289
+
290
+ unpartitioned_numel = shape.numel()
291
+ total_numel += unpartitioned_numel
292
+ total_params += 1
293
+
294
+ if debug:
295
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
296
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
297
+ offset += unpartitioned_numel
298
+
299
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
300
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
301
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
302
+ # live optimizer object, so we are checking that the numbers are within the right range
303
+ align_to = 2 * world_size
304
+
305
+ def zero2_align(x):
306
+ return align_to * math.ceil(x / align_to)
307
+
308
+ if debug:
309
+ print(f"original offset={offset}, avail_numel={avail_numel}")
310
+
311
+ offset = zero2_align(offset)
312
+ avail_numel = zero2_align(avail_numel)
313
+
314
+ if debug:
315
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
316
+
317
+ # Sanity check
318
+ if offset != avail_numel:
319
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
320
+
321
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
322
+
323
+
324
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
325
+ state_dict = OrderedDict()
326
+
327
+ # buffers
328
+ buffers = zero_model_states[0].buffers
329
+ state_dict.update(buffers)
330
+ if debug:
331
+ print(f"added {len(buffers)} buffers")
332
+
333
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
334
+
335
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
336
+
337
+ # recover shared parameters
338
+ for pair in zero_model_states[0].shared_params:
339
+ if pair[1] in state_dict:
340
+ state_dict[pair[0]] = state_dict[pair[1]]
341
+
342
+ return state_dict
343
+
344
+
345
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
346
+ remainder = unpartitioned_numel % world_size
347
+ padding_numel = (world_size - remainder) if remainder else 0
348
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
349
+ return partitioned_numel, padding_numel
350
+
351
+
352
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
353
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
354
+ return
355
+
356
+ if debug:
357
+ for i in range(world_size):
358
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
359
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
360
+
361
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
362
+ wanted_params = len(frozen_param_shapes)
363
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
364
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
365
+ print(f'Frozen params: Have {avail_numel} numels to process.')
366
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
367
+
368
+ total_params = 0
369
+ total_numel = 0
370
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
371
+ total_params += 1
372
+ unpartitioned_numel = shape.numel()
373
+ total_numel += unpartitioned_numel
374
+
375
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
376
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
377
+
378
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
379
+
380
+ if debug:
381
+ print(
382
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
383
+ )
384
+
385
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
386
+
387
+
388
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
389
+ param_shapes = zero_model_states[0].param_shapes
390
+ avail_numel = fp32_flat_groups[0].numel() * world_size
391
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
392
+ # param, re-consolidating each param, while dealing with padding if any
393
+
394
+ # merge list of dicts, preserving order
395
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
396
+
397
+ if debug:
398
+ for i in range(world_size):
399
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
400
+
401
+ wanted_params = len(param_shapes)
402
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
403
+ # not asserting if there is a mismatch due to possible padding
404
+ avail_numel = fp32_flat_groups[0].numel() * world_size
405
+ print(f"Trainable params: Have {avail_numel} numels to process.")
406
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
407
+
408
+ # params
409
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
410
+ # out-of-core computing solution
411
+ offset = 0
412
+ total_numel = 0
413
+ total_params = 0
414
+ for name, shape in param_shapes.items():
415
+
416
+ unpartitioned_numel = shape.numel()
417
+ total_numel += unpartitioned_numel
418
+ total_params += 1
419
+
420
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
421
+
422
+ if debug:
423
+ print(
424
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
425
+ )
426
+
427
+ # XXX: memory usage doubles here
428
+ state_dict[name] = torch.cat(
429
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
430
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
431
+ offset += partitioned_numel
432
+
433
+ offset *= world_size
434
+
435
+ # Sanity check
436
+ if offset != avail_numel:
437
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
438
+
439
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
440
+
441
+
442
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
443
+ state_dict = OrderedDict()
444
+
445
+ # buffers
446
+ buffers = zero_model_states[0].buffers
447
+ state_dict.update(buffers)
448
+ if debug:
449
+ print(f"added {len(buffers)} buffers")
450
+
451
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
452
+
453
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
454
+
455
+ # recover shared parameters
456
+ for pair in zero_model_states[0].shared_params:
457
+ if pair[1] in state_dict:
458
+ state_dict[pair[0]] = state_dict[pair[1]]
459
+
460
+ return state_dict
461
+
462
+
463
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
464
+ """
465
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
466
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
467
+ via a model hub.
468
+
469
+ Args:
470
+ - ``checkpoint_dir``: path to the desired checkpoint folder
471
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
472
+
473
+ Returns:
474
+ - pytorch ``state_dict``
475
+
476
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
477
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
478
+ the checkpoint.
479
+
480
+ A typical usage might be ::
481
+
482
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
483
+ # do the training and checkpoint saving
484
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
485
+ model = model.cpu() # move to cpu
486
+ model.load_state_dict(state_dict)
487
+ # submit to model hub or save the model to share with others
488
+
489
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
490
+ application. i.e. you will need to re-initialize the deepspeed engine, since
491
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
492
+
493
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
494
+
495
+ """
496
+ if tag is None:
497
+ latest_path = os.path.join(checkpoint_dir, 'latest')
498
+ if os.path.isfile(latest_path):
499
+ with open(latest_path, 'r') as fd:
500
+ tag = fd.read().strip()
501
+ else:
502
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
503
+
504
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
505
+
506
+ if not os.path.isdir(ds_checkpoint_dir):
507
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
508
+
509
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
510
+
511
+
512
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
513
+ """
514
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
515
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
516
+
517
+ Args:
518
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
519
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
520
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
521
+ """
522
+
523
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
524
+ print(f"Saving fp32 state dict to {output_file}")
525
+ torch.save(state_dict, output_file)
526
+
527
+
528
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
529
+ """
530
+ 1. Put the provided model to cpu
531
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
532
+ 3. Load it into the provided model
533
+
534
+ Args:
535
+ - ``model``: the model object to update
536
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
537
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
538
+
539
+ Returns:
540
+ - ``model`: modified model
541
+
542
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
543
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
544
+ conveniently placed for you in the checkpoint folder.
545
+
546
+ A typical usage might be ::
547
+
548
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
549
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
550
+ # submit to model hub or save the model to share with others
551
+
552
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
553
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
554
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
555
+
556
+ """
557
+ logger.info(f"Extracting fp32 weights")
558
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
559
+
560
+ logger.info(f"Overwriting model with fp32 weights")
561
+ model = model.cpu()
562
+ model.load_state_dict(state_dict, strict=False)
563
+
564
+ return model
565
+
566
+
567
+ if __name__ == "__main__":
568
+
569
+ parser = argparse.ArgumentParser()
570
+ parser.add_argument("checkpoint_dir",
571
+ type=str,
572
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
573
+ parser.add_argument(
574
+ "output_file",
575
+ type=str,
576
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
577
+ parser.add_argument("-t",
578
+ "--tag",
579
+ type=str,
580
+ default=None,
581
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
582
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
583
+ args = parser.parse_args()
584
+
585
+ debug = args.debug
586
+
587
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)