Hugging Face's logo

gotzmann
/
v0.8.18-adapter

Safetensors
Model card Files Community
gotzmann commited on
Commit
c469d9c
•
1 Parent(s): 070ed0c

..

Browse files
Files changed (11) hide show
  1. 550/README.md +5 -0
  2. 550/adapter_config.json +30 -0
  3. 550/adapter_model.safetensors +3 -0
  4. 550/latest +1 -0
  5. 550/scheduler.pt +3 -0
  6. 550/special_tokens_map.json +24 -0
  7. 550/tokenizer.model +3 -0
  8. 550/tokenizer_config.json +45 -0
  9. 550/trainer_state.json +3871 -0
  10. 550/training_args.bin +3 -0
  11. 550/zero_to_fp32.py +604 -0
550/README.md ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: gotzmann/uni
4
+ ---
5
+
550/adapter_config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "gotzmann/uni",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "loftq_config": {},
12
+ "lora_alpha": 64,
13
+ "lora_dropout": 0.0,
14
+ "megatron_config": null,
15
+ "megatron_core": "megatron.core",
16
+ "modules_to_save": null,
17
+ "peft_type": "LORA",
18
+ "r": 64,
19
+ "rank_pattern": {},
20
+ "revision": null,
21
+ "target_modules": [
22
+ "k_proj",
23
+ "q_proj",
24
+ "o_proj",
25
+ "v_proj"
26
+ ],
27
+ "task_type": "CAUSAL_LM",
28
+ "use_dora": false,
29
+ "use_rslora": true
30
+ }
550/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9dec318f995b7373344d5e2162bfe856c84311ba9ae0b310aeebb3ea7d43b91
3
+ size 524376008
550/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step550
550/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40a12d9285b8b158862f33ffc490a848cc05f36eadf4b07505ee89c3e499e4c2
3
+ size 1064
550/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
550/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
550/tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": false,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ '<s>' + '### System:\\n\\n' + system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '\\n\\n### Human:\\n\\n' + content }}{% elif message['role'] == 'assistant' %}{{ '\\n\\n### Assistant:\\n\\n' + content + '</s>' }}{% endif %}{% endfor %}",
33
+ "clean_up_tokenization_spaces": false,
34
+ "eos_token": "</s>",
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "spaces_between_special_tokens": false,
41
+ "split_special_tokens": false,
42
+ "tokenizer_class": "LlamaTokenizer",
43
+ "unk_token": "<unk>",
44
+ "use_default_system_prompt": false
45
+ }
550/trainer_state.json ADDED
@@ -0,0 +1,3871 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0018214936247722,
5
+ "eval_steps": 500,
6
+ "global_step": 550,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "grad_norm": 0.46431864432643544,
14
+ "learning_rate": 1.2121212121212122e-06,
15
+ "loss": 1.4151,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.0,
20
+ "grad_norm": 0.42969176658539837,
21
+ "learning_rate": 2.4242424242424244e-06,
22
+ "loss": 1.3729,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.01,
27
+ "grad_norm": 0.5004307936270223,
28
+ "learning_rate": 3.636363636363636e-06,
29
+ "loss": 1.3989,
30
+ "step": 3
31
+ },
32
+ {
33
+ "epoch": 0.01,
34
+ "grad_norm": 0.43666634920041486,
35
+ "learning_rate": 4.848484848484849e-06,
36
+ "loss": 1.3363,
37
+ "step": 4
38
+ },
39
+ {
40
+ "epoch": 0.01,
41
+ "grad_norm": 0.4691114419353825,
42
+ "learning_rate": 6.060606060606061e-06,
43
+ "loss": 1.4293,
44
+ "step": 5
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "grad_norm": 0.4277596061377729,
49
+ "learning_rate": 7.272727272727272e-06,
50
+ "loss": 1.4343,
51
+ "step": 6
52
+ },
53
+ {
54
+ "epoch": 0.01,
55
+ "grad_norm": 0.4238339229382504,
56
+ "learning_rate": 8.484848484848486e-06,
57
+ "loss": 1.4462,
58
+ "step": 7
59
+ },
60
+ {
61
+ "epoch": 0.01,
62
+ "grad_norm": 0.40992048534183273,
63
+ "learning_rate": 9.696969696969698e-06,
64
+ "loss": 1.2756,
65
+ "step": 8
66
+ },
67
+ {
68
+ "epoch": 0.02,
69
+ "grad_norm": 0.37885700313540693,
70
+ "learning_rate": 1.0909090909090909e-05,
71
+ "loss": 1.3464,
72
+ "step": 9
73
+ },
74
+ {
75
+ "epoch": 0.02,
76
+ "grad_norm": 0.36066141170123023,
77
+ "learning_rate": 1.2121212121212122e-05,
78
+ "loss": 1.3419,
79
+ "step": 10
80
+ },
81
+ {
82
+ "epoch": 0.02,
83
+ "grad_norm": 0.35617169386863406,
84
+ "learning_rate": 1.3333333333333333e-05,
85
+ "loss": 1.3533,
86
+ "step": 11
87
+ },
88
+ {
89
+ "epoch": 0.02,
90
+ "grad_norm": 0.3040188564782602,
91
+ "learning_rate": 1.4545454545454545e-05,
92
+ "loss": 1.2395,
93
+ "step": 12
94
+ },
95
+ {
96
+ "epoch": 0.02,
97
+ "grad_norm": 0.31038319439216566,
98
+ "learning_rate": 1.5757575757575756e-05,
99
+ "loss": 1.3082,
100
+ "step": 13
101
+ },
102
+ {
103
+ "epoch": 0.03,
104
+ "grad_norm": 0.26683768372135835,
105
+ "learning_rate": 1.6969696969696972e-05,
106
+ "loss": 1.3063,
107
+ "step": 14
108
+ },
109
+ {
110
+ "epoch": 0.03,
111
+ "grad_norm": 0.3652323682563078,
112
+ "learning_rate": 1.8181818181818182e-05,
113
+ "loss": 1.3045,
114
+ "step": 15
115
+ },
116
+ {
117
+ "epoch": 0.03,
118
+ "grad_norm": 0.23559121485457843,
119
+ "learning_rate": 1.9393939393939395e-05,
120
+ "loss": 1.2366,
121
+ "step": 16
122
+ },
123
+ {
124
+ "epoch": 0.03,
125
+ "grad_norm": 0.2342299313020104,
126
+ "learning_rate": 2.0606060606060608e-05,
127
+ "loss": 1.2831,
128
+ "step": 17
129
+ },
130
+ {
131
+ "epoch": 0.03,
132
+ "grad_norm": 0.2202931700357255,
133
+ "learning_rate": 2.1818181818181818e-05,
134
+ "loss": 1.3064,
135
+ "step": 18
136
+ },
137
+ {
138
+ "epoch": 0.03,
139
+ "grad_norm": 0.2097660599292375,
140
+ "learning_rate": 2.3030303030303034e-05,
141
+ "loss": 1.2376,
142
+ "step": 19
143
+ },
144
+ {
145
+ "epoch": 0.04,
146
+ "grad_norm": 0.2356785314652122,
147
+ "learning_rate": 2.4242424242424244e-05,
148
+ "loss": 1.2802,
149
+ "step": 20
150
+ },
151
+ {
152
+ "epoch": 0.04,
153
+ "grad_norm": 0.24639302530564244,
154
+ "learning_rate": 2.5454545454545454e-05,
155
+ "loss": 1.3016,
156
+ "step": 21
157
+ },
158
+ {
159
+ "epoch": 0.04,
160
+ "grad_norm": 0.24373126133228787,
161
+ "learning_rate": 2.6666666666666667e-05,
162
+ "loss": 1.3407,
163
+ "step": 22
164
+ },
165
+ {
166
+ "epoch": 0.04,
167
+ "grad_norm": 0.24488805144123432,
168
+ "learning_rate": 2.7878787878787883e-05,
169
+ "loss": 1.3325,
170
+ "step": 23
171
+ },
172
+ {
173
+ "epoch": 0.04,
174
+ "grad_norm": 0.2653033507571198,
175
+ "learning_rate": 2.909090909090909e-05,
176
+ "loss": 1.2811,
177
+ "step": 24
178
+ },
179
+ {
180
+ "epoch": 0.05,
181
+ "grad_norm": 1.2841724819336817,
182
+ "learning_rate": 3.0303030303030306e-05,
183
+ "loss": 1.2837,
184
+ "step": 25
185
+ },
186
+ {
187
+ "epoch": 0.05,
188
+ "grad_norm": 0.2183883020111492,
189
+ "learning_rate": 3.151515151515151e-05,
190
+ "loss": 1.2472,
191
+ "step": 26
192
+ },
193
+ {
194
+ "epoch": 0.05,
195
+ "grad_norm": 0.2137995163762026,
196
+ "learning_rate": 3.272727272727273e-05,
197
+ "loss": 1.2854,
198
+ "step": 27
199
+ },
200
+ {
201
+ "epoch": 0.05,
202
+ "grad_norm": 0.19499006223503876,
203
+ "learning_rate": 3.3939393939393945e-05,
204
+ "loss": 1.3018,
205
+ "step": 28
206
+ },
207
+ {
208
+ "epoch": 0.05,
209
+ "grad_norm": 0.17367919355340256,
210
+ "learning_rate": 3.515151515151515e-05,
211
+ "loss": 1.2824,
212
+ "step": 29
213
+ },
214
+ {
215
+ "epoch": 0.05,
216
+ "grad_norm": 0.18326045693683557,
217
+ "learning_rate": 3.6363636363636364e-05,
218
+ "loss": 1.2192,
219
+ "step": 30
220
+ },
221
+ {
222
+ "epoch": 0.06,
223
+ "grad_norm": 0.17474388188066411,
224
+ "learning_rate": 3.757575757575758e-05,
225
+ "loss": 1.2078,
226
+ "step": 31
227
+ },
228
+ {
229
+ "epoch": 0.06,
230
+ "grad_norm": 0.17856970178098716,
231
+ "learning_rate": 3.878787878787879e-05,
232
+ "loss": 1.2683,
233
+ "step": 32
234
+ },
235
+ {
236
+ "epoch": 0.06,
237
+ "grad_norm": 0.18617589704298348,
238
+ "learning_rate": 4e-05,
239
+ "loss": 1.2265,
240
+ "step": 33
241
+ },
242
+ {
243
+ "epoch": 0.06,
244
+ "grad_norm": 0.17653209733215317,
245
+ "learning_rate": 4.1212121212121216e-05,
246
+ "loss": 1.319,
247
+ "step": 34
248
+ },
249
+ {
250
+ "epoch": 0.06,
251
+ "grad_norm": 0.1722921367585233,
252
+ "learning_rate": 4.242424242424243e-05,
253
+ "loss": 1.2117,
254
+ "step": 35
255
+ },
256
+ {
257
+ "epoch": 0.07,
258
+ "grad_norm": 0.176642606378719,
259
+ "learning_rate": 4.3636363636363636e-05,
260
+ "loss": 1.2512,
261
+ "step": 36
262
+ },
263
+ {
264
+ "epoch": 0.07,
265
+ "grad_norm": 0.16696442324691066,
266
+ "learning_rate": 4.484848484848485e-05,
267
+ "loss": 1.2637,
268
+ "step": 37
269
+ },
270
+ {
271
+ "epoch": 0.07,
272
+ "grad_norm": 0.17035384059517106,
273
+ "learning_rate": 4.606060606060607e-05,
274
+ "loss": 1.2699,
275
+ "step": 38
276
+ },
277
+ {
278
+ "epoch": 0.07,
279
+ "grad_norm": 0.15545801881444482,
280
+ "learning_rate": 4.7272727272727275e-05,
281
+ "loss": 1.2939,
282
+ "step": 39
283
+ },
284
+ {
285
+ "epoch": 0.07,
286
+ "grad_norm": 0.17111439344347512,
287
+ "learning_rate": 4.848484848484849e-05,
288
+ "loss": 1.3033,
289
+ "step": 40
290
+ },
291
+ {
292
+ "epoch": 0.07,
293
+ "grad_norm": 0.16994151343455458,
294
+ "learning_rate": 4.9696969696969694e-05,
295
+ "loss": 1.2603,
296
+ "step": 41
297
+ },
298
+ {
299
+ "epoch": 0.08,
300
+ "grad_norm": 0.15929214926453447,
301
+ "learning_rate": 5.090909090909091e-05,
302
+ "loss": 1.2626,
303
+ "step": 42
304
+ },
305
+ {
306
+ "epoch": 0.08,
307
+ "grad_norm": 0.16761261516238699,
308
+ "learning_rate": 5.212121212121213e-05,
309
+ "loss": 1.296,
310
+ "step": 43
311
+ },
312
+ {
313
+ "epoch": 0.08,
314
+ "grad_norm": 0.15754700542426123,
315
+ "learning_rate": 5.333333333333333e-05,
316
+ "loss": 1.278,
317
+ "step": 44
318
+ },
319
+ {
320
+ "epoch": 0.08,
321
+ "grad_norm": 0.15522526683877644,
322
+ "learning_rate": 5.4545454545454546e-05,
323
+ "loss": 1.2355,
324
+ "step": 45
325
+ },
326
+ {
327
+ "epoch": 0.08,
328
+ "grad_norm": 0.1577929926930023,
329
+ "learning_rate": 5.5757575757575766e-05,
330
+ "loss": 1.2879,
331
+ "step": 46
332
+ },
333
+ {
334
+ "epoch": 0.09,
335
+ "grad_norm": 0.31075066632858317,
336
+ "learning_rate": 5.696969696969697e-05,
337
+ "loss": 1.2202,
338
+ "step": 47
339
+ },
340
+ {
341
+ "epoch": 0.09,
342
+ "grad_norm": 0.1663780653395111,
343
+ "learning_rate": 5.818181818181818e-05,
344
+ "loss": 1.2319,
345
+ "step": 48
346
+ },
347
+ {
348
+ "epoch": 0.09,
349
+ "grad_norm": 0.16049499655883026,
350
+ "learning_rate": 5.93939393939394e-05,
351
+ "loss": 1.2801,
352
+ "step": 49
353
+ },
354
+ {
355
+ "epoch": 0.09,
356
+ "grad_norm": 0.14515773124436285,
357
+ "learning_rate": 6.060606060606061e-05,
358
+ "loss": 1.2588,
359
+ "step": 50
360
+ },
361
+ {
362
+ "epoch": 0.09,
363
+ "grad_norm": 0.14653064850325623,
364
+ "learning_rate": 6.181818181818182e-05,
365
+ "loss": 1.2677,
366
+ "step": 51
367
+ },
368
+ {
369
+ "epoch": 0.09,
370
+ "grad_norm": 0.17193239746689878,
371
+ "learning_rate": 6.303030303030302e-05,
372
+ "loss": 1.2742,
373
+ "step": 52
374
+ },
375
+ {
376
+ "epoch": 0.1,
377
+ "grad_norm": 0.1967020450342533,
378
+ "learning_rate": 6.424242424242424e-05,
379
+ "loss": 1.1545,
380
+ "step": 53
381
+ },
382
+ {
383
+ "epoch": 0.1,
384
+ "grad_norm": 0.16247531997247225,
385
+ "learning_rate": 6.545454545454546e-05,
386
+ "loss": 1.222,
387
+ "step": 54
388
+ },
389
+ {
390
+ "epoch": 0.1,
391
+ "grad_norm": 0.14990706377244528,
392
+ "learning_rate": 6.666666666666667e-05,
393
+ "loss": 1.2103,
394
+ "step": 55
395
+ },
396
+ {
397
+ "epoch": 0.1,
398
+ "grad_norm": 0.1412817445239095,
399
+ "learning_rate": 6.787878787878789e-05,
400
+ "loss": 1.2169,
401
+ "step": 56
402
+ },
403
+ {
404
+ "epoch": 0.1,
405
+ "grad_norm": 0.14575971073482757,
406
+ "learning_rate": 6.90909090909091e-05,
407
+ "loss": 1.2751,
408
+ "step": 57
409
+ },
410
+ {
411
+ "epoch": 0.11,
412
+ "grad_norm": 0.13714747569950891,
413
+ "learning_rate": 7.03030303030303e-05,
414
+ "loss": 1.2508,
415
+ "step": 58
416
+ },
417
+ {
418
+ "epoch": 0.11,
419
+ "grad_norm": 0.14334695156859903,
420
+ "learning_rate": 7.151515151515152e-05,
421
+ "loss": 1.2721,
422
+ "step": 59
423
+ },
424
+ {
425
+ "epoch": 0.11,
426
+ "grad_norm": 0.1456824177522916,
427
+ "learning_rate": 7.272727272727273e-05,
428
+ "loss": 1.2649,
429
+ "step": 60
430
+ },
431
+ {
432
+ "epoch": 0.11,
433
+ "grad_norm": 0.15030318240210044,
434
+ "learning_rate": 7.393939393939395e-05,
435
+ "loss": 1.2167,
436
+ "step": 61
437
+ },
438
+ {
439
+ "epoch": 0.11,
440
+ "grad_norm": 0.1651326066719482,
441
+ "learning_rate": 7.515151515151515e-05,
442
+ "loss": 1.3126,
443
+ "step": 62
444
+ },
445
+ {
446
+ "epoch": 0.11,
447
+ "grad_norm": 0.1408250406479118,
448
+ "learning_rate": 7.636363636363637e-05,
449
+ "loss": 1.2891,
450
+ "step": 63
451
+ },
452
+ {
453
+ "epoch": 0.12,
454
+ "grad_norm": 0.21501384376905694,
455
+ "learning_rate": 7.757575757575758e-05,
456
+ "loss": 1.3019,
457
+ "step": 64
458
+ },
459
+ {
460
+ "epoch": 0.12,
461
+ "grad_norm": 0.1365168726167339,
462
+ "learning_rate": 7.878787878787879e-05,
463
+ "loss": 1.2498,
464
+ "step": 65
465
+ },
466
+ {
467
+ "epoch": 0.12,
468
+ "grad_norm": 0.1431463689660936,
469
+ "learning_rate": 8e-05,
470
+ "loss": 1.2793,
471
+ "step": 66
472
+ },
473
+ {
474
+ "epoch": 0.12,
475
+ "grad_norm": 0.13689045214286194,
476
+ "learning_rate": 8.121212121212121e-05,
477
+ "loss": 1.2295,
478
+ "step": 67
479
+ },
480
+ {
481
+ "epoch": 0.12,
482
+ "grad_norm": 0.13483608710081227,
483
+ "learning_rate": 8.242424242424243e-05,
484
+ "loss": 1.2258,
485
+ "step": 68
486
+ },
487
+ {
488
+ "epoch": 0.13,
489
+ "grad_norm": 0.13707618564415613,
490
+ "learning_rate": 8.363636363636364e-05,
491
+ "loss": 1.2252,
492
+ "step": 69
493
+ },
494
+ {
495
+ "epoch": 0.13,
496
+ "grad_norm": 0.13780236215967515,
497
+ "learning_rate": 8.484848484848486e-05,
498
+ "loss": 1.2565,
499
+ "step": 70
500
+ },
501
+ {
502
+ "epoch": 0.13,
503
+ "grad_norm": 0.14036805493494423,
504
+ "learning_rate": 8.606060606060606e-05,
505
+ "loss": 1.3023,
506
+ "step": 71
507
+ },
508
+ {
509
+ "epoch": 0.13,
510
+ "grad_norm": 0.12776919439147982,
511
+ "learning_rate": 8.727272727272727e-05,
512
+ "loss": 1.2292,
513
+ "step": 72
514
+ },
515
+ {
516
+ "epoch": 0.13,
517
+ "grad_norm": 0.1289941815481437,
518
+ "learning_rate": 8.848484848484849e-05,
519
+ "loss": 1.2191,
520
+ "step": 73
521
+ },
522
+ {
523
+ "epoch": 0.13,
524
+ "grad_norm": 0.13943952294847306,
525
+ "learning_rate": 8.96969696969697e-05,
526
+ "loss": 1.2915,
527
+ "step": 74
528
+ },
529
+ {
530
+ "epoch": 0.14,
531
+ "grad_norm": 0.1493528502117281,
532
+ "learning_rate": 9.090909090909092e-05,
533
+ "loss": 1.2797,
534
+ "step": 75
535
+ },
536
+ {
537
+ "epoch": 0.14,
538
+ "grad_norm": 0.1252401242451818,
539
+ "learning_rate": 9.212121212121214e-05,
540
+ "loss": 1.2552,
541
+ "step": 76
542
+ },
543
+ {
544
+ "epoch": 0.14,
545
+ "grad_norm": 0.13969800467546992,
546
+ "learning_rate": 9.333333333333334e-05,
547
+ "loss": 1.3147,
548
+ "step": 77
549
+ },
550
+ {
551
+ "epoch": 0.14,
552
+ "grad_norm": 0.1277258491470434,
553
+ "learning_rate": 9.454545454545455e-05,
554
+ "loss": 1.2089,
555
+ "step": 78
556
+ },
557
+ {
558
+ "epoch": 0.14,
559
+ "grad_norm": 0.133041369314817,
560
+ "learning_rate": 9.575757575757576e-05,
561
+ "loss": 1.2761,
562
+ "step": 79
563
+ },
564
+ {
565
+ "epoch": 0.15,
566
+ "grad_norm": 0.14564572037181842,
567
+ "learning_rate": 9.696969696969698e-05,
568
+ "loss": 1.1901,
569
+ "step": 80
570
+ },
571
+ {
572
+ "epoch": 0.15,
573
+ "grad_norm": 0.13666505656492195,
574
+ "learning_rate": 9.818181818181818e-05,
575
+ "loss": 1.2615,
576
+ "step": 81
577
+ },
578
+ {
579
+ "epoch": 0.15,
580
+ "grad_norm": 0.135007805210003,
581
+ "learning_rate": 9.939393939393939e-05,
582
+ "loss": 1.2669,
583
+ "step": 82
584
+ },
585
+ {
586
+ "epoch": 0.15,
587
+ "grad_norm": 0.17287563365884975,
588
+ "learning_rate": 0.00010060606060606062,
589
+ "loss": 1.2669,
590
+ "step": 83
591
+ },
592
+ {
593
+ "epoch": 0.15,
594
+ "grad_norm": 0.12934306326048103,
595
+ "learning_rate": 0.00010181818181818181,
596
+ "loss": 1.1979,
597
+ "step": 84
598
+ },
599
+ {
600
+ "epoch": 0.15,
601
+ "grad_norm": 0.13517436169178096,
602
+ "learning_rate": 0.00010303030303030303,
603
+ "loss": 1.2226,
604
+ "step": 85
605
+ },
606
+ {
607
+ "epoch": 0.16,
608
+ "grad_norm": 0.12105351159271568,
609
+ "learning_rate": 0.00010424242424242425,
610
+ "loss": 1.1172,
611
+ "step": 86
612
+ },
613
+ {
614
+ "epoch": 0.16,
615
+ "grad_norm": 0.1281676431775383,
616
+ "learning_rate": 0.00010545454545454545,
617
+ "loss": 1.2046,
618
+ "step": 87
619
+ },
620
+ {
621
+ "epoch": 0.16,
622
+ "grad_norm": 0.11730963057933333,
623
+ "learning_rate": 0.00010666666666666667,
624
+ "loss": 1.1883,
625
+ "step": 88
626
+ },
627
+ {
628
+ "epoch": 0.16,
629
+ "grad_norm": 0.12655235108503246,
630
+ "learning_rate": 0.00010787878787878789,
631
+ "loss": 1.1331,
632
+ "step": 89
633
+ },
634
+ {
635
+ "epoch": 0.16,
636
+ "grad_norm": 0.13047560307970027,
637
+ "learning_rate": 0.00010909090909090909,
638
+ "loss": 1.2731,
639
+ "step": 90
640
+ },
641
+ {
642
+ "epoch": 0.17,
643
+ "grad_norm": 0.12193522973752649,
644
+ "learning_rate": 0.00011030303030303031,
645
+ "loss": 1.2161,
646
+ "step": 91
647
+ },
648
+ {
649
+ "epoch": 0.17,
650
+ "grad_norm": 0.12804360300116346,
651
+ "learning_rate": 0.00011151515151515153,
652
+ "loss": 1.3062,
653
+ "step": 92
654
+ },
655
+ {
656
+ "epoch": 0.17,
657
+ "grad_norm": 0.15991741754516206,
658
+ "learning_rate": 0.00011272727272727272,
659
+ "loss": 1.239,
660
+ "step": 93
661
+ },
662
+ {
663
+ "epoch": 0.17,
664
+ "grad_norm": 0.15140182244454561,
665
+ "learning_rate": 0.00011393939393939394,
666
+ "loss": 1.2349,
667
+ "step": 94
668
+ },
669
+ {
670
+ "epoch": 0.17,
671
+ "grad_norm": 0.12320241076263434,
672
+ "learning_rate": 0.00011515151515151516,
673
+ "loss": 1.2875,
674
+ "step": 95
675
+ },
676
+ {
677
+ "epoch": 0.17,
678
+ "grad_norm": 0.13235998458230466,
679
+ "learning_rate": 0.00011636363636363636,
680
+ "loss": 1.2218,
681
+ "step": 96
682
+ },
683
+ {
684
+ "epoch": 0.18,
685
+ "grad_norm": 0.11783688734798668,
686
+ "learning_rate": 0.00011757575757575758,
687
+ "loss": 1.1864,
688
+ "step": 97
689
+ },
690
+ {
691
+ "epoch": 0.18,
692
+ "grad_norm": 0.3151933420750235,
693
+ "learning_rate": 0.0001187878787878788,
694
+ "loss": 1.3023,
695
+ "step": 98
696
+ },
697
+ {
698
+ "epoch": 0.18,
699
+ "grad_norm": 0.12665632567219295,
700
+ "learning_rate": 0.00012,
701
+ "loss": 1.2249,
702
+ "step": 99
703
+ },
704
+ {
705
+ "epoch": 0.18,
706
+ "grad_norm": 0.1228886740460738,
707
+ "learning_rate": 0.00012121212121212122,
708
+ "loss": 1.2517,
709
+ "step": 100
710
+ },
711
+ {
712
+ "epoch": 0.18,
713
+ "grad_norm": 0.11892005244989344,
714
+ "learning_rate": 0.00012242424242424243,
715
+ "loss": 1.2586,
716
+ "step": 101
717
+ },
718
+ {
719
+ "epoch": 0.19,
720
+ "grad_norm": 0.1232340827222201,
721
+ "learning_rate": 0.00012363636363636364,
722
+ "loss": 1.3217,
723
+ "step": 102
724
+ },
725
+ {
726
+ "epoch": 0.19,
727
+ "grad_norm": 0.13837226869323116,
728
+ "learning_rate": 0.00012484848484848487,
729
+ "loss": 1.2693,
730
+ "step": 103
731
+ },
732
+ {
733
+ "epoch": 0.19,
734
+ "grad_norm": 0.12068217991774362,
735
+ "learning_rate": 0.00012606060606060605,
736
+ "loss": 1.2623,
737
+ "step": 104
738
+ },
739
+ {
740
+ "epoch": 0.19,
741
+ "grad_norm": 0.16779277284606545,
742
+ "learning_rate": 0.00012727272727272728,
743
+ "loss": 1.2415,
744
+ "step": 105
745
+ },
746
+ {
747
+ "epoch": 0.19,
748
+ "grad_norm": 0.13396891539963085,
749
+ "learning_rate": 0.0001284848484848485,
750
+ "loss": 1.2313,
751
+ "step": 106
752
+ },
753
+ {
754
+ "epoch": 0.19,
755
+ "grad_norm": 0.12457104490772812,
756
+ "learning_rate": 0.0001296969696969697,
757
+ "loss": 1.1758,
758
+ "step": 107
759
+ },
760
+ {
761
+ "epoch": 0.2,
762
+ "grad_norm": 0.12676816816563452,
763
+ "learning_rate": 0.00013090909090909093,
764
+ "loss": 1.2478,
765
+ "step": 108
766
+ },
767
+ {
768
+ "epoch": 0.2,
769
+ "grad_norm": 0.11973639622066906,
770
+ "learning_rate": 0.00013212121212121213,
771
+ "loss": 1.2335,
772
+ "step": 109
773
+ },
774
+ {
775
+ "epoch": 0.2,
776
+ "grad_norm": 0.1330159646034068,
777
+ "learning_rate": 0.00013333333333333334,
778
+ "loss": 1.26,
779
+ "step": 110
780
+ },
781
+ {
782
+ "epoch": 0.2,
783
+ "grad_norm": 0.1298003025338099,
784
+ "learning_rate": 0.00013454545454545455,
785
+ "loss": 1.1907,
786
+ "step": 111
787
+ },
788
+ {
789
+ "epoch": 0.2,
790
+ "grad_norm": 0.1226154813287666,
791
+ "learning_rate": 0.00013575757575757578,
792
+ "loss": 1.1807,
793
+ "step": 112
794
+ },
795
+ {
796
+ "epoch": 0.21,
797
+ "grad_norm": 0.12533753244302145,
798
+ "learning_rate": 0.00013696969696969696,
799
+ "loss": 1.2098,
800
+ "step": 113
801
+ },
802
+ {
803
+ "epoch": 0.21,
804
+ "grad_norm": 0.12673266503840944,
805
+ "learning_rate": 0.0001381818181818182,
806
+ "loss": 1.2265,
807
+ "step": 114
808
+ },
809
+ {
810
+ "epoch": 0.21,
811
+ "grad_norm": 0.1299039569361384,
812
+ "learning_rate": 0.0001393939393939394,
813
+ "loss": 1.2534,
814
+ "step": 115
815
+ },
816
+ {
817
+ "epoch": 0.21,
818
+ "grad_norm": 0.13023496663090803,
819
+ "learning_rate": 0.0001406060606060606,
820
+ "loss": 1.2453,
821
+ "step": 116
822
+ },
823
+ {
824
+ "epoch": 0.21,
825
+ "grad_norm": 0.12001793500864573,
826
+ "learning_rate": 0.00014181818181818184,
827
+ "loss": 1.1608,
828
+ "step": 117
829
+ },
830
+ {
831
+ "epoch": 0.21,
832
+ "grad_norm": 0.14561862193041028,
833
+ "learning_rate": 0.00014303030303030304,
834
+ "loss": 1.2233,
835
+ "step": 118
836
+ },
837
+ {
838
+ "epoch": 0.22,
839
+ "grad_norm": 0.12636130876430832,
840
+ "learning_rate": 0.00014424242424242425,
841
+ "loss": 1.2833,
842
+ "step": 119
843
+ },
844
+ {
845
+ "epoch": 0.22,
846
+ "grad_norm": 0.189556849271166,
847
+ "learning_rate": 0.00014545454545454546,
848
+ "loss": 1.3105,
849
+ "step": 120
850
+ },
851
+ {
852
+ "epoch": 0.22,
853
+ "grad_norm": 0.12409073764495662,
854
+ "learning_rate": 0.00014666666666666666,
855
+ "loss": 1.1534,
856
+ "step": 121
857
+ },
858
+ {
859
+ "epoch": 0.22,
860
+ "grad_norm": 0.12149212466969316,
861
+ "learning_rate": 0.0001478787878787879,
862
+ "loss": 1.3039,
863
+ "step": 122
864
+ },
865
+ {
866
+ "epoch": 0.22,
867
+ "grad_norm": 0.12147336887953522,
868
+ "learning_rate": 0.0001490909090909091,
869
+ "loss": 1.326,
870
+ "step": 123
871
+ },
872
+ {
873
+ "epoch": 0.23,
874
+ "grad_norm": 0.1176585016163167,
875
+ "learning_rate": 0.0001503030303030303,
876
+ "loss": 1.191,
877
+ "step": 124
878
+ },
879
+ {
880
+ "epoch": 0.23,
881
+ "grad_norm": 0.2066428974234372,
882
+ "learning_rate": 0.00015151515151515152,
883
+ "loss": 1.3054,
884
+ "step": 125
885
+ },
886
+ {
887
+ "epoch": 0.23,
888
+ "grad_norm": 0.29582724255710047,
889
+ "learning_rate": 0.00015272727272727275,
890
+ "loss": 1.2032,
891
+ "step": 126
892
+ },
893
+ {
894
+ "epoch": 0.23,
895
+ "grad_norm": 0.13084381204119358,
896
+ "learning_rate": 0.00015393939393939393,
897
+ "loss": 1.2289,
898
+ "step": 127
899
+ },
900
+ {
901
+ "epoch": 0.23,
902
+ "grad_norm": 0.1294157600411397,
903
+ "learning_rate": 0.00015515151515151516,
904
+ "loss": 1.2561,
905
+ "step": 128
906
+ },
907
+ {
908
+ "epoch": 0.23,
909
+ "grad_norm": 0.14039614543447027,
910
+ "learning_rate": 0.00015636363636363637,
911
+ "loss": 1.243,
912
+ "step": 129
913
+ },
914
+ {
915
+ "epoch": 0.24,
916
+ "grad_norm": 0.19939984917282128,
917
+ "learning_rate": 0.00015757575757575757,
918
+ "loss": 1.1286,
919
+ "step": 130
920
+ },
921
+ {
922
+ "epoch": 0.24,
923
+ "grad_norm": 0.14402764349968203,
924
+ "learning_rate": 0.0001587878787878788,
925
+ "loss": 1.1959,
926
+ "step": 131
927
+ },
928
+ {
929
+ "epoch": 0.24,
930
+ "grad_norm": 0.13970978861500938,
931
+ "learning_rate": 0.00016,
932
+ "loss": 1.1814,
933
+ "step": 132
934
+ },
935
+ {
936
+ "epoch": 0.24,
937
+ "grad_norm": 0.14539538472563127,
938
+ "learning_rate": 0.00016121212121212122,
939
+ "loss": 1.2317,
940
+ "step": 133
941
+ },
942
+ {
943
+ "epoch": 0.24,
944
+ "grad_norm": 0.13456425455391557,
945
+ "learning_rate": 0.00016242424242424243,
946
+ "loss": 1.2239,
947
+ "step": 134
948
+ },
949
+ {
950
+ "epoch": 0.25,
951
+ "grad_norm": 0.1314997837157779,
952
+ "learning_rate": 0.00016363636363636366,
953
+ "loss": 1.1986,
954
+ "step": 135
955
+ },
956
+ {
957
+ "epoch": 0.25,
958
+ "grad_norm": 0.14046946525591422,
959
+ "learning_rate": 0.00016484848484848487,
960
+ "loss": 1.2238,
961
+ "step": 136
962
+ },
963
+ {
964
+ "epoch": 0.25,
965
+ "grad_norm": 0.6095538041505763,
966
+ "learning_rate": 0.00016606060606060607,
967
+ "loss": 1.2332,
968
+ "step": 137
969
+ },
970
+ {
971
+ "epoch": 0.25,
972
+ "grad_norm": 0.17707289712054367,
973
+ "learning_rate": 0.00016727272727272728,
974
+ "loss": 1.2401,
975
+ "step": 138
976
+ },
977
+ {
978
+ "epoch": 0.25,
979
+ "grad_norm": 0.19335172179099247,
980
+ "learning_rate": 0.00016848484848484848,
981
+ "loss": 1.2361,
982
+ "step": 139
983
+ },
984
+ {
985
+ "epoch": 0.26,
986
+ "grad_norm": 0.13725591818701255,
987
+ "learning_rate": 0.00016969696969696972,
988
+ "loss": 1.193,
989
+ "step": 140
990
+ },
991
+ {
992
+ "epoch": 0.26,
993
+ "grad_norm": 0.15535575462507384,
994
+ "learning_rate": 0.0001709090909090909,
995
+ "loss": 1.2769,
996
+ "step": 141
997
+ },
998
+ {
999
+ "epoch": 0.26,
1000
+ "grad_norm": 0.14909436560898923,
1001
+ "learning_rate": 0.00017212121212121213,
1002
+ "loss": 1.2602,
1003
+ "step": 142
1004
+ },
1005
+ {
1006
+ "epoch": 0.26,
1007
+ "grad_norm": 0.15054368082407957,
1008
+ "learning_rate": 0.00017333333333333334,
1009
+ "loss": 1.2607,
1010
+ "step": 143
1011
+ },
1012
+ {
1013
+ "epoch": 0.26,
1014
+ "grad_norm": 0.13386897838741724,
1015
+ "learning_rate": 0.00017454545454545454,
1016
+ "loss": 1.168,
1017
+ "step": 144
1018
+ },
1019
+ {
1020
+ "epoch": 0.26,
1021
+ "grad_norm": 0.13567889528730145,
1022
+ "learning_rate": 0.00017575757575757578,
1023
+ "loss": 1.1984,
1024
+ "step": 145
1025
+ },
1026
+ {
1027
+ "epoch": 0.27,
1028
+ "grad_norm": 0.13994382298003089,
1029
+ "learning_rate": 0.00017696969696969698,
1030
+ "loss": 1.2795,
1031
+ "step": 146
1032
+ },
1033
+ {
1034
+ "epoch": 0.27,
1035
+ "grad_norm": 0.13941573210713187,
1036
+ "learning_rate": 0.0001781818181818182,
1037
+ "loss": 1.2303,
1038
+ "step": 147
1039
+ },
1040
+ {
1041
+ "epoch": 0.27,
1042
+ "grad_norm": 0.18302605925485763,
1043
+ "learning_rate": 0.0001793939393939394,
1044
+ "loss": 1.2696,
1045
+ "step": 148
1046
+ },
1047
+ {
1048
+ "epoch": 0.27,
1049
+ "grad_norm": 0.1547402223275396,
1050
+ "learning_rate": 0.00018060606060606063,
1051
+ "loss": 1.1276,
1052
+ "step": 149
1053
+ },
1054
+ {
1055
+ "epoch": 0.27,
1056
+ "grad_norm": 0.19947594494850646,
1057
+ "learning_rate": 0.00018181818181818183,
1058
+ "loss": 1.271,
1059
+ "step": 150
1060
+ },
1061
+ {
1062
+ "epoch": 0.28,
1063
+ "grad_norm": 0.1517101450465788,
1064
+ "learning_rate": 0.00018303030303030304,
1065
+ "loss": 1.2193,
1066
+ "step": 151
1067
+ },
1068
+ {
1069
+ "epoch": 0.28,
1070
+ "grad_norm": 0.19251063116857103,
1071
+ "learning_rate": 0.00018424242424242427,
1072
+ "loss": 1.2703,
1073
+ "step": 152
1074
+ },
1075
+ {
1076
+ "epoch": 0.28,
1077
+ "grad_norm": 0.16789099560498666,
1078
+ "learning_rate": 0.00018545454545454545,
1079
+ "loss": 1.2244,
1080
+ "step": 153
1081
+ },
1082
+ {
1083
+ "epoch": 0.28,
1084
+ "grad_norm": 0.14907376557922342,
1085
+ "learning_rate": 0.0001866666666666667,
1086
+ "loss": 1.264,
1087
+ "step": 154
1088
+ },
1089
+ {
1090
+ "epoch": 0.28,
1091
+ "grad_norm": 0.14276598263036905,
1092
+ "learning_rate": 0.0001878787878787879,
1093
+ "loss": 1.2545,
1094
+ "step": 155
1095
+ },
1096
+ {
1097
+ "epoch": 0.28,
1098
+ "grad_norm": 0.14526753816999002,
1099
+ "learning_rate": 0.0001890909090909091,
1100
+ "loss": 1.2912,
1101
+ "step": 156
1102
+ },
1103
+ {
1104
+ "epoch": 0.29,
1105
+ "grad_norm": 0.1627048894660859,
1106
+ "learning_rate": 0.0001903030303030303,
1107
+ "loss": 1.2573,
1108
+ "step": 157
1109
+ },
1110
+ {
1111
+ "epoch": 0.29,
1112
+ "grad_norm": 0.16405036632332695,
1113
+ "learning_rate": 0.0001915151515151515,
1114
+ "loss": 1.2359,
1115
+ "step": 158
1116
+ },
1117
+ {
1118
+ "epoch": 0.29,
1119
+ "grad_norm": 0.14533427219788658,
1120
+ "learning_rate": 0.00019272727272727274,
1121
+ "loss": 1.1718,
1122
+ "step": 159
1123
+ },
1124
+ {
1125
+ "epoch": 0.29,
1126
+ "grad_norm": 0.13802382666732702,
1127
+ "learning_rate": 0.00019393939393939395,
1128
+ "loss": 1.2297,
1129
+ "step": 160
1130
+ },
1131
+ {
1132
+ "epoch": 0.29,
1133
+ "grad_norm": 0.15620193618511755,
1134
+ "learning_rate": 0.00019515151515151516,
1135
+ "loss": 1.2287,
1136
+ "step": 161
1137
+ },
1138
+ {
1139
+ "epoch": 0.3,
1140
+ "grad_norm": 0.1401696295700075,
1141
+ "learning_rate": 0.00019636363636363636,
1142
+ "loss": 1.2231,
1143
+ "step": 162
1144
+ },
1145
+ {
1146
+ "epoch": 0.3,
1147
+ "grad_norm": 0.15816133304035035,
1148
+ "learning_rate": 0.0001975757575757576,
1149
+ "loss": 1.2804,
1150
+ "step": 163
1151
+ },
1152
+ {
1153
+ "epoch": 0.3,
1154
+ "grad_norm": 0.14626275180535692,
1155
+ "learning_rate": 0.00019878787878787878,
1156
+ "loss": 1.2115,
1157
+ "step": 164
1158
+ },
1159
+ {
1160
+ "epoch": 0.3,
1161
+ "grad_norm": 0.13100680398305042,
1162
+ "learning_rate": 0.0002,
1163
+ "loss": 1.2524,
1164
+ "step": 165
1165
+ },
1166
+ {
1167
+ "epoch": 0.3,
1168
+ "grad_norm": 0.14849458896148926,
1169
+ "learning_rate": 0.00019999977531546566,
1170
+ "loss": 1.2161,
1171
+ "step": 166
1172
+ },
1173
+ {
1174
+ "epoch": 0.3,
1175
+ "grad_norm": 0.13628125499037252,
1176
+ "learning_rate": 0.0001999991012628722,
1177
+ "loss": 1.2452,
1178
+ "step": 167
1179
+ },
1180
+ {
1181
+ "epoch": 0.31,
1182
+ "grad_norm": 0.18617698759086793,
1183
+ "learning_rate": 0.00019999797784524866,
1184
+ "loss": 1.2197,
1185
+ "step": 168
1186
+ },
1187
+ {
1188
+ "epoch": 0.31,
1189
+ "grad_norm": 0.14416004826313944,
1190
+ "learning_rate": 0.00019999640506764336,
1191
+ "loss": 1.2796,
1192
+ "step": 169
1193
+ },
1194
+ {
1195
+ "epoch": 0.31,
1196
+ "grad_norm": 0.13807081386834757,
1197
+ "learning_rate": 0.0001999943829371238,
1198
+ "loss": 1.2732,
1199
+ "step": 170
1200
+ },
1201
+ {
1202
+ "epoch": 0.31,
1203
+ "grad_norm": 0.16526927436841996,
1204
+ "learning_rate": 0.0001999919114627769,
1205
+ "loss": 1.3016,
1206
+ "step": 171
1207
+ },
1208
+ {
1209
+ "epoch": 0.31,
1210
+ "grad_norm": 0.14479672734919855,
1211
+ "learning_rate": 0.0001999889906557086,
1212
+ "loss": 1.3106,
1213
+ "step": 172
1214
+ },
1215
+ {
1216
+ "epoch": 0.32,
1217
+ "grad_norm": 0.13829284006072087,
1218
+ "learning_rate": 0.00019998562052904418,
1219
+ "loss": 1.3355,
1220
+ "step": 173
1221
+ },
1222
+ {
1223
+ "epoch": 0.32,
1224
+ "grad_norm": 0.13484630104616105,
1225
+ "learning_rate": 0.0001999818010979279,
1226
+ "loss": 1.1928,
1227
+ "step": 174
1228
+ },
1229
+ {
1230
+ "epoch": 0.32,
1231
+ "grad_norm": 0.14972770674556948,
1232
+ "learning_rate": 0.00019997753237952317,
1233
+ "loss": 1.2559,
1234
+ "step": 175
1235
+ },
1236
+ {
1237
+ "epoch": 0.32,
1238
+ "grad_norm": 0.13378525020528342,
1239
+ "learning_rate": 0.00019997281439301218,
1240
+ "loss": 1.2673,
1241
+ "step": 176
1242
+ },
1243
+ {
1244
+ "epoch": 0.32,
1245
+ "grad_norm": 0.13242998699125438,
1246
+ "learning_rate": 0.00019996764715959618,
1247
+ "loss": 1.2272,
1248
+ "step": 177
1249
+ },
1250
+ {
1251
+ "epoch": 0.32,
1252
+ "grad_norm": 0.12938881004364342,
1253
+ "learning_rate": 0.00019996203070249516,
1254
+ "loss": 1.2035,
1255
+ "step": 178
1256
+ },
1257
+ {
1258
+ "epoch": 0.33,
1259
+ "grad_norm": 0.13388032350164566,
1260
+ "learning_rate": 0.00019995596504694763,
1261
+ "loss": 1.2642,
1262
+ "step": 179
1263
+ },
1264
+ {
1265
+ "epoch": 0.33,
1266
+ "grad_norm": 0.13893372222140873,
1267
+ "learning_rate": 0.00019994945022021082,
1268
+ "loss": 1.2235,
1269
+ "step": 180
1270
+ },
1271
+ {
1272
+ "epoch": 0.33,
1273
+ "grad_norm": 0.14131710715500717,
1274
+ "learning_rate": 0.00019994248625156038,
1275
+ "loss": 1.1095,
1276
+ "step": 181
1277
+ },
1278
+ {
1279
+ "epoch": 0.33,
1280
+ "grad_norm": 0.13448100369103572,
1281
+ "learning_rate": 0.0001999350731722902,
1282
+ "loss": 1.1879,
1283
+ "step": 182
1284
+ },
1285
+ {
1286
+ "epoch": 0.33,
1287
+ "grad_norm": 0.13862444003216381,
1288
+ "learning_rate": 0.00019992721101571236,
1289
+ "loss": 1.2227,
1290
+ "step": 183
1291
+ },
1292
+ {
1293
+ "epoch": 0.34,
1294
+ "grad_norm": 0.13506115547921224,
1295
+ "learning_rate": 0.00019991889981715698,
1296
+ "loss": 1.2833,
1297
+ "step": 184
1298
+ },
1299
+ {
1300
+ "epoch": 0.34,
1301
+ "grad_norm": 0.13174857502600473,
1302
+ "learning_rate": 0.00019991013961397197,
1303
+ "loss": 1.2394,
1304
+ "step": 185
1305
+ },
1306
+ {
1307
+ "epoch": 0.34,
1308
+ "grad_norm": 0.1290276308949748,
1309
+ "learning_rate": 0.00019990093044552304,
1310
+ "loss": 1.2659,
1311
+ "step": 186
1312
+ },
1313
+ {
1314
+ "epoch": 0.34,
1315
+ "grad_norm": 0.1388159912078538,
1316
+ "learning_rate": 0.0001998912723531933,
1317
+ "loss": 1.3052,
1318
+ "step": 187
1319
+ },
1320
+ {
1321
+ "epoch": 0.34,
1322
+ "grad_norm": 0.1256806205303357,
1323
+ "learning_rate": 0.00019988116538038325,
1324
+ "loss": 1.2031,
1325
+ "step": 188
1326
+ },
1327
+ {
1328
+ "epoch": 0.34,
1329
+ "grad_norm": 0.13256850855084143,
1330
+ "learning_rate": 0.00019987060957251047,
1331
+ "loss": 1.211,
1332
+ "step": 189
1333
+ },
1334
+ {
1335
+ "epoch": 0.35,
1336
+ "grad_norm": 0.13197363789890235,
1337
+ "learning_rate": 0.0001998596049770095,
1338
+ "loss": 1.2256,
1339
+ "step": 190
1340
+ },
1341
+ {
1342
+ "epoch": 0.35,
1343
+ "grad_norm": 0.13277364593883098,
1344
+ "learning_rate": 0.00019984815164333163,
1345
+ "loss": 1.2174,
1346
+ "step": 191
1347
+ },
1348
+ {
1349
+ "epoch": 0.35,
1350
+ "grad_norm": 0.13838072824574454,
1351
+ "learning_rate": 0.00019983624962294458,
1352
+ "loss": 1.3128,
1353
+ "step": 192
1354
+ },
1355
+ {
1356
+ "epoch": 0.35,
1357
+ "grad_norm": 0.13524759737199996,
1358
+ "learning_rate": 0.0001998238989693323,
1359
+ "loss": 1.1806,
1360
+ "step": 193
1361
+ },
1362
+ {
1363
+ "epoch": 0.35,
1364
+ "grad_norm": 0.12669987683723832,
1365
+ "learning_rate": 0.0001998110997379949,
1366
+ "loss": 1.2171,
1367
+ "step": 194
1368
+ },
1369
+ {
1370
+ "epoch": 0.36,
1371
+ "grad_norm": 0.1461834612451898,
1372
+ "learning_rate": 0.00019979785198644806,
1373
+ "loss": 1.2231,
1374
+ "step": 195
1375
+ },
1376
+ {
1377
+ "epoch": 0.36,
1378
+ "grad_norm": 0.13265793664862735,
1379
+ "learning_rate": 0.0001997841557742232,
1380
+ "loss": 1.1718,
1381
+ "step": 196
1382
+ },
1383
+ {
1384
+ "epoch": 0.36,
1385
+ "grad_norm": 0.12842971557690963,
1386
+ "learning_rate": 0.00019977001116286674,
1387
+ "loss": 1.2758,
1388
+ "step": 197
1389
+ },
1390
+ {
1391
+ "epoch": 0.36,
1392
+ "grad_norm": 0.12188365921206967,
1393
+ "learning_rate": 0.00019975541821594026,
1394
+ "loss": 1.2457,
1395
+ "step": 198
1396
+ },
1397
+ {
1398
+ "epoch": 0.36,
1399
+ "grad_norm": 0.12679949330022622,
1400
+ "learning_rate": 0.00019974037699901993,
1401
+ "loss": 1.1825,
1402
+ "step": 199
1403
+ },
1404
+ {
1405
+ "epoch": 0.36,
1406
+ "grad_norm": 0.12949746150357985,
1407
+ "learning_rate": 0.00019972488757969635,
1408
+ "loss": 1.2666,
1409
+ "step": 200
1410
+ },
1411
+ {
1412
+ "epoch": 0.37,
1413
+ "grad_norm": 0.1363496149379173,
1414
+ "learning_rate": 0.00019970895002757413,
1415
+ "loss": 1.2031,
1416
+ "step": 201
1417
+ },
1418
+ {
1419
+ "epoch": 0.37,
1420
+ "grad_norm": 0.14218340110669314,
1421
+ "learning_rate": 0.0001996925644142717,
1422
+ "loss": 1.3073,
1423
+ "step": 202
1424
+ },
1425
+ {
1426
+ "epoch": 0.37,
1427
+ "grad_norm": 0.14234535389443218,
1428
+ "learning_rate": 0.00019967573081342103,
1429
+ "loss": 1.2444,
1430
+ "step": 203
1431
+ },
1432
+ {
1433
+ "epoch": 0.37,
1434
+ "grad_norm": 0.12866113026310516,
1435
+ "learning_rate": 0.000199658449300667,
1436
+ "loss": 1.2257,
1437
+ "step": 204
1438
+ },
1439
+ {
1440
+ "epoch": 0.37,
1441
+ "grad_norm": 0.1324053366295965,
1442
+ "learning_rate": 0.00019964071995366744,
1443
+ "loss": 1.2374,
1444
+ "step": 205
1445
+ },
1446
+ {
1447
+ "epoch": 0.38,
1448
+ "grad_norm": 0.12906841330218152,
1449
+ "learning_rate": 0.00019962254285209254,
1450
+ "loss": 1.2334,
1451
+ "step": 206
1452
+ },
1453
+ {
1454
+ "epoch": 0.38,
1455
+ "grad_norm": 0.13620873131846425,
1456
+ "learning_rate": 0.00019960391807762463,
1457
+ "loss": 1.242,
1458
+ "step": 207
1459
+ },
1460
+ {
1461
+ "epoch": 0.38,
1462
+ "grad_norm": 0.14877366842835116,
1463
+ "learning_rate": 0.00019958484571395757,
1464
+ "loss": 1.1772,
1465
+ "step": 208
1466
+ },
1467
+ {
1468
+ "epoch": 0.38,
1469
+ "grad_norm": 0.13914108740445985,
1470
+ "learning_rate": 0.00019956532584679675,
1471
+ "loss": 1.2734,
1472
+ "step": 209
1473
+ },
1474
+ {
1475
+ "epoch": 0.38,
1476
+ "grad_norm": 0.13198394930310692,
1477
+ "learning_rate": 0.00019954535856385837,
1478
+ "loss": 1.1728,
1479
+ "step": 210
1480
+ },
1481
+ {
1482
+ "epoch": 0.38,
1483
+ "grad_norm": 0.3807736597404611,
1484
+ "learning_rate": 0.0001995249439548693,
1485
+ "loss": 1.2089,
1486
+ "step": 211
1487
+ },
1488
+ {
1489
+ "epoch": 0.39,
1490
+ "grad_norm": 0.1682550557564819,
1491
+ "learning_rate": 0.00019950408211156636,
1492
+ "loss": 1.2423,
1493
+ "step": 212
1494
+ },
1495
+ {
1496
+ "epoch": 0.39,
1497
+ "grad_norm": 0.2102196862007261,
1498
+ "learning_rate": 0.0001994827731276963,
1499
+ "loss": 1.2096,
1500
+ "step": 213
1501
+ },
1502
+ {
1503
+ "epoch": 0.39,
1504
+ "grad_norm": 0.154346739470422,
1505
+ "learning_rate": 0.00019946101709901514,
1506
+ "loss": 1.2847,
1507
+ "step": 214
1508
+ },
1509
+ {
1510
+ "epoch": 0.39,
1511
+ "grad_norm": 0.16416668358293746,
1512
+ "learning_rate": 0.0001994388141232876,
1513
+ "loss": 1.2503,
1514
+ "step": 215
1515
+ },
1516
+ {
1517
+ "epoch": 0.39,
1518
+ "grad_norm": 0.13134349458231093,
1519
+ "learning_rate": 0.0001994161643002871,
1520
+ "loss": 1.1231,
1521
+ "step": 216
1522
+ },
1523
+ {
1524
+ "epoch": 0.4,
1525
+ "grad_norm": 0.15083246389185287,
1526
+ "learning_rate": 0.00019939306773179497,
1527
+ "loss": 1.1614,
1528
+ "step": 217
1529
+ },
1530
+ {
1531
+ "epoch": 0.4,
1532
+ "grad_norm": 0.1742387260929692,
1533
+ "learning_rate": 0.00019936952452159995,
1534
+ "loss": 1.3568,
1535
+ "step": 218
1536
+ },
1537
+ {
1538
+ "epoch": 0.4,
1539
+ "grad_norm": 0.18146911432436974,
1540
+ "learning_rate": 0.00019934553477549794,
1541
+ "loss": 1.2686,
1542
+ "step": 219
1543
+ },
1544
+ {
1545
+ "epoch": 0.4,
1546
+ "grad_norm": 0.1393593447949332,
1547
+ "learning_rate": 0.00019932109860129154,
1548
+ "loss": 1.1141,
1549
+ "step": 220
1550
+ },
1551
+ {
1552
+ "epoch": 0.4,
1553
+ "grad_norm": 0.14856124153987935,
1554
+ "learning_rate": 0.00019929621610878927,
1555
+ "loss": 1.234,
1556
+ "step": 221
1557
+ },
1558
+ {
1559
+ "epoch": 0.4,
1560
+ "grad_norm": 0.14820851831477327,
1561
+ "learning_rate": 0.0001992708874098054,
1562
+ "loss": 1.2069,
1563
+ "step": 222
1564
+ },
1565
+ {
1566
+ "epoch": 0.41,
1567
+ "grad_norm": 0.17893142790958147,
1568
+ "learning_rate": 0.00019924511261815926,
1569
+ "loss": 1.1278,
1570
+ "step": 223
1571
+ },
1572
+ {
1573
+ "epoch": 0.41,
1574
+ "grad_norm": 0.14573658703265605,
1575
+ "learning_rate": 0.00019921889184967476,
1576
+ "loss": 1.2292,
1577
+ "step": 224
1578
+ },
1579
+ {
1580
+ "epoch": 0.41,
1581
+ "grad_norm": 0.15282321197574994,
1582
+ "learning_rate": 0.00019919222522217996,
1583
+ "loss": 1.2482,
1584
+ "step": 225
1585
+ },
1586
+ {
1587
+ "epoch": 0.41,
1588
+ "grad_norm": 0.16342112084119492,
1589
+ "learning_rate": 0.00019916511285550642,
1590
+ "loss": 1.2172,
1591
+ "step": 226
1592
+ },
1593
+ {
1594
+ "epoch": 0.41,
1595
+ "grad_norm": 0.1475889153814455,
1596
+ "learning_rate": 0.00019913755487148876,
1597
+ "loss": 1.1747,
1598
+ "step": 227
1599
+ },
1600
+ {
1601
+ "epoch": 0.42,
1602
+ "grad_norm": 0.163738064491857,
1603
+ "learning_rate": 0.00019910955139396396,
1604
+ "loss": 1.3007,
1605
+ "step": 228
1606
+ },
1607
+ {
1608
+ "epoch": 0.42,
1609
+ "grad_norm": 0.14427856196022704,
1610
+ "learning_rate": 0.00019908110254877106,
1611
+ "loss": 1.2464,
1612
+ "step": 229
1613
+ },
1614
+ {
1615
+ "epoch": 0.42,
1616
+ "grad_norm": 0.20204742660246344,
1617
+ "learning_rate": 0.00019905220846375032,
1618
+ "loss": 1.2515,
1619
+ "step": 230
1620
+ },
1621
+ {
1622
+ "epoch": 0.42,
1623
+ "grad_norm": 0.15134144918251685,
1624
+ "learning_rate": 0.0001990228692687429,
1625
+ "loss": 1.1786,
1626
+ "step": 231
1627
+ },
1628
+ {
1629
+ "epoch": 0.42,
1630
+ "grad_norm": 0.1636590177812163,
1631
+ "learning_rate": 0.00019899308509558998,
1632
+ "loss": 1.1974,
1633
+ "step": 232
1634
+ },
1635
+ {
1636
+ "epoch": 0.42,
1637
+ "grad_norm": 0.15552319776955892,
1638
+ "learning_rate": 0.00019896285607813244,
1639
+ "loss": 1.2308,
1640
+ "step": 233
1641
+ },
1642
+ {
1643
+ "epoch": 0.43,
1644
+ "grad_norm": 0.17104898009833774,
1645
+ "learning_rate": 0.00019893218235221015,
1646
+ "loss": 1.2828,
1647
+ "step": 234
1648
+ },
1649
+ {
1650
+ "epoch": 0.43,
1651
+ "grad_norm": 0.16387378763964267,
1652
+ "learning_rate": 0.00019890106405566138,
1653
+ "loss": 1.2779,
1654
+ "step": 235
1655
+ },
1656
+ {
1657
+ "epoch": 0.43,
1658
+ "grad_norm": 0.14622126798612248,
1659
+ "learning_rate": 0.00019886950132832207,
1660
+ "loss": 1.2894,
1661
+ "step": 236
1662
+ },
1663
+ {
1664
+ "epoch": 0.43,
1665
+ "grad_norm": 0.16619841547518147,
1666
+ "learning_rate": 0.0001988374943120254,
1667
+ "loss": 1.2133,
1668
+ "step": 237
1669
+ },
1670
+ {
1671
+ "epoch": 0.43,
1672
+ "grad_norm": 0.12664832399697545,
1673
+ "learning_rate": 0.00019880504315060096,
1674
+ "loss": 1.1807,
1675
+ "step": 238
1676
+ },
1677
+ {
1678
+ "epoch": 0.44,
1679
+ "grad_norm": 0.2015108381613456,
1680
+ "learning_rate": 0.00019877214798987426,
1681
+ "loss": 1.1876,
1682
+ "step": 239
1683
+ },
1684
+ {
1685
+ "epoch": 0.44,
1686
+ "grad_norm": 0.14468620723711506,
1687
+ "learning_rate": 0.00019873880897766598,
1688
+ "loss": 1.1883,
1689
+ "step": 240
1690
+ },
1691
+ {
1692
+ "epoch": 0.44,
1693
+ "grad_norm": 0.1549018650770757,
1694
+ "learning_rate": 0.00019870502626379127,
1695
+ "loss": 1.2896,
1696
+ "step": 241
1697
+ },
1698
+ {
1699
+ "epoch": 0.44,
1700
+ "grad_norm": 0.1492917963684983,
1701
+ "learning_rate": 0.0001986708000000593,
1702
+ "loss": 1.2102,
1703
+ "step": 242
1704
+ },
1705
+ {
1706
+ "epoch": 0.44,
1707
+ "grad_norm": 0.178606606459489,
1708
+ "learning_rate": 0.00019863613034027224,
1709
+ "loss": 1.2292,
1710
+ "step": 243
1711
+ },
1712
+ {
1713
+ "epoch": 0.44,
1714
+ "grad_norm": 0.206170239681528,
1715
+ "learning_rate": 0.00019860101744022485,
1716
+ "loss": 1.2666,
1717
+ "step": 244
1718
+ },
1719
+ {
1720
+ "epoch": 0.45,
1721
+ "grad_norm": 0.13741043007948167,
1722
+ "learning_rate": 0.0001985654614577036,
1723
+ "loss": 1.2022,
1724
+ "step": 245
1725
+ },
1726
+ {
1727
+ "epoch": 0.45,
1728
+ "grad_norm": 0.1595080658199459,
1729
+ "learning_rate": 0.0001985294625524861,
1730
+ "loss": 1.1203,
1731
+ "step": 246
1732
+ },
1733
+ {
1734
+ "epoch": 0.45,
1735
+ "grad_norm": 0.13929705183853777,
1736
+ "learning_rate": 0.00019849302088634034,
1737
+ "loss": 1.1505,
1738
+ "step": 247
1739
+ },
1740
+ {
1741
+ "epoch": 0.45,
1742
+ "grad_norm": 0.14045247607912964,
1743
+ "learning_rate": 0.00019845613662302383,
1744
+ "loss": 1.1897,
1745
+ "step": 248
1746
+ },
1747
+ {
1748
+ "epoch": 0.45,
1749
+ "grad_norm": 0.15002651347444407,
1750
+ "learning_rate": 0.00019841880992828306,
1751
+ "loss": 1.2133,
1752
+ "step": 249
1753
+ },
1754
+ {
1755
+ "epoch": 0.46,
1756
+ "grad_norm": 0.1567929487810952,
1757
+ "learning_rate": 0.00019838104096985267,
1758
+ "loss": 1.129,
1759
+ "step": 250
1760
+ },
1761
+ {
1762
+ "epoch": 0.46,
1763
+ "grad_norm": 0.15240634543877116,
1764
+ "learning_rate": 0.00019834282991745464,
1765
+ "loss": 1.1995,
1766
+ "step": 251
1767
+ },
1768
+ {
1769
+ "epoch": 0.46,
1770
+ "grad_norm": 0.151807679821367,
1771
+ "learning_rate": 0.00019830417694279766,
1772
+ "loss": 1.25,
1773
+ "step": 252
1774
+ },
1775
+ {
1776
+ "epoch": 0.46,
1777
+ "grad_norm": 0.1648599156208311,
1778
+ "learning_rate": 0.0001982650822195762,
1779
+ "loss": 1.2511,
1780
+ "step": 253
1781
+ },
1782
+ {
1783
+ "epoch": 0.46,
1784
+ "grad_norm": 0.15363401233808713,
1785
+ "learning_rate": 0.00019822554592346993,
1786
+ "loss": 1.1794,
1787
+ "step": 254
1788
+ },
1789
+ {
1790
+ "epoch": 0.46,
1791
+ "grad_norm": 0.1569644350778875,
1792
+ "learning_rate": 0.00019818556823214268,
1793
+ "loss": 1.2033,
1794
+ "step": 255
1795
+ },
1796
+ {
1797
+ "epoch": 0.47,
1798
+ "grad_norm": 0.15996552747294254,
1799
+ "learning_rate": 0.0001981451493252418,
1800
+ "loss": 1.2809,
1801
+ "step": 256
1802
+ },
1803
+ {
1804
+ "epoch": 0.47,
1805
+ "grad_norm": 0.15863104885072635,
1806
+ "learning_rate": 0.0001981042893843974,
1807
+ "loss": 1.1667,
1808
+ "step": 257
1809
+ },
1810
+ {
1811
+ "epoch": 0.47,
1812
+ "grad_norm": 0.2887466971861171,
1813
+ "learning_rate": 0.0001980629885932214,
1814
+ "loss": 1.1915,
1815
+ "step": 258
1816
+ },
1817
+ {
1818
+ "epoch": 0.47,
1819
+ "grad_norm": 0.15233015979193984,
1820
+ "learning_rate": 0.00019802124713730681,
1821
+ "loss": 1.1734,
1822
+ "step": 259
1823
+ },
1824
+ {
1825
+ "epoch": 0.47,
1826
+ "grad_norm": 0.18207884538436447,
1827
+ "learning_rate": 0.00019797906520422677,
1828
+ "loss": 1.2575,
1829
+ "step": 260
1830
+ },
1831
+ {
1832
+ "epoch": 0.48,
1833
+ "grad_norm": 0.17323546756038308,
1834
+ "learning_rate": 0.0001979364429835339,
1835
+ "loss": 1.1704,
1836
+ "step": 261
1837
+ },
1838
+ {
1839
+ "epoch": 0.48,
1840
+ "grad_norm": 0.14592153602263633,
1841
+ "learning_rate": 0.00019789338066675922,
1842
+ "loss": 1.192,
1843
+ "step": 262
1844
+ },
1845
+ {
1846
+ "epoch": 0.48,
1847
+ "grad_norm": 0.19250697792287097,
1848
+ "learning_rate": 0.0001978498784474115,
1849
+ "loss": 1.2779,
1850
+ "step": 263
1851
+ },
1852
+ {
1853
+ "epoch": 0.48,
1854
+ "grad_norm": 0.1429107680887097,
1855
+ "learning_rate": 0.0001978059365209762,
1856
+ "loss": 1.2529,
1857
+ "step": 264
1858
+ },
1859
+ {
1860
+ "epoch": 0.48,
1861
+ "grad_norm": 0.48514081074992116,
1862
+ "learning_rate": 0.00019776155508491482,
1863
+ "loss": 1.1917,
1864
+ "step": 265
1865
+ },
1866
+ {
1867
+ "epoch": 0.48,
1868
+ "grad_norm": 0.1534376167748161,
1869
+ "learning_rate": 0.0001977167343386638,
1870
+ "loss": 1.2384,
1871
+ "step": 266
1872
+ },
1873
+ {
1874
+ "epoch": 0.49,
1875
+ "grad_norm": 0.16744875760032166,
1876
+ "learning_rate": 0.00019767147448363366,
1877
+ "loss": 1.1744,
1878
+ "step": 267
1879
+ },
1880
+ {
1881
+ "epoch": 0.49,
1882
+ "grad_norm": 0.29195538170738244,
1883
+ "learning_rate": 0.00019762577572320824,
1884
+ "loss": 1.1418,
1885
+ "step": 268
1886
+ },
1887
+ {
1888
+ "epoch": 0.49,
1889
+ "grad_norm": 0.1820804717651353,
1890
+ "learning_rate": 0.00019757963826274357,
1891
+ "loss": 1.2815,
1892
+ "step": 269
1893
+ },
1894
+ {
1895
+ "epoch": 0.49,
1896
+ "grad_norm": 0.17522345110441973,
1897
+ "learning_rate": 0.00019753306230956718,
1898
+ "loss": 1.2363,
1899
+ "step": 270
1900
+ },
1901
+ {
1902
+ "epoch": 0.49,
1903
+ "grad_norm": 0.16354388270886613,
1904
+ "learning_rate": 0.000197486048072977,
1905
+ "loss": 1.2845,
1906
+ "step": 271
1907
+ },
1908
+ {
1909
+ "epoch": 0.5,
1910
+ "grad_norm": 0.17590082756401024,
1911
+ "learning_rate": 0.0001974385957642404,
1912
+ "loss": 1.192,
1913
+ "step": 272
1914
+ },
1915
+ {
1916
+ "epoch": 0.5,
1917
+ "grad_norm": 0.17345720403188775,
1918
+ "learning_rate": 0.00019739070559659347,
1919
+ "loss": 1.2068,
1920
+ "step": 273
1921
+ },
1922
+ {
1923
+ "epoch": 0.5,
1924
+ "grad_norm": 0.16070434867766506,
1925
+ "learning_rate": 0.00019734237778523976,
1926
+ "loss": 1.189,
1927
+ "step": 274
1928
+ },
1929
+ {
1930
+ "epoch": 0.5,
1931
+ "grad_norm": 0.18983443066710415,
1932
+ "learning_rate": 0.0001972936125473495,
1933
+ "loss": 1.2223,
1934
+ "step": 275
1935
+ },
1936
+ {
1937
+ "epoch": 0.5,
1938
+ "grad_norm": 0.15724400187981355,
1939
+ "learning_rate": 0.00019724441010205863,
1940
+ "loss": 1.2292,
1941
+ "step": 276
1942
+ },
1943
+ {
1944
+ "epoch": 0.5,
1945
+ "grad_norm": 0.14570729442956004,
1946
+ "learning_rate": 0.00019719477067046766,
1947
+ "loss": 1.1421,
1948
+ "step": 277
1949
+ },
1950
+ {
1951
+ "epoch": 0.51,
1952
+ "grad_norm": 0.1559242881177266,
1953
+ "learning_rate": 0.00019714469447564088,
1954
+ "loss": 1.2598,
1955
+ "step": 278
1956
+ },
1957
+ {
1958
+ "epoch": 0.51,
1959
+ "grad_norm": 0.16621830243096108,
1960
+ "learning_rate": 0.0001970941817426052,
1961
+ "loss": 1.3038,
1962
+ "step": 279
1963
+ },
1964
+ {
1965
+ "epoch": 0.51,
1966
+ "grad_norm": 4.675483994100576,
1967
+ "learning_rate": 0.00019704323269834927,
1968
+ "loss": 1.2298,
1969
+ "step": 280
1970
+ },
1971
+ {
1972
+ "epoch": 0.51,
1973
+ "grad_norm": 0.2769699381619058,
1974
+ "learning_rate": 0.00019699184757182225,
1975
+ "loss": 1.2566,
1976
+ "step": 281
1977
+ },
1978
+ {
1979
+ "epoch": 0.51,
1980
+ "grad_norm": 0.20189839889100783,
1981
+ "learning_rate": 0.00019694002659393305,
1982
+ "loss": 1.3181,
1983
+ "step": 282
1984
+ },
1985
+ {
1986
+ "epoch": 0.52,
1987
+ "grad_norm": 0.19497107359413876,
1988
+ "learning_rate": 0.00019688776999754912,
1989
+ "loss": 1.1502,
1990
+ "step": 283
1991
+ },
1992
+ {
1993
+ "epoch": 0.52,
1994
+ "grad_norm": 0.1982266815755412,
1995
+ "learning_rate": 0.00019683507801749545,
1996
+ "loss": 1.2053,
1997
+ "step": 284
1998
+ },
1999
+ {
2000
+ "epoch": 0.52,
2001
+ "grad_norm": 0.1924340950322314,
2002
+ "learning_rate": 0.00019678195089055346,
2003
+ "loss": 1.2149,
2004
+ "step": 285
2005
+ },
2006
+ {
2007
+ "epoch": 0.52,
2008
+ "grad_norm": 0.1725322346446431,
2009
+ "learning_rate": 0.00019672838885546008,
2010
+ "loss": 1.2553,
2011
+ "step": 286
2012
+ },
2013
+ {
2014
+ "epoch": 0.52,
2015
+ "grad_norm": 0.2535488743520272,
2016
+ "learning_rate": 0.00019667439215290648,
2017
+ "loss": 1.2576,
2018
+ "step": 287
2019
+ },
2020
+ {
2021
+ "epoch": 0.52,
2022
+ "grad_norm": 0.37837586860064026,
2023
+ "learning_rate": 0.00019661996102553718,
2024
+ "loss": 1.1815,
2025
+ "step": 288
2026
+ },
2027
+ {
2028
+ "epoch": 0.53,
2029
+ "grad_norm": 0.17520419597901843,
2030
+ "learning_rate": 0.00019656509571794878,
2031
+ "loss": 1.1932,
2032
+ "step": 289
2033
+ },
2034
+ {
2035
+ "epoch": 0.53,
2036
+ "grad_norm": 0.17056234784450633,
2037
+ "learning_rate": 0.00019650979647668906,
2038
+ "loss": 1.163,
2039
+ "step": 290
2040
+ },
2041
+ {
2042
+ "epoch": 0.53,
2043
+ "grad_norm": 0.18272246580207432,
2044
+ "learning_rate": 0.00019645406355025565,
2045
+ "loss": 1.1887,
2046
+ "step": 291
2047
+ },
2048
+ {
2049
+ "epoch": 0.53,
2050
+ "grad_norm": 0.17889037954429915,
2051
+ "learning_rate": 0.00019639789718909508,
2052
+ "loss": 1.2126,
2053
+ "step": 292
2054
+ },
2055
+ {
2056
+ "epoch": 0.53,
2057
+ "grad_norm": 0.23993734971101424,
2058
+ "learning_rate": 0.00019634129764560168,
2059
+ "loss": 1.2485,
2060
+ "step": 293
2061
+ },
2062
+ {
2063
+ "epoch": 0.54,
2064
+ "grad_norm": 0.1847578318208199,
2065
+ "learning_rate": 0.00019628426517411625,
2066
+ "loss": 1.2549,
2067
+ "step": 294
2068
+ },
2069
+ {
2070
+ "epoch": 0.54,
2071
+ "grad_norm": 0.23185098827091005,
2072
+ "learning_rate": 0.00019622680003092503,
2073
+ "loss": 1.1599,
2074
+ "step": 295
2075
+ },
2076
+ {
2077
+ "epoch": 0.54,
2078
+ "grad_norm": 0.220638044092583,
2079
+ "learning_rate": 0.00019616890247425866,
2080
+ "loss": 1.2281,
2081
+ "step": 296
2082
+ },
2083
+ {
2084
+ "epoch": 0.54,
2085
+ "grad_norm": 0.2303439219825616,
2086
+ "learning_rate": 0.00019611057276429085,
2087
+ "loss": 1.2208,
2088
+ "step": 297
2089
+ },
2090
+ {
2091
+ "epoch": 0.54,
2092
+ "grad_norm": 0.1744807302230573,
2093
+ "learning_rate": 0.00019605181116313724,
2094
+ "loss": 1.2303,
2095
+ "step": 298
2096
+ },
2097
+ {
2098
+ "epoch": 0.54,
2099
+ "grad_norm": 0.17510946821872422,
2100
+ "learning_rate": 0.0001959926179348543,
2101
+ "loss": 1.2385,
2102
+ "step": 299
2103
+ },
2104
+ {
2105
+ "epoch": 0.55,
2106
+ "grad_norm": 0.2218474349751746,
2107
+ "learning_rate": 0.00019593299334543808,
2108
+ "loss": 1.2153,
2109
+ "step": 300
2110
+ },
2111
+ {
2112
+ "epoch": 0.55,
2113
+ "grad_norm": 0.1742070481516402,
2114
+ "learning_rate": 0.00019587293766282308,
2115
+ "loss": 1.1628,
2116
+ "step": 301
2117
+ },
2118
+ {
2119
+ "epoch": 0.55,
2120
+ "grad_norm": 0.15250311715180823,
2121
+ "learning_rate": 0.00019581245115688094,
2122
+ "loss": 1.1632,
2123
+ "step": 302
2124
+ },
2125
+ {
2126
+ "epoch": 0.55,
2127
+ "grad_norm": 0.1744397677094501,
2128
+ "learning_rate": 0.0001957515340994193,
2129
+ "loss": 1.254,
2130
+ "step": 303
2131
+ },
2132
+ {
2133
+ "epoch": 0.55,
2134
+ "grad_norm": 0.1686772182789891,
2135
+ "learning_rate": 0.00019569018676418053,
2136
+ "loss": 1.2169,
2137
+ "step": 304
2138
+ },
2139
+ {
2140
+ "epoch": 0.56,
2141
+ "grad_norm": 0.16404966161017623,
2142
+ "learning_rate": 0.00019562840942684067,
2143
+ "loss": 1.2221,
2144
+ "step": 305
2145
+ },
2146
+ {
2147
+ "epoch": 0.56,
2148
+ "grad_norm": 0.16052011449463713,
2149
+ "learning_rate": 0.00019556620236500793,
2150
+ "loss": 1.2045,
2151
+ "step": 306
2152
+ },
2153
+ {
2154
+ "epoch": 0.56,
2155
+ "grad_norm": 0.16343251390831215,
2156
+ "learning_rate": 0.0001955035658582216,
2157
+ "loss": 1.2289,
2158
+ "step": 307
2159
+ },
2160
+ {
2161
+ "epoch": 0.56,
2162
+ "grad_norm": 0.14387162360389305,
2163
+ "learning_rate": 0.00019544050018795075,
2164
+ "loss": 1.1365,
2165
+ "step": 308
2166
+ },
2167
+ {
2168
+ "epoch": 0.56,
2169
+ "grad_norm": 0.15304461439740238,
2170
+ "learning_rate": 0.00019537700563759304,
2171
+ "loss": 1.1931,
2172
+ "step": 309
2173
+ },
2174
+ {
2175
+ "epoch": 0.56,
2176
+ "grad_norm": 0.17059958050065627,
2177
+ "learning_rate": 0.00019531308249247327,
2178
+ "loss": 1.2166,
2179
+ "step": 310
2180
+ },
2181
+ {
2182
+ "epoch": 0.57,
2183
+ "grad_norm": 0.17633385530926995,
2184
+ "learning_rate": 0.00019524873103984235,
2185
+ "loss": 1.2604,
2186
+ "step": 311
2187
+ },
2188
+ {
2189
+ "epoch": 0.57,
2190
+ "grad_norm": 0.17855814403303746,
2191
+ "learning_rate": 0.00019518395156887576,
2192
+ "loss": 1.1615,
2193
+ "step": 312
2194
+ },
2195
+ {
2196
+ "epoch": 0.57,
2197
+ "grad_norm": 0.19823982444256988,
2198
+ "learning_rate": 0.00019511874437067243,
2199
+ "loss": 1.2153,
2200
+ "step": 313
2201
+ },
2202
+ {
2203
+ "epoch": 0.57,
2204
+ "grad_norm": 0.1570784627362585,
2205
+ "learning_rate": 0.0001950531097382533,
2206
+ "loss": 1.2788,
2207
+ "step": 314
2208
+ },
2209
+ {
2210
+ "epoch": 0.57,
2211
+ "grad_norm": 0.2183125402112695,
2212
+ "learning_rate": 0.00019498704796656018,
2213
+ "loss": 1.2966,
2214
+ "step": 315
2215
+ },
2216
+ {
2217
+ "epoch": 0.58,
2218
+ "grad_norm": 0.18173933276147194,
2219
+ "learning_rate": 0.00019492055935245418,
2220
+ "loss": 1.2978,
2221
+ "step": 316
2222
+ },
2223
+ {
2224
+ "epoch": 0.58,
2225
+ "grad_norm": 0.17483116680914407,
2226
+ "learning_rate": 0.00019485364419471454,
2227
+ "loss": 1.258,
2228
+ "step": 317
2229
+ },
2230
+ {
2231
+ "epoch": 0.58,
2232
+ "grad_norm": 0.15490767356815494,
2233
+ "learning_rate": 0.0001947863027940374,
2234
+ "loss": 1.2088,
2235
+ "step": 318
2236
+ },
2237
+ {
2238
+ "epoch": 0.58,
2239
+ "grad_norm": 0.14703966491934156,
2240
+ "learning_rate": 0.00019471853545303405,
2241
+ "loss": 1.2355,
2242
+ "step": 319
2243
+ },
2244
+ {
2245
+ "epoch": 0.58,
2246
+ "grad_norm": 0.14386689086661608,
2247
+ "learning_rate": 0.00019465034247623003,
2248
+ "loss": 1.2583,
2249
+ "step": 320
2250
+ },
2251
+ {
2252
+ "epoch": 0.58,
2253
+ "grad_norm": 0.18818904376313625,
2254
+ "learning_rate": 0.00019458172417006347,
2255
+ "loss": 1.2181,
2256
+ "step": 321
2257
+ },
2258
+ {
2259
+ "epoch": 0.59,
2260
+ "grad_norm": 0.17393313719202513,
2261
+ "learning_rate": 0.00019451268084288385,
2262
+ "loss": 1.3453,
2263
+ "step": 322
2264
+ },
2265
+ {
2266
+ "epoch": 0.59,
2267
+ "grad_norm": 0.14706823379985753,
2268
+ "learning_rate": 0.00019444321280495043,
2269
+ "loss": 1.2234,
2270
+ "step": 323
2271
+ },
2272
+ {
2273
+ "epoch": 0.59,
2274
+ "grad_norm": 0.15282014755252687,
2275
+ "learning_rate": 0.00019437332036843118,
2276
+ "loss": 1.1262,
2277
+ "step": 324
2278
+ },
2279
+ {
2280
+ "epoch": 0.59,
2281
+ "grad_norm": 0.1618727884326225,
2282
+ "learning_rate": 0.00019430300384740105,
2283
+ "loss": 1.3136,
2284
+ "step": 325
2285
+ },
2286
+ {
2287
+ "epoch": 0.59,
2288
+ "grad_norm": 0.16090758705378874,
2289
+ "learning_rate": 0.00019423226355784077,
2290
+ "loss": 1.2055,
2291
+ "step": 326
2292
+ },
2293
+ {
2294
+ "epoch": 0.6,
2295
+ "grad_norm": 0.15241156801091013,
2296
+ "learning_rate": 0.00019416109981763526,
2297
+ "loss": 1.2678,
2298
+ "step": 327
2299
+ },
2300
+ {
2301
+ "epoch": 0.6,
2302
+ "grad_norm": 0.14216697909809062,
2303
+ "learning_rate": 0.0001940895129465724,
2304
+ "loss": 1.2841,
2305
+ "step": 328
2306
+ },
2307
+ {
2308
+ "epoch": 0.6,
2309
+ "grad_norm": 0.15790232415414485,
2310
+ "learning_rate": 0.00019401750326634144,
2311
+ "loss": 1.3119,
2312
+ "step": 329
2313
+ },
2314
+ {
2315
+ "epoch": 0.6,
2316
+ "grad_norm": 0.13322691961062616,
2317
+ "learning_rate": 0.0001939450711005316,
2318
+ "loss": 1.1293,
2319
+ "step": 330
2320
+ },
2321
+ {
2322
+ "epoch": 0.6,
2323
+ "grad_norm": 0.14075018938835404,
2324
+ "learning_rate": 0.00019387221677463062,
2325
+ "loss": 1.2176,
2326
+ "step": 331
2327
+ },
2328
+ {
2329
+ "epoch": 0.6,
2330
+ "grad_norm": 0.21565975459393052,
2331
+ "learning_rate": 0.00019379894061602335,
2332
+ "loss": 1.1723,
2333
+ "step": 332
2334
+ },
2335
+ {
2336
+ "epoch": 0.61,
2337
+ "grad_norm": 0.17967631394222838,
2338
+ "learning_rate": 0.00019372524295399013,
2339
+ "loss": 1.239,
2340
+ "step": 333
2341
+ },
2342
+ {
2343
+ "epoch": 0.61,
2344
+ "grad_norm": 0.21187969201978435,
2345
+ "learning_rate": 0.0001936511241197055,
2346
+ "loss": 1.2207,
2347
+ "step": 334
2348
+ },
2349
+ {
2350
+ "epoch": 0.61,
2351
+ "grad_norm": 0.16967789022974608,
2352
+ "learning_rate": 0.00019357658444623654,
2353
+ "loss": 1.2478,
2354
+ "step": 335
2355
+ },
2356
+ {
2357
+ "epoch": 0.61,
2358
+ "grad_norm": 0.14810621660374448,
2359
+ "learning_rate": 0.0001935016242685415,
2360
+ "loss": 1.1223,
2361
+ "step": 336
2362
+ },
2363
+ {
2364
+ "epoch": 0.61,
2365
+ "grad_norm": 0.1489106421847434,
2366
+ "learning_rate": 0.00019342624392346824,
2367
+ "loss": 1.1592,
2368
+ "step": 337
2369
+ },
2370
+ {
2371
+ "epoch": 0.62,
2372
+ "grad_norm": 0.17625176068748855,
2373
+ "learning_rate": 0.0001933504437497527,
2374
+ "loss": 1.2145,
2375
+ "step": 338
2376
+ },
2377
+ {
2378
+ "epoch": 0.62,
2379
+ "grad_norm": 0.17250255512763446,
2380
+ "learning_rate": 0.00019327422408801744,
2381
+ "loss": 1.2504,
2382
+ "step": 339
2383
+ },
2384
+ {
2385
+ "epoch": 0.62,
2386
+ "grad_norm": 0.16079375745566896,
2387
+ "learning_rate": 0.00019319758528077,
2388
+ "loss": 1.1795,
2389
+ "step": 340
2390
+ },
2391
+ {
2392
+ "epoch": 0.62,
2393
+ "grad_norm": 0.15454466809245995,
2394
+ "learning_rate": 0.0001931205276724015,
2395
+ "loss": 1.2123,
2396
+ "step": 341
2397
+ },
2398
+ {
2399
+ "epoch": 0.62,
2400
+ "grad_norm": 0.7021323604447972,
2401
+ "learning_rate": 0.000193043051609185,
2402
+ "loss": 1.2239,
2403
+ "step": 342
2404
+ },
2405
+ {
2406
+ "epoch": 0.62,
2407
+ "grad_norm": 0.1572764339385847,
2408
+ "learning_rate": 0.00019296515743927399,
2409
+ "loss": 1.2516,
2410
+ "step": 343
2411
+ },
2412
+ {
2413
+ "epoch": 0.63,
2414
+ "grad_norm": 0.2136637778252246,
2415
+ "learning_rate": 0.00019288684551270073,
2416
+ "loss": 1.2321,
2417
+ "step": 344
2418
+ },
2419
+ {
2420
+ "epoch": 0.63,
2421
+ "grad_norm": 0.4546540454773654,
2422
+ "learning_rate": 0.00019280811618137484,
2423
+ "loss": 1.18,
2424
+ "step": 345
2425
+ },
2426
+ {
2427
+ "epoch": 0.63,
2428
+ "grad_norm": 0.9809832576786297,
2429
+ "learning_rate": 0.00019272896979908154,
2430
+ "loss": 1.2081,
2431
+ "step": 346
2432
+ },
2433
+ {
2434
+ "epoch": 0.63,
2435
+ "grad_norm": 0.5246256133291822,
2436
+ "learning_rate": 0.00019264940672148018,
2437
+ "loss": 1.2722,
2438
+ "step": 347
2439
+ },
2440
+ {
2441
+ "epoch": 0.63,
2442
+ "grad_norm": 0.24941717134878091,
2443
+ "learning_rate": 0.00019256942730610268,
2444
+ "loss": 1.2352,
2445
+ "step": 348
2446
+ },
2447
+ {
2448
+ "epoch": 0.64,
2449
+ "grad_norm": 0.3356068462072784,
2450
+ "learning_rate": 0.00019248903191235176,
2451
+ "loss": 1.2225,
2452
+ "step": 349
2453
+ },
2454
+ {
2455
+ "epoch": 0.64,
2456
+ "grad_norm": 0.19535845221880543,
2457
+ "learning_rate": 0.00019240822090149944,
2458
+ "loss": 1.1669,
2459
+ "step": 350
2460
+ },
2461
+ {
2462
+ "epoch": 0.64,
2463
+ "grad_norm": 0.22306941566416597,
2464
+ "learning_rate": 0.00019232699463668542,
2465
+ "loss": 1.2281,
2466
+ "step": 351
2467
+ },
2468
+ {
2469
+ "epoch": 0.64,
2470
+ "grad_norm": 0.2700134013989352,
2471
+ "learning_rate": 0.00019224535348291542,
2472
+ "loss": 1.1939,
2473
+ "step": 352
2474
+ },
2475
+ {
2476
+ "epoch": 0.64,
2477
+ "grad_norm": 0.24406908935562743,
2478
+ "learning_rate": 0.00019216329780705953,
2479
+ "loss": 1.1839,
2480
+ "step": 353
2481
+ },
2482
+ {
2483
+ "epoch": 0.64,
2484
+ "grad_norm": 0.20465183000217488,
2485
+ "learning_rate": 0.00019208082797785055,
2486
+ "loss": 1.2277,
2487
+ "step": 354
2488
+ },
2489
+ {
2490
+ "epoch": 0.65,
2491
+ "grad_norm": 0.21324820828129784,
2492
+ "learning_rate": 0.00019199794436588243,
2493
+ "loss": 1.2072,
2494
+ "step": 355
2495
+ },
2496
+ {
2497
+ "epoch": 0.65,
2498
+ "grad_norm": 0.1780562512431263,
2499
+ "learning_rate": 0.00019191464734360844,
2500
+ "loss": 1.2082,
2501
+ "step": 356
2502
+ },
2503
+ {
2504
+ "epoch": 0.65,
2505
+ "grad_norm": 0.16547971467615655,
2506
+ "learning_rate": 0.00019183093728533966,
2507
+ "loss": 1.1978,
2508
+ "step": 357
2509
+ },
2510
+ {
2511
+ "epoch": 0.65,
2512
+ "grad_norm": 0.22904664933247196,
2513
+ "learning_rate": 0.00019174681456724318,
2514
+ "loss": 1.1562,
2515
+ "step": 358
2516
+ },
2517
+ {
2518
+ "epoch": 0.65,
2519
+ "grad_norm": 0.1737397860007602,
2520
+ "learning_rate": 0.00019166227956734052,
2521
+ "loss": 1.2383,
2522
+ "step": 359
2523
+ },
2524
+ {
2525
+ "epoch": 0.66,
2526
+ "grad_norm": 0.1589465455917568,
2527
+ "learning_rate": 0.00019157733266550575,
2528
+ "loss": 1.2158,
2529
+ "step": 360
2530
+ },
2531
+ {
2532
+ "epoch": 0.66,
2533
+ "grad_norm": 0.16253126221999709,
2534
+ "learning_rate": 0.00019149197424346405,
2535
+ "loss": 1.1952,
2536
+ "step": 361
2537
+ },
2538
+ {
2539
+ "epoch": 0.66,
2540
+ "grad_norm": 0.22436676243032663,
2541
+ "learning_rate": 0.00019140620468478968,
2542
+ "loss": 1.2315,
2543
+ "step": 362
2544
+ },
2545
+ {
2546
+ "epoch": 0.66,
2547
+ "grad_norm": 0.19291682612950423,
2548
+ "learning_rate": 0.00019132002437490458,
2549
+ "loss": 1.2283,
2550
+ "step": 363
2551
+ },
2552
+ {
2553
+ "epoch": 0.66,
2554
+ "grad_norm": 0.1519191258459668,
2555
+ "learning_rate": 0.00019123343370107637,
2556
+ "loss": 1.1151,
2557
+ "step": 364
2558
+ },
2559
+ {
2560
+ "epoch": 0.66,
2561
+ "grad_norm": 0.17179909633547025,
2562
+ "learning_rate": 0.00019114643305241676,
2563
+ "loss": 1.1576,
2564
+ "step": 365
2565
+ },
2566
+ {
2567
+ "epoch": 0.67,
2568
+ "grad_norm": 0.17992599023321432,
2569
+ "learning_rate": 0.00019105902281987976,
2570
+ "loss": 1.2592,
2571
+ "step": 366
2572
+ },
2573
+ {
2574
+ "epoch": 0.67,
2575
+ "grad_norm": 0.17714099390314453,
2576
+ "learning_rate": 0.00019097120339625994,
2577
+ "loss": 1.2578,
2578
+ "step": 367
2579
+ },
2580
+ {
2581
+ "epoch": 0.67,
2582
+ "grad_norm": 0.2455577642687935,
2583
+ "learning_rate": 0.00019088297517619055,
2584
+ "loss": 1.2361,
2585
+ "step": 368
2586
+ },
2587
+ {
2588
+ "epoch": 0.67,
2589
+ "grad_norm": 0.18398518628783986,
2590
+ "learning_rate": 0.00019079433855614201,
2591
+ "loss": 1.1906,
2592
+ "step": 369
2593
+ },
2594
+ {
2595
+ "epoch": 0.67,
2596
+ "grad_norm": 0.18944067022821645,
2597
+ "learning_rate": 0.00019070529393441985,
2598
+ "loss": 1.237,
2599
+ "step": 370
2600
+ },
2601
+ {
2602
+ "epoch": 0.68,
2603
+ "grad_norm": 0.17639967519781063,
2604
+ "learning_rate": 0.00019061584171116303,
2605
+ "loss": 1.1841,
2606
+ "step": 371
2607
+ },
2608
+ {
2609
+ "epoch": 0.68,
2610
+ "grad_norm": 0.15947129998283005,
2611
+ "learning_rate": 0.00019052598228834217,
2612
+ "loss": 1.1722,
2613
+ "step": 372
2614
+ },
2615
+ {
2616
+ "epoch": 0.68,
2617
+ "grad_norm": 0.1693354353719105,
2618
+ "learning_rate": 0.00019043571606975777,
2619
+ "loss": 1.2204,
2620
+ "step": 373
2621
+ },
2622
+ {
2623
+ "epoch": 0.68,
2624
+ "grad_norm": 0.16236190451963983,
2625
+ "learning_rate": 0.00019034504346103823,
2626
+ "loss": 1.1778,
2627
+ "step": 374
2628
+ },
2629
+ {
2630
+ "epoch": 0.68,
2631
+ "grad_norm": 0.17702370729269964,
2632
+ "learning_rate": 0.00019025396486963827,
2633
+ "loss": 1.2065,
2634
+ "step": 375
2635
+ },
2636
+ {
2637
+ "epoch": 0.68,
2638
+ "grad_norm": 0.19388150596154238,
2639
+ "learning_rate": 0.00019016248070483687,
2640
+ "loss": 1.2942,
2641
+ "step": 376
2642
+ },
2643
+ {
2644
+ "epoch": 0.69,
2645
+ "grad_norm": 0.16152000400319103,
2646
+ "learning_rate": 0.0001900705913777356,
2647
+ "loss": 1.1784,
2648
+ "step": 377
2649
+ },
2650
+ {
2651
+ "epoch": 0.69,
2652
+ "grad_norm": 0.1545267913996029,
2653
+ "learning_rate": 0.00018997829730125663,
2654
+ "loss": 1.1829,
2655
+ "step": 378
2656
+ },
2657
+ {
2658
+ "epoch": 0.69,
2659
+ "grad_norm": 0.15421727704318197,
2660
+ "learning_rate": 0.000189885598890141,
2661
+ "loss": 1.177,
2662
+ "step": 379
2663
+ },
2664
+ {
2665
+ "epoch": 0.69,
2666
+ "grad_norm": 0.1624966073814206,
2667
+ "learning_rate": 0.00018979249656094673,
2668
+ "loss": 1.2439,
2669
+ "step": 380
2670
+ },
2671
+ {
2672
+ "epoch": 0.69,
2673
+ "grad_norm": 0.9490737312904575,
2674
+ "learning_rate": 0.00018969899073204686,
2675
+ "loss": 1.2085,
2676
+ "step": 381
2677
+ },
2678
+ {
2679
+ "epoch": 0.7,
2680
+ "grad_norm": 0.8982903208613089,
2681
+ "learning_rate": 0.00018960508182362768,
2682
+ "loss": 1.2347,
2683
+ "step": 382
2684
+ },
2685
+ {
2686
+ "epoch": 0.7,
2687
+ "grad_norm": 0.3771428474797688,
2688
+ "learning_rate": 0.00018951077025768678,
2689
+ "loss": 1.2546,
2690
+ "step": 383
2691
+ },
2692
+ {
2693
+ "epoch": 0.7,
2694
+ "grad_norm": 0.4776152950069111,
2695
+ "learning_rate": 0.00018941605645803115,
2696
+ "loss": 1.2904,
2697
+ "step": 384
2698
+ },
2699
+ {
2700
+ "epoch": 0.7,
2701
+ "grad_norm": 0.18786943849618057,
2702
+ "learning_rate": 0.00018932094085027533,
2703
+ "loss": 1.2122,
2704
+ "step": 385
2705
+ },
2706
+ {
2707
+ "epoch": 0.7,
2708
+ "grad_norm": 1.6297025984167128,
2709
+ "learning_rate": 0.0001892254238618394,
2710
+ "loss": 1.171,
2711
+ "step": 386
2712
+ },
2713
+ {
2714
+ "epoch": 0.7,
2715
+ "grad_norm": 0.20382660707264952,
2716
+ "learning_rate": 0.0001891295059219472,
2717
+ "loss": 1.1874,
2718
+ "step": 387
2719
+ },
2720
+ {
2721
+ "epoch": 0.71,
2722
+ "grad_norm": 1.3580819775908755,
2723
+ "learning_rate": 0.00018903318746162429,
2724
+ "loss": 1.1531,
2725
+ "step": 388
2726
+ },
2727
+ {
2728
+ "epoch": 0.71,
2729
+ "grad_norm": 0.43619056173016185,
2730
+ "learning_rate": 0.00018893646891369602,
2731
+ "loss": 1.2289,
2732
+ "step": 389
2733
+ },
2734
+ {
2735
+ "epoch": 0.71,
2736
+ "grad_norm": 0.29385240705823723,
2737
+ "learning_rate": 0.0001888393507127856,
2738
+ "loss": 1.2073,
2739
+ "step": 390
2740
+ },
2741
+ {
2742
+ "epoch": 0.71,
2743
+ "grad_norm": 0.3136086850525623,
2744
+ "learning_rate": 0.00018874183329531223,
2745
+ "loss": 1.1898,
2746
+ "step": 391
2747
+ },
2748
+ {
2749
+ "epoch": 0.71,
2750
+ "grad_norm": 0.2307767217662562,
2751
+ "learning_rate": 0.000188643917099489,
2752
+ "loss": 1.207,
2753
+ "step": 392
2754
+ },
2755
+ {
2756
+ "epoch": 0.72,
2757
+ "grad_norm": 0.18703654518135468,
2758
+ "learning_rate": 0.000188545602565321,
2759
+ "loss": 1.1688,
2760
+ "step": 393
2761
+ },
2762
+ {
2763
+ "epoch": 0.72,
2764
+ "grad_norm": 0.4809351333934126,
2765
+ "learning_rate": 0.00018844689013460336,
2766
+ "loss": 1.2519,
2767
+ "step": 394
2768
+ },
2769
+ {
2770
+ "epoch": 0.72,
2771
+ "grad_norm": 0.40370101428544464,
2772
+ "learning_rate": 0.0001883477802509192,
2773
+ "loss": 1.2411,
2774
+ "step": 395
2775
+ },
2776
+ {
2777
+ "epoch": 0.72,
2778
+ "grad_norm": 0.2858848636432859,
2779
+ "learning_rate": 0.00018824827335963765,
2780
+ "loss": 1.194,
2781
+ "step": 396
2782
+ },
2783
+ {
2784
+ "epoch": 0.72,
2785
+ "grad_norm": 0.32195602638999565,
2786
+ "learning_rate": 0.000188148369907912,
2787
+ "loss": 1.0988,
2788
+ "step": 397
2789
+ },
2790
+ {
2791
+ "epoch": 0.72,
2792
+ "grad_norm": 0.23790306908901832,
2793
+ "learning_rate": 0.00018804807034467733,
2794
+ "loss": 1.2237,
2795
+ "step": 398
2796
+ },
2797
+ {
2798
+ "epoch": 0.73,
2799
+ "grad_norm": 0.20126988767112128,
2800
+ "learning_rate": 0.0001879473751206489,
2801
+ "loss": 1.2731,
2802
+ "step": 399
2803
+ },
2804
+ {
2805
+ "epoch": 0.73,
2806
+ "grad_norm": 0.3336380339194037,
2807
+ "learning_rate": 0.00018784628468831996,
2808
+ "loss": 1.2369,
2809
+ "step": 400
2810
+ },
2811
+ {
2812
+ "epoch": 0.73,
2813
+ "grad_norm": 0.5054330893305989,
2814
+ "learning_rate": 0.0001877447995019596,
2815
+ "loss": 1.2443,
2816
+ "step": 401
2817
+ },
2818
+ {
2819
+ "epoch": 0.73,
2820
+ "grad_norm": 0.2297866279715136,
2821
+ "learning_rate": 0.0001876429200176108,
2822
+ "loss": 1.2376,
2823
+ "step": 402
2824
+ },
2825
+ {
2826
+ "epoch": 0.73,
2827
+ "grad_norm": 0.39350567174184636,
2828
+ "learning_rate": 0.00018754064669308858,
2829
+ "loss": 1.2126,
2830
+ "step": 403
2831
+ },
2832
+ {
2833
+ "epoch": 0.74,
2834
+ "grad_norm": 0.2025361091435325,
2835
+ "learning_rate": 0.00018743797998797753,
2836
+ "loss": 1.2224,
2837
+ "step": 404
2838
+ },
2839
+ {
2840
+ "epoch": 0.74,
2841
+ "grad_norm": 0.31824903419753814,
2842
+ "learning_rate": 0.00018733492036363005,
2843
+ "loss": 1.2942,
2844
+ "step": 405
2845
+ },
2846
+ {
2847
+ "epoch": 0.74,
2848
+ "grad_norm": 2.4642066748643017,
2849
+ "learning_rate": 0.00018723146828316428,
2850
+ "loss": 1.2515,
2851
+ "step": 406
2852
+ },
2853
+ {
2854
+ "epoch": 0.74,
2855
+ "grad_norm": 0.7833055646295342,
2856
+ "learning_rate": 0.00018712762421146183,
2857
+ "loss": 1.2207,
2858
+ "step": 407
2859
+ },
2860
+ {
2861
+ "epoch": 0.74,
2862
+ "grad_norm": 0.2810249021786599,
2863
+ "learning_rate": 0.00018702338861516587,
2864
+ "loss": 1.2755,
2865
+ "step": 408
2866
+ },
2867
+ {
2868
+ "epoch": 0.74,
2869
+ "grad_norm": 0.460995724241333,
2870
+ "learning_rate": 0.0001869187619626789,
2871
+ "loss": 1.2856,
2872
+ "step": 409
2873
+ },
2874
+ {
2875
+ "epoch": 0.75,
2876
+ "grad_norm": 0.49139203044984286,
2877
+ "learning_rate": 0.00018681374472416073,
2878
+ "loss": 1.2392,
2879
+ "step": 410
2880
+ },
2881
+ {
2882
+ "epoch": 0.75,
2883
+ "grad_norm": 2.691604613969173,
2884
+ "learning_rate": 0.0001867083373715264,
2885
+ "loss": 1.2992,
2886
+ "step": 411
2887
+ },
2888
+ {
2889
+ "epoch": 0.75,
2890
+ "grad_norm": 0.8014112047318501,
2891
+ "learning_rate": 0.00018660254037844388,
2892
+ "loss": 1.2683,
2893
+ "step": 412
2894
+ },
2895
+ {
2896
+ "epoch": 0.75,
2897
+ "grad_norm": 0.31614342841331383,
2898
+ "learning_rate": 0.00018649635422033215,
2899
+ "loss": 1.2356,
2900
+ "step": 413
2901
+ },
2902
+ {
2903
+ "epoch": 0.75,
2904
+ "grad_norm": 0.2559855196513244,
2905
+ "learning_rate": 0.000186389779374359,
2906
+ "loss": 1.2053,
2907
+ "step": 414
2908
+ },
2909
+ {
2910
+ "epoch": 0.76,
2911
+ "grad_norm": 1.6613999986014714,
2912
+ "learning_rate": 0.0001862828163194388,
2913
+ "loss": 1.2568,
2914
+ "step": 415
2915
+ },
2916
+ {
2917
+ "epoch": 0.76,
2918
+ "grad_norm": 0.27190082167109786,
2919
+ "learning_rate": 0.0001861754655362304,
2920
+ "loss": 1.1288,
2921
+ "step": 416
2922
+ },
2923
+ {
2924
+ "epoch": 0.76,
2925
+ "grad_norm": 0.43819582203066043,
2926
+ "learning_rate": 0.00018606772750713504,
2927
+ "loss": 1.1758,
2928
+ "step": 417
2929
+ },
2930
+ {
2931
+ "epoch": 0.76,
2932
+ "grad_norm": 0.41738497400383384,
2933
+ "learning_rate": 0.0001859596027162941,
2934
+ "loss": 1.2993,
2935
+ "step": 418
2936
+ },
2937
+ {
2938
+ "epoch": 0.76,
2939
+ "grad_norm": 0.2595142634740817,
2940
+ "learning_rate": 0.000185851091649587,
2941
+ "loss": 1.269,
2942
+ "step": 419
2943
+ },
2944
+ {
2945
+ "epoch": 0.77,
2946
+ "grad_norm": 0.2795314201020271,
2947
+ "learning_rate": 0.00018574219479462878,
2948
+ "loss": 1.1915,
2949
+ "step": 420
2950
+ },
2951
+ {
2952
+ "epoch": 0.77,
2953
+ "grad_norm": 0.2502992494749938,
2954
+ "learning_rate": 0.00018563291264076835,
2955
+ "loss": 1.2157,
2956
+ "step": 421
2957
+ },
2958
+ {
2959
+ "epoch": 0.77,
2960
+ "grad_norm": 0.27422512335538374,
2961
+ "learning_rate": 0.00018552324567908585,
2962
+ "loss": 1.2541,
2963
+ "step": 422
2964
+ },
2965
+ {
2966
+ "epoch": 0.77,
2967
+ "grad_norm": 1.3360989016060905,
2968
+ "learning_rate": 0.00018541319440239066,
2969
+ "loss": 1.2666,
2970
+ "step": 423
2971
+ },
2972
+ {
2973
+ "epoch": 0.77,
2974
+ "grad_norm": 1.9441131913572127,
2975
+ "learning_rate": 0.00018530275930521924,
2976
+ "loss": 1.2924,
2977
+ "step": 424
2978
+ },
2979
+ {
2980
+ "epoch": 0.77,
2981
+ "grad_norm": 0.3772373301771213,
2982
+ "learning_rate": 0.00018519194088383273,
2983
+ "loss": 1.1952,
2984
+ "step": 425
2985
+ },
2986
+ {
2987
+ "epoch": 0.78,
2988
+ "grad_norm": 0.22091753616251295,
2989
+ "learning_rate": 0.0001850807396362148,
2990
+ "loss": 1.1858,
2991
+ "step": 426
2992
+ },
2993
+ {
2994
+ "epoch": 0.78,
2995
+ "grad_norm": 0.21423504993321807,
2996
+ "learning_rate": 0.00018496915606206951,
2997
+ "loss": 1.2245,
2998
+ "step": 427
2999
+ },
3000
+ {
3001
+ "epoch": 0.78,
3002
+ "grad_norm": 0.5238946238105926,
3003
+ "learning_rate": 0.00018485719066281892,
3004
+ "loss": 1.2351,
3005
+ "step": 428
3006
+ },
3007
+ {
3008
+ "epoch": 0.78,
3009
+ "grad_norm": 0.3037858949309141,
3010
+ "learning_rate": 0.0001847448439416009,
3011
+ "loss": 1.1669,
3012
+ "step": 429
3013
+ },
3014
+ {
3015
+ "epoch": 0.78,
3016
+ "grad_norm": 0.21553286799952254,
3017
+ "learning_rate": 0.00018463211640326686,
3018
+ "loss": 1.1454,
3019
+ "step": 430
3020
+ },
3021
+ {
3022
+ "epoch": 0.79,
3023
+ "grad_norm": 0.202875547805464,
3024
+ "learning_rate": 0.0001845190085543795,
3025
+ "loss": 1.188,
3026
+ "step": 431
3027
+ },
3028
+ {
3029
+ "epoch": 0.79,
3030
+ "grad_norm": 0.24385408620619278,
3031
+ "learning_rate": 0.00018440552090321047,
3032
+ "loss": 1.2307,
3033
+ "step": 432
3034
+ },
3035
+ {
3036
+ "epoch": 0.79,
3037
+ "grad_norm": 0.23793944272430378,
3038
+ "learning_rate": 0.0001842916539597382,
3039
+ "loss": 1.2253,
3040
+ "step": 433
3041
+ },
3042
+ {
3043
+ "epoch": 0.79,
3044
+ "grad_norm": 0.17062488448810784,
3045
+ "learning_rate": 0.0001841774082356455,
3046
+ "loss": 1.2681,
3047
+ "step": 434
3048
+ },
3049
+ {
3050
+ "epoch": 0.79,
3051
+ "grad_norm": 0.20003742001916064,
3052
+ "learning_rate": 0.00018406278424431736,
3053
+ "loss": 1.2428,
3054
+ "step": 435
3055
+ },
3056
+ {
3057
+ "epoch": 0.79,
3058
+ "grad_norm": 0.2696052831337752,
3059
+ "learning_rate": 0.0001839477825008385,
3060
+ "loss": 1.2945,
3061
+ "step": 436
3062
+ },
3063
+ {
3064
+ "epoch": 0.8,
3065
+ "grad_norm": 0.23302960820538443,
3066
+ "learning_rate": 0.00018383240352199117,
3067
+ "loss": 1.1718,
3068
+ "step": 437
3069
+ },
3070
+ {
3071
+ "epoch": 0.8,
3072
+ "grad_norm": 0.38187833239777536,
3073
+ "learning_rate": 0.00018371664782625287,
3074
+ "loss": 1.2311,
3075
+ "step": 438
3076
+ },
3077
+ {
3078
+ "epoch": 0.8,
3079
+ "grad_norm": 0.4052561772533732,
3080
+ "learning_rate": 0.00018360051593379383,
3081
+ "loss": 1.1639,
3082
+ "step": 439
3083
+ },
3084
+ {
3085
+ "epoch": 0.8,
3086
+ "grad_norm": 0.23379763821020377,
3087
+ "learning_rate": 0.0001834840083664749,
3088
+ "loss": 1.1809,
3089
+ "step": 440
3090
+ },
3091
+ {
3092
+ "epoch": 0.8,
3093
+ "grad_norm": 0.2368414607613928,
3094
+ "learning_rate": 0.00018336712564784503,
3095
+ "loss": 1.2357,
3096
+ "step": 441
3097
+ },
3098
+ {
3099
+ "epoch": 0.81,
3100
+ "grad_norm": 0.20230633988510938,
3101
+ "learning_rate": 0.000183249868303139,
3102
+ "loss": 1.1851,
3103
+ "step": 442
3104
+ },
3105
+ {
3106
+ "epoch": 0.81,
3107
+ "grad_norm": 0.170513157244292,
3108
+ "learning_rate": 0.00018313223685927505,
3109
+ "loss": 1.205,
3110
+ "step": 443
3111
+ },
3112
+ {
3113
+ "epoch": 0.81,
3114
+ "grad_norm": 0.18082295035256266,
3115
+ "learning_rate": 0.0001830142318448525,
3116
+ "loss": 1.2305,
3117
+ "step": 444
3118
+ },
3119
+ {
3120
+ "epoch": 0.81,
3121
+ "grad_norm": 0.18286299264146286,
3122
+ "learning_rate": 0.00018289585379014942,
3123
+ "loss": 1.23,
3124
+ "step": 445
3125
+ },
3126
+ {
3127
+ "epoch": 0.81,
3128
+ "grad_norm": 0.17868104103482751,
3129
+ "learning_rate": 0.00018277710322712012,
3130
+ "loss": 1.2894,
3131
+ "step": 446
3132
+ },
3133
+ {
3134
+ "epoch": 0.81,
3135
+ "grad_norm": 0.1820411127336495,
3136
+ "learning_rate": 0.00018265798068939294,
3137
+ "loss": 1.2395,
3138
+ "step": 447
3139
+ },
3140
+ {
3141
+ "epoch": 0.82,
3142
+ "grad_norm": 0.1738237541783663,
3143
+ "learning_rate": 0.0001825384867122677,
3144
+ "loss": 1.1576,
3145
+ "step": 448
3146
+ },
3147
+ {
3148
+ "epoch": 0.82,
3149
+ "grad_norm": 0.15693445967795147,
3150
+ "learning_rate": 0.0001824186218327134,
3151
+ "loss": 1.0809,
3152
+ "step": 449
3153
+ },
3154
+ {
3155
+ "epoch": 0.82,
3156
+ "grad_norm": 0.18509145652208978,
3157
+ "learning_rate": 0.00018229838658936564,
3158
+ "loss": 1.2717,
3159
+ "step": 450
3160
+ },
3161
+ {
3162
+ "epoch": 0.82,
3163
+ "grad_norm": 0.14702488366564262,
3164
+ "learning_rate": 0.0001821777815225245,
3165
+ "loss": 1.2236,
3166
+ "step": 451
3167
+ },
3168
+ {
3169
+ "epoch": 0.82,
3170
+ "grad_norm": 0.1828399354418095,
3171
+ "learning_rate": 0.00018205680717415187,
3172
+ "loss": 1.2565,
3173
+ "step": 452
3174
+ },
3175
+ {
3176
+ "epoch": 0.83,
3177
+ "grad_norm": 0.17460984182013486,
3178
+ "learning_rate": 0.00018193546408786898,
3179
+ "loss": 1.2474,
3180
+ "step": 453
3181
+ },
3182
+ {
3183
+ "epoch": 0.83,
3184
+ "grad_norm": 0.2001623109673152,
3185
+ "learning_rate": 0.00018181375280895416,
3186
+ "loss": 1.2544,
3187
+ "step": 454
3188
+ },
3189
+ {
3190
+ "epoch": 0.83,
3191
+ "grad_norm": 0.17228631742863837,
3192
+ "learning_rate": 0.00018169167388434025,
3193
+ "loss": 1.1851,
3194
+ "step": 455
3195
+ },
3196
+ {
3197
+ "epoch": 0.83,
3198
+ "grad_norm": 0.1644862232819482,
3199
+ "learning_rate": 0.00018156922786261216,
3200
+ "loss": 1.1817,
3201
+ "step": 456
3202
+ },
3203
+ {
3204
+ "epoch": 0.83,
3205
+ "grad_norm": 0.19775186397477057,
3206
+ "learning_rate": 0.00018144641529400446,
3207
+ "loss": 1.257,
3208
+ "step": 457
3209
+ },
3210
+ {
3211
+ "epoch": 0.83,
3212
+ "grad_norm": 0.1626281991220394,
3213
+ "learning_rate": 0.00018132323673039885,
3214
+ "loss": 1.2277,
3215
+ "step": 458
3216
+ },
3217
+ {
3218
+ "epoch": 0.84,
3219
+ "grad_norm": 0.16158256707311264,
3220
+ "learning_rate": 0.00018119969272532166,
3221
+ "loss": 1.1624,
3222
+ "step": 459
3223
+ },
3224
+ {
3225
+ "epoch": 0.84,
3226
+ "grad_norm": 0.17705809207051687,
3227
+ "learning_rate": 0.00018107578383394146,
3228
+ "loss": 1.2421,
3229
+ "step": 460
3230
+ },
3231
+ {
3232
+ "epoch": 0.84,
3233
+ "grad_norm": 0.17639060401882287,
3234
+ "learning_rate": 0.00018095151061306645,
3235
+ "loss": 1.285,
3236
+ "step": 461
3237
+ },
3238
+ {
3239
+ "epoch": 0.84,
3240
+ "grad_norm": 0.16918796486576196,
3241
+ "learning_rate": 0.00018082687362114212,
3242
+ "loss": 1.2606,
3243
+ "step": 462
3244
+ },
3245
+ {
3246
+ "epoch": 0.84,
3247
+ "grad_norm": 0.15968377185965665,
3248
+ "learning_rate": 0.0001807018734182485,
3249
+ "loss": 1.194,
3250
+ "step": 463
3251
+ },
3252
+ {
3253
+ "epoch": 0.85,
3254
+ "grad_norm": 0.17537027967397978,
3255
+ "learning_rate": 0.00018057651056609784,
3256
+ "loss": 1.1594,
3257
+ "step": 464
3258
+ },
3259
+ {
3260
+ "epoch": 0.85,
3261
+ "grad_norm": 0.15753665403127565,
3262
+ "learning_rate": 0.00018045078562803203,
3263
+ "loss": 1.1382,
3264
+ "step": 465
3265
+ },
3266
+ {
3267
+ "epoch": 0.85,
3268
+ "grad_norm": 0.17121200763916436,
3269
+ "learning_rate": 0.00018032469916902003,
3270
+ "loss": 1.2286,
3271
+ "step": 466
3272
+ },
3273
+ {
3274
+ "epoch": 0.85,
3275
+ "grad_norm": 0.19120510133331003,
3276
+ "learning_rate": 0.00018019825175565542,
3277
+ "loss": 1.2835,
3278
+ "step": 467
3279
+ },
3280
+ {
3281
+ "epoch": 0.85,
3282
+ "grad_norm": 0.1671735980123817,
3283
+ "learning_rate": 0.0001800714439561538,
3284
+ "loss": 1.2201,
3285
+ "step": 468
3286
+ },
3287
+ {
3288
+ "epoch": 0.85,
3289
+ "grad_norm": 0.1579098534969056,
3290
+ "learning_rate": 0.00017994427634035015,
3291
+ "loss": 1.2156,
3292
+ "step": 469
3293
+ },
3294
+ {
3295
+ "epoch": 0.86,
3296
+ "grad_norm": 0.1746075421158512,
3297
+ "learning_rate": 0.00017981674947969636,
3298
+ "loss": 1.2049,
3299
+ "step": 470
3300
+ },
3301
+ {
3302
+ "epoch": 0.86,
3303
+ "grad_norm": 0.16878182886737042,
3304
+ "learning_rate": 0.00017968886394725874,
3305
+ "loss": 1.2204,
3306
+ "step": 471
3307
+ },
3308
+ {
3309
+ "epoch": 0.86,
3310
+ "grad_norm": 0.16725956538286493,
3311
+ "learning_rate": 0.00017956062031771535,
3312
+ "loss": 1.2091,
3313
+ "step": 472
3314
+ },
3315
+ {
3316
+ "epoch": 0.86,
3317
+ "grad_norm": 0.18877845951705005,
3318
+ "learning_rate": 0.00017943201916735335,
3319
+ "loss": 1.241,
3320
+ "step": 473
3321
+ },
3322
+ {
3323
+ "epoch": 0.86,
3324
+ "grad_norm": 0.180337447476004,
3325
+ "learning_rate": 0.00017930306107406653,
3326
+ "loss": 1.2253,
3327
+ "step": 474
3328
+ },
3329
+ {
3330
+ "epoch": 0.87,
3331
+ "grad_norm": 0.16688572366717752,
3332
+ "learning_rate": 0.0001791737466173527,
3333
+ "loss": 1.239,
3334
+ "step": 475
3335
+ },
3336
+ {
3337
+ "epoch": 0.87,
3338
+ "grad_norm": 0.15385917621135983,
3339
+ "learning_rate": 0.00017904407637831099,
3340
+ "loss": 1.2476,
3341
+ "step": 476
3342
+ },
3343
+ {
3344
+ "epoch": 0.87,
3345
+ "grad_norm": 0.17725645269055587,
3346
+ "learning_rate": 0.00017891405093963938,
3347
+ "loss": 1.2599,
3348
+ "step": 477
3349
+ },
3350
+ {
3351
+ "epoch": 0.87,
3352
+ "grad_norm": 0.14758551718901028,
3353
+ "learning_rate": 0.00017878367088563195,
3354
+ "loss": 1.2249,
3355
+ "step": 478
3356
+ },
3357
+ {
3358
+ "epoch": 0.87,
3359
+ "grad_norm": 0.15216962408661316,
3360
+ "learning_rate": 0.00017865293680217637,
3361
+ "loss": 1.2346,
3362
+ "step": 479
3363
+ },
3364
+ {
3365
+ "epoch": 0.87,
3366
+ "grad_norm": 0.16679282848599514,
3367
+ "learning_rate": 0.00017852184927675112,
3368
+ "loss": 1.2443,
3369
+ "step": 480
3370
+ },
3371
+ {
3372
+ "epoch": 0.88,
3373
+ "grad_norm": 0.16723562739069214,
3374
+ "learning_rate": 0.00017839040889842305,
3375
+ "loss": 1.224,
3376
+ "step": 481
3377
+ },
3378
+ {
3379
+ "epoch": 0.88,
3380
+ "grad_norm": 0.15922276239929914,
3381
+ "learning_rate": 0.00017825861625784455,
3382
+ "loss": 1.2739,
3383
+ "step": 482
3384
+ },
3385
+ {
3386
+ "epoch": 0.88,
3387
+ "grad_norm": 0.1510107938469514,
3388
+ "learning_rate": 0.00017812647194725094,
3389
+ "loss": 1.1764,
3390
+ "step": 483
3391
+ },
3392
+ {
3393
+ "epoch": 0.88,
3394
+ "grad_norm": 0.16446999054333494,
3395
+ "learning_rate": 0.00017799397656045792,
3396
+ "loss": 1.2498,
3397
+ "step": 484
3398
+ },
3399
+ {
3400
+ "epoch": 0.88,
3401
+ "grad_norm": 0.18566301651865832,
3402
+ "learning_rate": 0.00017786113069285874,
3403
+ "loss": 1.232,
3404
+ "step": 485
3405
+ },
3406
+ {
3407
+ "epoch": 0.89,
3408
+ "grad_norm": 0.20592971655306183,
3409
+ "learning_rate": 0.00017772793494142167,
3410
+ "loss": 1.1586,
3411
+ "step": 486
3412
+ },
3413
+ {
3414
+ "epoch": 0.89,
3415
+ "grad_norm": 0.1581947714375729,
3416
+ "learning_rate": 0.00017759438990468725,
3417
+ "loss": 1.2502,
3418
+ "step": 487
3419
+ },
3420
+ {
3421
+ "epoch": 0.89,
3422
+ "grad_norm": 0.15466760695169174,
3423
+ "learning_rate": 0.00017746049618276545,
3424
+ "loss": 1.1605,
3425
+ "step": 488
3426
+ },
3427
+ {
3428
+ "epoch": 0.89,
3429
+ "grad_norm": 0.16041506222444918,
3430
+ "learning_rate": 0.00017732625437733335,
3431
+ "loss": 1.2778,
3432
+ "step": 489
3433
+ },
3434
+ {
3435
+ "epoch": 0.89,
3436
+ "grad_norm": 0.17168109661676773,
3437
+ "learning_rate": 0.0001771916650916321,
3438
+ "loss": 1.262,
3439
+ "step": 490
3440
+ },
3441
+ {
3442
+ "epoch": 0.89,
3443
+ "grad_norm": 0.1788973186498254,
3444
+ "learning_rate": 0.00017705672893046425,
3445
+ "loss": 1.2111,
3446
+ "step": 491
3447
+ },
3448
+ {
3449
+ "epoch": 0.9,
3450
+ "grad_norm": 0.1759644359346382,
3451
+ "learning_rate": 0.00017692144650019125,
3452
+ "loss": 1.2546,
3453
+ "step": 492
3454
+ },
3455
+ {
3456
+ "epoch": 0.9,
3457
+ "grad_norm": 0.15710749736088767,
3458
+ "learning_rate": 0.0001767858184087304,
3459
+ "loss": 1.2487,
3460
+ "step": 493
3461
+ },
3462
+ {
3463
+ "epoch": 0.9,
3464
+ "grad_norm": 0.1648235522911144,
3465
+ "learning_rate": 0.00017664984526555248,
3466
+ "loss": 1.2469,
3467
+ "step": 494
3468
+ },
3469
+ {
3470
+ "epoch": 0.9,
3471
+ "grad_norm": 0.15452607969890703,
3472
+ "learning_rate": 0.0001765135276816787,
3473
+ "loss": 1.1855,
3474
+ "step": 495
3475
+ },
3476
+ {
3477
+ "epoch": 0.9,
3478
+ "grad_norm": 0.1837695597880219,
3479
+ "learning_rate": 0.00017637686626967812,
3480
+ "loss": 1.2185,
3481
+ "step": 496
3482
+ },
3483
+ {
3484
+ "epoch": 0.91,
3485
+ "grad_norm": 0.15861390725762364,
3486
+ "learning_rate": 0.00017623986164366486,
3487
+ "loss": 1.2056,
3488
+ "step": 497
3489
+ },
3490
+ {
3491
+ "epoch": 0.91,
3492
+ "grad_norm": 0.1663260460966887,
3493
+ "learning_rate": 0.00017610251441929533,
3494
+ "loss": 1.1242,
3495
+ "step": 498
3496
+ },
3497
+ {
3498
+ "epoch": 0.91,
3499
+ "grad_norm": 0.1803309720529981,
3500
+ "learning_rate": 0.00017596482521376546,
3501
+ "loss": 1.2938,
3502
+ "step": 499
3503
+ },
3504
+ {
3505
+ "epoch": 0.91,
3506
+ "grad_norm": 0.14909085011764342,
3507
+ "learning_rate": 0.00017582679464580797,
3508
+ "loss": 1.1953,
3509
+ "step": 500
3510
+ },
3511
+ {
3512
+ "epoch": 0.91,
3513
+ "grad_norm": 0.15779022242482527,
3514
+ "learning_rate": 0.00017568842333568952,
3515
+ "loss": 1.2792,
3516
+ "step": 501
3517
+ },
3518
+ {
3519
+ "epoch": 0.91,
3520
+ "grad_norm": 0.1553327313967345,
3521
+ "learning_rate": 0.00017554971190520798,
3522
+ "loss": 1.2286,
3523
+ "step": 502
3524
+ },
3525
+ {
3526
+ "epoch": 0.92,
3527
+ "grad_norm": 0.16363964666273684,
3528
+ "learning_rate": 0.00017541066097768963,
3529
+ "loss": 1.2753,
3530
+ "step": 503
3531
+ },
3532
+ {
3533
+ "epoch": 0.92,
3534
+ "grad_norm": 0.16668099163659675,
3535
+ "learning_rate": 0.00017527127117798635,
3536
+ "loss": 1.185,
3537
+ "step": 504
3538
+ },
3539
+ {
3540
+ "epoch": 0.92,
3541
+ "grad_norm": 0.13957568397594883,
3542
+ "learning_rate": 0.0001751315431324727,
3543
+ "loss": 1.143,
3544
+ "step": 505
3545
+ },
3546
+ {
3547
+ "epoch": 0.92,
3548
+ "grad_norm": 0.1553111736740035,
3549
+ "learning_rate": 0.00017499147746904335,
3550
+ "loss": 1.2492,
3551
+ "step": 506
3552
+ },
3553
+ {
3554
+ "epoch": 0.92,
3555
+ "grad_norm": 0.1691517335818193,
3556
+ "learning_rate": 0.00017485107481711012,
3557
+ "loss": 1.2619,
3558
+ "step": 507
3559
+ },
3560
+ {
3561
+ "epoch": 0.93,
3562
+ "grad_norm": 0.15480883994395986,
3563
+ "learning_rate": 0.00017471033580759903,
3564
+ "loss": 1.2396,
3565
+ "step": 508
3566
+ },
3567
+ {
3568
+ "epoch": 0.93,
3569
+ "grad_norm": 0.1451690143792058,
3570
+ "learning_rate": 0.00017456926107294765,
3571
+ "loss": 1.1732,
3572
+ "step": 509
3573
+ },
3574
+ {
3575
+ "epoch": 0.93,
3576
+ "grad_norm": 0.1524398957482947,
3577
+ "learning_rate": 0.00017442785124710227,
3578
+ "loss": 1.2083,
3579
+ "step": 510
3580
+ },
3581
+ {
3582
+ "epoch": 0.93,
3583
+ "grad_norm": 0.16790264977550012,
3584
+ "learning_rate": 0.0001742861069655148,
3585
+ "loss": 1.2201,
3586
+ "step": 511
3587
+ },
3588
+ {
3589
+ "epoch": 0.93,
3590
+ "grad_norm": 0.1529847047636337,
3591
+ "learning_rate": 0.0001741440288651403,
3592
+ "loss": 1.243,
3593
+ "step": 512
3594
+ },
3595
+ {
3596
+ "epoch": 0.93,
3597
+ "grad_norm": 0.1485875402374676,
3598
+ "learning_rate": 0.00017400161758443375,
3599
+ "loss": 1.2053,
3600
+ "step": 513
3601
+ },
3602
+ {
3603
+ "epoch": 0.94,
3604
+ "grad_norm": 0.16950094279079617,
3605
+ "learning_rate": 0.00017385887376334742,
3606
+ "loss": 1.1944,
3607
+ "step": 514
3608
+ },
3609
+ {
3610
+ "epoch": 0.94,
3611
+ "grad_norm": 0.15289337084330445,
3612
+ "learning_rate": 0.00017371579804332789,
3613
+ "loss": 1.2503,
3614
+ "step": 515
3615
+ },
3616
+ {
3617
+ "epoch": 0.94,
3618
+ "grad_norm": 0.15337063655317973,
3619
+ "learning_rate": 0.00017357239106731317,
3620
+ "loss": 1.3092,
3621
+ "step": 516
3622
+ },
3623
+ {
3624
+ "epoch": 0.94,
3625
+ "grad_norm": 0.1458937961897621,
3626
+ "learning_rate": 0.00017342865347972988,
3627
+ "loss": 1.2244,
3628
+ "step": 517
3629
+ },
3630
+ {
3631
+ "epoch": 0.94,
3632
+ "grad_norm": 0.19897118610161338,
3633
+ "learning_rate": 0.00017328458592649027,
3634
+ "loss": 1.2238,
3635
+ "step": 518
3636
+ },
3637
+ {
3638
+ "epoch": 0.95,
3639
+ "grad_norm": 0.15850805264911003,
3640
+ "learning_rate": 0.00017314018905498931,
3641
+ "loss": 1.195,
3642
+ "step": 519
3643
+ },
3644
+ {
3645
+ "epoch": 0.95,
3646
+ "grad_norm": 0.14445183074519347,
3647
+ "learning_rate": 0.00017299546351410197,
3648
+ "loss": 1.1974,
3649
+ "step": 520
3650
+ },
3651
+ {
3652
+ "epoch": 0.95,
3653
+ "grad_norm": 0.18180731722745677,
3654
+ "learning_rate": 0.00017285040995418,
3655
+ "loss": 1.2107,
3656
+ "step": 521
3657
+ },
3658
+ {
3659
+ "epoch": 0.95,
3660
+ "grad_norm": 0.14943874953193587,
3661
+ "learning_rate": 0.00017270502902704926,
3662
+ "loss": 1.1843,
3663
+ "step": 522
3664
+ },
3665
+ {
3666
+ "epoch": 0.95,
3667
+ "grad_norm": 0.15767466790910512,
3668
+ "learning_rate": 0.00017255932138600665,
3669
+ "loss": 1.1409,
3670
+ "step": 523
3671
+ },
3672
+ {
3673
+ "epoch": 0.95,
3674
+ "grad_norm": 0.16402921378654775,
3675
+ "learning_rate": 0.00017241328768581726,
3676
+ "loss": 1.2135,
3677
+ "step": 524
3678
+ },
3679
+ {
3680
+ "epoch": 0.96,
3681
+ "grad_norm": 0.15526246786505485,
3682
+ "learning_rate": 0.00017226692858271134,
3683
+ "loss": 1.2255,
3684
+ "step": 525
3685
+ },
3686
+ {
3687
+ "epoch": 0.96,
3688
+ "grad_norm": 0.16608155892622348,
3689
+ "learning_rate": 0.00017212024473438147,
3690
+ "loss": 1.2691,
3691
+ "step": 526
3692
+ },
3693
+ {
3694
+ "epoch": 0.96,
3695
+ "grad_norm": 0.14913271520144072,
3696
+ "learning_rate": 0.00017197323679997943,
3697
+ "loss": 1.1574,
3698
+ "step": 527
3699
+ },
3700
+ {
3701
+ "epoch": 0.96,
3702
+ "grad_norm": 0.1471910610421707,
3703
+ "learning_rate": 0.00017182590544011347,
3704
+ "loss": 1.2774,
3705
+ "step": 528
3706
+ },
3707
+ {
3708
+ "epoch": 0.96,
3709
+ "grad_norm": 0.1417464185073962,
3710
+ "learning_rate": 0.00017167825131684513,
3711
+ "loss": 1.2446,
3712
+ "step": 529
3713
+ },
3714
+ {
3715
+ "epoch": 0.97,
3716
+ "grad_norm": 0.1610488125634495,
3717
+ "learning_rate": 0.0001715302750936864,
3718
+ "loss": 1.2862,
3719
+ "step": 530
3720
+ },
3721
+ {
3722
+ "epoch": 0.97,
3723
+ "grad_norm": 0.20227974555123074,
3724
+ "learning_rate": 0.00017138197743559654,
3725
+ "loss": 1.207,
3726
+ "step": 531
3727
+ },
3728
+ {
3729
+ "epoch": 0.97,
3730
+ "grad_norm": 0.1355502559749413,
3731
+ "learning_rate": 0.00017123335900897946,
3732
+ "loss": 1.1019,
3733
+ "step": 532
3734
+ },
3735
+ {
3736
+ "epoch": 0.97,
3737
+ "grad_norm": 0.1559423167028215,
3738
+ "learning_rate": 0.00017108442048168038,
3739
+ "loss": 1.2549,
3740
+ "step": 533
3741
+ },
3742
+ {
3743
+ "epoch": 0.97,
3744
+ "grad_norm": 0.15898973818185586,
3745
+ "learning_rate": 0.00017093516252298296,
3746
+ "loss": 1.2705,
3747
+ "step": 534
3748
+ },
3749
+ {
3750
+ "epoch": 0.97,
3751
+ "grad_norm": 0.15169569998999652,
3752
+ "learning_rate": 0.00017078558580360632,
3753
+ "loss": 1.2454,
3754
+ "step": 535
3755
+ },
3756
+ {
3757
+ "epoch": 0.98,
3758
+ "grad_norm": 0.15976111665597925,
3759
+ "learning_rate": 0.00017063569099570196,
3760
+ "loss": 1.2585,
3761
+ "step": 536
3762
+ },
3763
+ {
3764
+ "epoch": 0.98,
3765
+ "grad_norm": 0.14488877221999352,
3766
+ "learning_rate": 0.00017048547877285077,
3767
+ "loss": 1.2169,
3768
+ "step": 537
3769
+ },
3770
+ {
3771
+ "epoch": 0.98,
3772
+ "grad_norm": 0.14919533098974924,
3773
+ "learning_rate": 0.00017033494981006002,
3774
+ "loss": 1.2358,
3775
+ "step": 538
3776
+ },
3777
+ {
3778
+ "epoch": 0.98,
3779
+ "grad_norm": 0.15251746717084805,
3780
+ "learning_rate": 0.00017018410478376032,
3781
+ "loss": 1.2241,
3782
+ "step": 539
3783
+ },
3784
+ {
3785
+ "epoch": 0.98,
3786
+ "grad_norm": 0.1456060482002663,
3787
+ "learning_rate": 0.00017003294437180255,
3788
+ "loss": 1.2298,
3789
+ "step": 540
3790
+ },
3791
+ {
3792
+ "epoch": 0.99,
3793
+ "grad_norm": 0.17048886778787248,
3794
+ "learning_rate": 0.00016988146925345484,
3795
+ "loss": 1.2707,
3796
+ "step": 541
3797
+ },
3798
+ {
3799
+ "epoch": 0.99,
3800
+ "grad_norm": 0.15304381059310815,
3801
+ "learning_rate": 0.00016972968010939954,
3802
+ "loss": 1.1498,
3803
+ "step": 542
3804
+ },
3805
+ {
3806
+ "epoch": 0.99,
3807
+ "grad_norm": 0.16590055969071696,
3808
+ "learning_rate": 0.0001695775776217301,
3809
+ "loss": 1.2481,
3810
+ "step": 543
3811
+ },
3812
+ {
3813
+ "epoch": 0.99,
3814
+ "grad_norm": 0.14299575837437278,
3815
+ "learning_rate": 0.00016942516247394807,
3816
+ "loss": 1.2058,
3817
+ "step": 544
3818
+ },
3819
+ {
3820
+ "epoch": 0.99,
3821
+ "grad_norm": 0.14275107775859475,
3822
+ "learning_rate": 0.00016927243535095997,
3823
+ "loss": 1.2178,
3824
+ "step": 545
3825
+ },
3826
+ {
3827
+ "epoch": 0.99,
3828
+ "grad_norm": 0.1554250137491414,
3829
+ "learning_rate": 0.0001691193969390742,
3830
+ "loss": 1.1197,
3831
+ "step": 546
3832
+ },
3833
+ {
3834
+ "epoch": 1.0,
3835
+ "grad_norm": 0.16958418467021688,
3836
+ "learning_rate": 0.0001689660479259981,
3837
+ "loss": 1.1768,
3838
+ "step": 547
3839
+ },
3840
+ {
3841
+ "epoch": 1.0,
3842
+ "grad_norm": 0.1546216583314497,
3843
+ "learning_rate": 0.00016881238900083473,
3844
+ "loss": 1.1741,
3845
+ "step": 548
3846
+ },
3847
+ {
3848
+ "epoch": 1.0,
3849
+ "grad_norm": 0.15287056494787424,
3850
+ "learning_rate": 0.0001686584208540797,
3851
+ "loss": 1.2328,
3852
+ "step": 549
3853
+ },
3854
+ {
3855
+ "epoch": 1.0,
3856
+ "grad_norm": 0.1419329373337611,
3857
+ "learning_rate": 0.0001685041441776183,
3858
+ "loss": 1.1743,
3859
+ "step": 550
3860
+ }
3861
+ ],
3862
+ "logging_steps": 1.0,
3863
+ "max_steps": 1647,
3864
+ "num_input_tokens_seen": 0,
3865
+ "num_train_epochs": 3,
3866
+ "save_steps": 50,
3867
+ "total_flos": 5103643602714624.0,
3868
+ "train_batch_size": 1,
3869
+ "trial_name": null,
3870
+ "trial_params": null
3871
+ }
550/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c6c256082b8d848443372d8fc6c44427099f0a764e044ba63c75a1f81db6afd
3
+ size 6712
550/zero_to_fp32.py ADDED
@@ -0,0 +1,604 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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, exclude_frozen_parameters):
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
+ exclude_frozen_parameters)
216
+ elif zero_stage == 3:
217
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
218
+ exclude_frozen_parameters)
219
+
220
+
221
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
222
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
223
+ return
224
+
225
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
226
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
227
+
228
+ if debug:
229
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
230
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
231
+
232
+ wanted_params = len(frozen_param_shapes)
233
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
234
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
235
+ print(f'Frozen params: Have {avail_numel} numels to process.')
236
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
237
+
238
+ total_params = 0
239
+ total_numel = 0
240
+ for name, shape in frozen_param_shapes.items():
241
+ total_params += 1
242
+ unpartitioned_numel = shape.numel()
243
+ total_numel += unpartitioned_numel
244
+
245
+ state_dict[name] = frozen_param_fragments[name]
246
+
247
+ if debug:
248
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
249
+
250
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
251
+
252
+
253
+ def _has_callable(obj, fn):
254
+ attr = getattr(obj, fn, None)
255
+ return callable(attr)
256
+
257
+
258
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
259
+ param_shapes = zero_model_states[0].param_shapes
260
+
261
+ # Reconstruction protocol:
262
+ #
263
+ # XXX: document this
264
+
265
+ if debug:
266
+ for i in range(world_size):
267
+ for j in range(len(fp32_flat_groups[0])):
268
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
269
+
270
+ # XXX: memory usage doubles here (zero2)
271
+ num_param_groups = len(fp32_flat_groups[0])
272
+ merged_single_partition_of_fp32_groups = []
273
+ for i in range(num_param_groups):
274
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
275
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
276
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
277
+ avail_numel = sum(
278
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
279
+
280
+ if debug:
281
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
282
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
283
+ # not asserting if there is a mismatch due to possible padding
284
+ print(f"Have {avail_numel} numels to process.")
285
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
286
+
287
+ # params
288
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
289
+ # out-of-core computing solution
290
+ total_numel = 0
291
+ total_params = 0
292
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
293
+ offset = 0
294
+ avail_numel = full_single_fp32_vector.numel()
295
+ for name, shape in shapes.items():
296
+
297
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
298
+ total_numel += unpartitioned_numel
299
+ total_params += 1
300
+
301
+ if debug:
302
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
303
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
304
+ offset += unpartitioned_numel
305
+
306
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
307
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
308
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
309
+ # live optimizer object, so we are checking that the numbers are within the right range
310
+ align_to = 2 * world_size
311
+
312
+ def zero2_align(x):
313
+ return align_to * math.ceil(x / align_to)
314
+
315
+ if debug:
316
+ print(f"original offset={offset}, avail_numel={avail_numel}")
317
+
318
+ offset = zero2_align(offset)
319
+ avail_numel = zero2_align(avail_numel)
320
+
321
+ if debug:
322
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
323
+
324
+ # Sanity check
325
+ if offset != avail_numel:
326
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
327
+
328
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
329
+
330
+
331
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
332
+ exclude_frozen_parameters):
333
+ state_dict = OrderedDict()
334
+
335
+ # buffers
336
+ buffers = zero_model_states[0].buffers
337
+ state_dict.update(buffers)
338
+ if debug:
339
+ print(f"added {len(buffers)} buffers")
340
+
341
+ if not exclude_frozen_parameters:
342
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
343
+
344
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
345
+
346
+ # recover shared parameters
347
+ for pair in zero_model_states[0].shared_params:
348
+ if pair[1] in state_dict:
349
+ state_dict[pair[0]] = state_dict[pair[1]]
350
+
351
+ return state_dict
352
+
353
+
354
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
355
+ remainder = unpartitioned_numel % world_size
356
+ padding_numel = (world_size - remainder) if remainder else 0
357
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
358
+ return partitioned_numel, padding_numel
359
+
360
+
361
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
362
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
363
+ return
364
+
365
+ if debug:
366
+ for i in range(world_size):
367
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
368
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
369
+
370
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
371
+ wanted_params = len(frozen_param_shapes)
372
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
373
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
374
+ print(f'Frozen params: Have {avail_numel} numels to process.')
375
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
376
+
377
+ total_params = 0
378
+ total_numel = 0
379
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
380
+ total_params += 1
381
+ unpartitioned_numel = shape.numel()
382
+ total_numel += unpartitioned_numel
383
+
384
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
385
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
386
+
387
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
388
+
389
+ if debug:
390
+ print(
391
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
392
+ )
393
+
394
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
395
+
396
+
397
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
398
+ param_shapes = zero_model_states[0].param_shapes
399
+ avail_numel = fp32_flat_groups[0].numel() * world_size
400
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
401
+ # param, re-consolidating each param, while dealing with padding if any
402
+
403
+ # merge list of dicts, preserving order
404
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
405
+
406
+ if debug:
407
+ for i in range(world_size):
408
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
409
+
410
+ wanted_params = len(param_shapes)
411
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
412
+ # not asserting if there is a mismatch due to possible padding
413
+ avail_numel = fp32_flat_groups[0].numel() * world_size
414
+ print(f"Trainable params: Have {avail_numel} numels to process.")
415
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
416
+
417
+ # params
418
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
419
+ # out-of-core computing solution
420
+ offset = 0
421
+ total_numel = 0
422
+ total_params = 0
423
+ for name, shape in param_shapes.items():
424
+
425
+ unpartitioned_numel = shape.numel()
426
+ total_numel += unpartitioned_numel
427
+ total_params += 1
428
+
429
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
430
+
431
+ if debug:
432
+ print(
433
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
434
+ )
435
+
436
+ # XXX: memory usage doubles here
437
+ state_dict[name] = torch.cat(
438
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
439
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
440
+ offset += partitioned_numel
441
+
442
+ offset *= world_size
443
+
444
+ # Sanity check
445
+ if offset != avail_numel:
446
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
447
+
448
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
449
+
450
+
451
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
452
+ exclude_frozen_parameters):
453
+ state_dict = OrderedDict()
454
+
455
+ # buffers
456
+ buffers = zero_model_states[0].buffers
457
+ state_dict.update(buffers)
458
+ if debug:
459
+ print(f"added {len(buffers)} buffers")
460
+
461
+ if not exclude_frozen_parameters:
462
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
463
+
464
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
465
+
466
+ # recover shared parameters
467
+ for pair in zero_model_states[0].shared_params:
468
+ if pair[1] in state_dict:
469
+ state_dict[pair[0]] = state_dict[pair[1]]
470
+
471
+ return state_dict
472
+
473
+
474
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
475
+ """
476
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
477
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
478
+ via a model hub.
479
+
480
+ Args:
481
+ - ``checkpoint_dir``: path to the desired checkpoint folder
482
+ - ``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``
483
+ - ``exclude_frozen_parameters``: exclude frozen parameters
484
+
485
+ Returns:
486
+ - pytorch ``state_dict``
487
+
488
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
489
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
490
+ the checkpoint.
491
+
492
+ A typical usage might be ::
493
+
494
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
495
+ # do the training and checkpoint saving
496
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
497
+ model = model.cpu() # move to cpu
498
+ model.load_state_dict(state_dict)
499
+ # submit to model hub or save the model to share with others
500
+
501
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
502
+ application. i.e. you will need to re-initialize the deepspeed engine, since
503
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
504
+
505
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
506
+
507
+ """
508
+ if tag is None:
509
+ latest_path = os.path.join(checkpoint_dir, 'latest')
510
+ if os.path.isfile(latest_path):
511
+ with open(latest_path, 'r') as fd:
512
+ tag = fd.read().strip()
513
+ else:
514
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
515
+
516
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
517
+
518
+ if not os.path.isdir(ds_checkpoint_dir):
519
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
520
+
521
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
522
+
523
+
524
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
525
+ """
526
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
527
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
528
+
529
+ Args:
530
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
531
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
532
+ - ``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``
533
+ - ``exclude_frozen_parameters``: exclude frozen parameters
534
+ """
535
+
536
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
537
+ print(f"Saving fp32 state dict to {output_file}")
538
+ torch.save(state_dict, output_file)
539
+
540
+
541
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
542
+ """
543
+ 1. Put the provided model to cpu
544
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
545
+ 3. Load it into the provided model
546
+
547
+ Args:
548
+ - ``model``: the model object to update
549
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
550
+ - ``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``
551
+
552
+ Returns:
553
+ - ``model`: modified model
554
+
555
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
556
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
557
+ conveniently placed for you in the checkpoint folder.
558
+
559
+ A typical usage might be ::
560
+
561
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
562
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
563
+ # submit to model hub or save the model to share with others
564
+
565
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
566
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
567
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
568
+
569
+ """
570
+ logger.info(f"Extracting fp32 weights")
571
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
572
+
573
+ logger.info(f"Overwriting model with fp32 weights")
574
+ model = model.cpu()
575
+ model.load_state_dict(state_dict, strict=False)
576
+
577
+ return model
578
+
579
+
580
+ if __name__ == "__main__":
581
+
582
+ parser = argparse.ArgumentParser()
583
+ parser.add_argument("checkpoint_dir",
584
+ type=str,
585
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
586
+ parser.add_argument(
587
+ "output_file",
588
+ type=str,
589
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
590
+ parser.add_argument("-t",
591
+ "--tag",
592
+ type=str,
593
+ default=None,
594
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
595
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
596
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
597
+ args = parser.parse_args()
598
+
599
+ debug = args.debug
600
+
601
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
602
+ args.output_file,
603
+ tag=args.tag,
604
+ exclude_frozen_parameters=args.exclude_frozen_parameters)