Hugging Face's logo

Vividbot
/
Vivid-7B-Instruct-Lora-PTQF250-525

PEFT
Safetensors
mistral
Model card Files Community
dminhvu02 commited on
Commit
b165130
·
verified ·
1 Parent(s): 0ff6012

Upload folder using huggingface_hub

Browse files
Files changed (17) hide show
  1. README.md +10 -0
  2. adapter_config.json +26 -0
  3. adapter_model.bin +3 -0
  4. adapter_model.safetensors +3 -0
  5. config.json +43 -0
  6. global_step525/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  7. global_step525/mp_rank_00_model_states.pt +3 -0
  8. latest +1 -0
  9. non_lora_trainables.bin +3 -0
  10. rng_state.pth +3 -0
  11. scheduler.pt +3 -0
  12. special_tokens_map.json +24 -0
  13. tokenizer.model +3 -0
  14. tokenizer_config.json +45 -0
  15. trainer_state.json +3696 -0
  16. training_args.bin +3 -0
  17. zero_to_fp32.py +592 -0
README.md ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+ ### Framework versions
7
+
8
+ - PEFT 0.4.0
9
+
10
+ - PEFT 0.4.0
adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "/model_zoo/Vivid-7B-base",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 64,
11
+ "lora_dropout": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 32,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "v_proj",
18
+ "k_proj",
19
+ "down_proj",
20
+ "gate_proj",
21
+ "q_proj",
22
+ "up_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf246c435bb0761c494e9584c22a75d4f6f39822bf1de50757ea16deecd32a0b
3
+ size 167927754
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:460329d78814996d0825afd1e6be5faa13feb1966e1b749d471469532246268d
3
+ size 167832688
config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/model_zoo/Vivid-7B-base",
3
+ "architectures": [
4
+ "MistralForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bert_type": "qformer_layer:12",
8
+ "bos_token_id": 1,
9
+ "compress_type": "mean",
10
+ "eos_token_id": 2,
11
+ "freeze_mm_mlp_adapter": false,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 4096,
14
+ "image_aspect_ratio": "pad",
15
+ "image_grid_pinpoints": null,
16
+ "image_processor": "./llamavid/processor/intern-vit",
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 14336,
19
+ "max_position_embeddings": 32768,
20
+ "max_token": 4096,
21
+ "mm_hidden_size": 1024,
22
+ "mm_projector_type": "mlp2x_gelu",
23
+ "mm_use_im_patch_token": false,
24
+ "mm_use_im_start_end": false,
25
+ "mm_vision_select_feature": "patch",
26
+ "mm_vision_select_layer": -1,
27
+ "mm_vision_tower": "/model_zoo/intern-vit",
28
+ "model_type": "mistral",
29
+ "num_attention_heads": 32,
30
+ "num_hidden_layers": 32,
31
+ "num_key_value_heads": 8,
32
+ "num_query": 32,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_theta": 10000.0,
35
+ "sliding_window": 4096,
36
+ "tie_word_embeddings": false,
37
+ "torch_dtype": "bfloat16",
38
+ "transformers_version": "4.38.2",
39
+ "tune_mm_mlp_adapter": false,
40
+ "use_cache": false,
41
+ "use_mm_proj": true,
42
+ "vocab_size": 48384
43
+ }
global_step525/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30ad26a2b4aeadcd8f57d8eb5f3a21c4f9b791fe59ed5cba807543b60444df8c
3
+ size 7471050560
global_step525/mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4742a87baf94651655803534b535661ac260b2b4f1b8e1868633ee07a9c41394
3
+ size 1245738500
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step525
non_lora_trainables.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3ebdec47215a9bf60f43652cbfb5c4d3dc784812049117f1cb636133c313dfe4
3
+ size 1077631656
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab079d3977b35da0c70c2f86c7b434635b4e725a1b585958f017e16d6008b9c8
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3eb458428fb4d710cc6c3e0aba863084014be047e8d6d62df77ae7d341858cd2
3
+ size 1064
special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<unk>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d88bdadaa2a065aa7c6e18a4b5999ce4c76cec14d9fea882102e7b4931d7ef0
3
+ size 779539
tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
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
+ "additional_special_tokens": [],
32
+ "bos_token": "<s>",
33
+ "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{{ bos_token }}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '</s>'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
34
+ "clean_up_tokenization_spaces": false,
35
+ "eos_token": "</s>",
36
+ "legacy": true,
37
+ "model_max_length": 4096,
38
+ "pad_token": "<unk>",
39
+ "padding_side": "right",
40
+ "sp_model_kwargs": {},
41
+ "spaces_between_special_tokens": false,
42
+ "tokenizer_class": "LlamaTokenizer",
43
+ "unk_token": "<unk>",
44
+ "use_default_system_prompt": false
45
+ }
trainer_state.json ADDED
@@ -0,0 +1,3696 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.3879728419010669,
5
+ "eval_steps": 500,
6
+ "global_step": 525,
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": 1.9372913499539586,
14
+ "learning_rate": 2.4390243902439027e-06,
15
+ "loss": 1.6191,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.0,
20
+ "grad_norm": 1.8879048490473127,
21
+ "learning_rate": 4.8780487804878055e-06,
22
+ "loss": 1.6982,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.0,
27
+ "grad_norm": 1.857307355920545,
28
+ "learning_rate": 7.317073170731707e-06,
29
+ "loss": 1.6724,
30
+ "step": 3
31
+ },
32
+ {
33
+ "epoch": 0.0,
34
+ "grad_norm": 1.808250401091683,
35
+ "learning_rate": 9.756097560975611e-06,
36
+ "loss": 1.647,
37
+ "step": 4
38
+ },
39
+ {
40
+ "epoch": 0.0,
41
+ "grad_norm": 2.5133500505596453,
42
+ "learning_rate": 1.2195121951219513e-05,
43
+ "loss": 1.6079,
44
+ "step": 5
45
+ },
46
+ {
47
+ "epoch": 0.0,
48
+ "grad_norm": 1.2734146289947597,
49
+ "learning_rate": 1.4634146341463415e-05,
50
+ "loss": 1.5908,
51
+ "step": 6
52
+ },
53
+ {
54
+ "epoch": 0.01,
55
+ "grad_norm": 1.1812917040861377,
56
+ "learning_rate": 1.707317073170732e-05,
57
+ "loss": 1.5518,
58
+ "step": 7
59
+ },
60
+ {
61
+ "epoch": 0.01,
62
+ "grad_norm": 1.293637431287248,
63
+ "learning_rate": 1.9512195121951222e-05,
64
+ "loss": 1.5952,
65
+ "step": 8
66
+ },
67
+ {
68
+ "epoch": 0.01,
69
+ "grad_norm": 1.1620676440097686,
70
+ "learning_rate": 2.1951219512195124e-05,
71
+ "loss": 1.5493,
72
+ "step": 9
73
+ },
74
+ {
75
+ "epoch": 0.01,
76
+ "grad_norm": 1.3191260666446372,
77
+ "learning_rate": 2.4390243902439026e-05,
78
+ "loss": 1.5625,
79
+ "step": 10
80
+ },
81
+ {
82
+ "epoch": 0.01,
83
+ "grad_norm": 1.182981202097366,
84
+ "learning_rate": 2.682926829268293e-05,
85
+ "loss": 1.5498,
86
+ "step": 11
87
+ },
88
+ {
89
+ "epoch": 0.01,
90
+ "grad_norm": 1.0724491677903074,
91
+ "learning_rate": 2.926829268292683e-05,
92
+ "loss": 1.5547,
93
+ "step": 12
94
+ },
95
+ {
96
+ "epoch": 0.01,
97
+ "grad_norm": 0.9434780094091623,
98
+ "learning_rate": 3.170731707317073e-05,
99
+ "loss": 1.5327,
100
+ "step": 13
101
+ },
102
+ {
103
+ "epoch": 0.01,
104
+ "grad_norm": 1.0202543546064133,
105
+ "learning_rate": 3.414634146341464e-05,
106
+ "loss": 1.5933,
107
+ "step": 14
108
+ },
109
+ {
110
+ "epoch": 0.01,
111
+ "grad_norm": 0.996865818341891,
112
+ "learning_rate": 3.6585365853658535e-05,
113
+ "loss": 1.5796,
114
+ "step": 15
115
+ },
116
+ {
117
+ "epoch": 0.01,
118
+ "grad_norm": 0.9288105887086908,
119
+ "learning_rate": 3.9024390243902444e-05,
120
+ "loss": 1.4609,
121
+ "step": 16
122
+ },
123
+ {
124
+ "epoch": 0.01,
125
+ "grad_norm": 0.9726608694986103,
126
+ "learning_rate": 4.146341463414634e-05,
127
+ "loss": 1.5161,
128
+ "step": 17
129
+ },
130
+ {
131
+ "epoch": 0.01,
132
+ "grad_norm": 0.8619245275928736,
133
+ "learning_rate": 4.390243902439025e-05,
134
+ "loss": 1.5122,
135
+ "step": 18
136
+ },
137
+ {
138
+ "epoch": 0.01,
139
+ "grad_norm": 0.9215398746800475,
140
+ "learning_rate": 4.634146341463415e-05,
141
+ "loss": 1.5078,
142
+ "step": 19
143
+ },
144
+ {
145
+ "epoch": 0.01,
146
+ "grad_norm": 0.903097203515963,
147
+ "learning_rate": 4.878048780487805e-05,
148
+ "loss": 1.4502,
149
+ "step": 20
150
+ },
151
+ {
152
+ "epoch": 0.02,
153
+ "grad_norm": 0.8761498232482394,
154
+ "learning_rate": 5.121951219512195e-05,
155
+ "loss": 1.4893,
156
+ "step": 21
157
+ },
158
+ {
159
+ "epoch": 0.02,
160
+ "grad_norm": 0.8353385747464918,
161
+ "learning_rate": 5.365853658536586e-05,
162
+ "loss": 1.4717,
163
+ "step": 22
164
+ },
165
+ {
166
+ "epoch": 0.02,
167
+ "grad_norm": 0.8000291372477917,
168
+ "learning_rate": 5.6097560975609764e-05,
169
+ "loss": 1.481,
170
+ "step": 23
171
+ },
172
+ {
173
+ "epoch": 0.02,
174
+ "grad_norm": 0.8452088500727898,
175
+ "learning_rate": 5.853658536585366e-05,
176
+ "loss": 1.4644,
177
+ "step": 24
178
+ },
179
+ {
180
+ "epoch": 0.02,
181
+ "grad_norm": 0.8829309199222577,
182
+ "learning_rate": 6.097560975609756e-05,
183
+ "loss": 1.4868,
184
+ "step": 25
185
+ },
186
+ {
187
+ "epoch": 0.02,
188
+ "grad_norm": 0.8162327363449975,
189
+ "learning_rate": 6.341463414634146e-05,
190
+ "loss": 1.4883,
191
+ "step": 26
192
+ },
193
+ {
194
+ "epoch": 0.02,
195
+ "grad_norm": 0.7987925882960866,
196
+ "learning_rate": 6.585365853658538e-05,
197
+ "loss": 1.4268,
198
+ "step": 27
199
+ },
200
+ {
201
+ "epoch": 0.02,
202
+ "grad_norm": 0.7909140922467949,
203
+ "learning_rate": 6.829268292682928e-05,
204
+ "loss": 1.4873,
205
+ "step": 28
206
+ },
207
+ {
208
+ "epoch": 0.02,
209
+ "grad_norm": 0.7560592825415925,
210
+ "learning_rate": 7.073170731707317e-05,
211
+ "loss": 1.4116,
212
+ "step": 29
213
+ },
214
+ {
215
+ "epoch": 0.02,
216
+ "grad_norm": 0.7058796878894483,
217
+ "learning_rate": 7.317073170731707e-05,
218
+ "loss": 1.4023,
219
+ "step": 30
220
+ },
221
+ {
222
+ "epoch": 0.02,
223
+ "grad_norm": 0.7614550996113684,
224
+ "learning_rate": 7.560975609756099e-05,
225
+ "loss": 1.4312,
226
+ "step": 31
227
+ },
228
+ {
229
+ "epoch": 0.02,
230
+ "grad_norm": 0.7531993296256376,
231
+ "learning_rate": 7.804878048780489e-05,
232
+ "loss": 1.5024,
233
+ "step": 32
234
+ },
235
+ {
236
+ "epoch": 0.02,
237
+ "grad_norm": 0.7475795582718757,
238
+ "learning_rate": 8.048780487804879e-05,
239
+ "loss": 1.4363,
240
+ "step": 33
241
+ },
242
+ {
243
+ "epoch": 0.03,
244
+ "grad_norm": 0.7561530704205457,
245
+ "learning_rate": 8.292682926829268e-05,
246
+ "loss": 1.4873,
247
+ "step": 34
248
+ },
249
+ {
250
+ "epoch": 0.03,
251
+ "grad_norm": 0.7606234092420118,
252
+ "learning_rate": 8.53658536585366e-05,
253
+ "loss": 1.4204,
254
+ "step": 35
255
+ },
256
+ {
257
+ "epoch": 0.03,
258
+ "grad_norm": 0.7078849092381325,
259
+ "learning_rate": 8.78048780487805e-05,
260
+ "loss": 1.418,
261
+ "step": 36
262
+ },
263
+ {
264
+ "epoch": 0.03,
265
+ "grad_norm": 0.7583459620401868,
266
+ "learning_rate": 9.02439024390244e-05,
267
+ "loss": 1.4365,
268
+ "step": 37
269
+ },
270
+ {
271
+ "epoch": 0.03,
272
+ "grad_norm": 0.6479336734201823,
273
+ "learning_rate": 9.26829268292683e-05,
274
+ "loss": 1.3911,
275
+ "step": 38
276
+ },
277
+ {
278
+ "epoch": 0.03,
279
+ "grad_norm": 0.7138445522030739,
280
+ "learning_rate": 9.51219512195122e-05,
281
+ "loss": 1.4287,
282
+ "step": 39
283
+ },
284
+ {
285
+ "epoch": 0.03,
286
+ "grad_norm": 0.6772243082870256,
287
+ "learning_rate": 9.75609756097561e-05,
288
+ "loss": 1.3779,
289
+ "step": 40
290
+ },
291
+ {
292
+ "epoch": 0.03,
293
+ "grad_norm": 0.7001769060106223,
294
+ "learning_rate": 0.0001,
295
+ "loss": 1.3623,
296
+ "step": 41
297
+ },
298
+ {
299
+ "epoch": 0.03,
300
+ "grad_norm": 0.6593306891481673,
301
+ "learning_rate": 9.999985665852258e-05,
302
+ "loss": 1.3745,
303
+ "step": 42
304
+ },
305
+ {
306
+ "epoch": 0.03,
307
+ "grad_norm": 0.7111159325021309,
308
+ "learning_rate": 9.999942663491213e-05,
309
+ "loss": 1.3799,
310
+ "step": 43
311
+ },
312
+ {
313
+ "epoch": 0.03,
314
+ "grad_norm": 0.7023696510759943,
315
+ "learning_rate": 9.999870993163431e-05,
316
+ "loss": 1.4399,
317
+ "step": 44
318
+ },
319
+ {
320
+ "epoch": 0.03,
321
+ "grad_norm": 0.6736689337950041,
322
+ "learning_rate": 9.999770655279843e-05,
323
+ "loss": 1.4106,
324
+ "step": 45
325
+ },
326
+ {
327
+ "epoch": 0.03,
328
+ "grad_norm": 0.6746379997849087,
329
+ "learning_rate": 9.999641650415752e-05,
330
+ "loss": 1.4409,
331
+ "step": 46
332
+ },
333
+ {
334
+ "epoch": 0.03,
335
+ "grad_norm": 0.6615592598917496,
336
+ "learning_rate": 9.99948397931083e-05,
337
+ "loss": 1.3984,
338
+ "step": 47
339
+ },
340
+ {
341
+ "epoch": 0.04,
342
+ "grad_norm": 0.6538222984665192,
343
+ "learning_rate": 9.999297642869105e-05,
344
+ "loss": 1.4031,
345
+ "step": 48
346
+ },
347
+ {
348
+ "epoch": 0.04,
349
+ "grad_norm": 0.6129031974400467,
350
+ "learning_rate": 9.999082642158973e-05,
351
+ "loss": 1.396,
352
+ "step": 49
353
+ },
354
+ {
355
+ "epoch": 0.04,
356
+ "grad_norm": 0.6148818612628825,
357
+ "learning_rate": 9.998838978413168e-05,
358
+ "loss": 1.3574,
359
+ "step": 50
360
+ },
361
+ {
362
+ "epoch": 0.04,
363
+ "grad_norm": 0.6869612852614861,
364
+ "learning_rate": 9.99856665302878e-05,
365
+ "loss": 1.3762,
366
+ "step": 51
367
+ },
368
+ {
369
+ "epoch": 0.04,
370
+ "grad_norm": 0.7410178778694718,
371
+ "learning_rate": 9.998265667567226e-05,
372
+ "loss": 1.3481,
373
+ "step": 52
374
+ },
375
+ {
376
+ "epoch": 0.04,
377
+ "grad_norm": 0.6380516168920353,
378
+ "learning_rate": 9.997936023754257e-05,
379
+ "loss": 1.3513,
380
+ "step": 53
381
+ },
382
+ {
383
+ "epoch": 0.04,
384
+ "grad_norm": 0.6192351492724488,
385
+ "learning_rate": 9.997577723479938e-05,
386
+ "loss": 1.3662,
387
+ "step": 54
388
+ },
389
+ {
390
+ "epoch": 0.04,
391
+ "grad_norm": 0.633774941417789,
392
+ "learning_rate": 9.997190768798639e-05,
393
+ "loss": 1.3457,
394
+ "step": 55
395
+ },
396
+ {
397
+ "epoch": 0.04,
398
+ "grad_norm": 0.6016840416873676,
399
+ "learning_rate": 9.996775161929027e-05,
400
+ "loss": 1.3877,
401
+ "step": 56
402
+ },
403
+ {
404
+ "epoch": 0.04,
405
+ "grad_norm": 0.638026596140304,
406
+ "learning_rate": 9.99633090525405e-05,
407
+ "loss": 1.3892,
408
+ "step": 57
409
+ },
410
+ {
411
+ "epoch": 0.04,
412
+ "grad_norm": 0.5934027179170136,
413
+ "learning_rate": 9.995858001320926e-05,
414
+ "loss": 1.3223,
415
+ "step": 58
416
+ },
417
+ {
418
+ "epoch": 0.04,
419
+ "grad_norm": 0.6143195436309025,
420
+ "learning_rate": 9.995356452841122e-05,
421
+ "loss": 1.3862,
422
+ "step": 59
423
+ },
424
+ {
425
+ "epoch": 0.04,
426
+ "grad_norm": 0.6076935190423259,
427
+ "learning_rate": 9.994826262690347e-05,
428
+ "loss": 1.3584,
429
+ "step": 60
430
+ },
431
+ {
432
+ "epoch": 0.05,
433
+ "grad_norm": 0.6239965555110781,
434
+ "learning_rate": 9.994267433908533e-05,
435
+ "loss": 1.2771,
436
+ "step": 61
437
+ },
438
+ {
439
+ "epoch": 0.05,
440
+ "grad_norm": 0.5469871219286494,
441
+ "learning_rate": 9.99367996969981e-05,
442
+ "loss": 1.3579,
443
+ "step": 62
444
+ },
445
+ {
446
+ "epoch": 0.05,
447
+ "grad_norm": 0.5975500231663011,
448
+ "learning_rate": 9.9930638734325e-05,
449
+ "loss": 1.3872,
450
+ "step": 63
451
+ },
452
+ {
453
+ "epoch": 0.05,
454
+ "grad_norm": 0.6160102854784424,
455
+ "learning_rate": 9.992419148639087e-05,
456
+ "loss": 1.3831,
457
+ "step": 64
458
+ },
459
+ {
460
+ "epoch": 0.05,
461
+ "grad_norm": 0.5815474376554662,
462
+ "learning_rate": 9.991745799016206e-05,
463
+ "loss": 1.3745,
464
+ "step": 65
465
+ },
466
+ {
467
+ "epoch": 0.05,
468
+ "grad_norm": 0.5994591436721235,
469
+ "learning_rate": 9.991043828424612e-05,
470
+ "loss": 1.396,
471
+ "step": 66
472
+ },
473
+ {
474
+ "epoch": 0.05,
475
+ "grad_norm": 0.5896523240727669,
476
+ "learning_rate": 9.990313240889167e-05,
477
+ "loss": 1.3608,
478
+ "step": 67
479
+ },
480
+ {
481
+ "epoch": 0.05,
482
+ "grad_norm": 0.6062100949214702,
483
+ "learning_rate": 9.989554040598807e-05,
484
+ "loss": 1.2996,
485
+ "step": 68
486
+ },
487
+ {
488
+ "epoch": 0.05,
489
+ "grad_norm": 0.5941049216825265,
490
+ "learning_rate": 9.988766231906533e-05,
491
+ "loss": 1.4106,
492
+ "step": 69
493
+ },
494
+ {
495
+ "epoch": 0.05,
496
+ "grad_norm": 0.5604128113953568,
497
+ "learning_rate": 9.987949819329365e-05,
498
+ "loss": 1.3931,
499
+ "step": 70
500
+ },
501
+ {
502
+ "epoch": 0.05,
503
+ "grad_norm": 0.5519277490096212,
504
+ "learning_rate": 9.98710480754834e-05,
505
+ "loss": 1.3691,
506
+ "step": 71
507
+ },
508
+ {
509
+ "epoch": 0.05,
510
+ "grad_norm": 0.5900021330626725,
511
+ "learning_rate": 9.986231201408467e-05,
512
+ "loss": 1.4058,
513
+ "step": 72
514
+ },
515
+ {
516
+ "epoch": 0.05,
517
+ "grad_norm": 0.5699754681306506,
518
+ "learning_rate": 9.985329005918702e-05,
519
+ "loss": 1.355,
520
+ "step": 73
521
+ },
522
+ {
523
+ "epoch": 0.05,
524
+ "grad_norm": 0.593149750992695,
525
+ "learning_rate": 9.98439822625193e-05,
526
+ "loss": 1.3545,
527
+ "step": 74
528
+ },
529
+ {
530
+ "epoch": 0.06,
531
+ "grad_norm": 0.5824626045065218,
532
+ "learning_rate": 9.983438867744923e-05,
533
+ "loss": 1.3896,
534
+ "step": 75
535
+ },
536
+ {
537
+ "epoch": 0.06,
538
+ "grad_norm": 0.5900786393120402,
539
+ "learning_rate": 9.982450935898316e-05,
540
+ "loss": 1.3716,
541
+ "step": 76
542
+ },
543
+ {
544
+ "epoch": 0.06,
545
+ "grad_norm": 0.5688141367114475,
546
+ "learning_rate": 9.981434436376572e-05,
547
+ "loss": 1.3921,
548
+ "step": 77
549
+ },
550
+ {
551
+ "epoch": 0.06,
552
+ "grad_norm": 0.557565379686218,
553
+ "learning_rate": 9.980389375007955e-05,
554
+ "loss": 1.3506,
555
+ "step": 78
556
+ },
557
+ {
558
+ "epoch": 0.06,
559
+ "grad_norm": 0.5740715320740841,
560
+ "learning_rate": 9.979315757784488e-05,
561
+ "loss": 1.2917,
562
+ "step": 79
563
+ },
564
+ {
565
+ "epoch": 0.06,
566
+ "grad_norm": 0.5717745274109229,
567
+ "learning_rate": 9.97821359086193e-05,
568
+ "loss": 1.3154,
569
+ "step": 80
570
+ },
571
+ {
572
+ "epoch": 0.06,
573
+ "grad_norm": 0.609615875256831,
574
+ "learning_rate": 9.977082880559725e-05,
575
+ "loss": 1.3328,
576
+ "step": 81
577
+ },
578
+ {
579
+ "epoch": 0.06,
580
+ "grad_norm": 0.5777864702702744,
581
+ "learning_rate": 9.975923633360985e-05,
582
+ "loss": 1.3599,
583
+ "step": 82
584
+ },
585
+ {
586
+ "epoch": 0.06,
587
+ "grad_norm": 0.575948499045498,
588
+ "learning_rate": 9.974735855912436e-05,
589
+ "loss": 1.4038,
590
+ "step": 83
591
+ },
592
+ {
593
+ "epoch": 0.06,
594
+ "grad_norm": 0.550693122074238,
595
+ "learning_rate": 9.97351955502439e-05,
596
+ "loss": 1.3203,
597
+ "step": 84
598
+ },
599
+ {
600
+ "epoch": 0.06,
601
+ "grad_norm": 0.5561601283605949,
602
+ "learning_rate": 9.972274737670701e-05,
603
+ "loss": 1.3477,
604
+ "step": 85
605
+ },
606
+ {
607
+ "epoch": 0.06,
608
+ "grad_norm": 0.5601251180421914,
609
+ "learning_rate": 9.971001410988728e-05,
610
+ "loss": 1.333,
611
+ "step": 86
612
+ },
613
+ {
614
+ "epoch": 0.06,
615
+ "grad_norm": 0.6207004745075507,
616
+ "learning_rate": 9.969699582279292e-05,
617
+ "loss": 1.4048,
618
+ "step": 87
619
+ },
620
+ {
621
+ "epoch": 0.07,
622
+ "grad_norm": 0.5475040554880181,
623
+ "learning_rate": 9.968369259006634e-05,
624
+ "loss": 1.3208,
625
+ "step": 88
626
+ },
627
+ {
628
+ "epoch": 0.07,
629
+ "grad_norm": 0.6054670378552847,
630
+ "learning_rate": 9.967010448798375e-05,
631
+ "loss": 1.4131,
632
+ "step": 89
633
+ },
634
+ {
635
+ "epoch": 0.07,
636
+ "grad_norm": 0.5486336748948858,
637
+ "learning_rate": 9.965623159445471e-05,
638
+ "loss": 1.3843,
639
+ "step": 90
640
+ },
641
+ {
642
+ "epoch": 0.07,
643
+ "grad_norm": 0.585603864758025,
644
+ "learning_rate": 9.964207398902163e-05,
645
+ "loss": 1.3186,
646
+ "step": 91
647
+ },
648
+ {
649
+ "epoch": 0.07,
650
+ "grad_norm": 0.5412960874208915,
651
+ "learning_rate": 9.96276317528594e-05,
652
+ "loss": 1.2861,
653
+ "step": 92
654
+ },
655
+ {
656
+ "epoch": 0.07,
657
+ "grad_norm": 0.5442105369162202,
658
+ "learning_rate": 9.96129049687749e-05,
659
+ "loss": 1.3262,
660
+ "step": 93
661
+ },
662
+ {
663
+ "epoch": 0.07,
664
+ "grad_norm": 0.5816978676309428,
665
+ "learning_rate": 9.959789372120649e-05,
666
+ "loss": 1.3279,
667
+ "step": 94
668
+ },
669
+ {
670
+ "epoch": 0.07,
671
+ "grad_norm": 0.5557519862862452,
672
+ "learning_rate": 9.958259809622352e-05,
673
+ "loss": 1.3672,
674
+ "step": 95
675
+ },
676
+ {
677
+ "epoch": 0.07,
678
+ "grad_norm": 0.5666965195077155,
679
+ "learning_rate": 9.956701818152591e-05,
680
+ "loss": 1.3203,
681
+ "step": 96
682
+ },
683
+ {
684
+ "epoch": 0.07,
685
+ "grad_norm": 0.5354511291609182,
686
+ "learning_rate": 9.955115406644356e-05,
687
+ "loss": 1.3081,
688
+ "step": 97
689
+ },
690
+ {
691
+ "epoch": 0.07,
692
+ "grad_norm": 0.5685729288533676,
693
+ "learning_rate": 9.953500584193592e-05,
694
+ "loss": 1.3452,
695
+ "step": 98
696
+ },
697
+ {
698
+ "epoch": 0.07,
699
+ "grad_norm": 0.5922446508548838,
700
+ "learning_rate": 9.95185736005914e-05,
701
+ "loss": 1.3682,
702
+ "step": 99
703
+ },
704
+ {
705
+ "epoch": 0.07,
706
+ "grad_norm": 0.5837642463681222,
707
+ "learning_rate": 9.950185743662685e-05,
708
+ "loss": 1.3691,
709
+ "step": 100
710
+ },
711
+ {
712
+ "epoch": 0.07,
713
+ "grad_norm": 0.5761448966076219,
714
+ "learning_rate": 9.948485744588709e-05,
715
+ "loss": 1.3281,
716
+ "step": 101
717
+ },
718
+ {
719
+ "epoch": 0.08,
720
+ "grad_norm": 0.553490008569796,
721
+ "learning_rate": 9.946757372584423e-05,
722
+ "loss": 1.292,
723
+ "step": 102
724
+ },
725
+ {
726
+ "epoch": 0.08,
727
+ "grad_norm": 0.5686477341821499,
728
+ "learning_rate": 9.945000637559727e-05,
729
+ "loss": 1.3486,
730
+ "step": 103
731
+ },
732
+ {
733
+ "epoch": 0.08,
734
+ "grad_norm": 0.5772487636958804,
735
+ "learning_rate": 9.943215549587138e-05,
736
+ "loss": 1.3425,
737
+ "step": 104
738
+ },
739
+ {
740
+ "epoch": 0.08,
741
+ "grad_norm": 0.5758207849461601,
742
+ "learning_rate": 9.941402118901744e-05,
743
+ "loss": 1.3701,
744
+ "step": 105
745
+ },
746
+ {
747
+ "epoch": 0.08,
748
+ "grad_norm": 0.5582099537521159,
749
+ "learning_rate": 9.939560355901136e-05,
750
+ "loss": 1.3794,
751
+ "step": 106
752
+ },
753
+ {
754
+ "epoch": 0.08,
755
+ "grad_norm": 0.5336197399728324,
756
+ "learning_rate": 9.937690271145354e-05,
757
+ "loss": 1.3179,
758
+ "step": 107
759
+ },
760
+ {
761
+ "epoch": 0.08,
762
+ "grad_norm": 0.539037656457371,
763
+ "learning_rate": 9.935791875356832e-05,
764
+ "loss": 1.3071,
765
+ "step": 108
766
+ },
767
+ {
768
+ "epoch": 0.08,
769
+ "grad_norm": 0.5584770963502244,
770
+ "learning_rate": 9.933865179420321e-05,
771
+ "loss": 1.3945,
772
+ "step": 109
773
+ },
774
+ {
775
+ "epoch": 0.08,
776
+ "grad_norm": 0.5364047388288558,
777
+ "learning_rate": 9.931910194382837e-05,
778
+ "loss": 1.3462,
779
+ "step": 110
780
+ },
781
+ {
782
+ "epoch": 0.08,
783
+ "grad_norm": 0.5956933567804931,
784
+ "learning_rate": 9.929926931453599e-05,
785
+ "loss": 1.2585,
786
+ "step": 111
787
+ },
788
+ {
789
+ "epoch": 0.08,
790
+ "grad_norm": 0.5548298244830802,
791
+ "learning_rate": 9.927915402003964e-05,
792
+ "loss": 1.3765,
793
+ "step": 112
794
+ },
795
+ {
796
+ "epoch": 0.08,
797
+ "grad_norm": 0.5528131728204222,
798
+ "learning_rate": 9.92587561756735e-05,
799
+ "loss": 1.3452,
800
+ "step": 113
801
+ },
802
+ {
803
+ "epoch": 0.08,
804
+ "grad_norm": 0.5181397205586854,
805
+ "learning_rate": 9.92380758983919e-05,
806
+ "loss": 1.2671,
807
+ "step": 114
808
+ },
809
+ {
810
+ "epoch": 0.08,
811
+ "grad_norm": 0.5429954425262675,
812
+ "learning_rate": 9.921711330676848e-05,
813
+ "loss": 1.3574,
814
+ "step": 115
815
+ },
816
+ {
817
+ "epoch": 0.09,
818
+ "grad_norm": 0.5523231773869766,
819
+ "learning_rate": 9.919586852099562e-05,
820
+ "loss": 1.3184,
821
+ "step": 116
822
+ },
823
+ {
824
+ "epoch": 0.09,
825
+ "grad_norm": 0.5583959107787768,
826
+ "learning_rate": 9.917434166288364e-05,
827
+ "loss": 1.3442,
828
+ "step": 117
829
+ },
830
+ {
831
+ "epoch": 0.09,
832
+ "grad_norm": 0.5850081526075311,
833
+ "learning_rate": 9.915253285586024e-05,
834
+ "loss": 1.3477,
835
+ "step": 118
836
+ },
837
+ {
838
+ "epoch": 0.09,
839
+ "grad_norm": 0.5498743192645993,
840
+ "learning_rate": 9.913044222496966e-05,
841
+ "loss": 1.3398,
842
+ "step": 119
843
+ },
844
+ {
845
+ "epoch": 0.09,
846
+ "grad_norm": 0.5853233345937257,
847
+ "learning_rate": 9.910806989687206e-05,
848
+ "loss": 1.3276,
849
+ "step": 120
850
+ },
851
+ {
852
+ "epoch": 0.09,
853
+ "grad_norm": 0.559389561256856,
854
+ "learning_rate": 9.908541599984276e-05,
855
+ "loss": 1.3462,
856
+ "step": 121
857
+ },
858
+ {
859
+ "epoch": 0.09,
860
+ "grad_norm": 0.5298088621667728,
861
+ "learning_rate": 9.906248066377143e-05,
862
+ "loss": 1.2568,
863
+ "step": 122
864
+ },
865
+ {
866
+ "epoch": 0.09,
867
+ "grad_norm": 0.5731884986496186,
868
+ "learning_rate": 9.903926402016153e-05,
869
+ "loss": 1.3394,
870
+ "step": 123
871
+ },
872
+ {
873
+ "epoch": 0.09,
874
+ "grad_norm": 0.5549155957971303,
875
+ "learning_rate": 9.901576620212933e-05,
876
+ "loss": 1.311,
877
+ "step": 124
878
+ },
879
+ {
880
+ "epoch": 0.09,
881
+ "grad_norm": 0.5620092141236146,
882
+ "learning_rate": 9.899198734440335e-05,
883
+ "loss": 1.291,
884
+ "step": 125
885
+ },
886
+ {
887
+ "epoch": 0.09,
888
+ "grad_norm": 0.5405164924320079,
889
+ "learning_rate": 9.896792758332341e-05,
890
+ "loss": 1.248,
891
+ "step": 126
892
+ },
893
+ {
894
+ "epoch": 0.09,
895
+ "grad_norm": 0.5602202105737174,
896
+ "learning_rate": 9.894358705684002e-05,
897
+ "loss": 1.3115,
898
+ "step": 127
899
+ },
900
+ {
901
+ "epoch": 0.09,
902
+ "grad_norm": 0.5580296998093701,
903
+ "learning_rate": 9.891896590451344e-05,
904
+ "loss": 1.2947,
905
+ "step": 128
906
+ },
907
+ {
908
+ "epoch": 0.1,
909
+ "grad_norm": 0.5755635897570144,
910
+ "learning_rate": 9.889406426751296e-05,
911
+ "loss": 1.3086,
912
+ "step": 129
913
+ },
914
+ {
915
+ "epoch": 0.1,
916
+ "grad_norm": 0.6025851962917577,
917
+ "learning_rate": 9.886888228861608e-05,
918
+ "loss": 1.3447,
919
+ "step": 130
920
+ },
921
+ {
922
+ "epoch": 0.1,
923
+ "grad_norm": 0.5660419268974345,
924
+ "learning_rate": 9.88434201122077e-05,
925
+ "loss": 1.3232,
926
+ "step": 131
927
+ },
928
+ {
929
+ "epoch": 0.1,
930
+ "grad_norm": 0.5495648120402916,
931
+ "learning_rate": 9.881767788427925e-05,
932
+ "loss": 1.3096,
933
+ "step": 132
934
+ },
935
+ {
936
+ "epoch": 0.1,
937
+ "grad_norm": 0.5577872798163368,
938
+ "learning_rate": 9.879165575242787e-05,
939
+ "loss": 1.291,
940
+ "step": 133
941
+ },
942
+ {
943
+ "epoch": 0.1,
944
+ "grad_norm": 0.5540620803629338,
945
+ "learning_rate": 9.876535386585561e-05,
946
+ "loss": 1.335,
947
+ "step": 134
948
+ },
949
+ {
950
+ "epoch": 0.1,
951
+ "grad_norm": 0.5573425731012122,
952
+ "learning_rate": 9.873877237536853e-05,
953
+ "loss": 1.2327,
954
+ "step": 135
955
+ },
956
+ {
957
+ "epoch": 0.1,
958
+ "grad_norm": 0.5827857038389533,
959
+ "learning_rate": 9.871191143337582e-05,
960
+ "loss": 1.3333,
961
+ "step": 136
962
+ },
963
+ {
964
+ "epoch": 0.1,
965
+ "grad_norm": 0.5897883061496167,
966
+ "learning_rate": 9.868477119388896e-05,
967
+ "loss": 1.3076,
968
+ "step": 137
969
+ },
970
+ {
971
+ "epoch": 0.1,
972
+ "grad_norm": 0.5800275384221499,
973
+ "learning_rate": 9.865735181252085e-05,
974
+ "loss": 1.3188,
975
+ "step": 138
976
+ },
977
+ {
978
+ "epoch": 0.1,
979
+ "grad_norm": 0.5605765677262206,
980
+ "learning_rate": 9.862965344648485e-05,
981
+ "loss": 1.3086,
982
+ "step": 139
983
+ },
984
+ {
985
+ "epoch": 0.1,
986
+ "grad_norm": 0.5432447170586258,
987
+ "learning_rate": 9.860167625459398e-05,
988
+ "loss": 1.2861,
989
+ "step": 140
990
+ },
991
+ {
992
+ "epoch": 0.1,
993
+ "grad_norm": 0.5687257803544524,
994
+ "learning_rate": 9.85734203972599e-05,
995
+ "loss": 1.2839,
996
+ "step": 141
997
+ },
998
+ {
999
+ "epoch": 0.1,
1000
+ "grad_norm": 0.5475328993701518,
1001
+ "learning_rate": 9.854488603649206e-05,
1002
+ "loss": 1.3169,
1003
+ "step": 142
1004
+ },
1005
+ {
1006
+ "epoch": 0.11,
1007
+ "grad_norm": 0.5408143803639806,
1008
+ "learning_rate": 9.851607333589677e-05,
1009
+ "loss": 1.3374,
1010
+ "step": 143
1011
+ },
1012
+ {
1013
+ "epoch": 0.11,
1014
+ "grad_norm": 0.5350053494827027,
1015
+ "learning_rate": 9.848698246067623e-05,
1016
+ "loss": 1.2888,
1017
+ "step": 144
1018
+ },
1019
+ {
1020
+ "epoch": 0.11,
1021
+ "grad_norm": 0.5642075781884446,
1022
+ "learning_rate": 9.84576135776276e-05,
1023
+ "loss": 1.3105,
1024
+ "step": 145
1025
+ },
1026
+ {
1027
+ "epoch": 0.11,
1028
+ "grad_norm": 0.5725161088840623,
1029
+ "learning_rate": 9.842796685514203e-05,
1030
+ "loss": 1.3516,
1031
+ "step": 146
1032
+ },
1033
+ {
1034
+ "epoch": 0.11,
1035
+ "grad_norm": 0.5837888943455876,
1036
+ "learning_rate": 9.839804246320375e-05,
1037
+ "loss": 1.2871,
1038
+ "step": 147
1039
+ },
1040
+ {
1041
+ "epoch": 0.11,
1042
+ "grad_norm": 0.5833329842842448,
1043
+ "learning_rate": 9.836784057338899e-05,
1044
+ "loss": 1.3232,
1045
+ "step": 148
1046
+ },
1047
+ {
1048
+ "epoch": 0.11,
1049
+ "grad_norm": 0.5244172538585695,
1050
+ "learning_rate": 9.833736135886512e-05,
1051
+ "loss": 1.2568,
1052
+ "step": 149
1053
+ },
1054
+ {
1055
+ "epoch": 0.11,
1056
+ "grad_norm": 0.5163576076330887,
1057
+ "learning_rate": 9.830660499438955e-05,
1058
+ "loss": 1.2759,
1059
+ "step": 150
1060
+ },
1061
+ {
1062
+ "epoch": 0.11,
1063
+ "grad_norm": 0.5617840717093857,
1064
+ "learning_rate": 9.827557165630879e-05,
1065
+ "loss": 1.2524,
1066
+ "step": 151
1067
+ },
1068
+ {
1069
+ "epoch": 0.11,
1070
+ "grad_norm": 0.547220410155329,
1071
+ "learning_rate": 9.824426152255741e-05,
1072
+ "loss": 1.312,
1073
+ "step": 152
1074
+ },
1075
+ {
1076
+ "epoch": 0.11,
1077
+ "grad_norm": 0.5715922980351898,
1078
+ "learning_rate": 9.821267477265705e-05,
1079
+ "loss": 1.335,
1080
+ "step": 153
1081
+ },
1082
+ {
1083
+ "epoch": 0.11,
1084
+ "grad_norm": 0.5626236612178414,
1085
+ "learning_rate": 9.818081158771538e-05,
1086
+ "loss": 1.3633,
1087
+ "step": 154
1088
+ },
1089
+ {
1090
+ "epoch": 0.11,
1091
+ "grad_norm": 0.556817713740677,
1092
+ "learning_rate": 9.814867215042502e-05,
1093
+ "loss": 1.3345,
1094
+ "step": 155
1095
+ },
1096
+ {
1097
+ "epoch": 0.12,
1098
+ "grad_norm": 0.5658424328358594,
1099
+ "learning_rate": 9.811625664506259e-05,
1100
+ "loss": 1.3325,
1101
+ "step": 156
1102
+ },
1103
+ {
1104
+ "epoch": 0.12,
1105
+ "grad_norm": 0.5518987143292007,
1106
+ "learning_rate": 9.808356525748748e-05,
1107
+ "loss": 1.3179,
1108
+ "step": 157
1109
+ },
1110
+ {
1111
+ "epoch": 0.12,
1112
+ "grad_norm": 0.5509045139485853,
1113
+ "learning_rate": 9.805059817514101e-05,
1114
+ "loss": 1.3276,
1115
+ "step": 158
1116
+ },
1117
+ {
1118
+ "epoch": 0.12,
1119
+ "grad_norm": 0.5612999607711056,
1120
+ "learning_rate": 9.801735558704517e-05,
1121
+ "loss": 1.2192,
1122
+ "step": 159
1123
+ },
1124
+ {
1125
+ "epoch": 0.12,
1126
+ "grad_norm": 0.530326353544212,
1127
+ "learning_rate": 9.798383768380164e-05,
1128
+ "loss": 1.2988,
1129
+ "step": 160
1130
+ },
1131
+ {
1132
+ "epoch": 0.12,
1133
+ "grad_norm": 0.5524425336112486,
1134
+ "learning_rate": 9.795004465759065e-05,
1135
+ "loss": 1.2622,
1136
+ "step": 161
1137
+ },
1138
+ {
1139
+ "epoch": 0.12,
1140
+ "grad_norm": 0.5121240819278214,
1141
+ "learning_rate": 9.791597670216989e-05,
1142
+ "loss": 1.2603,
1143
+ "step": 162
1144
+ },
1145
+ {
1146
+ "epoch": 0.12,
1147
+ "grad_norm": 0.5262701595678754,
1148
+ "learning_rate": 9.78816340128734e-05,
1149
+ "loss": 1.22,
1150
+ "step": 163
1151
+ },
1152
+ {
1153
+ "epoch": 0.12,
1154
+ "grad_norm": 0.5866254674193113,
1155
+ "learning_rate": 9.784701678661045e-05,
1156
+ "loss": 1.311,
1157
+ "step": 164
1158
+ },
1159
+ {
1160
+ "epoch": 0.12,
1161
+ "grad_norm": 0.567120419528464,
1162
+ "learning_rate": 9.781212522186443e-05,
1163
+ "loss": 1.3145,
1164
+ "step": 165
1165
+ },
1166
+ {
1167
+ "epoch": 0.12,
1168
+ "grad_norm": 0.5704512174009239,
1169
+ "learning_rate": 9.777695951869164e-05,
1170
+ "loss": 1.2612,
1171
+ "step": 166
1172
+ },
1173
+ {
1174
+ "epoch": 0.12,
1175
+ "grad_norm": 0.5359884622353506,
1176
+ "learning_rate": 9.774151987872027e-05,
1177
+ "loss": 1.2117,
1178
+ "step": 167
1179
+ },
1180
+ {
1181
+ "epoch": 0.12,
1182
+ "grad_norm": 0.5772321074843504,
1183
+ "learning_rate": 9.770580650514914e-05,
1184
+ "loss": 1.3525,
1185
+ "step": 168
1186
+ },
1187
+ {
1188
+ "epoch": 0.12,
1189
+ "grad_norm": 0.5316876920831217,
1190
+ "learning_rate": 9.766981960274653e-05,
1191
+ "loss": 1.3442,
1192
+ "step": 169
1193
+ },
1194
+ {
1195
+ "epoch": 0.13,
1196
+ "grad_norm": 0.5622203218145027,
1197
+ "learning_rate": 9.763355937784909e-05,
1198
+ "loss": 1.2964,
1199
+ "step": 170
1200
+ },
1201
+ {
1202
+ "epoch": 0.13,
1203
+ "grad_norm": 0.5614932814360857,
1204
+ "learning_rate": 9.759702603836059e-05,
1205
+ "loss": 1.3389,
1206
+ "step": 171
1207
+ },
1208
+ {
1209
+ "epoch": 0.13,
1210
+ "grad_norm": 0.568962837143467,
1211
+ "learning_rate": 9.756021979375071e-05,
1212
+ "loss": 1.3174,
1213
+ "step": 172
1214
+ },
1215
+ {
1216
+ "epoch": 0.13,
1217
+ "grad_norm": 0.5382419139994956,
1218
+ "learning_rate": 9.752314085505395e-05,
1219
+ "loss": 1.3125,
1220
+ "step": 173
1221
+ },
1222
+ {
1223
+ "epoch": 0.13,
1224
+ "grad_norm": 0.5677837729549118,
1225
+ "learning_rate": 9.748578943486828e-05,
1226
+ "loss": 1.2871,
1227
+ "step": 174
1228
+ },
1229
+ {
1230
+ "epoch": 0.13,
1231
+ "grad_norm": 0.5602612877442024,
1232
+ "learning_rate": 9.744816574735405e-05,
1233
+ "loss": 1.3438,
1234
+ "step": 175
1235
+ },
1236
+ {
1237
+ "epoch": 0.13,
1238
+ "grad_norm": 0.5735194400650546,
1239
+ "learning_rate": 9.74102700082326e-05,
1240
+ "loss": 1.3208,
1241
+ "step": 176
1242
+ },
1243
+ {
1244
+ "epoch": 0.13,
1245
+ "grad_norm": 0.5670876099448275,
1246
+ "learning_rate": 9.737210243478521e-05,
1247
+ "loss": 1.2969,
1248
+ "step": 177
1249
+ },
1250
+ {
1251
+ "epoch": 0.13,
1252
+ "grad_norm": 0.5450536272385241,
1253
+ "learning_rate": 9.733366324585175e-05,
1254
+ "loss": 1.2673,
1255
+ "step": 178
1256
+ },
1257
+ {
1258
+ "epoch": 0.13,
1259
+ "grad_norm": 0.5340701964695135,
1260
+ "learning_rate": 9.72949526618294e-05,
1261
+ "loss": 1.3403,
1262
+ "step": 179
1263
+ },
1264
+ {
1265
+ "epoch": 0.13,
1266
+ "grad_norm": 0.5422933717116616,
1267
+ "learning_rate": 9.725597090467144e-05,
1268
+ "loss": 1.2539,
1269
+ "step": 180
1270
+ },
1271
+ {
1272
+ "epoch": 0.13,
1273
+ "grad_norm": 0.5680150103490264,
1274
+ "learning_rate": 9.721671819788602e-05,
1275
+ "loss": 1.3149,
1276
+ "step": 181
1277
+ },
1278
+ {
1279
+ "epoch": 0.13,
1280
+ "grad_norm": 0.560101859043945,
1281
+ "learning_rate": 9.717719476653475e-05,
1282
+ "loss": 1.321,
1283
+ "step": 182
1284
+ },
1285
+ {
1286
+ "epoch": 0.14,
1287
+ "grad_norm": 0.5267278121510764,
1288
+ "learning_rate": 9.71374008372315e-05,
1289
+ "loss": 1.2227,
1290
+ "step": 183
1291
+ },
1292
+ {
1293
+ "epoch": 0.14,
1294
+ "grad_norm": 0.5687530339596342,
1295
+ "learning_rate": 9.709733663814113e-05,
1296
+ "loss": 1.3159,
1297
+ "step": 184
1298
+ },
1299
+ {
1300
+ "epoch": 0.14,
1301
+ "grad_norm": 0.5321503974993333,
1302
+ "learning_rate": 9.705700239897809e-05,
1303
+ "loss": 1.3188,
1304
+ "step": 185
1305
+ },
1306
+ {
1307
+ "epoch": 0.14,
1308
+ "grad_norm": 0.5593956329311583,
1309
+ "learning_rate": 9.701639835100513e-05,
1310
+ "loss": 1.249,
1311
+ "step": 186
1312
+ },
1313
+ {
1314
+ "epoch": 0.14,
1315
+ "grad_norm": 0.5591047172889141,
1316
+ "learning_rate": 9.697552472703205e-05,
1317
+ "loss": 1.2756,
1318
+ "step": 187
1319
+ },
1320
+ {
1321
+ "epoch": 0.14,
1322
+ "grad_norm": 0.5543029039316694,
1323
+ "learning_rate": 9.693438176141425e-05,
1324
+ "loss": 1.2915,
1325
+ "step": 188
1326
+ },
1327
+ {
1328
+ "epoch": 0.14,
1329
+ "grad_norm": 0.5494961227055172,
1330
+ "learning_rate": 9.68929696900515e-05,
1331
+ "loss": 1.313,
1332
+ "step": 189
1333
+ },
1334
+ {
1335
+ "epoch": 0.14,
1336
+ "grad_norm": 0.5541252042617403,
1337
+ "learning_rate": 9.685128875038647e-05,
1338
+ "loss": 1.2754,
1339
+ "step": 190
1340
+ },
1341
+ {
1342
+ "epoch": 0.14,
1343
+ "grad_norm": 0.5163534781462605,
1344
+ "learning_rate": 9.680933918140348e-05,
1345
+ "loss": 1.2681,
1346
+ "step": 191
1347
+ },
1348
+ {
1349
+ "epoch": 0.14,
1350
+ "grad_norm": 0.537157272716453,
1351
+ "learning_rate": 9.676712122362706e-05,
1352
+ "loss": 1.2551,
1353
+ "step": 192
1354
+ },
1355
+ {
1356
+ "epoch": 0.14,
1357
+ "grad_norm": 0.5397175193183968,
1358
+ "learning_rate": 9.672463511912055e-05,
1359
+ "loss": 1.2822,
1360
+ "step": 193
1361
+ },
1362
+ {
1363
+ "epoch": 0.14,
1364
+ "grad_norm": 0.5488691397441863,
1365
+ "learning_rate": 9.668188111148484e-05,
1366
+ "loss": 1.283,
1367
+ "step": 194
1368
+ },
1369
+ {
1370
+ "epoch": 0.14,
1371
+ "grad_norm": 0.5905761212464122,
1372
+ "learning_rate": 9.66388594458568e-05,
1373
+ "loss": 1.2896,
1374
+ "step": 195
1375
+ },
1376
+ {
1377
+ "epoch": 0.14,
1378
+ "grad_norm": 0.580369444338734,
1379
+ "learning_rate": 9.659557036890801e-05,
1380
+ "loss": 1.3416,
1381
+ "step": 196
1382
+ },
1383
+ {
1384
+ "epoch": 0.15,
1385
+ "grad_norm": 0.5262728809847318,
1386
+ "learning_rate": 9.655201412884327e-05,
1387
+ "loss": 1.2554,
1388
+ "step": 197
1389
+ },
1390
+ {
1391
+ "epoch": 0.15,
1392
+ "grad_norm": 0.5375550652008795,
1393
+ "learning_rate": 9.650819097539922e-05,
1394
+ "loss": 1.2612,
1395
+ "step": 198
1396
+ },
1397
+ {
1398
+ "epoch": 0.15,
1399
+ "grad_norm": 0.5208197207069616,
1400
+ "learning_rate": 9.646410115984289e-05,
1401
+ "loss": 1.2358,
1402
+ "step": 199
1403
+ },
1404
+ {
1405
+ "epoch": 0.15,
1406
+ "grad_norm": 0.5409371788748774,
1407
+ "learning_rate": 9.641974493497024e-05,
1408
+ "loss": 1.3262,
1409
+ "step": 200
1410
+ },
1411
+ {
1412
+ "epoch": 0.15,
1413
+ "grad_norm": 0.5389211233425135,
1414
+ "learning_rate": 9.637512255510475e-05,
1415
+ "loss": 1.2729,
1416
+ "step": 201
1417
+ },
1418
+ {
1419
+ "epoch": 0.15,
1420
+ "grad_norm": 0.5501782779153785,
1421
+ "learning_rate": 9.633023427609591e-05,
1422
+ "loss": 1.2322,
1423
+ "step": 202
1424
+ },
1425
+ {
1426
+ "epoch": 0.15,
1427
+ "grad_norm": 0.5678681105856288,
1428
+ "learning_rate": 9.628508035531785e-05,
1429
+ "loss": 1.3721,
1430
+ "step": 203
1431
+ },
1432
+ {
1433
+ "epoch": 0.15,
1434
+ "grad_norm": 0.5559621306210715,
1435
+ "learning_rate": 9.623966105166772e-05,
1436
+ "loss": 1.3267,
1437
+ "step": 204
1438
+ },
1439
+ {
1440
+ "epoch": 0.15,
1441
+ "grad_norm": 0.5417687907113425,
1442
+ "learning_rate": 9.619397662556435e-05,
1443
+ "loss": 1.2666,
1444
+ "step": 205
1445
+ },
1446
+ {
1447
+ "epoch": 0.15,
1448
+ "grad_norm": 0.5546614199696198,
1449
+ "learning_rate": 9.614802733894665e-05,
1450
+ "loss": 1.3389,
1451
+ "step": 206
1452
+ },
1453
+ {
1454
+ "epoch": 0.15,
1455
+ "grad_norm": 0.5594799442475286,
1456
+ "learning_rate": 9.610181345527217e-05,
1457
+ "loss": 1.2671,
1458
+ "step": 207
1459
+ },
1460
+ {
1461
+ "epoch": 0.15,
1462
+ "grad_norm": 0.5852167375394156,
1463
+ "learning_rate": 9.605533523951558e-05,
1464
+ "loss": 1.3335,
1465
+ "step": 208
1466
+ },
1467
+ {
1468
+ "epoch": 0.15,
1469
+ "grad_norm": 0.5465110917787175,
1470
+ "learning_rate": 9.600859295816708e-05,
1471
+ "loss": 1.3096,
1472
+ "step": 209
1473
+ },
1474
+ {
1475
+ "epoch": 0.16,
1476
+ "grad_norm": 0.5704616015169348,
1477
+ "learning_rate": 9.596158687923104e-05,
1478
+ "loss": 1.3022,
1479
+ "step": 210
1480
+ },
1481
+ {
1482
+ "epoch": 0.16,
1483
+ "grad_norm": 0.5617616139462727,
1484
+ "learning_rate": 9.591431727222424e-05,
1485
+ "loss": 1.3159,
1486
+ "step": 211
1487
+ },
1488
+ {
1489
+ "epoch": 0.16,
1490
+ "grad_norm": 0.5465602681324426,
1491
+ "learning_rate": 9.586678440817453e-05,
1492
+ "loss": 1.2708,
1493
+ "step": 212
1494
+ },
1495
+ {
1496
+ "epoch": 0.16,
1497
+ "grad_norm": 0.5864421378413351,
1498
+ "learning_rate": 9.581898855961912e-05,
1499
+ "loss": 1.2607,
1500
+ "step": 213
1501
+ },
1502
+ {
1503
+ "epoch": 0.16,
1504
+ "grad_norm": 0.556548001041405,
1505
+ "learning_rate": 9.577093000060312e-05,
1506
+ "loss": 1.3081,
1507
+ "step": 214
1508
+ },
1509
+ {
1510
+ "epoch": 0.16,
1511
+ "grad_norm": 0.5642842704902283,
1512
+ "learning_rate": 9.572260900667794e-05,
1513
+ "loss": 1.2759,
1514
+ "step": 215
1515
+ },
1516
+ {
1517
+ "epoch": 0.16,
1518
+ "grad_norm": 0.5486665255067006,
1519
+ "learning_rate": 9.567402585489963e-05,
1520
+ "loss": 1.2104,
1521
+ "step": 216
1522
+ },
1523
+ {
1524
+ "epoch": 0.16,
1525
+ "grad_norm": 0.5361207508020517,
1526
+ "learning_rate": 9.56251808238275e-05,
1527
+ "loss": 1.2451,
1528
+ "step": 217
1529
+ },
1530
+ {
1531
+ "epoch": 0.16,
1532
+ "grad_norm": 0.5149380805556683,
1533
+ "learning_rate": 9.557607419352226e-05,
1534
+ "loss": 1.2778,
1535
+ "step": 218
1536
+ },
1537
+ {
1538
+ "epoch": 0.16,
1539
+ "grad_norm": 0.5469266902951428,
1540
+ "learning_rate": 9.552670624554461e-05,
1541
+ "loss": 1.2617,
1542
+ "step": 219
1543
+ },
1544
+ {
1545
+ "epoch": 0.16,
1546
+ "grad_norm": 0.5430295319416,
1547
+ "learning_rate": 9.54770772629535e-05,
1548
+ "loss": 1.2915,
1549
+ "step": 220
1550
+ },
1551
+ {
1552
+ "epoch": 0.16,
1553
+ "grad_norm": 0.5744217791056692,
1554
+ "learning_rate": 9.542718753030463e-05,
1555
+ "loss": 1.3281,
1556
+ "step": 221
1557
+ },
1558
+ {
1559
+ "epoch": 0.16,
1560
+ "grad_norm": 0.5587545969611539,
1561
+ "learning_rate": 9.537703733364871e-05,
1562
+ "loss": 1.2837,
1563
+ "step": 222
1564
+ },
1565
+ {
1566
+ "epoch": 0.16,
1567
+ "grad_norm": 0.5288053303373643,
1568
+ "learning_rate": 9.532662696052985e-05,
1569
+ "loss": 1.2949,
1570
+ "step": 223
1571
+ },
1572
+ {
1573
+ "epoch": 0.17,
1574
+ "grad_norm": 0.5791175310063906,
1575
+ "learning_rate": 9.527595669998399e-05,
1576
+ "loss": 1.2917,
1577
+ "step": 224
1578
+ },
1579
+ {
1580
+ "epoch": 0.17,
1581
+ "grad_norm": 0.5250029719207272,
1582
+ "learning_rate": 9.522502684253709e-05,
1583
+ "loss": 1.2375,
1584
+ "step": 225
1585
+ },
1586
+ {
1587
+ "epoch": 0.17,
1588
+ "grad_norm": 0.5177601049436101,
1589
+ "learning_rate": 9.517383768020361e-05,
1590
+ "loss": 1.2695,
1591
+ "step": 226
1592
+ },
1593
+ {
1594
+ "epoch": 0.17,
1595
+ "grad_norm": 0.5554993860583297,
1596
+ "learning_rate": 9.512238950648474e-05,
1597
+ "loss": 1.2917,
1598
+ "step": 227
1599
+ },
1600
+ {
1601
+ "epoch": 0.17,
1602
+ "grad_norm": 0.5738329488665082,
1603
+ "learning_rate": 9.507068261636679e-05,
1604
+ "loss": 1.2944,
1605
+ "step": 228
1606
+ },
1607
+ {
1608
+ "epoch": 0.17,
1609
+ "grad_norm": 0.5562896023700302,
1610
+ "learning_rate": 9.501871730631942e-05,
1611
+ "loss": 1.3296,
1612
+ "step": 229
1613
+ },
1614
+ {
1615
+ "epoch": 0.17,
1616
+ "grad_norm": 0.5416347008024398,
1617
+ "learning_rate": 9.496649387429404e-05,
1618
+ "loss": 1.2437,
1619
+ "step": 230
1620
+ },
1621
+ {
1622
+ "epoch": 0.17,
1623
+ "grad_norm": 0.5699356753997783,
1624
+ "learning_rate": 9.491401261972195e-05,
1625
+ "loss": 1.2705,
1626
+ "step": 231
1627
+ },
1628
+ {
1629
+ "epoch": 0.17,
1630
+ "grad_norm": 0.5481624625613764,
1631
+ "learning_rate": 9.486127384351282e-05,
1632
+ "loss": 1.3779,
1633
+ "step": 232
1634
+ },
1635
+ {
1636
+ "epoch": 0.17,
1637
+ "grad_norm": 0.5688206917165098,
1638
+ "learning_rate": 9.480827784805278e-05,
1639
+ "loss": 1.2754,
1640
+ "step": 233
1641
+ },
1642
+ {
1643
+ "epoch": 0.17,
1644
+ "grad_norm": 0.5490377714658476,
1645
+ "learning_rate": 9.475502493720283e-05,
1646
+ "loss": 1.3125,
1647
+ "step": 234
1648
+ },
1649
+ {
1650
+ "epoch": 0.17,
1651
+ "grad_norm": 0.5355672804730123,
1652
+ "learning_rate": 9.470151541629699e-05,
1653
+ "loss": 1.2627,
1654
+ "step": 235
1655
+ },
1656
+ {
1657
+ "epoch": 0.17,
1658
+ "grad_norm": 0.5905840590902287,
1659
+ "learning_rate": 9.464774959214063e-05,
1660
+ "loss": 1.3027,
1661
+ "step": 236
1662
+ },
1663
+ {
1664
+ "epoch": 0.18,
1665
+ "grad_norm": 0.56064622426517,
1666
+ "learning_rate": 9.459372777300864e-05,
1667
+ "loss": 1.2065,
1668
+ "step": 237
1669
+ },
1670
+ {
1671
+ "epoch": 0.18,
1672
+ "grad_norm": 0.5568610691565873,
1673
+ "learning_rate": 9.45394502686437e-05,
1674
+ "loss": 1.3223,
1675
+ "step": 238
1676
+ },
1677
+ {
1678
+ "epoch": 0.18,
1679
+ "grad_norm": 0.5300725401389981,
1680
+ "learning_rate": 9.448491739025454e-05,
1681
+ "loss": 1.2805,
1682
+ "step": 239
1683
+ },
1684
+ {
1685
+ "epoch": 0.18,
1686
+ "grad_norm": 0.5519662242216672,
1687
+ "learning_rate": 9.44301294505141e-05,
1688
+ "loss": 1.2371,
1689
+ "step": 240
1690
+ },
1691
+ {
1692
+ "epoch": 0.18,
1693
+ "grad_norm": 0.5402101018249572,
1694
+ "learning_rate": 9.437508676355773e-05,
1695
+ "loss": 1.2749,
1696
+ "step": 241
1697
+ },
1698
+ {
1699
+ "epoch": 0.18,
1700
+ "grad_norm": 0.5389383005608104,
1701
+ "learning_rate": 9.431978964498143e-05,
1702
+ "loss": 1.2876,
1703
+ "step": 242
1704
+ },
1705
+ {
1706
+ "epoch": 0.18,
1707
+ "grad_norm": 0.5310718244911751,
1708
+ "learning_rate": 9.426423841184005e-05,
1709
+ "loss": 1.3057,
1710
+ "step": 243
1711
+ },
1712
+ {
1713
+ "epoch": 0.18,
1714
+ "grad_norm": 0.5454082533825911,
1715
+ "learning_rate": 9.420843338264542e-05,
1716
+ "loss": 1.2578,
1717
+ "step": 244
1718
+ },
1719
+ {
1720
+ "epoch": 0.18,
1721
+ "grad_norm": 0.565349361879851,
1722
+ "learning_rate": 9.415237487736452e-05,
1723
+ "loss": 1.3306,
1724
+ "step": 245
1725
+ },
1726
+ {
1727
+ "epoch": 0.18,
1728
+ "grad_norm": 0.5224746893789486,
1729
+ "learning_rate": 9.409606321741775e-05,
1730
+ "loss": 1.2598,
1731
+ "step": 246
1732
+ },
1733
+ {
1734
+ "epoch": 0.18,
1735
+ "grad_norm": 0.5440997273729092,
1736
+ "learning_rate": 9.403949872567695e-05,
1737
+ "loss": 1.2749,
1738
+ "step": 247
1739
+ },
1740
+ {
1741
+ "epoch": 0.18,
1742
+ "grad_norm": 0.5668696203741111,
1743
+ "learning_rate": 9.398268172646365e-05,
1744
+ "loss": 1.2739,
1745
+ "step": 248
1746
+ },
1747
+ {
1748
+ "epoch": 0.18,
1749
+ "grad_norm": 0.538410569856225,
1750
+ "learning_rate": 9.392561254554713e-05,
1751
+ "loss": 1.2734,
1752
+ "step": 249
1753
+ },
1754
+ {
1755
+ "epoch": 0.18,
1756
+ "grad_norm": 0.5458663263053075,
1757
+ "learning_rate": 9.386829151014262e-05,
1758
+ "loss": 1.3101,
1759
+ "step": 250
1760
+ },
1761
+ {
1762
+ "epoch": 0.19,
1763
+ "grad_norm": 0.537905713825921,
1764
+ "learning_rate": 9.381071894890941e-05,
1765
+ "loss": 1.2666,
1766
+ "step": 251
1767
+ },
1768
+ {
1769
+ "epoch": 0.19,
1770
+ "grad_norm": 0.5288916095430457,
1771
+ "learning_rate": 9.375289519194894e-05,
1772
+ "loss": 1.2666,
1773
+ "step": 252
1774
+ },
1775
+ {
1776
+ "epoch": 0.19,
1777
+ "grad_norm": 0.5335913282729025,
1778
+ "learning_rate": 9.369482057080292e-05,
1779
+ "loss": 1.2886,
1780
+ "step": 253
1781
+ },
1782
+ {
1783
+ "epoch": 0.19,
1784
+ "grad_norm": 0.5523824410197196,
1785
+ "learning_rate": 9.363649541845142e-05,
1786
+ "loss": 1.2571,
1787
+ "step": 254
1788
+ },
1789
+ {
1790
+ "epoch": 0.19,
1791
+ "grad_norm": 0.5912264857528259,
1792
+ "learning_rate": 9.357792006931098e-05,
1793
+ "loss": 1.261,
1794
+ "step": 255
1795
+ },
1796
+ {
1797
+ "epoch": 0.19,
1798
+ "grad_norm": 0.5594499774840426,
1799
+ "learning_rate": 9.35190948592327e-05,
1800
+ "loss": 1.3027,
1801
+ "step": 256
1802
+ },
1803
+ {
1804
+ "epoch": 0.19,
1805
+ "grad_norm": 0.5379207919206825,
1806
+ "learning_rate": 9.346002012550027e-05,
1807
+ "loss": 1.2983,
1808
+ "step": 257
1809
+ },
1810
+ {
1811
+ "epoch": 0.19,
1812
+ "grad_norm": 0.5455629199690059,
1813
+ "learning_rate": 9.340069620682806e-05,
1814
+ "loss": 1.2695,
1815
+ "step": 258
1816
+ },
1817
+ {
1818
+ "epoch": 0.19,
1819
+ "grad_norm": 0.5471737544580354,
1820
+ "learning_rate": 9.334112344335924e-05,
1821
+ "loss": 1.3047,
1822
+ "step": 259
1823
+ },
1824
+ {
1825
+ "epoch": 0.19,
1826
+ "grad_norm": 0.5397100655209365,
1827
+ "learning_rate": 9.328130217666366e-05,
1828
+ "loss": 1.2896,
1829
+ "step": 260
1830
+ },
1831
+ {
1832
+ "epoch": 0.19,
1833
+ "grad_norm": 0.5636004509867364,
1834
+ "learning_rate": 9.322123274973613e-05,
1835
+ "loss": 1.3501,
1836
+ "step": 261
1837
+ },
1838
+ {
1839
+ "epoch": 0.19,
1840
+ "grad_norm": 0.5605154015144495,
1841
+ "learning_rate": 9.316091550699424e-05,
1842
+ "loss": 1.2983,
1843
+ "step": 262
1844
+ },
1845
+ {
1846
+ "epoch": 0.19,
1847
+ "grad_norm": 0.5461515781521593,
1848
+ "learning_rate": 9.310035079427651e-05,
1849
+ "loss": 1.269,
1850
+ "step": 263
1851
+ },
1852
+ {
1853
+ "epoch": 0.2,
1854
+ "grad_norm": 0.5175024878789147,
1855
+ "learning_rate": 9.303953895884033e-05,
1856
+ "loss": 1.1653,
1857
+ "step": 264
1858
+ },
1859
+ {
1860
+ "epoch": 0.2,
1861
+ "grad_norm": 0.5224669601631107,
1862
+ "learning_rate": 9.297848034936006e-05,
1863
+ "loss": 1.2554,
1864
+ "step": 265
1865
+ },
1866
+ {
1867
+ "epoch": 0.2,
1868
+ "grad_norm": 0.5444106809363777,
1869
+ "learning_rate": 9.291717531592494e-05,
1870
+ "loss": 1.293,
1871
+ "step": 266
1872
+ },
1873
+ {
1874
+ "epoch": 0.2,
1875
+ "grad_norm": 0.5287552712313793,
1876
+ "learning_rate": 9.285562421003715e-05,
1877
+ "loss": 1.2651,
1878
+ "step": 267
1879
+ },
1880
+ {
1881
+ "epoch": 0.2,
1882
+ "grad_norm": 0.5381309609110954,
1883
+ "learning_rate": 9.279382738460971e-05,
1884
+ "loss": 1.2812,
1885
+ "step": 268
1886
+ },
1887
+ {
1888
+ "epoch": 0.2,
1889
+ "grad_norm": 0.5528803396804242,
1890
+ "learning_rate": 9.273178519396459e-05,
1891
+ "loss": 1.3149,
1892
+ "step": 269
1893
+ },
1894
+ {
1895
+ "epoch": 0.2,
1896
+ "grad_norm": 0.5270531797880375,
1897
+ "learning_rate": 9.266949799383053e-05,
1898
+ "loss": 1.2615,
1899
+ "step": 270
1900
+ },
1901
+ {
1902
+ "epoch": 0.2,
1903
+ "grad_norm": 0.5488129774725259,
1904
+ "learning_rate": 9.260696614134114e-05,
1905
+ "loss": 1.2837,
1906
+ "step": 271
1907
+ },
1908
+ {
1909
+ "epoch": 0.2,
1910
+ "grad_norm": 0.5335083589116082,
1911
+ "learning_rate": 9.254418999503271e-05,
1912
+ "loss": 1.2339,
1913
+ "step": 272
1914
+ },
1915
+ {
1916
+ "epoch": 0.2,
1917
+ "grad_norm": 0.5974061497388541,
1918
+ "learning_rate": 9.248116991484229e-05,
1919
+ "loss": 1.2825,
1920
+ "step": 273
1921
+ },
1922
+ {
1923
+ "epoch": 0.2,
1924
+ "grad_norm": 0.5381713380415607,
1925
+ "learning_rate": 9.241790626210549e-05,
1926
+ "loss": 1.1895,
1927
+ "step": 274
1928
+ },
1929
+ {
1930
+ "epoch": 0.2,
1931
+ "grad_norm": 0.5384430847504001,
1932
+ "learning_rate": 9.235439939955457e-05,
1933
+ "loss": 1.2358,
1934
+ "step": 275
1935
+ },
1936
+ {
1937
+ "epoch": 0.2,
1938
+ "grad_norm": 0.5256588888016233,
1939
+ "learning_rate": 9.229064969131621e-05,
1940
+ "loss": 1.2407,
1941
+ "step": 276
1942
+ },
1943
+ {
1944
+ "epoch": 0.2,
1945
+ "grad_norm": 0.5242296953154587,
1946
+ "learning_rate": 9.222665750290953e-05,
1947
+ "loss": 1.2832,
1948
+ "step": 277
1949
+ },
1950
+ {
1951
+ "epoch": 0.21,
1952
+ "grad_norm": 0.5224106607183625,
1953
+ "learning_rate": 9.216242320124388e-05,
1954
+ "loss": 1.2388,
1955
+ "step": 278
1956
+ },
1957
+ {
1958
+ "epoch": 0.21,
1959
+ "grad_norm": 0.540400861953043,
1960
+ "learning_rate": 9.20979471546169e-05,
1961
+ "loss": 1.2695,
1962
+ "step": 279
1963
+ },
1964
+ {
1965
+ "epoch": 0.21,
1966
+ "grad_norm": 0.5289483661482471,
1967
+ "learning_rate": 9.203322973271223e-05,
1968
+ "loss": 1.2832,
1969
+ "step": 280
1970
+ },
1971
+ {
1972
+ "epoch": 0.21,
1973
+ "grad_norm": 0.5376637104674151,
1974
+ "learning_rate": 9.19682713065975e-05,
1975
+ "loss": 1.2783,
1976
+ "step": 281
1977
+ },
1978
+ {
1979
+ "epoch": 0.21,
1980
+ "grad_norm": 0.5547766359095799,
1981
+ "learning_rate": 9.19030722487222e-05,
1982
+ "loss": 1.2515,
1983
+ "step": 282
1984
+ },
1985
+ {
1986
+ "epoch": 0.21,
1987
+ "grad_norm": 0.5431030883095361,
1988
+ "learning_rate": 9.183763293291549e-05,
1989
+ "loss": 1.2346,
1990
+ "step": 283
1991
+ },
1992
+ {
1993
+ "epoch": 0.21,
1994
+ "grad_norm": 0.5767856753870191,
1995
+ "learning_rate": 9.17719537343841e-05,
1996
+ "loss": 1.2974,
1997
+ "step": 284
1998
+ },
1999
+ {
2000
+ "epoch": 0.21,
2001
+ "grad_norm": 0.5356401648893151,
2002
+ "learning_rate": 9.170603502971016e-05,
2003
+ "loss": 1.2532,
2004
+ "step": 285
2005
+ },
2006
+ {
2007
+ "epoch": 0.21,
2008
+ "grad_norm": 0.5528695803408737,
2009
+ "learning_rate": 9.163987719684907e-05,
2010
+ "loss": 1.3442,
2011
+ "step": 286
2012
+ },
2013
+ {
2014
+ "epoch": 0.21,
2015
+ "grad_norm": 0.5356080125920785,
2016
+ "learning_rate": 9.157348061512727e-05,
2017
+ "loss": 1.2686,
2018
+ "step": 287
2019
+ },
2020
+ {
2021
+ "epoch": 0.21,
2022
+ "grad_norm": 0.5778656916381988,
2023
+ "learning_rate": 9.150684566524012e-05,
2024
+ "loss": 1.2041,
2025
+ "step": 288
2026
+ },
2027
+ {
2028
+ "epoch": 0.21,
2029
+ "grad_norm": 0.5328749801157324,
2030
+ "learning_rate": 9.143997272924973e-05,
2031
+ "loss": 1.2437,
2032
+ "step": 289
2033
+ },
2034
+ {
2035
+ "epoch": 0.21,
2036
+ "grad_norm": 0.5656275076768376,
2037
+ "learning_rate": 9.13728621905827e-05,
2038
+ "loss": 1.2886,
2039
+ "step": 290
2040
+ },
2041
+ {
2042
+ "epoch": 0.22,
2043
+ "grad_norm": 0.5655646337419664,
2044
+ "learning_rate": 9.130551443402799e-05,
2045
+ "loss": 1.2783,
2046
+ "step": 291
2047
+ },
2048
+ {
2049
+ "epoch": 0.22,
2050
+ "grad_norm": 0.567975953014803,
2051
+ "learning_rate": 9.123792984573466e-05,
2052
+ "loss": 1.3223,
2053
+ "step": 292
2054
+ },
2055
+ {
2056
+ "epoch": 0.22,
2057
+ "grad_norm": 0.5361585380833186,
2058
+ "learning_rate": 9.117010881320973e-05,
2059
+ "loss": 1.2231,
2060
+ "step": 293
2061
+ },
2062
+ {
2063
+ "epoch": 0.22,
2064
+ "grad_norm": 0.5527612532950269,
2065
+ "learning_rate": 9.110205172531585e-05,
2066
+ "loss": 1.3506,
2067
+ "step": 294
2068
+ },
2069
+ {
2070
+ "epoch": 0.22,
2071
+ "grad_norm": 0.5330323483779986,
2072
+ "learning_rate": 9.103375897226918e-05,
2073
+ "loss": 1.2974,
2074
+ "step": 295
2075
+ },
2076
+ {
2077
+ "epoch": 0.22,
2078
+ "grad_norm": 0.541076058179259,
2079
+ "learning_rate": 9.096523094563708e-05,
2080
+ "loss": 1.2617,
2081
+ "step": 296
2082
+ },
2083
+ {
2084
+ "epoch": 0.22,
2085
+ "grad_norm": 0.5340836977689315,
2086
+ "learning_rate": 9.089646803833589e-05,
2087
+ "loss": 1.2603,
2088
+ "step": 297
2089
+ },
2090
+ {
2091
+ "epoch": 0.22,
2092
+ "grad_norm": 0.5383753245320845,
2093
+ "learning_rate": 9.082747064462867e-05,
2094
+ "loss": 1.2583,
2095
+ "step": 298
2096
+ },
2097
+ {
2098
+ "epoch": 0.22,
2099
+ "grad_norm": 0.5192836861689345,
2100
+ "learning_rate": 9.075823916012298e-05,
2101
+ "loss": 1.2568,
2102
+ "step": 299
2103
+ },
2104
+ {
2105
+ "epoch": 0.22,
2106
+ "grad_norm": 0.5744817919271316,
2107
+ "learning_rate": 9.068877398176852e-05,
2108
+ "loss": 1.2131,
2109
+ "step": 300
2110
+ },
2111
+ {
2112
+ "epoch": 0.22,
2113
+ "grad_norm": 0.5323047093147705,
2114
+ "learning_rate": 9.061907550785498e-05,
2115
+ "loss": 1.2783,
2116
+ "step": 301
2117
+ },
2118
+ {
2119
+ "epoch": 0.22,
2120
+ "grad_norm": 0.5607328564400242,
2121
+ "learning_rate": 9.054914413800961e-05,
2122
+ "loss": 1.3398,
2123
+ "step": 302
2124
+ },
2125
+ {
2126
+ "epoch": 0.22,
2127
+ "grad_norm": 0.5782257895199574,
2128
+ "learning_rate": 9.047898027319507e-05,
2129
+ "loss": 1.2759,
2130
+ "step": 303
2131
+ },
2132
+ {
2133
+ "epoch": 0.22,
2134
+ "grad_norm": 0.546644793451931,
2135
+ "learning_rate": 9.040858431570702e-05,
2136
+ "loss": 1.2632,
2137
+ "step": 304
2138
+ },
2139
+ {
2140
+ "epoch": 0.23,
2141
+ "grad_norm": 0.5535852227341702,
2142
+ "learning_rate": 9.033795666917191e-05,
2143
+ "loss": 1.312,
2144
+ "step": 305
2145
+ },
2146
+ {
2147
+ "epoch": 0.23,
2148
+ "grad_norm": 0.5371002551511538,
2149
+ "learning_rate": 9.026709773854457e-05,
2150
+ "loss": 1.2593,
2151
+ "step": 306
2152
+ },
2153
+ {
2154
+ "epoch": 0.23,
2155
+ "grad_norm": 0.5394441228369942,
2156
+ "learning_rate": 9.019600793010597e-05,
2157
+ "loss": 1.269,
2158
+ "step": 307
2159
+ },
2160
+ {
2161
+ "epoch": 0.23,
2162
+ "grad_norm": 0.5512445550522174,
2163
+ "learning_rate": 9.012468765146079e-05,
2164
+ "loss": 1.2686,
2165
+ "step": 308
2166
+ },
2167
+ {
2168
+ "epoch": 0.23,
2169
+ "grad_norm": 0.5043850111181398,
2170
+ "learning_rate": 9.005313731153524e-05,
2171
+ "loss": 1.2363,
2172
+ "step": 309
2173
+ },
2174
+ {
2175
+ "epoch": 0.23,
2176
+ "grad_norm": 0.5294693808157453,
2177
+ "learning_rate": 8.998135732057458e-05,
2178
+ "loss": 1.2725,
2179
+ "step": 310
2180
+ },
2181
+ {
2182
+ "epoch": 0.23,
2183
+ "grad_norm": 0.5235449664008548,
2184
+ "learning_rate": 8.990934809014077e-05,
2185
+ "loss": 1.249,
2186
+ "step": 311
2187
+ },
2188
+ {
2189
+ "epoch": 0.23,
2190
+ "grad_norm": 0.5228082226582549,
2191
+ "learning_rate": 8.983711003311024e-05,
2192
+ "loss": 1.2153,
2193
+ "step": 312
2194
+ },
2195
+ {
2196
+ "epoch": 0.23,
2197
+ "grad_norm": 0.5525620828249341,
2198
+ "learning_rate": 8.976464356367134e-05,
2199
+ "loss": 1.2136,
2200
+ "step": 313
2201
+ },
2202
+ {
2203
+ "epoch": 0.23,
2204
+ "grad_norm": 0.5605215996168639,
2205
+ "learning_rate": 8.96919490973221e-05,
2206
+ "loss": 1.271,
2207
+ "step": 314
2208
+ },
2209
+ {
2210
+ "epoch": 0.23,
2211
+ "grad_norm": 0.5277359930208506,
2212
+ "learning_rate": 8.961902705086785e-05,
2213
+ "loss": 1.1836,
2214
+ "step": 315
2215
+ },
2216
+ {
2217
+ "epoch": 0.23,
2218
+ "grad_norm": 0.5405930304733125,
2219
+ "learning_rate": 8.954587784241871e-05,
2220
+ "loss": 1.2705,
2221
+ "step": 316
2222
+ },
2223
+ {
2224
+ "epoch": 0.23,
2225
+ "grad_norm": 0.5248476194932483,
2226
+ "learning_rate": 8.947250189138731e-05,
2227
+ "loss": 1.2607,
2228
+ "step": 317
2229
+ },
2230
+ {
2231
+ "epoch": 0.24,
2232
+ "grad_norm": 0.573678896783169,
2233
+ "learning_rate": 8.939889961848634e-05,
2234
+ "loss": 1.2727,
2235
+ "step": 318
2236
+ },
2237
+ {
2238
+ "epoch": 0.24,
2239
+ "grad_norm": 0.5773485095137408,
2240
+ "learning_rate": 8.932507144572616e-05,
2241
+ "loss": 1.2607,
2242
+ "step": 319
2243
+ },
2244
+ {
2245
+ "epoch": 0.24,
2246
+ "grad_norm": 0.5633980526681968,
2247
+ "learning_rate": 8.925101779641232e-05,
2248
+ "loss": 1.1917,
2249
+ "step": 320
2250
+ },
2251
+ {
2252
+ "epoch": 0.24,
2253
+ "grad_norm": 0.5300371631849218,
2254
+ "learning_rate": 8.917673909514322e-05,
2255
+ "loss": 1.3105,
2256
+ "step": 321
2257
+ },
2258
+ {
2259
+ "epoch": 0.24,
2260
+ "grad_norm": 0.5310192196200603,
2261
+ "learning_rate": 8.910223576780758e-05,
2262
+ "loss": 1.2808,
2263
+ "step": 322
2264
+ },
2265
+ {
2266
+ "epoch": 0.24,
2267
+ "grad_norm": 0.5234569464366723,
2268
+ "learning_rate": 8.902750824158212e-05,
2269
+ "loss": 1.2468,
2270
+ "step": 323
2271
+ },
2272
+ {
2273
+ "epoch": 0.24,
2274
+ "grad_norm": 0.5473770126434013,
2275
+ "learning_rate": 8.895255694492896e-05,
2276
+ "loss": 1.2676,
2277
+ "step": 324
2278
+ },
2279
+ {
2280
+ "epoch": 0.24,
2281
+ "grad_norm": 0.5670393642092653,
2282
+ "learning_rate": 8.887738230759333e-05,
2283
+ "loss": 1.2456,
2284
+ "step": 325
2285
+ },
2286
+ {
2287
+ "epoch": 0.24,
2288
+ "grad_norm": 0.5484650752546845,
2289
+ "learning_rate": 8.880198476060095e-05,
2290
+ "loss": 1.251,
2291
+ "step": 326
2292
+ },
2293
+ {
2294
+ "epoch": 0.24,
2295
+ "grad_norm": 0.5569076336735002,
2296
+ "learning_rate": 8.872636473625565e-05,
2297
+ "loss": 1.272,
2298
+ "step": 327
2299
+ },
2300
+ {
2301
+ "epoch": 0.24,
2302
+ "grad_norm": 0.5237290090420638,
2303
+ "learning_rate": 8.865052266813685e-05,
2304
+ "loss": 1.2822,
2305
+ "step": 328
2306
+ },
2307
+ {
2308
+ "epoch": 0.24,
2309
+ "grad_norm": 0.5507489271814671,
2310
+ "learning_rate": 8.857445899109715e-05,
2311
+ "loss": 1.2783,
2312
+ "step": 329
2313
+ },
2314
+ {
2315
+ "epoch": 0.24,
2316
+ "grad_norm": 0.5527246685898635,
2317
+ "learning_rate": 8.849817414125973e-05,
2318
+ "loss": 1.2705,
2319
+ "step": 330
2320
+ },
2321
+ {
2322
+ "epoch": 0.24,
2323
+ "grad_norm": 0.5544016696123183,
2324
+ "learning_rate": 8.84216685560159e-05,
2325
+ "loss": 1.2856,
2326
+ "step": 331
2327
+ },
2328
+ {
2329
+ "epoch": 0.25,
2330
+ "grad_norm": 0.5424146088216879,
2331
+ "learning_rate": 8.834494267402263e-05,
2332
+ "loss": 1.2202,
2333
+ "step": 332
2334
+ },
2335
+ {
2336
+ "epoch": 0.25,
2337
+ "grad_norm": 0.5323806898987287,
2338
+ "learning_rate": 8.826799693519996e-05,
2339
+ "loss": 1.248,
2340
+ "step": 333
2341
+ },
2342
+ {
2343
+ "epoch": 0.25,
2344
+ "grad_norm": 0.5595146324987165,
2345
+ "learning_rate": 8.819083178072852e-05,
2346
+ "loss": 1.1672,
2347
+ "step": 334
2348
+ },
2349
+ {
2350
+ "epoch": 0.25,
2351
+ "grad_norm": 0.5854406580169095,
2352
+ "learning_rate": 8.811344765304698e-05,
2353
+ "loss": 1.2146,
2354
+ "step": 335
2355
+ },
2356
+ {
2357
+ "epoch": 0.25,
2358
+ "grad_norm": 0.5697562446019094,
2359
+ "learning_rate": 8.80358449958496e-05,
2360
+ "loss": 1.2568,
2361
+ "step": 336
2362
+ },
2363
+ {
2364
+ "epoch": 0.25,
2365
+ "grad_norm": 0.5538906977604374,
2366
+ "learning_rate": 8.795802425408352e-05,
2367
+ "loss": 1.2544,
2368
+ "step": 337
2369
+ },
2370
+ {
2371
+ "epoch": 0.25,
2372
+ "grad_norm": 0.5211793067308176,
2373
+ "learning_rate": 8.787998587394637e-05,
2374
+ "loss": 1.2183,
2375
+ "step": 338
2376
+ },
2377
+ {
2378
+ "epoch": 0.25,
2379
+ "grad_norm": 0.5732446722628473,
2380
+ "learning_rate": 8.780173030288359e-05,
2381
+ "loss": 1.3057,
2382
+ "step": 339
2383
+ },
2384
+ {
2385
+ "epoch": 0.25,
2386
+ "grad_norm": 0.5352980539739127,
2387
+ "learning_rate": 8.772325798958597e-05,
2388
+ "loss": 1.2598,
2389
+ "step": 340
2390
+ },
2391
+ {
2392
+ "epoch": 0.25,
2393
+ "grad_norm": 0.5234917926015726,
2394
+ "learning_rate": 8.7644569383987e-05,
2395
+ "loss": 1.1982,
2396
+ "step": 341
2397
+ },
2398
+ {
2399
+ "epoch": 0.25,
2400
+ "grad_norm": 0.5844314852721842,
2401
+ "learning_rate": 8.75656649372603e-05,
2402
+ "loss": 1.2656,
2403
+ "step": 342
2404
+ },
2405
+ {
2406
+ "epoch": 0.25,
2407
+ "grad_norm": 0.5646854448914282,
2408
+ "learning_rate": 8.748654510181709e-05,
2409
+ "loss": 1.21,
2410
+ "step": 343
2411
+ },
2412
+ {
2413
+ "epoch": 0.25,
2414
+ "grad_norm": 0.5216723813831847,
2415
+ "learning_rate": 8.740721033130352e-05,
2416
+ "loss": 1.2329,
2417
+ "step": 344
2418
+ },
2419
+ {
2420
+ "epoch": 0.25,
2421
+ "grad_norm": 0.5099027314874095,
2422
+ "learning_rate": 8.732766108059813e-05,
2423
+ "loss": 1.2236,
2424
+ "step": 345
2425
+ },
2426
+ {
2427
+ "epoch": 0.26,
2428
+ "grad_norm": 0.5188769999186538,
2429
+ "learning_rate": 8.72478978058092e-05,
2430
+ "loss": 1.2905,
2431
+ "step": 346
2432
+ },
2433
+ {
2434
+ "epoch": 0.26,
2435
+ "grad_norm": 0.5245157404984339,
2436
+ "learning_rate": 8.716792096427217e-05,
2437
+ "loss": 1.2339,
2438
+ "step": 347
2439
+ },
2440
+ {
2441
+ "epoch": 0.26,
2442
+ "grad_norm": 0.5160205485678449,
2443
+ "learning_rate": 8.708773101454697e-05,
2444
+ "loss": 1.2524,
2445
+ "step": 348
2446
+ },
2447
+ {
2448
+ "epoch": 0.26,
2449
+ "grad_norm": 0.510633107323387,
2450
+ "learning_rate": 8.700732841641542e-05,
2451
+ "loss": 1.2756,
2452
+ "step": 349
2453
+ },
2454
+ {
2455
+ "epoch": 0.26,
2456
+ "grad_norm": 0.5097028901140956,
2457
+ "learning_rate": 8.692671363087863e-05,
2458
+ "loss": 1.2539,
2459
+ "step": 350
2460
+ },
2461
+ {
2462
+ "epoch": 0.26,
2463
+ "grad_norm": 0.5506040438253419,
2464
+ "learning_rate": 8.68458871201543e-05,
2465
+ "loss": 1.1733,
2466
+ "step": 351
2467
+ },
2468
+ {
2469
+ "epoch": 0.26,
2470
+ "grad_norm": 0.5339837805003954,
2471
+ "learning_rate": 8.676484934767409e-05,
2472
+ "loss": 1.1919,
2473
+ "step": 352
2474
+ },
2475
+ {
2476
+ "epoch": 0.26,
2477
+ "grad_norm": 0.5243053855032012,
2478
+ "learning_rate": 8.668360077808093e-05,
2479
+ "loss": 1.2637,
2480
+ "step": 353
2481
+ },
2482
+ {
2483
+ "epoch": 0.26,
2484
+ "grad_norm": 0.5475923045103417,
2485
+ "learning_rate": 8.660214187722646e-05,
2486
+ "loss": 1.2583,
2487
+ "step": 354
2488
+ },
2489
+ {
2490
+ "epoch": 0.26,
2491
+ "grad_norm": 0.5139607250185231,
2492
+ "learning_rate": 8.652047311216822e-05,
2493
+ "loss": 1.2939,
2494
+ "step": 355
2495
+ },
2496
+ {
2497
+ "epoch": 0.26,
2498
+ "grad_norm": 0.5310090229071474,
2499
+ "learning_rate": 8.64385949511671e-05,
2500
+ "loss": 1.2788,
2501
+ "step": 356
2502
+ },
2503
+ {
2504
+ "epoch": 0.26,
2505
+ "grad_norm": 0.5531120494965365,
2506
+ "learning_rate": 8.635650786368452e-05,
2507
+ "loss": 1.25,
2508
+ "step": 357
2509
+ },
2510
+ {
2511
+ "epoch": 0.26,
2512
+ "grad_norm": 0.5315969577054235,
2513
+ "learning_rate": 8.627421232037989e-05,
2514
+ "loss": 1.2357,
2515
+ "step": 358
2516
+ },
2517
+ {
2518
+ "epoch": 0.27,
2519
+ "grad_norm": 0.5266216921573422,
2520
+ "learning_rate": 8.619170879310779e-05,
2521
+ "loss": 1.2729,
2522
+ "step": 359
2523
+ },
2524
+ {
2525
+ "epoch": 0.27,
2526
+ "grad_norm": 0.5593055072800345,
2527
+ "learning_rate": 8.61089977549153e-05,
2528
+ "loss": 1.2529,
2529
+ "step": 360
2530
+ },
2531
+ {
2532
+ "epoch": 0.27,
2533
+ "grad_norm": 0.5596710951308123,
2534
+ "learning_rate": 8.602607968003935e-05,
2535
+ "loss": 1.2725,
2536
+ "step": 361
2537
+ },
2538
+ {
2539
+ "epoch": 0.27,
2540
+ "grad_norm": 0.5433552854623133,
2541
+ "learning_rate": 8.59429550439039e-05,
2542
+ "loss": 1.2446,
2543
+ "step": 362
2544
+ },
2545
+ {
2546
+ "epoch": 0.27,
2547
+ "grad_norm": 0.5818949631250041,
2548
+ "learning_rate": 8.585962432311727e-05,
2549
+ "loss": 1.2998,
2550
+ "step": 363
2551
+ },
2552
+ {
2553
+ "epoch": 0.27,
2554
+ "grad_norm": 0.514243535892493,
2555
+ "learning_rate": 8.577608799546942e-05,
2556
+ "loss": 1.23,
2557
+ "step": 364
2558
+ },
2559
+ {
2560
+ "epoch": 0.27,
2561
+ "grad_norm": 0.5465838481685172,
2562
+ "learning_rate": 8.569234653992916e-05,
2563
+ "loss": 1.2532,
2564
+ "step": 365
2565
+ },
2566
+ {
2567
+ "epoch": 0.27,
2568
+ "grad_norm": 0.519563471824199,
2569
+ "learning_rate": 8.560840043664144e-05,
2570
+ "loss": 1.2607,
2571
+ "step": 366
2572
+ },
2573
+ {
2574
+ "epoch": 0.27,
2575
+ "grad_norm": 0.5334398982863738,
2576
+ "learning_rate": 8.552425016692464e-05,
2577
+ "loss": 1.2363,
2578
+ "step": 367
2579
+ },
2580
+ {
2581
+ "epoch": 0.27,
2582
+ "grad_norm": 0.5530652812053678,
2583
+ "learning_rate": 8.543989621326768e-05,
2584
+ "loss": 1.2681,
2585
+ "step": 368
2586
+ },
2587
+ {
2588
+ "epoch": 0.27,
2589
+ "grad_norm": 0.5502954863671434,
2590
+ "learning_rate": 8.535533905932738e-05,
2591
+ "loss": 1.1721,
2592
+ "step": 369
2593
+ },
2594
+ {
2595
+ "epoch": 0.27,
2596
+ "grad_norm": 0.5180001078920966,
2597
+ "learning_rate": 8.527057918992565e-05,
2598
+ "loss": 1.2139,
2599
+ "step": 370
2600
+ },
2601
+ {
2602
+ "epoch": 0.27,
2603
+ "grad_norm": 0.5333180911534254,
2604
+ "learning_rate": 8.518561709104667e-05,
2605
+ "loss": 1.2461,
2606
+ "step": 371
2607
+ },
2608
+ {
2609
+ "epoch": 0.27,
2610
+ "grad_norm": 0.5479350107655593,
2611
+ "learning_rate": 8.510045324983417e-05,
2612
+ "loss": 1.2512,
2613
+ "step": 372
2614
+ },
2615
+ {
2616
+ "epoch": 0.28,
2617
+ "grad_norm": 0.5246093324411485,
2618
+ "learning_rate": 8.501508815458855e-05,
2619
+ "loss": 1.1787,
2620
+ "step": 373
2621
+ },
2622
+ {
2623
+ "epoch": 0.28,
2624
+ "grad_norm": 0.50033135264865,
2625
+ "learning_rate": 8.492952229476421e-05,
2626
+ "loss": 1.2271,
2627
+ "step": 374
2628
+ },
2629
+ {
2630
+ "epoch": 0.28,
2631
+ "grad_norm": 0.5418162221365314,
2632
+ "learning_rate": 8.484375616096658e-05,
2633
+ "loss": 1.2383,
2634
+ "step": 375
2635
+ },
2636
+ {
2637
+ "epoch": 0.28,
2638
+ "grad_norm": 0.516783670359288,
2639
+ "learning_rate": 8.475779024494945e-05,
2640
+ "loss": 1.2681,
2641
+ "step": 376
2642
+ },
2643
+ {
2644
+ "epoch": 0.28,
2645
+ "grad_norm": 0.5298750460233759,
2646
+ "learning_rate": 8.467162503961208e-05,
2647
+ "loss": 1.2451,
2648
+ "step": 377
2649
+ },
2650
+ {
2651
+ "epoch": 0.28,
2652
+ "grad_norm": 0.5149476400550106,
2653
+ "learning_rate": 8.45852610389964e-05,
2654
+ "loss": 1.23,
2655
+ "step": 378
2656
+ },
2657
+ {
2658
+ "epoch": 0.28,
2659
+ "grad_norm": 0.5268563601419046,
2660
+ "learning_rate": 8.449869873828411e-05,
2661
+ "loss": 1.2129,
2662
+ "step": 379
2663
+ },
2664
+ {
2665
+ "epoch": 0.28,
2666
+ "grad_norm": 0.5357435202461692,
2667
+ "learning_rate": 8.441193863379396e-05,
2668
+ "loss": 1.2881,
2669
+ "step": 380
2670
+ },
2671
+ {
2672
+ "epoch": 0.28,
2673
+ "grad_norm": 0.5407114377511073,
2674
+ "learning_rate": 8.432498122297878e-05,
2675
+ "loss": 1.2559,
2676
+ "step": 381
2677
+ },
2678
+ {
2679
+ "epoch": 0.28,
2680
+ "grad_norm": 0.5376253272809564,
2681
+ "learning_rate": 8.423782700442277e-05,
2682
+ "loss": 1.2346,
2683
+ "step": 382
2684
+ },
2685
+ {
2686
+ "epoch": 0.28,
2687
+ "grad_norm": 0.5378153063595059,
2688
+ "learning_rate": 8.415047647783847e-05,
2689
+ "loss": 1.2031,
2690
+ "step": 383
2691
+ },
2692
+ {
2693
+ "epoch": 0.28,
2694
+ "grad_norm": 0.514779002563088,
2695
+ "learning_rate": 8.406293014406403e-05,
2696
+ "loss": 1.2056,
2697
+ "step": 384
2698
+ },
2699
+ {
2700
+ "epoch": 0.28,
2701
+ "grad_norm": 0.5659231392943161,
2702
+ "learning_rate": 8.397518850506028e-05,
2703
+ "loss": 1.2346,
2704
+ "step": 385
2705
+ },
2706
+ {
2707
+ "epoch": 0.29,
2708
+ "grad_norm": 0.5483974446090379,
2709
+ "learning_rate": 8.388725206390788e-05,
2710
+ "loss": 1.2974,
2711
+ "step": 386
2712
+ },
2713
+ {
2714
+ "epoch": 0.29,
2715
+ "grad_norm": 0.5297423113703096,
2716
+ "learning_rate": 8.379912132480441e-05,
2717
+ "loss": 1.2427,
2718
+ "step": 387
2719
+ },
2720
+ {
2721
+ "epoch": 0.29,
2722
+ "grad_norm": 0.5339239833592698,
2723
+ "learning_rate": 8.371079679306146e-05,
2724
+ "loss": 1.2788,
2725
+ "step": 388
2726
+ },
2727
+ {
2728
+ "epoch": 0.29,
2729
+ "grad_norm": 0.5346762752364651,
2730
+ "learning_rate": 8.36222789751018e-05,
2731
+ "loss": 1.2329,
2732
+ "step": 389
2733
+ },
2734
+ {
2735
+ "epoch": 0.29,
2736
+ "grad_norm": 0.5267945253503268,
2737
+ "learning_rate": 8.353356837845642e-05,
2738
+ "loss": 1.3101,
2739
+ "step": 390
2740
+ },
2741
+ {
2742
+ "epoch": 0.29,
2743
+ "grad_norm": 0.5227678407329124,
2744
+ "learning_rate": 8.344466551176164e-05,
2745
+ "loss": 1.2544,
2746
+ "step": 391
2747
+ },
2748
+ {
2749
+ "epoch": 0.29,
2750
+ "grad_norm": 0.5351886972585579,
2751
+ "learning_rate": 8.335557088475618e-05,
2752
+ "loss": 1.2036,
2753
+ "step": 392
2754
+ },
2755
+ {
2756
+ "epoch": 0.29,
2757
+ "grad_norm": 0.547855768363372,
2758
+ "learning_rate": 8.326628500827826e-05,
2759
+ "loss": 1.2256,
2760
+ "step": 393
2761
+ },
2762
+ {
2763
+ "epoch": 0.29,
2764
+ "grad_norm": 0.5232912428703006,
2765
+ "learning_rate": 8.31768083942627e-05,
2766
+ "loss": 1.2524,
2767
+ "step": 394
2768
+ },
2769
+ {
2770
+ "epoch": 0.29,
2771
+ "grad_norm": 0.5355407135538937,
2772
+ "learning_rate": 8.308714155573785e-05,
2773
+ "loss": 1.1904,
2774
+ "step": 395
2775
+ },
2776
+ {
2777
+ "epoch": 0.29,
2778
+ "grad_norm": 0.5398818834520477,
2779
+ "learning_rate": 8.29972850068228e-05,
2780
+ "loss": 1.2544,
2781
+ "step": 396
2782
+ },
2783
+ {
2784
+ "epoch": 0.29,
2785
+ "grad_norm": 0.5365767973671521,
2786
+ "learning_rate": 8.290723926272439e-05,
2787
+ "loss": 1.2378,
2788
+ "step": 397
2789
+ },
2790
+ {
2791
+ "epoch": 0.29,
2792
+ "grad_norm": 0.5505960932890972,
2793
+ "learning_rate": 8.281700483973421e-05,
2794
+ "loss": 1.2471,
2795
+ "step": 398
2796
+ },
2797
+ {
2798
+ "epoch": 0.29,
2799
+ "grad_norm": 0.5479428166637395,
2800
+ "learning_rate": 8.272658225522569e-05,
2801
+ "loss": 1.2607,
2802
+ "step": 399
2803
+ },
2804
+ {
2805
+ "epoch": 0.3,
2806
+ "grad_norm": 0.5764125413085645,
2807
+ "learning_rate": 8.263597202765109e-05,
2808
+ "loss": 1.2888,
2809
+ "step": 400
2810
+ },
2811
+ {
2812
+ "epoch": 0.3,
2813
+ "grad_norm": 0.5193462362673806,
2814
+ "learning_rate": 8.254517467653858e-05,
2815
+ "loss": 1.1882,
2816
+ "step": 401
2817
+ },
2818
+ {
2819
+ "epoch": 0.3,
2820
+ "grad_norm": 0.5374168368793678,
2821
+ "learning_rate": 8.245419072248919e-05,
2822
+ "loss": 1.2358,
2823
+ "step": 402
2824
+ },
2825
+ {
2826
+ "epoch": 0.3,
2827
+ "grad_norm": 0.5560345573494497,
2828
+ "learning_rate": 8.236302068717392e-05,
2829
+ "loss": 1.3,
2830
+ "step": 403
2831
+ },
2832
+ {
2833
+ "epoch": 0.3,
2834
+ "grad_norm": 0.5223138605512301,
2835
+ "learning_rate": 8.227166509333068e-05,
2836
+ "loss": 1.2559,
2837
+ "step": 404
2838
+ },
2839
+ {
2840
+ "epoch": 0.3,
2841
+ "grad_norm": 0.5009208364979428,
2842
+ "learning_rate": 8.218012446476128e-05,
2843
+ "loss": 1.2617,
2844
+ "step": 405
2845
+ },
2846
+ {
2847
+ "epoch": 0.3,
2848
+ "grad_norm": 0.509867725986647,
2849
+ "learning_rate": 8.208839932632849e-05,
2850
+ "loss": 1.2715,
2851
+ "step": 406
2852
+ },
2853
+ {
2854
+ "epoch": 0.3,
2855
+ "grad_norm": 0.5190782935920448,
2856
+ "learning_rate": 8.199649020395298e-05,
2857
+ "loss": 1.2183,
2858
+ "step": 407
2859
+ },
2860
+ {
2861
+ "epoch": 0.3,
2862
+ "grad_norm": 0.551317848502644,
2863
+ "learning_rate": 8.190439762461033e-05,
2864
+ "loss": 1.2241,
2865
+ "step": 408
2866
+ },
2867
+ {
2868
+ "epoch": 0.3,
2869
+ "grad_norm": 0.5299140869699253,
2870
+ "learning_rate": 8.181212211632799e-05,
2871
+ "loss": 1.1746,
2872
+ "step": 409
2873
+ },
2874
+ {
2875
+ "epoch": 0.3,
2876
+ "grad_norm": 0.5161200175965883,
2877
+ "learning_rate": 8.171966420818228e-05,
2878
+ "loss": 1.2544,
2879
+ "step": 410
2880
+ },
2881
+ {
2882
+ "epoch": 0.3,
2883
+ "grad_norm": 0.5368310977870265,
2884
+ "learning_rate": 8.162702443029531e-05,
2885
+ "loss": 1.2505,
2886
+ "step": 411
2887
+ },
2888
+ {
2889
+ "epoch": 0.3,
2890
+ "grad_norm": 0.5392135585371384,
2891
+ "learning_rate": 8.153420331383199e-05,
2892
+ "loss": 1.2378,
2893
+ "step": 412
2894
+ },
2895
+ {
2896
+ "epoch": 0.31,
2897
+ "grad_norm": 0.5652426070182841,
2898
+ "learning_rate": 8.144120139099697e-05,
2899
+ "loss": 1.2788,
2900
+ "step": 413
2901
+ },
2902
+ {
2903
+ "epoch": 0.31,
2904
+ "grad_norm": 0.5264883521440279,
2905
+ "learning_rate": 8.134801919503154e-05,
2906
+ "loss": 1.2432,
2907
+ "step": 414
2908
+ },
2909
+ {
2910
+ "epoch": 0.31,
2911
+ "grad_norm": 0.5391198787958846,
2912
+ "learning_rate": 8.125465726021069e-05,
2913
+ "loss": 1.2642,
2914
+ "step": 415
2915
+ },
2916
+ {
2917
+ "epoch": 0.31,
2918
+ "grad_norm": 0.5447234901673647,
2919
+ "learning_rate": 8.116111612183989e-05,
2920
+ "loss": 1.2598,
2921
+ "step": 416
2922
+ },
2923
+ {
2924
+ "epoch": 0.31,
2925
+ "grad_norm": 0.5239448356746366,
2926
+ "learning_rate": 8.106739631625217e-05,
2927
+ "loss": 1.2383,
2928
+ "step": 417
2929
+ },
2930
+ {
2931
+ "epoch": 0.31,
2932
+ "grad_norm": 0.522466994953917,
2933
+ "learning_rate": 8.09734983808049e-05,
2934
+ "loss": 1.21,
2935
+ "step": 418
2936
+ },
2937
+ {
2938
+ "epoch": 0.31,
2939
+ "grad_norm": 0.49320728726020635,
2940
+ "learning_rate": 8.087942285387688e-05,
2941
+ "loss": 1.1643,
2942
+ "step": 419
2943
+ },
2944
+ {
2945
+ "epoch": 0.31,
2946
+ "grad_norm": 0.538615135680076,
2947
+ "learning_rate": 8.07851702748651e-05,
2948
+ "loss": 1.2485,
2949
+ "step": 420
2950
+ },
2951
+ {
2952
+ "epoch": 0.31,
2953
+ "grad_norm": 0.5546864636999657,
2954
+ "learning_rate": 8.06907411841817e-05,
2955
+ "loss": 1.1887,
2956
+ "step": 421
2957
+ },
2958
+ {
2959
+ "epoch": 0.31,
2960
+ "grad_norm": 0.5337150121699967,
2961
+ "learning_rate": 8.05961361232509e-05,
2962
+ "loss": 1.2378,
2963
+ "step": 422
2964
+ },
2965
+ {
2966
+ "epoch": 0.31,
2967
+ "grad_norm": 0.5548120199862732,
2968
+ "learning_rate": 8.050135563450587e-05,
2969
+ "loss": 1.2129,
2970
+ "step": 423
2971
+ },
2972
+ {
2973
+ "epoch": 0.31,
2974
+ "grad_norm": 0.5491477319207145,
2975
+ "learning_rate": 8.040640026138562e-05,
2976
+ "loss": 1.2615,
2977
+ "step": 424
2978
+ },
2979
+ {
2980
+ "epoch": 0.31,
2981
+ "grad_norm": 0.5292609791678348,
2982
+ "learning_rate": 8.03112705483319e-05,
2983
+ "loss": 1.1963,
2984
+ "step": 425
2985
+ },
2986
+ {
2987
+ "epoch": 0.31,
2988
+ "grad_norm": 0.5386073890465884,
2989
+ "learning_rate": 8.021596704078605e-05,
2990
+ "loss": 1.2822,
2991
+ "step": 426
2992
+ },
2993
+ {
2994
+ "epoch": 0.32,
2995
+ "grad_norm": 0.5208877771953219,
2996
+ "learning_rate": 8.012049028518589e-05,
2997
+ "loss": 1.2468,
2998
+ "step": 427
2999
+ },
3000
+ {
3001
+ "epoch": 0.32,
3002
+ "grad_norm": 0.5300893442105213,
3003
+ "learning_rate": 8.002484082896257e-05,
3004
+ "loss": 1.2141,
3005
+ "step": 428
3006
+ },
3007
+ {
3008
+ "epoch": 0.32,
3009
+ "grad_norm": 0.5426660622332912,
3010
+ "learning_rate": 7.992901922053752e-05,
3011
+ "loss": 1.2083,
3012
+ "step": 429
3013
+ },
3014
+ {
3015
+ "epoch": 0.32,
3016
+ "grad_norm": 0.5280778314237736,
3017
+ "learning_rate": 7.983302600931911e-05,
3018
+ "loss": 1.2556,
3019
+ "step": 430
3020
+ },
3021
+ {
3022
+ "epoch": 0.32,
3023
+ "grad_norm": 0.5303015472910759,
3024
+ "learning_rate": 7.973686174569972e-05,
3025
+ "loss": 1.2246,
3026
+ "step": 431
3027
+ },
3028
+ {
3029
+ "epoch": 0.32,
3030
+ "grad_norm": 0.5385117857553907,
3031
+ "learning_rate": 7.964052698105247e-05,
3032
+ "loss": 1.2544,
3033
+ "step": 432
3034
+ },
3035
+ {
3036
+ "epoch": 0.32,
3037
+ "grad_norm": 0.5175160927509813,
3038
+ "learning_rate": 7.954402226772804e-05,
3039
+ "loss": 1.1724,
3040
+ "step": 433
3041
+ },
3042
+ {
3043
+ "epoch": 0.32,
3044
+ "grad_norm": 0.5167307050244405,
3045
+ "learning_rate": 7.944734815905154e-05,
3046
+ "loss": 1.228,
3047
+ "step": 434
3048
+ },
3049
+ {
3050
+ "epoch": 0.32,
3051
+ "grad_norm": 0.533666702216578,
3052
+ "learning_rate": 7.93505052093194e-05,
3053
+ "loss": 1.2349,
3054
+ "step": 435
3055
+ },
3056
+ {
3057
+ "epoch": 0.32,
3058
+ "grad_norm": 0.5259498652131873,
3059
+ "learning_rate": 7.925349397379604e-05,
3060
+ "loss": 1.2415,
3061
+ "step": 436
3062
+ },
3063
+ {
3064
+ "epoch": 0.32,
3065
+ "grad_norm": 0.5445977576017799,
3066
+ "learning_rate": 7.915631500871083e-05,
3067
+ "loss": 1.2065,
3068
+ "step": 437
3069
+ },
3070
+ {
3071
+ "epoch": 0.32,
3072
+ "grad_norm": 0.5649990455410109,
3073
+ "learning_rate": 7.905896887125482e-05,
3074
+ "loss": 1.2417,
3075
+ "step": 438
3076
+ },
3077
+ {
3078
+ "epoch": 0.32,
3079
+ "grad_norm": 0.5260513948557283,
3080
+ "learning_rate": 7.896145611957759e-05,
3081
+ "loss": 1.1918,
3082
+ "step": 439
3083
+ },
3084
+ {
3085
+ "epoch": 0.33,
3086
+ "grad_norm": 0.5258410063287358,
3087
+ "learning_rate": 7.8863777312784e-05,
3088
+ "loss": 1.2124,
3089
+ "step": 440
3090
+ },
3091
+ {
3092
+ "epoch": 0.33,
3093
+ "grad_norm": 0.5434644442116746,
3094
+ "learning_rate": 7.876593301093104e-05,
3095
+ "loss": 1.2349,
3096
+ "step": 441
3097
+ },
3098
+ {
3099
+ "epoch": 0.33,
3100
+ "grad_norm": 0.5462561748612222,
3101
+ "learning_rate": 7.866792377502457e-05,
3102
+ "loss": 1.2373,
3103
+ "step": 442
3104
+ },
3105
+ {
3106
+ "epoch": 0.33,
3107
+ "grad_norm": 0.5661256454549024,
3108
+ "learning_rate": 7.856975016701615e-05,
3109
+ "loss": 1.2334,
3110
+ "step": 443
3111
+ },
3112
+ {
3113
+ "epoch": 0.33,
3114
+ "grad_norm": 0.5517524055311237,
3115
+ "learning_rate": 7.847141274979977e-05,
3116
+ "loss": 1.2549,
3117
+ "step": 444
3118
+ },
3119
+ {
3120
+ "epoch": 0.33,
3121
+ "grad_norm": 0.5588533911643465,
3122
+ "learning_rate": 7.837291208720866e-05,
3123
+ "loss": 1.248,
3124
+ "step": 445
3125
+ },
3126
+ {
3127
+ "epoch": 0.33,
3128
+ "grad_norm": 0.5432341108696274,
3129
+ "learning_rate": 7.827424874401203e-05,
3130
+ "loss": 1.207,
3131
+ "step": 446
3132
+ },
3133
+ {
3134
+ "epoch": 0.33,
3135
+ "grad_norm": 0.5185655878803792,
3136
+ "learning_rate": 7.81754232859119e-05,
3137
+ "loss": 1.2087,
3138
+ "step": 447
3139
+ },
3140
+ {
3141
+ "epoch": 0.33,
3142
+ "grad_norm": 0.546989000271988,
3143
+ "learning_rate": 7.807643627953969e-05,
3144
+ "loss": 1.2852,
3145
+ "step": 448
3146
+ },
3147
+ {
3148
+ "epoch": 0.33,
3149
+ "grad_norm": 0.5609807732483688,
3150
+ "learning_rate": 7.797728829245321e-05,
3151
+ "loss": 1.23,
3152
+ "step": 449
3153
+ },
3154
+ {
3155
+ "epoch": 0.33,
3156
+ "grad_norm": 0.5290536891546959,
3157
+ "learning_rate": 7.787797989313317e-05,
3158
+ "loss": 1.1687,
3159
+ "step": 450
3160
+ },
3161
+ {
3162
+ "epoch": 0.33,
3163
+ "grad_norm": 0.527486366572943,
3164
+ "learning_rate": 7.777851165098012e-05,
3165
+ "loss": 1.2349,
3166
+ "step": 451
3167
+ },
3168
+ {
3169
+ "epoch": 0.33,
3170
+ "grad_norm": 0.5444668761845415,
3171
+ "learning_rate": 7.767888413631101e-05,
3172
+ "loss": 1.248,
3173
+ "step": 452
3174
+ },
3175
+ {
3176
+ "epoch": 0.33,
3177
+ "grad_norm": 0.5194113505588946,
3178
+ "learning_rate": 7.757909792035608e-05,
3179
+ "loss": 1.3081,
3180
+ "step": 453
3181
+ },
3182
+ {
3183
+ "epoch": 0.34,
3184
+ "grad_norm": 0.5174613130879753,
3185
+ "learning_rate": 7.747915357525545e-05,
3186
+ "loss": 1.2046,
3187
+ "step": 454
3188
+ },
3189
+ {
3190
+ "epoch": 0.34,
3191
+ "grad_norm": 0.5535670191712191,
3192
+ "learning_rate": 7.737905167405595e-05,
3193
+ "loss": 1.2136,
3194
+ "step": 455
3195
+ },
3196
+ {
3197
+ "epoch": 0.34,
3198
+ "grad_norm": 0.546209627520353,
3199
+ "learning_rate": 7.727879279070773e-05,
3200
+ "loss": 1.2097,
3201
+ "step": 456
3202
+ },
3203
+ {
3204
+ "epoch": 0.34,
3205
+ "grad_norm": 0.5221397456131871,
3206
+ "learning_rate": 7.717837750006106e-05,
3207
+ "loss": 1.2832,
3208
+ "step": 457
3209
+ },
3210
+ {
3211
+ "epoch": 0.34,
3212
+ "grad_norm": 0.5380906003507856,
3213
+ "learning_rate": 7.7077806377863e-05,
3214
+ "loss": 1.1807,
3215
+ "step": 458
3216
+ },
3217
+ {
3218
+ "epoch": 0.34,
3219
+ "grad_norm": 0.546159089637007,
3220
+ "learning_rate": 7.697708000075403e-05,
3221
+ "loss": 1.262,
3222
+ "step": 459
3223
+ },
3224
+ {
3225
+ "epoch": 0.34,
3226
+ "grad_norm": 0.5378903447286532,
3227
+ "learning_rate": 7.687619894626493e-05,
3228
+ "loss": 1.2639,
3229
+ "step": 460
3230
+ },
3231
+ {
3232
+ "epoch": 0.34,
3233
+ "grad_norm": 0.5183593724417229,
3234
+ "learning_rate": 7.677516379281321e-05,
3235
+ "loss": 1.2344,
3236
+ "step": 461
3237
+ },
3238
+ {
3239
+ "epoch": 0.34,
3240
+ "grad_norm": 0.5110004203240966,
3241
+ "learning_rate": 7.667397511970005e-05,
3242
+ "loss": 1.2144,
3243
+ "step": 462
3244
+ },
3245
+ {
3246
+ "epoch": 0.34,
3247
+ "grad_norm": 0.5237401648978784,
3248
+ "learning_rate": 7.657263350710676e-05,
3249
+ "loss": 1.1992,
3250
+ "step": 463
3251
+ },
3252
+ {
3253
+ "epoch": 0.34,
3254
+ "grad_norm": 0.5458624581753624,
3255
+ "learning_rate": 7.647113953609163e-05,
3256
+ "loss": 1.252,
3257
+ "step": 464
3258
+ },
3259
+ {
3260
+ "epoch": 0.34,
3261
+ "grad_norm": 0.55612272064723,
3262
+ "learning_rate": 7.636949378858646e-05,
3263
+ "loss": 1.188,
3264
+ "step": 465
3265
+ },
3266
+ {
3267
+ "epoch": 0.34,
3268
+ "grad_norm": 0.5578526299155908,
3269
+ "learning_rate": 7.626769684739337e-05,
3270
+ "loss": 1.1951,
3271
+ "step": 466
3272
+ },
3273
+ {
3274
+ "epoch": 0.35,
3275
+ "grad_norm": 0.5092511020982519,
3276
+ "learning_rate": 7.616574929618125e-05,
3277
+ "loss": 1.1543,
3278
+ "step": 467
3279
+ },
3280
+ {
3281
+ "epoch": 0.35,
3282
+ "grad_norm": 0.5348616024567703,
3283
+ "learning_rate": 7.606365171948267e-05,
3284
+ "loss": 1.2368,
3285
+ "step": 468
3286
+ },
3287
+ {
3288
+ "epoch": 0.35,
3289
+ "grad_norm": 0.532298079012496,
3290
+ "learning_rate": 7.596140470269029e-05,
3291
+ "loss": 1.2107,
3292
+ "step": 469
3293
+ },
3294
+ {
3295
+ "epoch": 0.35,
3296
+ "grad_norm": 0.5514395726265122,
3297
+ "learning_rate": 7.585900883205364e-05,
3298
+ "loss": 1.241,
3299
+ "step": 470
3300
+ },
3301
+ {
3302
+ "epoch": 0.35,
3303
+ "grad_norm": 0.5539874834294591,
3304
+ "learning_rate": 7.575646469467575e-05,
3305
+ "loss": 1.2249,
3306
+ "step": 471
3307
+ },
3308
+ {
3309
+ "epoch": 0.35,
3310
+ "grad_norm": 0.5141238427544136,
3311
+ "learning_rate": 7.565377287850977e-05,
3312
+ "loss": 1.21,
3313
+ "step": 472
3314
+ },
3315
+ {
3316
+ "epoch": 0.35,
3317
+ "grad_norm": 0.526119772429715,
3318
+ "learning_rate": 7.555093397235552e-05,
3319
+ "loss": 1.2141,
3320
+ "step": 473
3321
+ },
3322
+ {
3323
+ "epoch": 0.35,
3324
+ "grad_norm": 0.5239544155150679,
3325
+ "learning_rate": 7.544794856585626e-05,
3326
+ "loss": 1.2446,
3327
+ "step": 474
3328
+ },
3329
+ {
3330
+ "epoch": 0.35,
3331
+ "grad_norm": 0.5116743183638587,
3332
+ "learning_rate": 7.53448172494952e-05,
3333
+ "loss": 1.2251,
3334
+ "step": 475
3335
+ },
3336
+ {
3337
+ "epoch": 0.35,
3338
+ "grad_norm": 0.5465278452905271,
3339
+ "learning_rate": 7.524154061459215e-05,
3340
+ "loss": 1.1744,
3341
+ "step": 476
3342
+ },
3343
+ {
3344
+ "epoch": 0.35,
3345
+ "grad_norm": 0.5242898434746838,
3346
+ "learning_rate": 7.51381192533001e-05,
3347
+ "loss": 1.2305,
3348
+ "step": 477
3349
+ },
3350
+ {
3351
+ "epoch": 0.35,
3352
+ "grad_norm": 0.5524906450650563,
3353
+ "learning_rate": 7.503455375860192e-05,
3354
+ "loss": 1.271,
3355
+ "step": 478
3356
+ },
3357
+ {
3358
+ "epoch": 0.35,
3359
+ "grad_norm": 0.5422094091125237,
3360
+ "learning_rate": 7.493084472430682e-05,
3361
+ "loss": 1.2983,
3362
+ "step": 479
3363
+ },
3364
+ {
3365
+ "epoch": 0.35,
3366
+ "grad_norm": 0.5100606069460412,
3367
+ "learning_rate": 7.482699274504708e-05,
3368
+ "loss": 1.1914,
3369
+ "step": 480
3370
+ },
3371
+ {
3372
+ "epoch": 0.36,
3373
+ "grad_norm": 0.5258246755815246,
3374
+ "learning_rate": 7.472299841627451e-05,
3375
+ "loss": 1.1948,
3376
+ "step": 481
3377
+ },
3378
+ {
3379
+ "epoch": 0.36,
3380
+ "grad_norm": 0.5183104456102203,
3381
+ "learning_rate": 7.461886233425717e-05,
3382
+ "loss": 1.1658,
3383
+ "step": 482
3384
+ },
3385
+ {
3386
+ "epoch": 0.36,
3387
+ "grad_norm": 0.5283305385961874,
3388
+ "learning_rate": 7.451458509607582e-05,
3389
+ "loss": 1.2378,
3390
+ "step": 483
3391
+ },
3392
+ {
3393
+ "epoch": 0.36,
3394
+ "grad_norm": 0.5552677702446687,
3395
+ "learning_rate": 7.441016729962064e-05,
3396
+ "loss": 1.1938,
3397
+ "step": 484
3398
+ },
3399
+ {
3400
+ "epoch": 0.36,
3401
+ "grad_norm": 0.5198625616185957,
3402
+ "learning_rate": 7.430560954358764e-05,
3403
+ "loss": 1.2515,
3404
+ "step": 485
3405
+ },
3406
+ {
3407
+ "epoch": 0.36,
3408
+ "grad_norm": 0.524907115545136,
3409
+ "learning_rate": 7.420091242747536e-05,
3410
+ "loss": 1.2437,
3411
+ "step": 486
3412
+ },
3413
+ {
3414
+ "epoch": 0.36,
3415
+ "grad_norm": 0.520819742542826,
3416
+ "learning_rate": 7.409607655158139e-05,
3417
+ "loss": 1.2764,
3418
+ "step": 487
3419
+ },
3420
+ {
3421
+ "epoch": 0.36,
3422
+ "grad_norm": 0.5297968503831433,
3423
+ "learning_rate": 7.399110251699887e-05,
3424
+ "loss": 1.2529,
3425
+ "step": 488
3426
+ },
3427
+ {
3428
+ "epoch": 0.36,
3429
+ "grad_norm": 0.5214545833543685,
3430
+ "learning_rate": 7.388599092561315e-05,
3431
+ "loss": 1.2979,
3432
+ "step": 489
3433
+ },
3434
+ {
3435
+ "epoch": 0.36,
3436
+ "grad_norm": 0.5158994351772959,
3437
+ "learning_rate": 7.378074238009826e-05,
3438
+ "loss": 1.2363,
3439
+ "step": 490
3440
+ },
3441
+ {
3442
+ "epoch": 0.36,
3443
+ "grad_norm": 0.49265767229951024,
3444
+ "learning_rate": 7.367535748391349e-05,
3445
+ "loss": 1.228,
3446
+ "step": 491
3447
+ },
3448
+ {
3449
+ "epoch": 0.36,
3450
+ "grad_norm": 0.5308141896787576,
3451
+ "learning_rate": 7.35698368412999e-05,
3452
+ "loss": 1.2527,
3453
+ "step": 492
3454
+ },
3455
+ {
3456
+ "epoch": 0.36,
3457
+ "grad_norm": 0.5185543266636785,
3458
+ "learning_rate": 7.346418105727686e-05,
3459
+ "loss": 1.2192,
3460
+ "step": 493
3461
+ },
3462
+ {
3463
+ "epoch": 0.37,
3464
+ "grad_norm": 0.5231300605729964,
3465
+ "learning_rate": 7.335839073763865e-05,
3466
+ "loss": 1.2065,
3467
+ "step": 494
3468
+ },
3469
+ {
3470
+ "epoch": 0.37,
3471
+ "grad_norm": 0.5399567824066669,
3472
+ "learning_rate": 7.325246648895088e-05,
3473
+ "loss": 1.2563,
3474
+ "step": 495
3475
+ },
3476
+ {
3477
+ "epoch": 0.37,
3478
+ "grad_norm": 0.5239942836551379,
3479
+ "learning_rate": 7.31464089185471e-05,
3480
+ "loss": 1.2549,
3481
+ "step": 496
3482
+ },
3483
+ {
3484
+ "epoch": 0.37,
3485
+ "grad_norm": 0.5367247940798874,
3486
+ "learning_rate": 7.304021863452524e-05,
3487
+ "loss": 1.2061,
3488
+ "step": 497
3489
+ },
3490
+ {
3491
+ "epoch": 0.37,
3492
+ "grad_norm": 0.5404506218621764,
3493
+ "learning_rate": 7.293389624574422e-05,
3494
+ "loss": 1.2142,
3495
+ "step": 498
3496
+ },
3497
+ {
3498
+ "epoch": 0.37,
3499
+ "grad_norm": 0.5055969660442964,
3500
+ "learning_rate": 7.282744236182034e-05,
3501
+ "loss": 1.2451,
3502
+ "step": 499
3503
+ },
3504
+ {
3505
+ "epoch": 0.37,
3506
+ "grad_norm": 0.5423433133756662,
3507
+ "learning_rate": 7.27208575931239e-05,
3508
+ "loss": 1.2012,
3509
+ "step": 500
3510
+ },
3511
+ {
3512
+ "epoch": 0.37,
3513
+ "grad_norm": 0.5291351969461193,
3514
+ "learning_rate": 7.26141425507756e-05,
3515
+ "loss": 1.1768,
3516
+ "step": 501
3517
+ },
3518
+ {
3519
+ "epoch": 0.37,
3520
+ "grad_norm": 0.5217703642849318,
3521
+ "learning_rate": 7.250729784664316e-05,
3522
+ "loss": 1.209,
3523
+ "step": 502
3524
+ },
3525
+ {
3526
+ "epoch": 0.37,
3527
+ "grad_norm": 0.5201622197991884,
3528
+ "learning_rate": 7.240032409333764e-05,
3529
+ "loss": 1.2031,
3530
+ "step": 503
3531
+ },
3532
+ {
3533
+ "epoch": 0.37,
3534
+ "grad_norm": 0.5281271991799672,
3535
+ "learning_rate": 7.22932219042101e-05,
3536
+ "loss": 1.1987,
3537
+ "step": 504
3538
+ },
3539
+ {
3540
+ "epoch": 0.37,
3541
+ "grad_norm": 0.5573441678253518,
3542
+ "learning_rate": 7.218599189334799e-05,
3543
+ "loss": 1.2739,
3544
+ "step": 505
3545
+ },
3546
+ {
3547
+ "epoch": 0.37,
3548
+ "grad_norm": 0.5665017191299871,
3549
+ "learning_rate": 7.207863467557162e-05,
3550
+ "loss": 1.2773,
3551
+ "step": 506
3552
+ },
3553
+ {
3554
+ "epoch": 0.37,
3555
+ "grad_norm": 0.5325104774494102,
3556
+ "learning_rate": 7.19711508664307e-05,
3557
+ "loss": 1.209,
3558
+ "step": 507
3559
+ },
3560
+ {
3561
+ "epoch": 0.38,
3562
+ "grad_norm": 0.518792873366363,
3563
+ "learning_rate": 7.186354108220072e-05,
3564
+ "loss": 1.2173,
3565
+ "step": 508
3566
+ },
3567
+ {
3568
+ "epoch": 0.38,
3569
+ "grad_norm": 0.530762745727063,
3570
+ "learning_rate": 7.175580593987951e-05,
3571
+ "loss": 1.2466,
3572
+ "step": 509
3573
+ },
3574
+ {
3575
+ "epoch": 0.38,
3576
+ "grad_norm": 0.5140061528285057,
3577
+ "learning_rate": 7.164794605718366e-05,
3578
+ "loss": 1.2139,
3579
+ "step": 510
3580
+ },
3581
+ {
3582
+ "epoch": 0.38,
3583
+ "grad_norm": 0.5194168189274216,
3584
+ "learning_rate": 7.153996205254495e-05,
3585
+ "loss": 1.2476,
3586
+ "step": 511
3587
+ },
3588
+ {
3589
+ "epoch": 0.38,
3590
+ "grad_norm": 0.5487088087238914,
3591
+ "learning_rate": 7.143185454510686e-05,
3592
+ "loss": 1.2251,
3593
+ "step": 512
3594
+ },
3595
+ {
3596
+ "epoch": 0.38,
3597
+ "grad_norm": 0.49449833617368844,
3598
+ "learning_rate": 7.1323624154721e-05,
3599
+ "loss": 1.2021,
3600
+ "step": 513
3601
+ },
3602
+ {
3603
+ "epoch": 0.38,
3604
+ "grad_norm": 0.5209680110441622,
3605
+ "learning_rate": 7.121527150194349e-05,
3606
+ "loss": 1.229,
3607
+ "step": 514
3608
+ },
3609
+ {
3610
+ "epoch": 0.38,
3611
+ "grad_norm": 0.5179658980514732,
3612
+ "learning_rate": 7.110679720803156e-05,
3613
+ "loss": 1.2324,
3614
+ "step": 515
3615
+ },
3616
+ {
3617
+ "epoch": 0.38,
3618
+ "grad_norm": 0.5237224991500224,
3619
+ "learning_rate": 7.099820189493977e-05,
3620
+ "loss": 1.269,
3621
+ "step": 516
3622
+ },
3623
+ {
3624
+ "epoch": 0.38,
3625
+ "grad_norm": 0.5302189416292129,
3626
+ "learning_rate": 7.088948618531667e-05,
3627
+ "loss": 1.2041,
3628
+ "step": 517
3629
+ },
3630
+ {
3631
+ "epoch": 0.38,
3632
+ "grad_norm": 0.5384341108312423,
3633
+ "learning_rate": 7.078065070250106e-05,
3634
+ "loss": 1.1746,
3635
+ "step": 518
3636
+ },
3637
+ {
3638
+ "epoch": 0.38,
3639
+ "grad_norm": 0.5521437637462966,
3640
+ "learning_rate": 7.067169607051851e-05,
3641
+ "loss": 1.2886,
3642
+ "step": 519
3643
+ },
3644
+ {
3645
+ "epoch": 0.38,
3646
+ "grad_norm": 0.5328288678743964,
3647
+ "learning_rate": 7.056262291407772e-05,
3648
+ "loss": 1.1877,
3649
+ "step": 520
3650
+ },
3651
+ {
3652
+ "epoch": 0.39,
3653
+ "grad_norm": 0.5359494830051162,
3654
+ "learning_rate": 7.045343185856701e-05,
3655
+ "loss": 1.2202,
3656
+ "step": 521
3657
+ },
3658
+ {
3659
+ "epoch": 0.39,
3660
+ "grad_norm": 0.5288532232218185,
3661
+ "learning_rate": 7.034412353005063e-05,
3662
+ "loss": 1.21,
3663
+ "step": 522
3664
+ },
3665
+ {
3666
+ "epoch": 0.39,
3667
+ "grad_norm": 0.5512085122241619,
3668
+ "learning_rate": 7.02346985552653e-05,
3669
+ "loss": 1.2798,
3670
+ "step": 523
3671
+ },
3672
+ {
3673
+ "epoch": 0.39,
3674
+ "grad_norm": 0.533944460040126,
3675
+ "learning_rate": 7.01251575616165e-05,
3676
+ "loss": 1.2539,
3677
+ "step": 524
3678
+ },
3679
+ {
3680
+ "epoch": 0.39,
3681
+ "grad_norm": 0.5837632563221825,
3682
+ "learning_rate": 7.0015501177175e-05,
3683
+ "loss": 1.1335,
3684
+ "step": 525
3685
+ }
3686
+ ],
3687
+ "logging_steps": 1.0,
3688
+ "max_steps": 1353,
3689
+ "num_input_tokens_seen": 0,
3690
+ "num_train_epochs": 1,
3691
+ "save_steps": 25,
3692
+ "total_flos": 4.4085090777437307e+18,
3693
+ "train_batch_size": 16,
3694
+ "trial_name": null,
3695
+ "trial_params": null
3696
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91a5b0aeacf520d00ebefbd60b0266b0ab146e9ad9ce5cb558fac2a023344a6d
3
+ size 6584
zero_to_fp32.py ADDED
@@ -0,0 +1,592 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
215
+ elif zero_stage == 3:
216
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
217
+
218
+
219
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
220
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
221
+ return
222
+
223
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
224
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
225
+
226
+ if debug:
227
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
228
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
229
+
230
+ wanted_params = len(frozen_param_shapes)
231
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
232
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
233
+ print(f'Frozen params: Have {avail_numel} numels to process.')
234
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
235
+
236
+ total_params = 0
237
+ total_numel = 0
238
+ for name, shape in frozen_param_shapes.items():
239
+ total_params += 1
240
+ unpartitioned_numel = shape.numel()
241
+ total_numel += unpartitioned_numel
242
+
243
+ state_dict[name] = frozen_param_fragments[name]
244
+
245
+ if debug:
246
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
247
+
248
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
249
+
250
+
251
+ def _has_callable(obj, fn):
252
+ attr = getattr(obj, fn, None)
253
+ return callable(attr)
254
+
255
+
256
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
257
+ param_shapes = zero_model_states[0].param_shapes
258
+
259
+ # Reconstruction protocol:
260
+ #
261
+ # XXX: document this
262
+
263
+ if debug:
264
+ for i in range(world_size):
265
+ for j in range(len(fp32_flat_groups[0])):
266
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
267
+
268
+ # XXX: memory usage doubles here (zero2)
269
+ num_param_groups = len(fp32_flat_groups[0])
270
+ merged_single_partition_of_fp32_groups = []
271
+ for i in range(num_param_groups):
272
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
273
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
274
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
275
+ avail_numel = sum(
276
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
277
+
278
+ if debug:
279
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
280
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
281
+ # not asserting if there is a mismatch due to possible padding
282
+ print(f"Have {avail_numel} numels to process.")
283
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
284
+
285
+ # params
286
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
287
+ # out-of-core computing solution
288
+ total_numel = 0
289
+ total_params = 0
290
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
291
+ offset = 0
292
+ avail_numel = full_single_fp32_vector.numel()
293
+ for name, shape in shapes.items():
294
+
295
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
296
+ total_numel += unpartitioned_numel
297
+ total_params += 1
298
+
299
+ if debug:
300
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
301
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
302
+ offset += unpartitioned_numel
303
+
304
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
305
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
306
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
307
+ # live optimizer object, so we are checking that the numbers are within the right range
308
+ align_to = 2 * world_size
309
+
310
+ def zero2_align(x):
311
+ return align_to * math.ceil(x / align_to)
312
+
313
+ if debug:
314
+ print(f"original offset={offset}, avail_numel={avail_numel}")
315
+
316
+ offset = zero2_align(offset)
317
+ avail_numel = zero2_align(avail_numel)
318
+
319
+ if debug:
320
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
321
+
322
+ # Sanity check
323
+ if offset != avail_numel:
324
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
325
+
326
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
327
+
328
+
329
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
330
+ state_dict = OrderedDict()
331
+
332
+ # buffers
333
+ buffers = zero_model_states[0].buffers
334
+ state_dict.update(buffers)
335
+ if debug:
336
+ print(f"added {len(buffers)} buffers")
337
+
338
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
339
+
340
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
341
+
342
+ # recover shared parameters
343
+ for pair in zero_model_states[0].shared_params:
344
+ if pair[1] in state_dict:
345
+ state_dict[pair[0]] = state_dict[pair[1]]
346
+
347
+ return state_dict
348
+
349
+
350
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
351
+ remainder = unpartitioned_numel % world_size
352
+ padding_numel = (world_size - remainder) if remainder else 0
353
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
354
+ return partitioned_numel, padding_numel
355
+
356
+
357
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
358
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
359
+ return
360
+
361
+ if debug:
362
+ for i in range(world_size):
363
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
364
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
365
+
366
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
367
+ wanted_params = len(frozen_param_shapes)
368
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
369
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
370
+ print(f'Frozen params: Have {avail_numel} numels to process.')
371
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
372
+
373
+ total_params = 0
374
+ total_numel = 0
375
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
376
+ total_params += 1
377
+ unpartitioned_numel = shape.numel()
378
+ total_numel += unpartitioned_numel
379
+
380
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
381
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
382
+
383
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
384
+
385
+ if debug:
386
+ print(
387
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
388
+ )
389
+
390
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
391
+
392
+
393
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
394
+ param_shapes = zero_model_states[0].param_shapes
395
+ avail_numel = fp32_flat_groups[0].numel() * world_size
396
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
397
+ # param, re-consolidating each param, while dealing with padding if any
398
+
399
+ # merge list of dicts, preserving order
400
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
401
+
402
+ if debug:
403
+ for i in range(world_size):
404
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
405
+
406
+ wanted_params = len(param_shapes)
407
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
408
+ # not asserting if there is a mismatch due to possible padding
409
+ avail_numel = fp32_flat_groups[0].numel() * world_size
410
+ print(f"Trainable params: Have {avail_numel} numels to process.")
411
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
412
+
413
+ # params
414
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
415
+ # out-of-core computing solution
416
+ offset = 0
417
+ total_numel = 0
418
+ total_params = 0
419
+ for name, shape in param_shapes.items():
420
+
421
+ unpartitioned_numel = shape.numel()
422
+ total_numel += unpartitioned_numel
423
+ total_params += 1
424
+
425
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
426
+
427
+ if debug:
428
+ print(
429
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
430
+ )
431
+
432
+ # XXX: memory usage doubles here
433
+ state_dict[name] = torch.cat(
434
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
435
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
436
+ offset += partitioned_numel
437
+
438
+ offset *= world_size
439
+
440
+ # Sanity check
441
+ if offset != avail_numel:
442
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
443
+
444
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
445
+
446
+
447
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
448
+ state_dict = OrderedDict()
449
+
450
+ # buffers
451
+ buffers = zero_model_states[0].buffers
452
+ state_dict.update(buffers)
453
+ if debug:
454
+ print(f"added {len(buffers)} buffers")
455
+
456
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
457
+
458
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
459
+
460
+ # recover shared parameters
461
+ for pair in zero_model_states[0].shared_params:
462
+ if pair[1] in state_dict:
463
+ state_dict[pair[0]] = state_dict[pair[1]]
464
+
465
+ return state_dict
466
+
467
+
468
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
469
+ """
470
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
471
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
472
+ via a model hub.
473
+
474
+ Args:
475
+ - ``checkpoint_dir``: path to the desired checkpoint folder
476
+ - ``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``
477
+
478
+ Returns:
479
+ - pytorch ``state_dict``
480
+
481
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
482
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
483
+ the checkpoint.
484
+
485
+ A typical usage might be ::
486
+
487
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
488
+ # do the training and checkpoint saving
489
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
490
+ model = model.cpu() # move to cpu
491
+ model.load_state_dict(state_dict)
492
+ # submit to model hub or save the model to share with others
493
+
494
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
495
+ application. i.e. you will need to re-initialize the deepspeed engine, since
496
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
497
+
498
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
499
+
500
+ """
501
+ if tag is None:
502
+ latest_path = os.path.join(checkpoint_dir, 'latest')
503
+ if os.path.isfile(latest_path):
504
+ with open(latest_path, 'r') as fd:
505
+ tag = fd.read().strip()
506
+ else:
507
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
508
+
509
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
510
+
511
+ if not os.path.isdir(ds_checkpoint_dir):
512
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
513
+
514
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
515
+
516
+
517
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
518
+ """
519
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
520
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
521
+
522
+ Args:
523
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
524
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
525
+ - ``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``
526
+ """
527
+
528
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
529
+ print(f"Saving fp32 state dict to {output_file}")
530
+ torch.save(state_dict, output_file)
531
+
532
+
533
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
534
+ """
535
+ 1. Put the provided model to cpu
536
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
537
+ 3. Load it into the provided model
538
+
539
+ Args:
540
+ - ``model``: the model object to update
541
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
542
+ - ``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``
543
+
544
+ Returns:
545
+ - ``model`: modified model
546
+
547
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
548
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
549
+ conveniently placed for you in the checkpoint folder.
550
+
551
+ A typical usage might be ::
552
+
553
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
554
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
555
+ # submit to model hub or save the model to share with others
556
+
557
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
558
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
559
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
560
+
561
+ """
562
+ logger.info(f"Extracting fp32 weights")
563
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
564
+
565
+ logger.info(f"Overwriting model with fp32 weights")
566
+ model = model.cpu()
567
+ model.load_state_dict(state_dict, strict=False)
568
+
569
+ return model
570
+
571
+
572
+ if __name__ == "__main__":
573
+
574
+ parser = argparse.ArgumentParser()
575
+ parser.add_argument("checkpoint_dir",
576
+ type=str,
577
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
578
+ parser.add_argument(
579
+ "output_file",
580
+ type=str,
581
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
582
+ parser.add_argument("-t",
583
+ "--tag",
584
+ type=str,
585
+ default=None,
586
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
587
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
588
+ args = parser.parse_args()
589
+
590
+ debug = args.debug
591
+
592
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file, tag=args.tag)