Hugging Face's logo

lwhalen7
/
data4elm-SLaM-submission

PEFT
Safetensors
lora
causal-lm
elmb
function-calling
Model card Files
xet
Community
lwhalen7 commited on
Commit
f94d935
·
verified ·
1 Parent(s): 606e4c7

Migrate LoRA adapter from dataset repository

Browse files
Files changed (34) hide show
  1. README.md +63 -0
  2. adapter_config.json +41 -0
  3. adapter_model.safetensors +3 -0
  4. all_results.json +9 -0
  5. checkpoint-10000/adapter_config.json +41 -0
  6. checkpoint-10000/adapter_model.safetensors +3 -0
  7. checkpoint-10000/special_tokens_map.json +24 -0
  8. checkpoint-10000/tokenizer.json +0 -0
  9. checkpoint-10000/tokenizer_config.json +43 -0
  10. checkpoint-10000/trainer_state.json +3534 -0
  11. checkpoint-13563/adapter_config.json +41 -0
  12. checkpoint-13563/adapter_model.safetensors +3 -0
  13. checkpoint-13563/special_tokens_map.json +24 -0
  14. checkpoint-13563/tokenizer.json +0 -0
  15. checkpoint-13563/tokenizer_config.json +43 -0
  16. checkpoint-13563/trainer_state.json +0 -0
  17. checkpoint-3391/adapter_config.json +41 -0
  18. checkpoint-3391/adapter_model.safetensors +3 -0
  19. checkpoint-3391/special_tokens_map.json +24 -0
  20. checkpoint-3391/tokenizer.json +0 -0
  21. checkpoint-3391/tokenizer_config.json +43 -0
  22. checkpoint-3391/trainer_state.json +1217 -0
  23. checkpoint-5000/adapter_config.json +41 -0
  24. checkpoint-5000/adapter_model.safetensors +3 -0
  25. checkpoint-5000/special_tokens_map.json +24 -0
  26. checkpoint-5000/tokenizer.json +0 -0
  27. checkpoint-5000/tokenizer_config.json +43 -0
  28. checkpoint-5000/trainer_state.json +1784 -0
  29. dataset_info.json +7 -0
  30. special_tokens_map.json +24 -0
  31. tokenizer.json +0 -0
  32. tokenizer_config.json +43 -0
  33. train_results.json +9 -0
  34. trainer_state.json +0 -0
README.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: meta-llama/Llama-2-7b-hf
4
+ tags:
5
+ - lora
6
+ - causal-lm
7
+ - elmb
8
+ - function-calling
9
+ license: apache-2.0
10
+ ---
11
+
12
+ # data4elm-SLaM-submission
13
+
14
+ This is a LoRA adapter trained for the ELMB function calling task.
15
+
16
+ ## Usage
17
+
18
+ ### With PEFT
19
+
20
+ ```python
21
+ from transformers import AutoModelForCausalLM, AutoTokenizer
22
+ from peft import PeftModel
23
+
24
+ # Load base model
25
+ base_model = AutoModelForCausalLM.from_pretrained(
26
+ "meta-llama/Llama-2-7b-hf",
27
+ torch_dtype=torch.float16,
28
+ device_map="auto"
29
+ )
30
+
31
+ # Load LoRA adapter
32
+ model = PeftModel.from_pretrained(
33
+ base_model,
34
+ "lwhalen7/data4elm-SLaM-submission"
35
+ )
36
+
37
+ tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-hf")
38
+
39
+ # Use for inference
40
+ inputs = tokenizer("Your prompt here", return_tensors="pt")
41
+ outputs = model.generate(**inputs)
42
+ print(tokenizer.decode(outputs[0]))
43
+ ```
44
+
45
+ ### With lm-eval
46
+
47
+ ```bash
48
+ lm_eval --model hf \
49
+ --model_args pretrained=meta-llama/Llama-2-7b-hf,peft=lwhalen7/data4elm-SLaM-submission,trust_remote_code=True \
50
+ --tasks elmb_functioncalling_test \
51
+ --device cuda:0 \
52
+ --batch_size 1 \
53
+ --log_samples \
54
+ --output_path ./eval_results/elmb_test_set.jsonl
55
+ ```
56
+
57
+ ## Training Details
58
+
59
+ This adapter was migrated from a dataset repository to enable proper usage with inference tools.
60
+
61
+ ## Base Model
62
+
63
+ This adapter is compatible with: `meta-llama/Llama-2-7b-hf`
adapter_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "data4elm/Llama-400M-12L",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 32,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.1,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 16,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "embed_tokens",
28
+ "gate_proj",
29
+ "down_proj",
30
+ "v_proj",
31
+ "k_proj",
32
+ "up_proj",
33
+ "q_proj",
34
+ "lm_head",
35
+ "o_proj"
36
+ ],
37
+ "task_type": "CAUSAL_LM",
38
+ "trainable_token_indices": null,
39
+ "use_dora": true,
40
+ "use_rslora": false
41
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad8ca1b35f4a1ea23c04e2ec7703a98e2c472464a0abe7fe6711c8af1d70dc1b
3
+ size 177776400
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "total_flos": 5.80008589556659e+18,
4
+ "train_loss": 2.354225961330374,
5
+ "train_runtime": 12020.9549,
6
+ "train_samples": 2604072,
7
+ "train_samples_per_second": 216.628,
8
+ "train_steps_per_second": 1.128
9
+ }
checkpoint-10000/adapter_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "data4elm/Llama-400M-12L",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 32,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.1,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 16,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "embed_tokens",
28
+ "gate_proj",
29
+ "down_proj",
30
+ "v_proj",
31
+ "k_proj",
32
+ "up_proj",
33
+ "q_proj",
34
+ "lm_head",
35
+ "o_proj"
36
+ ],
37
+ "task_type": "CAUSAL_LM",
38
+ "trainable_token_indices": null,
39
+ "use_dora": true,
40
+ "use_rslora": false
41
+ }
checkpoint-10000/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:105ec703320ebc09dab64a2d405ee95a289c75557d303c219688d945017587a3
3
+ size 177776400
checkpoint-10000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-10000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
checkpoint-10000/trainer_state.json ADDED
@@ -0,0 +1,3534 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.737300007373,
6
+ "eval_steps": 500,
7
+ "global_step": 10000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0014746000147460002,
14
+ "grad_norm": 0.4404066503047943,
15
+ "learning_rate": 9.985991299859913e-06,
16
+ "loss": 2.4466,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.0029492000294920003,
21
+ "grad_norm": 0.5653194785118103,
22
+ "learning_rate": 9.971245299712454e-06,
23
+ "loss": 2.4307,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004423800044238001,
28
+ "grad_norm": 0.4618721902370453,
29
+ "learning_rate": 9.956499299564993e-06,
30
+ "loss": 2.4367,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005898400058984001,
35
+ "grad_norm": 0.4391142427921295,
36
+ "learning_rate": 9.941753299417534e-06,
37
+ "loss": 2.4183,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.007373000073730001,
42
+ "grad_norm": 0.474587619304657,
43
+ "learning_rate": 9.927007299270073e-06,
44
+ "loss": 2.42,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008847600088476001,
49
+ "grad_norm": 0.420256644487381,
50
+ "learning_rate": 9.912261299122614e-06,
51
+ "loss": 2.4164,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.010322200103222,
56
+ "grad_norm": 0.40069517493247986,
57
+ "learning_rate": 9.897515298975153e-06,
58
+ "loss": 2.3962,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.011796800117968001,
63
+ "grad_norm": 0.36043497920036316,
64
+ "learning_rate": 9.882769298827694e-06,
65
+ "loss": 2.4026,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.013271400132714002,
70
+ "grad_norm": 0.354769229888916,
71
+ "learning_rate": 9.868023298680233e-06,
72
+ "loss": 2.4013,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.014746000147460001,
77
+ "grad_norm": 0.40864863991737366,
78
+ "learning_rate": 9.853277298532774e-06,
79
+ "loss": 2.4077,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.016220600162206,
84
+ "grad_norm": 0.37636885046958923,
85
+ "learning_rate": 9.838531298385315e-06,
86
+ "loss": 2.4058,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.017695200176952003,
91
+ "grad_norm": 0.37770289182662964,
92
+ "learning_rate": 9.823785298237854e-06,
93
+ "loss": 2.3789,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.019169800191698002,
98
+ "grad_norm": 0.4997946321964264,
99
+ "learning_rate": 9.809039298090393e-06,
100
+ "loss": 2.3921,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.020644400206444,
105
+ "grad_norm": 0.42146745324134827,
106
+ "learning_rate": 9.794293297942934e-06,
107
+ "loss": 2.389,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.022119000221190004,
112
+ "grad_norm": 0.31817182898521423,
113
+ "learning_rate": 9.779547297795475e-06,
114
+ "loss": 2.3969,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.023593600235936003,
119
+ "grad_norm": 0.41462913155555725,
120
+ "learning_rate": 9.764801297648014e-06,
121
+ "loss": 2.3903,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.025068200250682002,
126
+ "grad_norm": 0.35277777910232544,
127
+ "learning_rate": 9.750055297500553e-06,
128
+ "loss": 2.3886,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.026542800265428004,
133
+ "grad_norm": 0.39531123638153076,
134
+ "learning_rate": 9.735309297353094e-06,
135
+ "loss": 2.3847,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.028017400280174003,
140
+ "grad_norm": 0.3687122166156769,
141
+ "learning_rate": 9.720563297205633e-06,
142
+ "loss": 2.3824,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.029492000294920002,
147
+ "grad_norm": 0.5362212061882019,
148
+ "learning_rate": 9.705817297058172e-06,
149
+ "loss": 2.3861,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.030966600309666,
154
+ "grad_norm": 0.5793041586875916,
155
+ "learning_rate": 9.691071296910713e-06,
156
+ "loss": 2.4069,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.032441200324412,
161
+ "grad_norm": 0.3618432581424713,
162
+ "learning_rate": 9.676325296763254e-06,
163
+ "loss": 2.379,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.033915800339158,
168
+ "grad_norm": 0.39441928267478943,
169
+ "learning_rate": 9.661579296615793e-06,
170
+ "loss": 2.3878,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.035390400353904006,
175
+ "grad_norm": 0.3998846113681793,
176
+ "learning_rate": 9.646833296468334e-06,
177
+ "loss": 2.4028,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03686500036865,
182
+ "grad_norm": 0.354809045791626,
183
+ "learning_rate": 9.632087296320873e-06,
184
+ "loss": 2.384,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.038339600383396004,
189
+ "grad_norm": 0.6007148623466492,
190
+ "learning_rate": 9.617341296173414e-06,
191
+ "loss": 2.3974,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.039814200398142006,
196
+ "grad_norm": 0.37758493423461914,
197
+ "learning_rate": 9.602595296025953e-06,
198
+ "loss": 2.3762,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.041288800412888,
203
+ "grad_norm": 0.4141014516353607,
204
+ "learning_rate": 9.587849295878494e-06,
205
+ "loss": 2.3831,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.042763400427634005,
210
+ "grad_norm": 0.48960772156715393,
211
+ "learning_rate": 9.573103295731033e-06,
212
+ "loss": 2.3961,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04423800044238001,
217
+ "grad_norm": 0.3811470568180084,
218
+ "learning_rate": 9.558357295583574e-06,
219
+ "loss": 2.3751,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.045712600457126,
224
+ "grad_norm": 0.35424861311912537,
225
+ "learning_rate": 9.543611295436113e-06,
226
+ "loss": 2.3725,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.047187200471872005,
231
+ "grad_norm": 0.36190420389175415,
232
+ "learning_rate": 9.528865295288654e-06,
233
+ "loss": 2.3944,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.04866180048661801,
238
+ "grad_norm": 0.42110538482666016,
239
+ "learning_rate": 9.514119295141195e-06,
240
+ "loss": 2.3909,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.050136400501364004,
245
+ "grad_norm": 0.4361230134963989,
246
+ "learning_rate": 9.499373294993734e-06,
247
+ "loss": 2.3752,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.051611000516110006,
252
+ "grad_norm": 0.3302260935306549,
253
+ "learning_rate": 9.484627294846273e-06,
254
+ "loss": 2.3699,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.05308560053085601,
259
+ "grad_norm": 0.3521312177181244,
260
+ "learning_rate": 9.469881294698814e-06,
261
+ "loss": 2.3772,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.054560200545602004,
266
+ "grad_norm": 0.3700699806213379,
267
+ "learning_rate": 9.455135294551355e-06,
268
+ "loss": 2.3751,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05603480056034801,
273
+ "grad_norm": 0.3583256006240845,
274
+ "learning_rate": 9.440389294403894e-06,
275
+ "loss": 2.3675,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.057509400575094,
280
+ "grad_norm": 0.33290043473243713,
281
+ "learning_rate": 9.425643294256433e-06,
282
+ "loss": 2.3784,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.058984000589840005,
287
+ "grad_norm": 0.3490481674671173,
288
+ "learning_rate": 9.410897294108974e-06,
289
+ "loss": 2.3597,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.06045860060458601,
294
+ "grad_norm": 0.479775071144104,
295
+ "learning_rate": 9.396151293961515e-06,
296
+ "loss": 2.3852,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.061933200619332,
301
+ "grad_norm": 0.36794590950012207,
302
+ "learning_rate": 9.381405293814054e-06,
303
+ "loss": 2.3748,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.06340780063407801,
308
+ "grad_norm": 0.38288605213165283,
309
+ "learning_rate": 9.366659293666593e-06,
310
+ "loss": 2.3673,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.064882400648824,
315
+ "grad_norm": 0.40629106760025024,
316
+ "learning_rate": 9.351913293519134e-06,
317
+ "loss": 2.3906,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06635700066357,
322
+ "grad_norm": 0.3594074249267578,
323
+ "learning_rate": 9.337167293371675e-06,
324
+ "loss": 2.3773,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.067831600678316,
329
+ "grad_norm": 0.463144451379776,
330
+ "learning_rate": 9.322421293224214e-06,
331
+ "loss": 2.3846,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06930620069306201,
336
+ "grad_norm": 0.35561177134513855,
337
+ "learning_rate": 9.307675293076753e-06,
338
+ "loss": 2.3723,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.07078080070780801,
343
+ "grad_norm": 0.3446957767009735,
344
+ "learning_rate": 9.292929292929294e-06,
345
+ "loss": 2.4011,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.07225540072255401,
350
+ "grad_norm": 0.5307037830352783,
351
+ "learning_rate": 9.278183292781835e-06,
352
+ "loss": 2.3747,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.0737300007373,
357
+ "grad_norm": 0.3605501055717468,
358
+ "learning_rate": 9.263437292634374e-06,
359
+ "loss": 2.3523,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.075204600752046,
364
+ "grad_norm": 0.3705900013446808,
365
+ "learning_rate": 9.248691292486913e-06,
366
+ "loss": 2.3693,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07667920076679201,
371
+ "grad_norm": 0.3397590219974518,
372
+ "learning_rate": 9.233945292339454e-06,
373
+ "loss": 2.362,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07815380078153801,
378
+ "grad_norm": 0.32325974106788635,
379
+ "learning_rate": 9.219199292191993e-06,
380
+ "loss": 2.3863,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07962840079628401,
385
+ "grad_norm": 0.3661843240261078,
386
+ "learning_rate": 9.204453292044534e-06,
387
+ "loss": 2.3769,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.08110300081103,
392
+ "grad_norm": 0.3777139186859131,
393
+ "learning_rate": 9.189707291897075e-06,
394
+ "loss": 2.3626,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.082577600825776,
399
+ "grad_norm": 0.37038654088974,
400
+ "learning_rate": 9.174961291749614e-06,
401
+ "loss": 2.3895,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.084052200840522,
406
+ "grad_norm": 0.3628358542919159,
407
+ "learning_rate": 9.160215291602153e-06,
408
+ "loss": 2.3649,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.08552680085526801,
413
+ "grad_norm": 0.3579985201358795,
414
+ "learning_rate": 9.145469291454694e-06,
415
+ "loss": 2.3696,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.08700140087001401,
420
+ "grad_norm": 0.42436483502388,
421
+ "learning_rate": 9.130723291307235e-06,
422
+ "loss": 2.3845,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08847600088476001,
427
+ "grad_norm": 0.3851209580898285,
428
+ "learning_rate": 9.115977291159774e-06,
429
+ "loss": 2.36,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.089950600899506,
434
+ "grad_norm": 0.5885825157165527,
435
+ "learning_rate": 9.101231291012313e-06,
436
+ "loss": 2.3776,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.091425200914252,
441
+ "grad_norm": 0.42304036021232605,
442
+ "learning_rate": 9.086485290864854e-06,
443
+ "loss": 2.3632,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.09289980092899801,
448
+ "grad_norm": 0.4534706473350525,
449
+ "learning_rate": 9.071739290717394e-06,
450
+ "loss": 2.3531,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.09437440094374401,
455
+ "grad_norm": 0.38540026545524597,
456
+ "learning_rate": 9.056993290569934e-06,
457
+ "loss": 2.3755,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.09584900095849001,
462
+ "grad_norm": 0.38401320576667786,
463
+ "learning_rate": 9.042247290422473e-06,
464
+ "loss": 2.3614,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.09732360097323602,
469
+ "grad_norm": 0.35363873839378357,
470
+ "learning_rate": 9.027501290275014e-06,
471
+ "loss": 2.3655,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.098798200987982,
476
+ "grad_norm": 0.36643800139427185,
477
+ "learning_rate": 9.012755290127554e-06,
478
+ "loss": 2.369,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.10027280100272801,
483
+ "grad_norm": 0.3596145212650299,
484
+ "learning_rate": 8.998009289980094e-06,
485
+ "loss": 2.3703,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.10174740101747401,
490
+ "grad_norm": 0.554706871509552,
491
+ "learning_rate": 8.983263289832633e-06,
492
+ "loss": 2.3631,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.10322200103222001,
497
+ "grad_norm": 0.3995848298072815,
498
+ "learning_rate": 8.968517289685174e-06,
499
+ "loss": 2.3671,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.10469660104696601,
504
+ "grad_norm": 0.3653299808502197,
505
+ "learning_rate": 8.953771289537714e-06,
506
+ "loss": 2.3783,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.10617120106171202,
511
+ "grad_norm": 0.39819827675819397,
512
+ "learning_rate": 8.939025289390254e-06,
513
+ "loss": 2.3617,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.107645801076458,
518
+ "grad_norm": 0.3512992262840271,
519
+ "learning_rate": 8.924279289242793e-06,
520
+ "loss": 2.3649,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.10912040109120401,
525
+ "grad_norm": 0.43283718824386597,
526
+ "learning_rate": 8.909533289095333e-06,
527
+ "loss": 2.3624,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.11059500110595001,
532
+ "grad_norm": 0.3857913613319397,
533
+ "learning_rate": 8.894787288947874e-06,
534
+ "loss": 2.3686,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.11206960112069601,
539
+ "grad_norm": 0.3893970251083374,
540
+ "learning_rate": 8.880041288800413e-06,
541
+ "loss": 2.3722,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.11354420113544202,
546
+ "grad_norm": 0.6659526228904724,
547
+ "learning_rate": 8.865295288652953e-06,
548
+ "loss": 2.347,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.115018801150188,
553
+ "grad_norm": 0.3801191449165344,
554
+ "learning_rate": 8.850549288505493e-06,
555
+ "loss": 2.3595,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.11649340116493401,
560
+ "grad_norm": 0.3556024432182312,
561
+ "learning_rate": 8.835803288358034e-06,
562
+ "loss": 2.3597,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.11796800117968001,
567
+ "grad_norm": 0.45088592171669006,
568
+ "learning_rate": 8.821057288210573e-06,
569
+ "loss": 2.3623,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.11944260119442601,
574
+ "grad_norm": 0.3952132761478424,
575
+ "learning_rate": 8.806311288063114e-06,
576
+ "loss": 2.3616,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.12091720120917201,
581
+ "grad_norm": 0.4083469808101654,
582
+ "learning_rate": 8.791565287915653e-06,
583
+ "loss": 2.3597,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.12239180122391802,
588
+ "grad_norm": 0.374976247549057,
589
+ "learning_rate": 8.776819287768194e-06,
590
+ "loss": 2.3656,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.123866401238664,
595
+ "grad_norm": 0.37194111943244934,
596
+ "learning_rate": 8.762073287620733e-06,
597
+ "loss": 2.368,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.12534100125341002,
602
+ "grad_norm": 0.35863035917282104,
603
+ "learning_rate": 8.747327287473274e-06,
604
+ "loss": 2.3648,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.12681560126815603,
609
+ "grad_norm": 0.43522682785987854,
610
+ "learning_rate": 8.732581287325813e-06,
611
+ "loss": 2.3663,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.128290201282902,
616
+ "grad_norm": 0.3582712411880493,
617
+ "learning_rate": 8.717835287178354e-06,
618
+ "loss": 2.3684,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.129764801297648,
623
+ "grad_norm": 0.40035149455070496,
624
+ "learning_rate": 8.703089287030893e-06,
625
+ "loss": 2.3679,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.131239401312394,
630
+ "grad_norm": 0.36125797033309937,
631
+ "learning_rate": 8.688343286883434e-06,
632
+ "loss": 2.3501,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.13271400132714,
637
+ "grad_norm": 0.3568665087223053,
638
+ "learning_rate": 8.673597286735973e-06,
639
+ "loss": 2.3594,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.134188601341886,
644
+ "grad_norm": 0.4135202169418335,
645
+ "learning_rate": 8.658851286588512e-06,
646
+ "loss": 2.387,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.135663201356632,
651
+ "grad_norm": 0.3586069345474243,
652
+ "learning_rate": 8.644105286441053e-06,
653
+ "loss": 2.3702,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.13713780137137802,
658
+ "grad_norm": 0.47354263067245483,
659
+ "learning_rate": 8.629359286293594e-06,
660
+ "loss": 2.3709,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.13861240138612402,
665
+ "grad_norm": 0.3735561668872833,
666
+ "learning_rate": 8.614613286146133e-06,
667
+ "loss": 2.3639,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.14008700140087002,
672
+ "grad_norm": 0.351646363735199,
673
+ "learning_rate": 8.599867285998672e-06,
674
+ "loss": 2.3678,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.14156160141561602,
679
+ "grad_norm": 0.3624926507472992,
680
+ "learning_rate": 8.585121285851213e-06,
681
+ "loss": 2.3725,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.14303620143036203,
686
+ "grad_norm": 0.41470956802368164,
687
+ "learning_rate": 8.570375285703754e-06,
688
+ "loss": 2.364,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.14451080144510803,
693
+ "grad_norm": 0.3892461657524109,
694
+ "learning_rate": 8.555629285556293e-06,
695
+ "loss": 2.3623,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.145985401459854,
700
+ "grad_norm": 0.37670454382896423,
701
+ "learning_rate": 8.540883285408832e-06,
702
+ "loss": 2.3649,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.1474600014746,
707
+ "grad_norm": 0.36948758363723755,
708
+ "learning_rate": 8.526137285261373e-06,
709
+ "loss": 2.3689,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.148934601489346,
714
+ "grad_norm": 0.3593231737613678,
715
+ "learning_rate": 8.511391285113914e-06,
716
+ "loss": 2.3693,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.150409201504092,
721
+ "grad_norm": 0.3645029366016388,
722
+ "learning_rate": 8.496645284966453e-06,
723
+ "loss": 2.3607,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.151883801518838,
728
+ "grad_norm": 0.761593759059906,
729
+ "learning_rate": 8.481899284818994e-06,
730
+ "loss": 2.362,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.15335840153358402,
735
+ "grad_norm": 0.3621225953102112,
736
+ "learning_rate": 8.467153284671533e-06,
737
+ "loss": 2.3566,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.15483300154833002,
742
+ "grad_norm": 0.353289395570755,
743
+ "learning_rate": 8.452407284524074e-06,
744
+ "loss": 2.3565,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.15630760156307602,
749
+ "grad_norm": 0.4106830060482025,
750
+ "learning_rate": 8.437661284376613e-06,
751
+ "loss": 2.3536,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.15778220157782202,
756
+ "grad_norm": 0.41519981622695923,
757
+ "learning_rate": 8.422915284229154e-06,
758
+ "loss": 2.3532,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.15925680159256803,
763
+ "grad_norm": 0.3569793701171875,
764
+ "learning_rate": 8.408169284081693e-06,
765
+ "loss": 2.3648,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.16073140160731403,
770
+ "grad_norm": 0.3916598856449127,
771
+ "learning_rate": 8.393423283934234e-06,
772
+ "loss": 2.3675,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.16220600162206,
777
+ "grad_norm": 0.3435116708278656,
778
+ "learning_rate": 8.378677283786773e-06,
779
+ "loss": 2.3685,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.163680601636806,
784
+ "grad_norm": 0.47191643714904785,
785
+ "learning_rate": 8.363931283639314e-06,
786
+ "loss": 2.3596,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.165155201651552,
791
+ "grad_norm": 0.54694002866745,
792
+ "learning_rate": 8.349185283491853e-06,
793
+ "loss": 2.3551,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.166629801666298,
798
+ "grad_norm": 0.9548392295837402,
799
+ "learning_rate": 8.334439283344394e-06,
800
+ "loss": 2.3784,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.168104401681044,
805
+ "grad_norm": 0.3765574097633362,
806
+ "learning_rate": 8.319693283196933e-06,
807
+ "loss": 2.3668,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.16957900169579002,
812
+ "grad_norm": 0.748131275177002,
813
+ "learning_rate": 8.304947283049474e-06,
814
+ "loss": 2.3492,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.17105360171053602,
819
+ "grad_norm": 0.3873535096645355,
820
+ "learning_rate": 8.290201282902015e-06,
821
+ "loss": 2.3592,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.17252820172528202,
826
+ "grad_norm": 0.3429403305053711,
827
+ "learning_rate": 8.275455282754554e-06,
828
+ "loss": 2.3688,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.17400280174002802,
833
+ "grad_norm": 0.3973939120769501,
834
+ "learning_rate": 8.260709282607093e-06,
835
+ "loss": 2.3666,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.17547740175477403,
840
+ "grad_norm": 0.35062074661254883,
841
+ "learning_rate": 8.245963282459634e-06,
842
+ "loss": 2.3534,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.17695200176952003,
847
+ "grad_norm": 0.3548290431499481,
848
+ "learning_rate": 8.231217282312175e-06,
849
+ "loss": 2.3477,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.17842660178426603,
854
+ "grad_norm": 0.4035143256187439,
855
+ "learning_rate": 8.216471282164714e-06,
856
+ "loss": 2.3601,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.179901201799012,
861
+ "grad_norm": 0.37481680512428284,
862
+ "learning_rate": 8.201725282017253e-06,
863
+ "loss": 2.3579,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.181375801813758,
868
+ "grad_norm": 0.5035015344619751,
869
+ "learning_rate": 8.186979281869794e-06,
870
+ "loss": 2.366,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.182850401828504,
875
+ "grad_norm": 0.3541349768638611,
876
+ "learning_rate": 8.172233281722333e-06,
877
+ "loss": 2.368,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.18432500184325,
882
+ "grad_norm": 0.3855954110622406,
883
+ "learning_rate": 8.157487281574874e-06,
884
+ "loss": 2.3542,
885
+ "step": 2500
886
+ },
887
+ {
888
+ "epoch": 0.18579960185799602,
889
+ "grad_norm": 0.4045879542827606,
890
+ "learning_rate": 8.142741281427413e-06,
891
+ "loss": 2.3579,
892
+ "step": 2520
893
+ },
894
+ {
895
+ "epoch": 0.18727420187274202,
896
+ "grad_norm": 0.3711334764957428,
897
+ "learning_rate": 8.127995281279954e-06,
898
+ "loss": 2.3711,
899
+ "step": 2540
900
+ },
901
+ {
902
+ "epoch": 0.18874880188748802,
903
+ "grad_norm": 0.3625940978527069,
904
+ "learning_rate": 8.113249281132493e-06,
905
+ "loss": 2.3573,
906
+ "step": 2560
907
+ },
908
+ {
909
+ "epoch": 0.19022340190223402,
910
+ "grad_norm": 0.6012184619903564,
911
+ "learning_rate": 8.098503280985034e-06,
912
+ "loss": 2.3705,
913
+ "step": 2580
914
+ },
915
+ {
916
+ "epoch": 0.19169800191698003,
917
+ "grad_norm": 0.40935376286506653,
918
+ "learning_rate": 8.083757280837573e-06,
919
+ "loss": 2.3615,
920
+ "step": 2600
921
+ },
922
+ {
923
+ "epoch": 0.19317260193172603,
924
+ "grad_norm": 0.34611454606056213,
925
+ "learning_rate": 8.069011280690114e-06,
926
+ "loss": 2.358,
927
+ "step": 2620
928
+ },
929
+ {
930
+ "epoch": 0.19464720194647203,
931
+ "grad_norm": 0.36542612314224243,
932
+ "learning_rate": 8.054265280542653e-06,
933
+ "loss": 2.3656,
934
+ "step": 2640
935
+ },
936
+ {
937
+ "epoch": 0.196121801961218,
938
+ "grad_norm": 0.38626089692115784,
939
+ "learning_rate": 8.039519280395194e-06,
940
+ "loss": 2.3643,
941
+ "step": 2660
942
+ },
943
+ {
944
+ "epoch": 0.197596401975964,
945
+ "grad_norm": 0.39764684438705444,
946
+ "learning_rate": 8.024773280247733e-06,
947
+ "loss": 2.3556,
948
+ "step": 2680
949
+ },
950
+ {
951
+ "epoch": 0.19907100199071,
952
+ "grad_norm": 0.3800354301929474,
953
+ "learning_rate": 8.010027280100274e-06,
954
+ "loss": 2.3696,
955
+ "step": 2700
956
+ },
957
+ {
958
+ "epoch": 0.20054560200545601,
959
+ "grad_norm": 0.37549829483032227,
960
+ "learning_rate": 7.995281279952813e-06,
961
+ "loss": 2.3623,
962
+ "step": 2720
963
+ },
964
+ {
965
+ "epoch": 0.20202020202020202,
966
+ "grad_norm": 0.3357870280742645,
967
+ "learning_rate": 7.980535279805354e-06,
968
+ "loss": 2.3609,
969
+ "step": 2740
970
+ },
971
+ {
972
+ "epoch": 0.20349480203494802,
973
+ "grad_norm": 0.38587677478790283,
974
+ "learning_rate": 7.965789279657895e-06,
975
+ "loss": 2.3648,
976
+ "step": 2760
977
+ },
978
+ {
979
+ "epoch": 0.20496940204969402,
980
+ "grad_norm": 0.3734722137451172,
981
+ "learning_rate": 7.951043279510434e-06,
982
+ "loss": 2.3637,
983
+ "step": 2780
984
+ },
985
+ {
986
+ "epoch": 0.20644400206444002,
987
+ "grad_norm": 0.37882205843925476,
988
+ "learning_rate": 7.936297279362973e-06,
989
+ "loss": 2.3638,
990
+ "step": 2800
991
+ },
992
+ {
993
+ "epoch": 0.20791860207918603,
994
+ "grad_norm": 0.3540538549423218,
995
+ "learning_rate": 7.921551279215514e-06,
996
+ "loss": 2.3483,
997
+ "step": 2820
998
+ },
999
+ {
1000
+ "epoch": 0.20939320209393203,
1001
+ "grad_norm": 0.3712068796157837,
1002
+ "learning_rate": 7.906805279068054e-06,
1003
+ "loss": 2.3553,
1004
+ "step": 2840
1005
+ },
1006
+ {
1007
+ "epoch": 0.21086780210867803,
1008
+ "grad_norm": 0.3830094039440155,
1009
+ "learning_rate": 7.892059278920594e-06,
1010
+ "loss": 2.3526,
1011
+ "step": 2860
1012
+ },
1013
+ {
1014
+ "epoch": 0.21234240212342403,
1015
+ "grad_norm": 0.37301984429359436,
1016
+ "learning_rate": 7.877313278773133e-06,
1017
+ "loss": 2.3597,
1018
+ "step": 2880
1019
+ },
1020
+ {
1021
+ "epoch": 0.21381700213817,
1022
+ "grad_norm": 0.3589264452457428,
1023
+ "learning_rate": 7.862567278625674e-06,
1024
+ "loss": 2.3526,
1025
+ "step": 2900
1026
+ },
1027
+ {
1028
+ "epoch": 0.215291602152916,
1029
+ "grad_norm": 0.33692067861557007,
1030
+ "learning_rate": 7.847821278478214e-06,
1031
+ "loss": 2.3529,
1032
+ "step": 2920
1033
+ },
1034
+ {
1035
+ "epoch": 0.21676620216766201,
1036
+ "grad_norm": 0.3636477589607239,
1037
+ "learning_rate": 7.833075278330754e-06,
1038
+ "loss": 2.3583,
1039
+ "step": 2940
1040
+ },
1041
+ {
1042
+ "epoch": 0.21824080218240802,
1043
+ "grad_norm": 0.39444780349731445,
1044
+ "learning_rate": 7.818329278183293e-06,
1045
+ "loss": 2.3448,
1046
+ "step": 2960
1047
+ },
1048
+ {
1049
+ "epoch": 0.21971540219715402,
1050
+ "grad_norm": 0.3866218626499176,
1051
+ "learning_rate": 7.803583278035834e-06,
1052
+ "loss": 2.3626,
1053
+ "step": 2980
1054
+ },
1055
+ {
1056
+ "epoch": 0.22119000221190002,
1057
+ "grad_norm": 0.37035319209098816,
1058
+ "learning_rate": 7.788837277888374e-06,
1059
+ "loss": 2.3672,
1060
+ "step": 3000
1061
+ },
1062
+ {
1063
+ "epoch": 0.22266460222664602,
1064
+ "grad_norm": 0.3661469519138336,
1065
+ "learning_rate": 7.774091277740914e-06,
1066
+ "loss": 2.3658,
1067
+ "step": 3020
1068
+ },
1069
+ {
1070
+ "epoch": 0.22413920224139203,
1071
+ "grad_norm": 0.391837477684021,
1072
+ "learning_rate": 7.759345277593453e-06,
1073
+ "loss": 2.3672,
1074
+ "step": 3040
1075
+ },
1076
+ {
1077
+ "epoch": 0.22561380225613803,
1078
+ "grad_norm": 0.36069273948669434,
1079
+ "learning_rate": 7.744599277445994e-06,
1080
+ "loss": 2.3685,
1081
+ "step": 3060
1082
+ },
1083
+ {
1084
+ "epoch": 0.22708840227088403,
1085
+ "grad_norm": 0.3551023602485657,
1086
+ "learning_rate": 7.729853277298534e-06,
1087
+ "loss": 2.3697,
1088
+ "step": 3080
1089
+ },
1090
+ {
1091
+ "epoch": 0.22856300228563003,
1092
+ "grad_norm": 0.3549511730670929,
1093
+ "learning_rate": 7.715107277151073e-06,
1094
+ "loss": 2.3708,
1095
+ "step": 3100
1096
+ },
1097
+ {
1098
+ "epoch": 0.230037602300376,
1099
+ "grad_norm": 0.3584568202495575,
1100
+ "learning_rate": 7.700361277003613e-06,
1101
+ "loss": 2.3491,
1102
+ "step": 3120
1103
+ },
1104
+ {
1105
+ "epoch": 0.231512202315122,
1106
+ "grad_norm": 0.3881671726703644,
1107
+ "learning_rate": 7.685615276856153e-06,
1108
+ "loss": 2.3603,
1109
+ "step": 3140
1110
+ },
1111
+ {
1112
+ "epoch": 0.23298680232986801,
1113
+ "grad_norm": 0.37565991282463074,
1114
+ "learning_rate": 7.670869276708693e-06,
1115
+ "loss": 2.349,
1116
+ "step": 3160
1117
+ },
1118
+ {
1119
+ "epoch": 0.23446140234461402,
1120
+ "grad_norm": 0.3800000548362732,
1121
+ "learning_rate": 7.656123276561233e-06,
1122
+ "loss": 2.3634,
1123
+ "step": 3180
1124
+ },
1125
+ {
1126
+ "epoch": 0.23593600235936002,
1127
+ "grad_norm": 0.3795914053916931,
1128
+ "learning_rate": 7.641377276413774e-06,
1129
+ "loss": 2.3541,
1130
+ "step": 3200
1131
+ },
1132
+ {
1133
+ "epoch": 0.23741060237410602,
1134
+ "grad_norm": 0.34739038348197937,
1135
+ "learning_rate": 7.626631276266313e-06,
1136
+ "loss": 2.364,
1137
+ "step": 3220
1138
+ },
1139
+ {
1140
+ "epoch": 0.23888520238885202,
1141
+ "grad_norm": 0.3631921112537384,
1142
+ "learning_rate": 7.611885276118853e-06,
1143
+ "loss": 2.3472,
1144
+ "step": 3240
1145
+ },
1146
+ {
1147
+ "epoch": 0.24035980240359803,
1148
+ "grad_norm": 0.34708309173583984,
1149
+ "learning_rate": 7.597139275971393e-06,
1150
+ "loss": 2.3745,
1151
+ "step": 3260
1152
+ },
1153
+ {
1154
+ "epoch": 0.24183440241834403,
1155
+ "grad_norm": 0.39002010226249695,
1156
+ "learning_rate": 7.582393275823933e-06,
1157
+ "loss": 2.3496,
1158
+ "step": 3280
1159
+ },
1160
+ {
1161
+ "epoch": 0.24330900243309003,
1162
+ "grad_norm": 0.3856920003890991,
1163
+ "learning_rate": 7.5676472756764725e-06,
1164
+ "loss": 2.3556,
1165
+ "step": 3300
1166
+ },
1167
+ {
1168
+ "epoch": 0.24478360244783604,
1169
+ "grad_norm": 0.3604213297367096,
1170
+ "learning_rate": 7.552901275529013e-06,
1171
+ "loss": 2.3588,
1172
+ "step": 3320
1173
+ },
1174
+ {
1175
+ "epoch": 0.24625820246258204,
1176
+ "grad_norm": 0.4610792100429535,
1177
+ "learning_rate": 7.538155275381553e-06,
1178
+ "loss": 2.3495,
1179
+ "step": 3340
1180
+ },
1181
+ {
1182
+ "epoch": 0.247732802477328,
1183
+ "grad_norm": 0.37130406498908997,
1184
+ "learning_rate": 7.523409275234093e-06,
1185
+ "loss": 2.3654,
1186
+ "step": 3360
1187
+ },
1188
+ {
1189
+ "epoch": 0.24920740249207401,
1190
+ "grad_norm": 0.4634559154510498,
1191
+ "learning_rate": 7.5086632750866325e-06,
1192
+ "loss": 2.3472,
1193
+ "step": 3380
1194
+ },
1195
+ {
1196
+ "epoch": 0.25068200250682005,
1197
+ "grad_norm": 0.3728873133659363,
1198
+ "learning_rate": 7.493917274939173e-06,
1199
+ "loss": 2.3601,
1200
+ "step": 3400
1201
+ },
1202
+ {
1203
+ "epoch": 0.252156602521566,
1204
+ "grad_norm": 0.35525408387184143,
1205
+ "learning_rate": 7.479171274791713e-06,
1206
+ "loss": 2.3483,
1207
+ "step": 3420
1208
+ },
1209
+ {
1210
+ "epoch": 0.25363120253631205,
1211
+ "grad_norm": 0.35072821378707886,
1212
+ "learning_rate": 7.464425274644253e-06,
1213
+ "loss": 2.3575,
1214
+ "step": 3440
1215
+ },
1216
+ {
1217
+ "epoch": 0.255105802551058,
1218
+ "grad_norm": 0.33984318375587463,
1219
+ "learning_rate": 7.449679274496794e-06,
1220
+ "loss": 2.3562,
1221
+ "step": 3460
1222
+ },
1223
+ {
1224
+ "epoch": 0.256580402565804,
1225
+ "grad_norm": 0.3893386423587799,
1226
+ "learning_rate": 7.434933274349333e-06,
1227
+ "loss": 2.3417,
1228
+ "step": 3480
1229
+ },
1230
+ {
1231
+ "epoch": 0.25805500258055003,
1232
+ "grad_norm": 0.3702589273452759,
1233
+ "learning_rate": 7.420187274201873e-06,
1234
+ "loss": 2.3575,
1235
+ "step": 3500
1236
+ },
1237
+ {
1238
+ "epoch": 0.259529602595296,
1239
+ "grad_norm": 0.37328028678894043,
1240
+ "learning_rate": 7.405441274054413e-06,
1241
+ "loss": 2.3691,
1242
+ "step": 3520
1243
+ },
1244
+ {
1245
+ "epoch": 0.26100420261004204,
1246
+ "grad_norm": 0.362596720457077,
1247
+ "learning_rate": 7.390695273906954e-06,
1248
+ "loss": 2.3516,
1249
+ "step": 3540
1250
+ },
1251
+ {
1252
+ "epoch": 0.262478802624788,
1253
+ "grad_norm": 0.35985130071640015,
1254
+ "learning_rate": 7.375949273759493e-06,
1255
+ "loss": 2.3579,
1256
+ "step": 3560
1257
+ },
1258
+ {
1259
+ "epoch": 0.26395340263953404,
1260
+ "grad_norm": 0.39259010553359985,
1261
+ "learning_rate": 7.361203273612033e-06,
1262
+ "loss": 2.3674,
1263
+ "step": 3580
1264
+ },
1265
+ {
1266
+ "epoch": 0.26542800265428,
1267
+ "grad_norm": 0.3379274904727936,
1268
+ "learning_rate": 7.346457273464573e-06,
1269
+ "loss": 2.3409,
1270
+ "step": 3600
1271
+ },
1272
+ {
1273
+ "epoch": 0.26690260266902605,
1274
+ "grad_norm": 0.38404515385627747,
1275
+ "learning_rate": 7.331711273317113e-06,
1276
+ "loss": 2.3493,
1277
+ "step": 3620
1278
+ },
1279
+ {
1280
+ "epoch": 0.268377202683772,
1281
+ "grad_norm": 0.3580029606819153,
1282
+ "learning_rate": 7.316965273169654e-06,
1283
+ "loss": 2.3426,
1284
+ "step": 3640
1285
+ },
1286
+ {
1287
+ "epoch": 0.26985180269851805,
1288
+ "grad_norm": 0.36124977469444275,
1289
+ "learning_rate": 7.302219273022193e-06,
1290
+ "loss": 2.3563,
1291
+ "step": 3660
1292
+ },
1293
+ {
1294
+ "epoch": 0.271326402713264,
1295
+ "grad_norm": 0.37032002210617065,
1296
+ "learning_rate": 7.287473272874733e-06,
1297
+ "loss": 2.3558,
1298
+ "step": 3680
1299
+ },
1300
+ {
1301
+ "epoch": 0.27280100272801,
1302
+ "grad_norm": 0.37544354796409607,
1303
+ "learning_rate": 7.272727272727273e-06,
1304
+ "loss": 2.353,
1305
+ "step": 3700
1306
+ },
1307
+ {
1308
+ "epoch": 0.27427560274275603,
1309
+ "grad_norm": 0.38063186407089233,
1310
+ "learning_rate": 7.257981272579814e-06,
1311
+ "loss": 2.355,
1312
+ "step": 3720
1313
+ },
1314
+ {
1315
+ "epoch": 0.275750202757502,
1316
+ "grad_norm": 0.34436681866645813,
1317
+ "learning_rate": 7.243235272432353e-06,
1318
+ "loss": 2.3671,
1319
+ "step": 3740
1320
+ },
1321
+ {
1322
+ "epoch": 0.27722480277224804,
1323
+ "grad_norm": 0.3745051622390747,
1324
+ "learning_rate": 7.228489272284893e-06,
1325
+ "loss": 2.343,
1326
+ "step": 3760
1327
+ },
1328
+ {
1329
+ "epoch": 0.278699402786994,
1330
+ "grad_norm": 0.36901938915252686,
1331
+ "learning_rate": 7.213743272137433e-06,
1332
+ "loss": 2.3367,
1333
+ "step": 3780
1334
+ },
1335
+ {
1336
+ "epoch": 0.28017400280174004,
1337
+ "grad_norm": 0.38288217782974243,
1338
+ "learning_rate": 7.198997271989974e-06,
1339
+ "loss": 2.3598,
1340
+ "step": 3800
1341
+ },
1342
+ {
1343
+ "epoch": 0.281648602816486,
1344
+ "grad_norm": 0.37386777997016907,
1345
+ "learning_rate": 7.184251271842513e-06,
1346
+ "loss": 2.3618,
1347
+ "step": 3820
1348
+ },
1349
+ {
1350
+ "epoch": 0.28312320283123205,
1351
+ "grad_norm": 0.35673725605010986,
1352
+ "learning_rate": 7.169505271695053e-06,
1353
+ "loss": 2.3514,
1354
+ "step": 3840
1355
+ },
1356
+ {
1357
+ "epoch": 0.284597802845978,
1358
+ "grad_norm": 0.38071954250335693,
1359
+ "learning_rate": 7.154759271547593e-06,
1360
+ "loss": 2.3564,
1361
+ "step": 3860
1362
+ },
1363
+ {
1364
+ "epoch": 0.28607240286072405,
1365
+ "grad_norm": 0.3805045187473297,
1366
+ "learning_rate": 7.140013271400134e-06,
1367
+ "loss": 2.3599,
1368
+ "step": 3880
1369
+ },
1370
+ {
1371
+ "epoch": 0.28754700287547,
1372
+ "grad_norm": 0.3432954251766205,
1373
+ "learning_rate": 7.125267271252674e-06,
1374
+ "loss": 2.3568,
1375
+ "step": 3900
1376
+ },
1377
+ {
1378
+ "epoch": 0.28902160289021606,
1379
+ "grad_norm": 0.35117459297180176,
1380
+ "learning_rate": 7.110521271105213e-06,
1381
+ "loss": 2.3566,
1382
+ "step": 3920
1383
+ },
1384
+ {
1385
+ "epoch": 0.29049620290496203,
1386
+ "grad_norm": 0.39861348271369934,
1387
+ "learning_rate": 7.095775270957753e-06,
1388
+ "loss": 2.3553,
1389
+ "step": 3940
1390
+ },
1391
+ {
1392
+ "epoch": 0.291970802919708,
1393
+ "grad_norm": 0.3653263747692108,
1394
+ "learning_rate": 7.081029270810294e-06,
1395
+ "loss": 2.3781,
1396
+ "step": 3960
1397
+ },
1398
+ {
1399
+ "epoch": 0.29344540293445404,
1400
+ "grad_norm": 0.3614581823348999,
1401
+ "learning_rate": 7.066283270662834e-06,
1402
+ "loss": 2.3407,
1403
+ "step": 3980
1404
+ },
1405
+ {
1406
+ "epoch": 0.2949200029492,
1407
+ "grad_norm": 0.39455854892730713,
1408
+ "learning_rate": 7.051537270515373e-06,
1409
+ "loss": 2.3541,
1410
+ "step": 4000
1411
+ },
1412
+ {
1413
+ "epoch": 0.29639460296394604,
1414
+ "grad_norm": 0.33767521381378174,
1415
+ "learning_rate": 7.036791270367913e-06,
1416
+ "loss": 2.3332,
1417
+ "step": 4020
1418
+ },
1419
+ {
1420
+ "epoch": 0.297869202978692,
1421
+ "grad_norm": 0.36079707741737366,
1422
+ "learning_rate": 7.022045270220453e-06,
1423
+ "loss": 2.3546,
1424
+ "step": 4040
1425
+ },
1426
+ {
1427
+ "epoch": 0.29934380299343805,
1428
+ "grad_norm": 0.395107626914978,
1429
+ "learning_rate": 7.007299270072994e-06,
1430
+ "loss": 2.3483,
1431
+ "step": 4060
1432
+ },
1433
+ {
1434
+ "epoch": 0.300818403008184,
1435
+ "grad_norm": 0.3456408381462097,
1436
+ "learning_rate": 6.992553269925533e-06,
1437
+ "loss": 2.3657,
1438
+ "step": 4080
1439
+ },
1440
+ {
1441
+ "epoch": 0.30229300302293005,
1442
+ "grad_norm": 0.4138198792934418,
1443
+ "learning_rate": 6.977807269778073e-06,
1444
+ "loss": 2.3616,
1445
+ "step": 4100
1446
+ },
1447
+ {
1448
+ "epoch": 0.303767603037676,
1449
+ "grad_norm": 0.382722944021225,
1450
+ "learning_rate": 6.963061269630613e-06,
1451
+ "loss": 2.3607,
1452
+ "step": 4120
1453
+ },
1454
+ {
1455
+ "epoch": 0.30524220305242206,
1456
+ "grad_norm": 0.4129588007926941,
1457
+ "learning_rate": 6.948315269483154e-06,
1458
+ "loss": 2.3582,
1459
+ "step": 4140
1460
+ },
1461
+ {
1462
+ "epoch": 0.30671680306716803,
1463
+ "grad_norm": 0.35810399055480957,
1464
+ "learning_rate": 6.933569269335694e-06,
1465
+ "loss": 2.345,
1466
+ "step": 4160
1467
+ },
1468
+ {
1469
+ "epoch": 0.308191403081914,
1470
+ "grad_norm": 0.3707892596721649,
1471
+ "learning_rate": 6.918823269188233e-06,
1472
+ "loss": 2.3389,
1473
+ "step": 4180
1474
+ },
1475
+ {
1476
+ "epoch": 0.30966600309666004,
1477
+ "grad_norm": 0.43102672696113586,
1478
+ "learning_rate": 6.904077269040773e-06,
1479
+ "loss": 2.3636,
1480
+ "step": 4200
1481
+ },
1482
+ {
1483
+ "epoch": 0.311140603111406,
1484
+ "grad_norm": 0.37081730365753174,
1485
+ "learning_rate": 6.889331268893314e-06,
1486
+ "loss": 2.3589,
1487
+ "step": 4220
1488
+ },
1489
+ {
1490
+ "epoch": 0.31261520312615204,
1491
+ "grad_norm": 0.3554447591304779,
1492
+ "learning_rate": 6.874585268745854e-06,
1493
+ "loss": 2.3639,
1494
+ "step": 4240
1495
+ },
1496
+ {
1497
+ "epoch": 0.314089803140898,
1498
+ "grad_norm": 0.3365320563316345,
1499
+ "learning_rate": 6.859839268598393e-06,
1500
+ "loss": 2.3604,
1501
+ "step": 4260
1502
+ },
1503
+ {
1504
+ "epoch": 0.31556440315564405,
1505
+ "grad_norm": 0.40808168053627014,
1506
+ "learning_rate": 6.845093268450933e-06,
1507
+ "loss": 2.3496,
1508
+ "step": 4280
1509
+ },
1510
+ {
1511
+ "epoch": 0.31703900317039,
1512
+ "grad_norm": 0.380753755569458,
1513
+ "learning_rate": 6.830347268303474e-06,
1514
+ "loss": 2.3478,
1515
+ "step": 4300
1516
+ },
1517
+ {
1518
+ "epoch": 0.31851360318513605,
1519
+ "grad_norm": 0.3863361179828644,
1520
+ "learning_rate": 6.815601268156014e-06,
1521
+ "loss": 2.3381,
1522
+ "step": 4320
1523
+ },
1524
+ {
1525
+ "epoch": 0.319988203199882,
1526
+ "grad_norm": 0.40554317831993103,
1527
+ "learning_rate": 6.800855268008554e-06,
1528
+ "loss": 2.3518,
1529
+ "step": 4340
1530
+ },
1531
+ {
1532
+ "epoch": 0.32146280321462806,
1533
+ "grad_norm": 0.39253705739974976,
1534
+ "learning_rate": 6.786109267861093e-06,
1535
+ "loss": 2.3567,
1536
+ "step": 4360
1537
+ },
1538
+ {
1539
+ "epoch": 0.32293740322937403,
1540
+ "grad_norm": 0.37108898162841797,
1541
+ "learning_rate": 6.771363267713633e-06,
1542
+ "loss": 2.3509,
1543
+ "step": 4380
1544
+ },
1545
+ {
1546
+ "epoch": 0.32441200324412,
1547
+ "grad_norm": 0.36466971039772034,
1548
+ "learning_rate": 6.756617267566174e-06,
1549
+ "loss": 2.3627,
1550
+ "step": 4400
1551
+ },
1552
+ {
1553
+ "epoch": 0.32588660325886604,
1554
+ "grad_norm": 0.37712493538856506,
1555
+ "learning_rate": 6.741871267418714e-06,
1556
+ "loss": 2.3754,
1557
+ "step": 4420
1558
+ },
1559
+ {
1560
+ "epoch": 0.327361203273612,
1561
+ "grad_norm": 0.3712570071220398,
1562
+ "learning_rate": 6.727125267271253e-06,
1563
+ "loss": 2.3619,
1564
+ "step": 4440
1565
+ },
1566
+ {
1567
+ "epoch": 0.32883580328835804,
1568
+ "grad_norm": 0.34880587458610535,
1569
+ "learning_rate": 6.712379267123793e-06,
1570
+ "loss": 2.3574,
1571
+ "step": 4460
1572
+ },
1573
+ {
1574
+ "epoch": 0.330310403303104,
1575
+ "grad_norm": 0.4109377861022949,
1576
+ "learning_rate": 6.697633266976334e-06,
1577
+ "loss": 2.3574,
1578
+ "step": 4480
1579
+ },
1580
+ {
1581
+ "epoch": 0.33178500331785005,
1582
+ "grad_norm": 0.3626454770565033,
1583
+ "learning_rate": 6.682887266828874e-06,
1584
+ "loss": 2.3527,
1585
+ "step": 4500
1586
+ },
1587
+ {
1588
+ "epoch": 0.333259603332596,
1589
+ "grad_norm": 0.3580588400363922,
1590
+ "learning_rate": 6.668141266681413e-06,
1591
+ "loss": 2.348,
1592
+ "step": 4520
1593
+ },
1594
+ {
1595
+ "epoch": 0.33473420334734205,
1596
+ "grad_norm": 0.3460945785045624,
1597
+ "learning_rate": 6.653395266533953e-06,
1598
+ "loss": 2.3461,
1599
+ "step": 4540
1600
+ },
1601
+ {
1602
+ "epoch": 0.336208803362088,
1603
+ "grad_norm": 0.39259958267211914,
1604
+ "learning_rate": 6.6386492663864936e-06,
1605
+ "loss": 2.3818,
1606
+ "step": 4560
1607
+ },
1608
+ {
1609
+ "epoch": 0.33768340337683406,
1610
+ "grad_norm": 0.44788721203804016,
1611
+ "learning_rate": 6.6239032662390335e-06,
1612
+ "loss": 2.3636,
1613
+ "step": 4580
1614
+ },
1615
+ {
1616
+ "epoch": 0.33915800339158003,
1617
+ "grad_norm": 0.40668484568595886,
1618
+ "learning_rate": 6.6091572660915735e-06,
1619
+ "loss": 2.3542,
1620
+ "step": 4600
1621
+ },
1622
+ {
1623
+ "epoch": 0.340632603406326,
1624
+ "grad_norm": 0.3672851026058197,
1625
+ "learning_rate": 6.594411265944113e-06,
1626
+ "loss": 2.3497,
1627
+ "step": 4620
1628
+ },
1629
+ {
1630
+ "epoch": 0.34210720342107204,
1631
+ "grad_norm": 0.3685692846775055,
1632
+ "learning_rate": 6.5796652657966535e-06,
1633
+ "loss": 2.361,
1634
+ "step": 4640
1635
+ },
1636
+ {
1637
+ "epoch": 0.343581803435818,
1638
+ "grad_norm": 0.4522005617618561,
1639
+ "learning_rate": 6.5649192656491935e-06,
1640
+ "loss": 2.3498,
1641
+ "step": 4660
1642
+ },
1643
+ {
1644
+ "epoch": 0.34505640345056404,
1645
+ "grad_norm": 0.4089388847351074,
1646
+ "learning_rate": 6.5501732655017335e-06,
1647
+ "loss": 2.3479,
1648
+ "step": 4680
1649
+ },
1650
+ {
1651
+ "epoch": 0.34653100346531,
1652
+ "grad_norm": 0.3874075710773468,
1653
+ "learning_rate": 6.535427265354273e-06,
1654
+ "loss": 2.3437,
1655
+ "step": 4700
1656
+ },
1657
+ {
1658
+ "epoch": 0.34800560348005605,
1659
+ "grad_norm": 0.3733789622783661,
1660
+ "learning_rate": 6.520681265206813e-06,
1661
+ "loss": 2.3524,
1662
+ "step": 4720
1663
+ },
1664
+ {
1665
+ "epoch": 0.349480203494802,
1666
+ "grad_norm": 0.3892102539539337,
1667
+ "learning_rate": 6.5059352650593535e-06,
1668
+ "loss": 2.346,
1669
+ "step": 4740
1670
+ },
1671
+ {
1672
+ "epoch": 0.35095480350954805,
1673
+ "grad_norm": 0.36850956082344055,
1674
+ "learning_rate": 6.4911892649118935e-06,
1675
+ "loss": 2.3532,
1676
+ "step": 4760
1677
+ },
1678
+ {
1679
+ "epoch": 0.352429403524294,
1680
+ "grad_norm": 0.3519984185695648,
1681
+ "learning_rate": 6.476443264764433e-06,
1682
+ "loss": 2.3429,
1683
+ "step": 4780
1684
+ },
1685
+ {
1686
+ "epoch": 0.35390400353904006,
1687
+ "grad_norm": 0.36120033264160156,
1688
+ "learning_rate": 6.461697264616973e-06,
1689
+ "loss": 2.3397,
1690
+ "step": 4800
1691
+ },
1692
+ {
1693
+ "epoch": 0.35537860355378603,
1694
+ "grad_norm": 0.40259799361228943,
1695
+ "learning_rate": 6.446951264469513e-06,
1696
+ "loss": 2.3443,
1697
+ "step": 4820
1698
+ },
1699
+ {
1700
+ "epoch": 0.35685320356853206,
1701
+ "grad_norm": 0.370540976524353,
1702
+ "learning_rate": 6.432205264322053e-06,
1703
+ "loss": 2.3494,
1704
+ "step": 4840
1705
+ },
1706
+ {
1707
+ "epoch": 0.35832780358327804,
1708
+ "grad_norm": 0.7460556626319885,
1709
+ "learning_rate": 6.417459264174593e-06,
1710
+ "loss": 2.3452,
1711
+ "step": 4860
1712
+ },
1713
+ {
1714
+ "epoch": 0.359802403598024,
1715
+ "grad_norm": 0.3665451109409332,
1716
+ "learning_rate": 6.4027132640271325e-06,
1717
+ "loss": 2.3693,
1718
+ "step": 4880
1719
+ },
1720
+ {
1721
+ "epoch": 0.36127700361277004,
1722
+ "grad_norm": 0.38104239106178284,
1723
+ "learning_rate": 6.387967263879673e-06,
1724
+ "loss": 2.3568,
1725
+ "step": 4900
1726
+ },
1727
+ {
1728
+ "epoch": 0.362751603627516,
1729
+ "grad_norm": 0.36739110946655273,
1730
+ "learning_rate": 6.373221263732213e-06,
1731
+ "loss": 2.361,
1732
+ "step": 4920
1733
+ },
1734
+ {
1735
+ "epoch": 0.36422620364226205,
1736
+ "grad_norm": 0.405453085899353,
1737
+ "learning_rate": 6.358475263584753e-06,
1738
+ "loss": 2.3448,
1739
+ "step": 4940
1740
+ },
1741
+ {
1742
+ "epoch": 0.365700803657008,
1743
+ "grad_norm": 0.349317342042923,
1744
+ "learning_rate": 6.3437292634372925e-06,
1745
+ "loss": 2.3569,
1746
+ "step": 4960
1747
+ },
1748
+ {
1749
+ "epoch": 0.36717540367175405,
1750
+ "grad_norm": 0.3882797360420227,
1751
+ "learning_rate": 6.328983263289833e-06,
1752
+ "loss": 2.3457,
1753
+ "step": 4980
1754
+ },
1755
+ {
1756
+ "epoch": 0.3686500036865,
1757
+ "grad_norm": 0.42195364832878113,
1758
+ "learning_rate": 6.314237263142373e-06,
1759
+ "loss": 2.364,
1760
+ "step": 5000
1761
+ },
1762
+ {
1763
+ "epoch": 0.37012460370124606,
1764
+ "grad_norm": 0.40603727102279663,
1765
+ "learning_rate": 6.299491262994913e-06,
1766
+ "loss": 2.3365,
1767
+ "step": 5020
1768
+ },
1769
+ {
1770
+ "epoch": 0.37159920371599203,
1771
+ "grad_norm": 0.3686840534210205,
1772
+ "learning_rate": 6.284745262847453e-06,
1773
+ "loss": 2.3624,
1774
+ "step": 5040
1775
+ },
1776
+ {
1777
+ "epoch": 0.37307380373073806,
1778
+ "grad_norm": 0.37392449378967285,
1779
+ "learning_rate": 6.2699992626999924e-06,
1780
+ "loss": 2.3597,
1781
+ "step": 5060
1782
+ },
1783
+ {
1784
+ "epoch": 0.37454840374548404,
1785
+ "grad_norm": 0.3410843014717102,
1786
+ "learning_rate": 6.255253262552533e-06,
1787
+ "loss": 2.3552,
1788
+ "step": 5080
1789
+ },
1790
+ {
1791
+ "epoch": 0.37602300376023,
1792
+ "grad_norm": 0.3532737195491791,
1793
+ "learning_rate": 6.240507262405073e-06,
1794
+ "loss": 2.3532,
1795
+ "step": 5100
1796
+ },
1797
+ {
1798
+ "epoch": 0.37749760377497604,
1799
+ "grad_norm": 0.3908025026321411,
1800
+ "learning_rate": 6.225761262257613e-06,
1801
+ "loss": 2.3624,
1802
+ "step": 5120
1803
+ },
1804
+ {
1805
+ "epoch": 0.378972203789722,
1806
+ "grad_norm": 0.3747566044330597,
1807
+ "learning_rate": 6.211015262110152e-06,
1808
+ "loss": 2.3542,
1809
+ "step": 5140
1810
+ },
1811
+ {
1812
+ "epoch": 0.38044680380446805,
1813
+ "grad_norm": 0.37590813636779785,
1814
+ "learning_rate": 6.196269261962693e-06,
1815
+ "loss": 2.3523,
1816
+ "step": 5160
1817
+ },
1818
+ {
1819
+ "epoch": 0.381921403819214,
1820
+ "grad_norm": 0.3960760533809662,
1821
+ "learning_rate": 6.181523261815233e-06,
1822
+ "loss": 2.367,
1823
+ "step": 5180
1824
+ },
1825
+ {
1826
+ "epoch": 0.38339600383396005,
1827
+ "grad_norm": 0.34570005536079407,
1828
+ "learning_rate": 6.166777261667773e-06,
1829
+ "loss": 2.354,
1830
+ "step": 5200
1831
+ },
1832
+ {
1833
+ "epoch": 0.384870603848706,
1834
+ "grad_norm": 0.3910704553127289,
1835
+ "learning_rate": 6.152031261520312e-06,
1836
+ "loss": 2.3657,
1837
+ "step": 5220
1838
+ },
1839
+ {
1840
+ "epoch": 0.38634520386345206,
1841
+ "grad_norm": 0.35407891869544983,
1842
+ "learning_rate": 6.137285261372853e-06,
1843
+ "loss": 2.3494,
1844
+ "step": 5240
1845
+ },
1846
+ {
1847
+ "epoch": 0.38781980387819803,
1848
+ "grad_norm": 0.37890905141830444,
1849
+ "learning_rate": 6.122539261225393e-06,
1850
+ "loss": 2.3476,
1851
+ "step": 5260
1852
+ },
1853
+ {
1854
+ "epoch": 0.38929440389294406,
1855
+ "grad_norm": 0.388045072555542,
1856
+ "learning_rate": 6.107793261077933e-06,
1857
+ "loss": 2.3593,
1858
+ "step": 5280
1859
+ },
1860
+ {
1861
+ "epoch": 0.39076900390769004,
1862
+ "grad_norm": 0.3695070743560791,
1863
+ "learning_rate": 6.093047260930474e-06,
1864
+ "loss": 2.3645,
1865
+ "step": 5300
1866
+ },
1867
+ {
1868
+ "epoch": 0.392243603922436,
1869
+ "grad_norm": 0.3945058584213257,
1870
+ "learning_rate": 6.078301260783013e-06,
1871
+ "loss": 2.3439,
1872
+ "step": 5320
1873
+ },
1874
+ {
1875
+ "epoch": 0.39371820393718204,
1876
+ "grad_norm": 0.3858231008052826,
1877
+ "learning_rate": 6.063555260635553e-06,
1878
+ "loss": 2.3648,
1879
+ "step": 5340
1880
+ },
1881
+ {
1882
+ "epoch": 0.395192803951928,
1883
+ "grad_norm": 0.3549276292324066,
1884
+ "learning_rate": 6.048809260488093e-06,
1885
+ "loss": 2.3514,
1886
+ "step": 5360
1887
+ },
1888
+ {
1889
+ "epoch": 0.39666740396667405,
1890
+ "grad_norm": 0.3770926296710968,
1891
+ "learning_rate": 6.034063260340633e-06,
1892
+ "loss": 2.3598,
1893
+ "step": 5380
1894
+ },
1895
+ {
1896
+ "epoch": 0.39814200398142,
1897
+ "grad_norm": 0.37677156925201416,
1898
+ "learning_rate": 6.019317260193172e-06,
1899
+ "loss": 2.3539,
1900
+ "step": 5400
1901
+ },
1902
+ {
1903
+ "epoch": 0.39961660399616605,
1904
+ "grad_norm": 0.3545861542224884,
1905
+ "learning_rate": 6.004571260045713e-06,
1906
+ "loss": 2.3535,
1907
+ "step": 5420
1908
+ },
1909
+ {
1910
+ "epoch": 0.40109120401091203,
1911
+ "grad_norm": 0.3972468078136444,
1912
+ "learning_rate": 5.989825259898253e-06,
1913
+ "loss": 2.3533,
1914
+ "step": 5440
1915
+ },
1916
+ {
1917
+ "epoch": 0.40256580402565806,
1918
+ "grad_norm": 0.368756502866745,
1919
+ "learning_rate": 5.975079259750793e-06,
1920
+ "loss": 2.3509,
1921
+ "step": 5460
1922
+ },
1923
+ {
1924
+ "epoch": 0.40404040404040403,
1925
+ "grad_norm": 0.40190038084983826,
1926
+ "learning_rate": 5.960333259603332e-06,
1927
+ "loss": 2.3375,
1928
+ "step": 5480
1929
+ },
1930
+ {
1931
+ "epoch": 0.40551500405515006,
1932
+ "grad_norm": 0.34512144327163696,
1933
+ "learning_rate": 5.945587259455873e-06,
1934
+ "loss": 2.3557,
1935
+ "step": 5500
1936
+ },
1937
+ {
1938
+ "epoch": 0.40698960406989604,
1939
+ "grad_norm": 0.3768044114112854,
1940
+ "learning_rate": 5.930841259308413e-06,
1941
+ "loss": 2.3466,
1942
+ "step": 5520
1943
+ },
1944
+ {
1945
+ "epoch": 0.40846420408464207,
1946
+ "grad_norm": 0.3744104504585266,
1947
+ "learning_rate": 5.916095259160953e-06,
1948
+ "loss": 2.3565,
1949
+ "step": 5540
1950
+ },
1951
+ {
1952
+ "epoch": 0.40993880409938804,
1953
+ "grad_norm": 0.35657835006713867,
1954
+ "learning_rate": 5.901349259013494e-06,
1955
+ "loss": 2.3474,
1956
+ "step": 5560
1957
+ },
1958
+ {
1959
+ "epoch": 0.411413404114134,
1960
+ "grad_norm": 0.40596744418144226,
1961
+ "learning_rate": 5.886603258866033e-06,
1962
+ "loss": 2.344,
1963
+ "step": 5580
1964
+ },
1965
+ {
1966
+ "epoch": 0.41288800412888005,
1967
+ "grad_norm": 0.3615570366382599,
1968
+ "learning_rate": 5.871857258718573e-06,
1969
+ "loss": 2.3375,
1970
+ "step": 5600
1971
+ },
1972
+ {
1973
+ "epoch": 0.414362604143626,
1974
+ "grad_norm": 0.3853433132171631,
1975
+ "learning_rate": 5.857111258571113e-06,
1976
+ "loss": 2.3536,
1977
+ "step": 5620
1978
+ },
1979
+ {
1980
+ "epoch": 0.41583720415837205,
1981
+ "grad_norm": 0.4154307544231415,
1982
+ "learning_rate": 5.842365258423654e-06,
1983
+ "loss": 2.3399,
1984
+ "step": 5640
1985
+ },
1986
+ {
1987
+ "epoch": 0.41731180417311803,
1988
+ "grad_norm": 0.4312744438648224,
1989
+ "learning_rate": 5.827619258276193e-06,
1990
+ "loss": 2.3442,
1991
+ "step": 5660
1992
+ },
1993
+ {
1994
+ "epoch": 0.41878640418786406,
1995
+ "grad_norm": 0.402468740940094,
1996
+ "learning_rate": 5.812873258128733e-06,
1997
+ "loss": 2.3522,
1998
+ "step": 5680
1999
+ },
2000
+ {
2001
+ "epoch": 0.42026100420261003,
2002
+ "grad_norm": 0.4803409278392792,
2003
+ "learning_rate": 5.798127257981273e-06,
2004
+ "loss": 2.3547,
2005
+ "step": 5700
2006
+ },
2007
+ {
2008
+ "epoch": 0.42173560421735606,
2009
+ "grad_norm": 0.3646136522293091,
2010
+ "learning_rate": 5.783381257833813e-06,
2011
+ "loss": 2.3431,
2012
+ "step": 5720
2013
+ },
2014
+ {
2015
+ "epoch": 0.42321020423210204,
2016
+ "grad_norm": 0.380462646484375,
2017
+ "learning_rate": 5.768635257686354e-06,
2018
+ "loss": 2.3564,
2019
+ "step": 5740
2020
+ },
2021
+ {
2022
+ "epoch": 0.42468480424684807,
2023
+ "grad_norm": 0.3754338324069977,
2024
+ "learning_rate": 5.753889257538893e-06,
2025
+ "loss": 2.3464,
2026
+ "step": 5760
2027
+ },
2028
+ {
2029
+ "epoch": 0.42615940426159404,
2030
+ "grad_norm": 0.3629322350025177,
2031
+ "learning_rate": 5.739143257391433e-06,
2032
+ "loss": 2.3563,
2033
+ "step": 5780
2034
+ },
2035
+ {
2036
+ "epoch": 0.42763400427634,
2037
+ "grad_norm": 0.4097397029399872,
2038
+ "learning_rate": 5.724397257243973e-06,
2039
+ "loss": 2.3488,
2040
+ "step": 5800
2041
+ },
2042
+ {
2043
+ "epoch": 0.42910860429108605,
2044
+ "grad_norm": 0.3598233163356781,
2045
+ "learning_rate": 5.709651257096514e-06,
2046
+ "loss": 2.3641,
2047
+ "step": 5820
2048
+ },
2049
+ {
2050
+ "epoch": 0.430583204305832,
2051
+ "grad_norm": 0.39495351910591125,
2052
+ "learning_rate": 5.694905256949053e-06,
2053
+ "loss": 2.3584,
2054
+ "step": 5840
2055
+ },
2056
+ {
2057
+ "epoch": 0.43205780432057805,
2058
+ "grad_norm": 0.3598216772079468,
2059
+ "learning_rate": 5.680159256801593e-06,
2060
+ "loss": 2.3519,
2061
+ "step": 5860
2062
+ },
2063
+ {
2064
+ "epoch": 0.43353240433532403,
2065
+ "grad_norm": 0.4005551338195801,
2066
+ "learning_rate": 5.665413256654133e-06,
2067
+ "loss": 2.3455,
2068
+ "step": 5880
2069
+ },
2070
+ {
2071
+ "epoch": 0.43500700435007006,
2072
+ "grad_norm": 0.39282020926475525,
2073
+ "learning_rate": 5.650667256506674e-06,
2074
+ "loss": 2.3562,
2075
+ "step": 5900
2076
+ },
2077
+ {
2078
+ "epoch": 0.43648160436481603,
2079
+ "grad_norm": 0.3693186044692993,
2080
+ "learning_rate": 5.635921256359213e-06,
2081
+ "loss": 2.3404,
2082
+ "step": 5920
2083
+ },
2084
+ {
2085
+ "epoch": 0.43795620437956206,
2086
+ "grad_norm": 0.3796662986278534,
2087
+ "learning_rate": 5.621175256211753e-06,
2088
+ "loss": 2.3545,
2089
+ "step": 5940
2090
+ },
2091
+ {
2092
+ "epoch": 0.43943080439430804,
2093
+ "grad_norm": 0.3727371394634247,
2094
+ "learning_rate": 5.606429256064293e-06,
2095
+ "loss": 2.3613,
2096
+ "step": 5960
2097
+ },
2098
+ {
2099
+ "epoch": 0.44090540440905407,
2100
+ "grad_norm": 0.38252395391464233,
2101
+ "learning_rate": 5.591683255916834e-06,
2102
+ "loss": 2.3477,
2103
+ "step": 5980
2104
+ },
2105
+ {
2106
+ "epoch": 0.44238000442380004,
2107
+ "grad_norm": 0.36342570185661316,
2108
+ "learning_rate": 5.576937255769374e-06,
2109
+ "loss": 2.361,
2110
+ "step": 6000
2111
+ },
2112
+ {
2113
+ "epoch": 0.443854604438546,
2114
+ "grad_norm": 0.37686145305633545,
2115
+ "learning_rate": 5.562191255621913e-06,
2116
+ "loss": 2.358,
2117
+ "step": 6020
2118
+ },
2119
+ {
2120
+ "epoch": 0.44532920445329205,
2121
+ "grad_norm": 0.373463898897171,
2122
+ "learning_rate": 5.547445255474453e-06,
2123
+ "loss": 2.3528,
2124
+ "step": 6040
2125
+ },
2126
+ {
2127
+ "epoch": 0.446803804468038,
2128
+ "grad_norm": 0.3747381567955017,
2129
+ "learning_rate": 5.532699255326993e-06,
2130
+ "loss": 2.3428,
2131
+ "step": 6060
2132
+ },
2133
+ {
2134
+ "epoch": 0.44827840448278405,
2135
+ "grad_norm": 0.3892759382724762,
2136
+ "learning_rate": 5.517953255179534e-06,
2137
+ "loss": 2.3551,
2138
+ "step": 6080
2139
+ },
2140
+ {
2141
+ "epoch": 0.44975300449753003,
2142
+ "grad_norm": 0.37567898631095886,
2143
+ "learning_rate": 5.503207255032073e-06,
2144
+ "loss": 2.3487,
2145
+ "step": 6100
2146
+ },
2147
+ {
2148
+ "epoch": 0.45122760451227606,
2149
+ "grad_norm": 0.3452893793582916,
2150
+ "learning_rate": 5.488461254884613e-06,
2151
+ "loss": 2.3427,
2152
+ "step": 6120
2153
+ },
2154
+ {
2155
+ "epoch": 0.45270220452702203,
2156
+ "grad_norm": 0.387379914522171,
2157
+ "learning_rate": 5.473715254737153e-06,
2158
+ "loss": 2.3445,
2159
+ "step": 6140
2160
+ },
2161
+ {
2162
+ "epoch": 0.45417680454176806,
2163
+ "grad_norm": 0.36834776401519775,
2164
+ "learning_rate": 5.4589692545896936e-06,
2165
+ "loss": 2.3483,
2166
+ "step": 6160
2167
+ },
2168
+ {
2169
+ "epoch": 0.45565140455651404,
2170
+ "grad_norm": 0.3641732633113861,
2171
+ "learning_rate": 5.4442232544422335e-06,
2172
+ "loss": 2.3514,
2173
+ "step": 6180
2174
+ },
2175
+ {
2176
+ "epoch": 0.45712600457126007,
2177
+ "grad_norm": 0.41028374433517456,
2178
+ "learning_rate": 5.429477254294773e-06,
2179
+ "loss": 2.3425,
2180
+ "step": 6200
2181
+ },
2182
+ {
2183
+ "epoch": 0.45860060458600604,
2184
+ "grad_norm": 0.3992222845554352,
2185
+ "learning_rate": 5.414731254147313e-06,
2186
+ "loss": 2.3565,
2187
+ "step": 6220
2188
+ },
2189
+ {
2190
+ "epoch": 0.460075204600752,
2191
+ "grad_norm": 0.377287358045578,
2192
+ "learning_rate": 5.3999852539998535e-06,
2193
+ "loss": 2.3562,
2194
+ "step": 6240
2195
+ },
2196
+ {
2197
+ "epoch": 0.46154980461549805,
2198
+ "grad_norm": 0.4046432077884674,
2199
+ "learning_rate": 5.3852392538523935e-06,
2200
+ "loss": 2.3483,
2201
+ "step": 6260
2202
+ },
2203
+ {
2204
+ "epoch": 0.463024404630244,
2205
+ "grad_norm": 0.35515862703323364,
2206
+ "learning_rate": 5.370493253704933e-06,
2207
+ "loss": 2.353,
2208
+ "step": 6280
2209
+ },
2210
+ {
2211
+ "epoch": 0.46449900464499005,
2212
+ "grad_norm": 0.36241665482521057,
2213
+ "learning_rate": 5.355747253557473e-06,
2214
+ "loss": 2.3389,
2215
+ "step": 6300
2216
+ },
2217
+ {
2218
+ "epoch": 0.46597360465973603,
2219
+ "grad_norm": 0.3970908522605896,
2220
+ "learning_rate": 5.3410012534100135e-06,
2221
+ "loss": 2.3639,
2222
+ "step": 6320
2223
+ },
2224
+ {
2225
+ "epoch": 0.46744820467448206,
2226
+ "grad_norm": 0.36760082840919495,
2227
+ "learning_rate": 5.3262552532625535e-06,
2228
+ "loss": 2.3546,
2229
+ "step": 6340
2230
+ },
2231
+ {
2232
+ "epoch": 0.46892280468922803,
2233
+ "grad_norm": 0.34653639793395996,
2234
+ "learning_rate": 5.311509253115093e-06,
2235
+ "loss": 2.3448,
2236
+ "step": 6360
2237
+ },
2238
+ {
2239
+ "epoch": 0.47039740470397406,
2240
+ "grad_norm": 0.35380202531814575,
2241
+ "learning_rate": 5.296763252967633e-06,
2242
+ "loss": 2.3461,
2243
+ "step": 6380
2244
+ },
2245
+ {
2246
+ "epoch": 0.47187200471872004,
2247
+ "grad_norm": 0.3386562764644623,
2248
+ "learning_rate": 5.282017252820173e-06,
2249
+ "loss": 2.3489,
2250
+ "step": 6400
2251
+ },
2252
+ {
2253
+ "epoch": 0.47334660473346607,
2254
+ "grad_norm": 0.3794664144515991,
2255
+ "learning_rate": 5.267271252672713e-06,
2256
+ "loss": 2.3565,
2257
+ "step": 6420
2258
+ },
2259
+ {
2260
+ "epoch": 0.47482120474821204,
2261
+ "grad_norm": 0.3754018545150757,
2262
+ "learning_rate": 5.252525252525253e-06,
2263
+ "loss": 2.3443,
2264
+ "step": 6440
2265
+ },
2266
+ {
2267
+ "epoch": 0.4762958047629581,
2268
+ "grad_norm": 0.41498491168022156,
2269
+ "learning_rate": 5.2377792523777926e-06,
2270
+ "loss": 2.3501,
2271
+ "step": 6460
2272
+ },
2273
+ {
2274
+ "epoch": 0.47777040477770405,
2275
+ "grad_norm": 0.4399779140949249,
2276
+ "learning_rate": 5.2230332522303325e-06,
2277
+ "loss": 2.3434,
2278
+ "step": 6480
2279
+ },
2280
+ {
2281
+ "epoch": 0.47924500479245,
2282
+ "grad_norm": 0.3858584463596344,
2283
+ "learning_rate": 5.208287252082873e-06,
2284
+ "loss": 2.3612,
2285
+ "step": 6500
2286
+ },
2287
+ {
2288
+ "epoch": 0.48071960480719605,
2289
+ "grad_norm": 0.379436731338501,
2290
+ "learning_rate": 5.193541251935413e-06,
2291
+ "loss": 2.3534,
2292
+ "step": 6520
2293
+ },
2294
+ {
2295
+ "epoch": 0.48219420482194203,
2296
+ "grad_norm": 0.36703166365623474,
2297
+ "learning_rate": 5.1787952517879525e-06,
2298
+ "loss": 2.3385,
2299
+ "step": 6540
2300
+ },
2301
+ {
2302
+ "epoch": 0.48366880483668806,
2303
+ "grad_norm": 0.365614652633667,
2304
+ "learning_rate": 5.1640492516404925e-06,
2305
+ "loss": 2.3417,
2306
+ "step": 6560
2307
+ },
2308
+ {
2309
+ "epoch": 0.48514340485143403,
2310
+ "grad_norm": 0.3665941655635834,
2311
+ "learning_rate": 5.149303251493033e-06,
2312
+ "loss": 2.3478,
2313
+ "step": 6580
2314
+ },
2315
+ {
2316
+ "epoch": 0.48661800486618007,
2317
+ "grad_norm": 0.37497478723526,
2318
+ "learning_rate": 5.134557251345573e-06,
2319
+ "loss": 2.3713,
2320
+ "step": 6600
2321
+ },
2322
+ {
2323
+ "epoch": 0.48809260488092604,
2324
+ "grad_norm": 0.37431296706199646,
2325
+ "learning_rate": 5.1198112511981125e-06,
2326
+ "loss": 2.3544,
2327
+ "step": 6620
2328
+ },
2329
+ {
2330
+ "epoch": 0.48956720489567207,
2331
+ "grad_norm": 0.3486333191394806,
2332
+ "learning_rate": 5.1050652510506525e-06,
2333
+ "loss": 2.3578,
2334
+ "step": 6640
2335
+ },
2336
+ {
2337
+ "epoch": 0.49104180491041804,
2338
+ "grad_norm": 0.36677151918411255,
2339
+ "learning_rate": 5.090319250903193e-06,
2340
+ "loss": 2.3437,
2341
+ "step": 6660
2342
+ },
2343
+ {
2344
+ "epoch": 0.4925164049251641,
2345
+ "grad_norm": 0.36574506759643555,
2346
+ "learning_rate": 5.075573250755733e-06,
2347
+ "loss": 2.3574,
2348
+ "step": 6680
2349
+ },
2350
+ {
2351
+ "epoch": 0.49399100493991005,
2352
+ "grad_norm": 0.47523924708366394,
2353
+ "learning_rate": 5.060827250608273e-06,
2354
+ "loss": 2.3535,
2355
+ "step": 6700
2356
+ },
2357
+ {
2358
+ "epoch": 0.495465604954656,
2359
+ "grad_norm": 0.39526769518852234,
2360
+ "learning_rate": 5.046081250460812e-06,
2361
+ "loss": 2.3536,
2362
+ "step": 6720
2363
+ },
2364
+ {
2365
+ "epoch": 0.49694020496940206,
2366
+ "grad_norm": 0.38120681047439575,
2367
+ "learning_rate": 5.031335250313352e-06,
2368
+ "loss": 2.342,
2369
+ "step": 6740
2370
+ },
2371
+ {
2372
+ "epoch": 0.49841480498414803,
2373
+ "grad_norm": 0.3704290986061096,
2374
+ "learning_rate": 5.016589250165893e-06,
2375
+ "loss": 2.3455,
2376
+ "step": 6760
2377
+ },
2378
+ {
2379
+ "epoch": 0.49988940499889406,
2380
+ "grad_norm": 0.3897051215171814,
2381
+ "learning_rate": 5.001843250018433e-06,
2382
+ "loss": 2.3538,
2383
+ "step": 6780
2384
+ },
2385
+ {
2386
+ "epoch": 0.5013640050136401,
2387
+ "grad_norm": 0.3682934045791626,
2388
+ "learning_rate": 4.987097249870973e-06,
2389
+ "loss": 2.3647,
2390
+ "step": 6800
2391
+ },
2392
+ {
2393
+ "epoch": 0.5028386050283861,
2394
+ "grad_norm": 0.370624840259552,
2395
+ "learning_rate": 4.972351249723512e-06,
2396
+ "loss": 2.3561,
2397
+ "step": 6820
2398
+ },
2399
+ {
2400
+ "epoch": 0.504313205043132,
2401
+ "grad_norm": 0.3740212321281433,
2402
+ "learning_rate": 4.957605249576053e-06,
2403
+ "loss": 2.3297,
2404
+ "step": 6840
2405
+ },
2406
+ {
2407
+ "epoch": 0.505787805057878,
2408
+ "grad_norm": 0.3610004484653473,
2409
+ "learning_rate": 4.942859249428592e-06,
2410
+ "loss": 2.3645,
2411
+ "step": 6860
2412
+ },
2413
+ {
2414
+ "epoch": 0.5072624050726241,
2415
+ "grad_norm": 0.47170379757881165,
2416
+ "learning_rate": 4.928113249281133e-06,
2417
+ "loss": 2.3497,
2418
+ "step": 6880
2419
+ },
2420
+ {
2421
+ "epoch": 0.5087370050873701,
2422
+ "grad_norm": 0.36539992690086365,
2423
+ "learning_rate": 4.913367249133673e-06,
2424
+ "loss": 2.357,
2425
+ "step": 6900
2426
+ },
2427
+ {
2428
+ "epoch": 0.510211605102116,
2429
+ "grad_norm": 0.4011884927749634,
2430
+ "learning_rate": 4.898621248986213e-06,
2431
+ "loss": 2.3517,
2432
+ "step": 6920
2433
+ },
2434
+ {
2435
+ "epoch": 0.511686205116862,
2436
+ "grad_norm": 0.37110114097595215,
2437
+ "learning_rate": 4.883875248838753e-06,
2438
+ "loss": 2.359,
2439
+ "step": 6940
2440
+ },
2441
+ {
2442
+ "epoch": 0.513160805131608,
2443
+ "grad_norm": 0.36303210258483887,
2444
+ "learning_rate": 4.869129248691293e-06,
2445
+ "loss": 2.3638,
2446
+ "step": 6960
2447
+ },
2448
+ {
2449
+ "epoch": 0.5146354051463541,
2450
+ "grad_norm": 0.39052852988243103,
2451
+ "learning_rate": 4.854383248543833e-06,
2452
+ "loss": 2.3573,
2453
+ "step": 6980
2454
+ },
2455
+ {
2456
+ "epoch": 0.5161100051611001,
2457
+ "grad_norm": 0.3620651066303253,
2458
+ "learning_rate": 4.839637248396373e-06,
2459
+ "loss": 2.3413,
2460
+ "step": 7000
2461
+ },
2462
+ {
2463
+ "epoch": 0.517584605175846,
2464
+ "grad_norm": 0.35670602321624756,
2465
+ "learning_rate": 4.824891248248913e-06,
2466
+ "loss": 2.358,
2467
+ "step": 7020
2468
+ },
2469
+ {
2470
+ "epoch": 0.519059205190592,
2471
+ "grad_norm": 0.3757779002189636,
2472
+ "learning_rate": 4.810145248101453e-06,
2473
+ "loss": 2.3408,
2474
+ "step": 7040
2475
+ },
2476
+ {
2477
+ "epoch": 0.5205338052053381,
2478
+ "grad_norm": 0.3540636897087097,
2479
+ "learning_rate": 4.795399247953993e-06,
2480
+ "loss": 2.3386,
2481
+ "step": 7060
2482
+ },
2483
+ {
2484
+ "epoch": 0.5220084052200841,
2485
+ "grad_norm": 0.3777630031108856,
2486
+ "learning_rate": 4.780653247806532e-06,
2487
+ "loss": 2.3498,
2488
+ "step": 7080
2489
+ },
2490
+ {
2491
+ "epoch": 0.52348300523483,
2492
+ "grad_norm": 0.38715338706970215,
2493
+ "learning_rate": 4.765907247659073e-06,
2494
+ "loss": 2.3357,
2495
+ "step": 7100
2496
+ },
2497
+ {
2498
+ "epoch": 0.524957605249576,
2499
+ "grad_norm": 0.37200599908828735,
2500
+ "learning_rate": 4.751161247511613e-06,
2501
+ "loss": 2.3409,
2502
+ "step": 7120
2503
+ },
2504
+ {
2505
+ "epoch": 0.5264322052643221,
2506
+ "grad_norm": 0.34266674518585205,
2507
+ "learning_rate": 4.736415247364153e-06,
2508
+ "loss": 2.3489,
2509
+ "step": 7140
2510
+ },
2511
+ {
2512
+ "epoch": 0.5279068052790681,
2513
+ "grad_norm": 0.39111700654029846,
2514
+ "learning_rate": 4.721669247216693e-06,
2515
+ "loss": 2.3399,
2516
+ "step": 7160
2517
+ },
2518
+ {
2519
+ "epoch": 0.5293814052938141,
2520
+ "grad_norm": 0.34332647919654846,
2521
+ "learning_rate": 4.706923247069233e-06,
2522
+ "loss": 2.3407,
2523
+ "step": 7180
2524
+ },
2525
+ {
2526
+ "epoch": 0.53085600530856,
2527
+ "grad_norm": 0.3694305717945099,
2528
+ "learning_rate": 4.692177246921773e-06,
2529
+ "loss": 2.3507,
2530
+ "step": 7200
2531
+ },
2532
+ {
2533
+ "epoch": 0.532330605323306,
2534
+ "grad_norm": 0.3671371042728424,
2535
+ "learning_rate": 4.677431246774313e-06,
2536
+ "loss": 2.3511,
2537
+ "step": 7220
2538
+ },
2539
+ {
2540
+ "epoch": 0.5338052053380521,
2541
+ "grad_norm": 0.35831600427627563,
2542
+ "learning_rate": 4.662685246626853e-06,
2543
+ "loss": 2.3409,
2544
+ "step": 7240
2545
+ },
2546
+ {
2547
+ "epoch": 0.5352798053527981,
2548
+ "grad_norm": 0.388938307762146,
2549
+ "learning_rate": 4.647939246479393e-06,
2550
+ "loss": 2.3491,
2551
+ "step": 7260
2552
+ },
2553
+ {
2554
+ "epoch": 0.536754405367544,
2555
+ "grad_norm": 0.3723289668560028,
2556
+ "learning_rate": 4.633193246331933e-06,
2557
+ "loss": 2.3538,
2558
+ "step": 7280
2559
+ },
2560
+ {
2561
+ "epoch": 0.53822900538229,
2562
+ "grad_norm": 0.3653968870639801,
2563
+ "learning_rate": 4.618447246184473e-06,
2564
+ "loss": 2.3477,
2565
+ "step": 7300
2566
+ },
2567
+ {
2568
+ "epoch": 0.5397036053970361,
2569
+ "grad_norm": 0.3933035731315613,
2570
+ "learning_rate": 4.603701246037013e-06,
2571
+ "loss": 2.3452,
2572
+ "step": 7320
2573
+ },
2574
+ {
2575
+ "epoch": 0.5411782054117821,
2576
+ "grad_norm": 0.36271172761917114,
2577
+ "learning_rate": 4.588955245889553e-06,
2578
+ "loss": 2.3538,
2579
+ "step": 7340
2580
+ },
2581
+ {
2582
+ "epoch": 0.542652805426528,
2583
+ "grad_norm": 0.3978424072265625,
2584
+ "learning_rate": 4.574209245742093e-06,
2585
+ "loss": 2.3462,
2586
+ "step": 7360
2587
+ },
2588
+ {
2589
+ "epoch": 0.544127405441274,
2590
+ "grad_norm": 0.35292622447013855,
2591
+ "learning_rate": 4.559463245594633e-06,
2592
+ "loss": 2.3517,
2593
+ "step": 7380
2594
+ },
2595
+ {
2596
+ "epoch": 0.54560200545602,
2597
+ "grad_norm": 0.3812173008918762,
2598
+ "learning_rate": 4.544717245447173e-06,
2599
+ "loss": 2.3494,
2600
+ "step": 7400
2601
+ },
2602
+ {
2603
+ "epoch": 0.5470766054707661,
2604
+ "grad_norm": 0.3730863034725189,
2605
+ "learning_rate": 4.529971245299713e-06,
2606
+ "loss": 2.3468,
2607
+ "step": 7420
2608
+ },
2609
+ {
2610
+ "epoch": 0.5485512054855121,
2611
+ "grad_norm": 0.3639664053916931,
2612
+ "learning_rate": 4.515225245152253e-06,
2613
+ "loss": 2.3509,
2614
+ "step": 7440
2615
+ },
2616
+ {
2617
+ "epoch": 0.550025805500258,
2618
+ "grad_norm": 0.3697253167629242,
2619
+ "learning_rate": 4.500479245004793e-06,
2620
+ "loss": 2.3563,
2621
+ "step": 7460
2622
+ },
2623
+ {
2624
+ "epoch": 0.551500405515004,
2625
+ "grad_norm": 0.44058653712272644,
2626
+ "learning_rate": 4.485733244857333e-06,
2627
+ "loss": 2.3463,
2628
+ "step": 7480
2629
+ },
2630
+ {
2631
+ "epoch": 0.5529750055297501,
2632
+ "grad_norm": 0.3621077835559845,
2633
+ "learning_rate": 4.470987244709873e-06,
2634
+ "loss": 2.3484,
2635
+ "step": 7500
2636
+ },
2637
+ {
2638
+ "epoch": 0.5544496055444961,
2639
+ "grad_norm": 0.3550095558166504,
2640
+ "learning_rate": 4.456241244562413e-06,
2641
+ "loss": 2.3503,
2642
+ "step": 7520
2643
+ },
2644
+ {
2645
+ "epoch": 0.555924205559242,
2646
+ "grad_norm": 0.40449610352516174,
2647
+ "learning_rate": 4.441495244414953e-06,
2648
+ "loss": 2.3392,
2649
+ "step": 7540
2650
+ },
2651
+ {
2652
+ "epoch": 0.557398805573988,
2653
+ "grad_norm": 0.41954267024993896,
2654
+ "learning_rate": 4.426749244267493e-06,
2655
+ "loss": 2.3544,
2656
+ "step": 7560
2657
+ },
2658
+ {
2659
+ "epoch": 0.5588734055887341,
2660
+ "grad_norm": 0.3714821934700012,
2661
+ "learning_rate": 4.412003244120033e-06,
2662
+ "loss": 2.3517,
2663
+ "step": 7580
2664
+ },
2665
+ {
2666
+ "epoch": 0.5603480056034801,
2667
+ "grad_norm": 0.35409021377563477,
2668
+ "learning_rate": 4.397257243972573e-06,
2669
+ "loss": 2.3368,
2670
+ "step": 7600
2671
+ },
2672
+ {
2673
+ "epoch": 0.5618226056182261,
2674
+ "grad_norm": 0.38718315958976746,
2675
+ "learning_rate": 4.382511243825113e-06,
2676
+ "loss": 2.349,
2677
+ "step": 7620
2678
+ },
2679
+ {
2680
+ "epoch": 0.563297205632972,
2681
+ "grad_norm": 0.3989495038986206,
2682
+ "learning_rate": 4.367765243677653e-06,
2683
+ "loss": 2.3622,
2684
+ "step": 7640
2685
+ },
2686
+ {
2687
+ "epoch": 0.564771805647718,
2688
+ "grad_norm": 0.38206225633621216,
2689
+ "learning_rate": 4.353019243530193e-06,
2690
+ "loss": 2.348,
2691
+ "step": 7660
2692
+ },
2693
+ {
2694
+ "epoch": 0.5662464056624641,
2695
+ "grad_norm": 0.4094175100326538,
2696
+ "learning_rate": 4.338273243382733e-06,
2697
+ "loss": 2.337,
2698
+ "step": 7680
2699
+ },
2700
+ {
2701
+ "epoch": 0.5677210056772101,
2702
+ "grad_norm": 0.3633534908294678,
2703
+ "learning_rate": 4.323527243235273e-06,
2704
+ "loss": 2.3413,
2705
+ "step": 7700
2706
+ },
2707
+ {
2708
+ "epoch": 0.569195605691956,
2709
+ "grad_norm": 0.3575294613838196,
2710
+ "learning_rate": 4.308781243087813e-06,
2711
+ "loss": 2.3388,
2712
+ "step": 7720
2713
+ },
2714
+ {
2715
+ "epoch": 0.570670205706702,
2716
+ "grad_norm": 0.3401530683040619,
2717
+ "learning_rate": 4.294035242940353e-06,
2718
+ "loss": 2.337,
2719
+ "step": 7740
2720
+ },
2721
+ {
2722
+ "epoch": 0.5721448057214481,
2723
+ "grad_norm": 0.37117597460746765,
2724
+ "learning_rate": 4.279289242792893e-06,
2725
+ "loss": 2.3489,
2726
+ "step": 7760
2727
+ },
2728
+ {
2729
+ "epoch": 0.5736194057361941,
2730
+ "grad_norm": 0.4254913330078125,
2731
+ "learning_rate": 4.264543242645433e-06,
2732
+ "loss": 2.3592,
2733
+ "step": 7780
2734
+ },
2735
+ {
2736
+ "epoch": 0.57509400575094,
2737
+ "grad_norm": 0.3747338354587555,
2738
+ "learning_rate": 4.249797242497973e-06,
2739
+ "loss": 2.3525,
2740
+ "step": 7800
2741
+ },
2742
+ {
2743
+ "epoch": 0.576568605765686,
2744
+ "grad_norm": 0.4081355035305023,
2745
+ "learning_rate": 4.235051242350513e-06,
2746
+ "loss": 2.3482,
2747
+ "step": 7820
2748
+ },
2749
+ {
2750
+ "epoch": 0.5780432057804321,
2751
+ "grad_norm": 0.3858603835105896,
2752
+ "learning_rate": 4.220305242203053e-06,
2753
+ "loss": 2.3443,
2754
+ "step": 7840
2755
+ },
2756
+ {
2757
+ "epoch": 0.5795178057951781,
2758
+ "grad_norm": 0.3556371033191681,
2759
+ "learning_rate": 4.205559242055593e-06,
2760
+ "loss": 2.3328,
2761
+ "step": 7860
2762
+ },
2763
+ {
2764
+ "epoch": 0.5809924058099241,
2765
+ "grad_norm": 0.3800757825374603,
2766
+ "learning_rate": 4.190813241908133e-06,
2767
+ "loss": 2.3494,
2768
+ "step": 7880
2769
+ },
2770
+ {
2771
+ "epoch": 0.58246700582467,
2772
+ "grad_norm": 0.3651157021522522,
2773
+ "learning_rate": 4.176067241760673e-06,
2774
+ "loss": 2.3476,
2775
+ "step": 7900
2776
+ },
2777
+ {
2778
+ "epoch": 0.583941605839416,
2779
+ "grad_norm": 0.35027289390563965,
2780
+ "learning_rate": 4.161321241613213e-06,
2781
+ "loss": 2.362,
2782
+ "step": 7920
2783
+ },
2784
+ {
2785
+ "epoch": 0.5854162058541621,
2786
+ "grad_norm": 0.37928810715675354,
2787
+ "learning_rate": 4.146575241465753e-06,
2788
+ "loss": 2.3471,
2789
+ "step": 7940
2790
+ },
2791
+ {
2792
+ "epoch": 0.5868908058689081,
2793
+ "grad_norm": 0.3614574372768402,
2794
+ "learning_rate": 4.131829241318293e-06,
2795
+ "loss": 2.3469,
2796
+ "step": 7960
2797
+ },
2798
+ {
2799
+ "epoch": 0.588365405883654,
2800
+ "grad_norm": 0.3692425489425659,
2801
+ "learning_rate": 4.117083241170833e-06,
2802
+ "loss": 2.3517,
2803
+ "step": 7980
2804
+ },
2805
+ {
2806
+ "epoch": 0.5898400058984,
2807
+ "grad_norm": 0.3615201413631439,
2808
+ "learning_rate": 4.102337241023373e-06,
2809
+ "loss": 2.3385,
2810
+ "step": 8000
2811
+ },
2812
+ {
2813
+ "epoch": 0.5913146059131461,
2814
+ "grad_norm": 0.37288710474967957,
2815
+ "learning_rate": 4.0875912408759126e-06,
2816
+ "loss": 2.3592,
2817
+ "step": 8020
2818
+ },
2819
+ {
2820
+ "epoch": 0.5927892059278921,
2821
+ "grad_norm": 0.3470204472541809,
2822
+ "learning_rate": 4.072845240728453e-06,
2823
+ "loss": 2.3546,
2824
+ "step": 8040
2825
+ },
2826
+ {
2827
+ "epoch": 0.5942638059426381,
2828
+ "grad_norm": 0.38719677925109863,
2829
+ "learning_rate": 4.0580992405809925e-06,
2830
+ "loss": 2.3439,
2831
+ "step": 8060
2832
+ },
2833
+ {
2834
+ "epoch": 0.595738405957384,
2835
+ "grad_norm": 0.3788948953151703,
2836
+ "learning_rate": 4.0433532404335325e-06,
2837
+ "loss": 2.3555,
2838
+ "step": 8080
2839
+ },
2840
+ {
2841
+ "epoch": 0.59721300597213,
2842
+ "grad_norm": 0.42110002040863037,
2843
+ "learning_rate": 4.0286072402860725e-06,
2844
+ "loss": 2.3502,
2845
+ "step": 8100
2846
+ },
2847
+ {
2848
+ "epoch": 0.5986876059868761,
2849
+ "grad_norm": 0.4007836580276489,
2850
+ "learning_rate": 4.0138612401386125e-06,
2851
+ "loss": 2.3571,
2852
+ "step": 8120
2853
+ },
2854
+ {
2855
+ "epoch": 0.6001622060016221,
2856
+ "grad_norm": 0.36043021082878113,
2857
+ "learning_rate": 3.9991152399911525e-06,
2858
+ "loss": 2.3589,
2859
+ "step": 8140
2860
+ },
2861
+ {
2862
+ "epoch": 0.601636806016368,
2863
+ "grad_norm": 0.4028465151786804,
2864
+ "learning_rate": 3.9843692398436925e-06,
2865
+ "loss": 2.3399,
2866
+ "step": 8160
2867
+ },
2868
+ {
2869
+ "epoch": 0.603111406031114,
2870
+ "grad_norm": 0.39534154534339905,
2871
+ "learning_rate": 3.9696232396962325e-06,
2872
+ "loss": 2.3571,
2873
+ "step": 8180
2874
+ },
2875
+ {
2876
+ "epoch": 0.6045860060458601,
2877
+ "grad_norm": 0.361286997795105,
2878
+ "learning_rate": 3.9548772395487725e-06,
2879
+ "loss": 2.3358,
2880
+ "step": 8200
2881
+ },
2882
+ {
2883
+ "epoch": 0.6060606060606061,
2884
+ "grad_norm": 0.3674760162830353,
2885
+ "learning_rate": 3.9401312394013125e-06,
2886
+ "loss": 2.3449,
2887
+ "step": 8220
2888
+ },
2889
+ {
2890
+ "epoch": 0.607535206075352,
2891
+ "grad_norm": 0.39205583930015564,
2892
+ "learning_rate": 3.9253852392538525e-06,
2893
+ "loss": 2.3558,
2894
+ "step": 8240
2895
+ },
2896
+ {
2897
+ "epoch": 0.609009806090098,
2898
+ "grad_norm": 0.3901711404323578,
2899
+ "learning_rate": 3.910639239106393e-06,
2900
+ "loss": 2.3565,
2901
+ "step": 8260
2902
+ },
2903
+ {
2904
+ "epoch": 0.6104844061048441,
2905
+ "grad_norm": 0.3791930377483368,
2906
+ "learning_rate": 3.8958932389589324e-06,
2907
+ "loss": 2.3425,
2908
+ "step": 8280
2909
+ },
2910
+ {
2911
+ "epoch": 0.6119590061195901,
2912
+ "grad_norm": 0.36277633905410767,
2913
+ "learning_rate": 3.881147238811473e-06,
2914
+ "loss": 2.3485,
2915
+ "step": 8300
2916
+ },
2917
+ {
2918
+ "epoch": 0.6134336061343361,
2919
+ "grad_norm": 0.35051462054252625,
2920
+ "learning_rate": 3.866401238664012e-06,
2921
+ "loss": 2.3486,
2922
+ "step": 8320
2923
+ },
2924
+ {
2925
+ "epoch": 0.614908206149082,
2926
+ "grad_norm": 0.3554931581020355,
2927
+ "learning_rate": 3.851655238516553e-06,
2928
+ "loss": 2.3626,
2929
+ "step": 8340
2930
+ },
2931
+ {
2932
+ "epoch": 0.616382806163828,
2933
+ "grad_norm": 0.36661261320114136,
2934
+ "learning_rate": 3.836909238369092e-06,
2935
+ "loss": 2.3405,
2936
+ "step": 8360
2937
+ },
2938
+ {
2939
+ "epoch": 0.6178574061785741,
2940
+ "grad_norm": 0.3876403570175171,
2941
+ "learning_rate": 3.822163238221633e-06,
2942
+ "loss": 2.3422,
2943
+ "step": 8380
2944
+ },
2945
+ {
2946
+ "epoch": 0.6193320061933201,
2947
+ "grad_norm": 0.3810766637325287,
2948
+ "learning_rate": 3.8074172380741724e-06,
2949
+ "loss": 2.3466,
2950
+ "step": 8400
2951
+ },
2952
+ {
2953
+ "epoch": 0.620806606208066,
2954
+ "grad_norm": 0.3949829638004303,
2955
+ "learning_rate": 3.7926712379267128e-06,
2956
+ "loss": 2.3501,
2957
+ "step": 8420
2958
+ },
2959
+ {
2960
+ "epoch": 0.622281206222812,
2961
+ "grad_norm": 0.39543384313583374,
2962
+ "learning_rate": 3.7779252377792523e-06,
2963
+ "loss": 2.3365,
2964
+ "step": 8440
2965
+ },
2966
+ {
2967
+ "epoch": 0.6237558062375581,
2968
+ "grad_norm": 0.3472473919391632,
2969
+ "learning_rate": 3.7631792376317928e-06,
2970
+ "loss": 2.3544,
2971
+ "step": 8460
2972
+ },
2973
+ {
2974
+ "epoch": 0.6252304062523041,
2975
+ "grad_norm": 0.3589697778224945,
2976
+ "learning_rate": 3.7484332374843323e-06,
2977
+ "loss": 2.3402,
2978
+ "step": 8480
2979
+ },
2980
+ {
2981
+ "epoch": 0.6267050062670501,
2982
+ "grad_norm": 0.37059295177459717,
2983
+ "learning_rate": 3.7336872373368727e-06,
2984
+ "loss": 2.3462,
2985
+ "step": 8500
2986
+ },
2987
+ {
2988
+ "epoch": 0.628179606281796,
2989
+ "grad_norm": 0.40565216541290283,
2990
+ "learning_rate": 3.7189412371894127e-06,
2991
+ "loss": 2.3501,
2992
+ "step": 8520
2993
+ },
2994
+ {
2995
+ "epoch": 0.6296542062965421,
2996
+ "grad_norm": 0.36283183097839355,
2997
+ "learning_rate": 3.7041952370419527e-06,
2998
+ "loss": 2.3423,
2999
+ "step": 8540
3000
+ },
3001
+ {
3002
+ "epoch": 0.6311288063112881,
3003
+ "grad_norm": 0.3886597156524658,
3004
+ "learning_rate": 3.6894492368944927e-06,
3005
+ "loss": 2.3386,
3006
+ "step": 8560
3007
+ },
3008
+ {
3009
+ "epoch": 0.6326034063260341,
3010
+ "grad_norm": 0.3656463027000427,
3011
+ "learning_rate": 3.6747032367470327e-06,
3012
+ "loss": 2.3526,
3013
+ "step": 8580
3014
+ },
3015
+ {
3016
+ "epoch": 0.63407800634078,
3017
+ "grad_norm": 0.4167777895927429,
3018
+ "learning_rate": 3.6599572365995727e-06,
3019
+ "loss": 2.3581,
3020
+ "step": 8600
3021
+ },
3022
+ {
3023
+ "epoch": 0.635552606355526,
3024
+ "grad_norm": 0.3539351522922516,
3025
+ "learning_rate": 3.6452112364521127e-06,
3026
+ "loss": 2.3396,
3027
+ "step": 8620
3028
+ },
3029
+ {
3030
+ "epoch": 0.6370272063702721,
3031
+ "grad_norm": 0.3550587296485901,
3032
+ "learning_rate": 3.6304652363046527e-06,
3033
+ "loss": 2.344,
3034
+ "step": 8640
3035
+ },
3036
+ {
3037
+ "epoch": 0.6385018063850181,
3038
+ "grad_norm": 0.3939066529273987,
3039
+ "learning_rate": 3.6157192361571927e-06,
3040
+ "loss": 2.357,
3041
+ "step": 8660
3042
+ },
3043
+ {
3044
+ "epoch": 0.639976406399764,
3045
+ "grad_norm": 0.3614286482334137,
3046
+ "learning_rate": 3.6009732360097326e-06,
3047
+ "loss": 2.3433,
3048
+ "step": 8680
3049
+ },
3050
+ {
3051
+ "epoch": 0.64145100641451,
3052
+ "grad_norm": 0.402174711227417,
3053
+ "learning_rate": 3.586227235862272e-06,
3054
+ "loss": 2.3329,
3055
+ "step": 8700
3056
+ },
3057
+ {
3058
+ "epoch": 0.6429256064292561,
3059
+ "grad_norm": 0.4092716574668884,
3060
+ "learning_rate": 3.5714812357148126e-06,
3061
+ "loss": 2.3489,
3062
+ "step": 8720
3063
+ },
3064
+ {
3065
+ "epoch": 0.6444002064440021,
3066
+ "grad_norm": 0.3816792666912079,
3067
+ "learning_rate": 3.556735235567353e-06,
3068
+ "loss": 2.3493,
3069
+ "step": 8740
3070
+ },
3071
+ {
3072
+ "epoch": 0.6458748064587481,
3073
+ "grad_norm": 0.34877556562423706,
3074
+ "learning_rate": 3.5419892354198926e-06,
3075
+ "loss": 2.3694,
3076
+ "step": 8760
3077
+ },
3078
+ {
3079
+ "epoch": 0.647349406473494,
3080
+ "grad_norm": 0.39949148893356323,
3081
+ "learning_rate": 3.527243235272433e-06,
3082
+ "loss": 2.3366,
3083
+ "step": 8780
3084
+ },
3085
+ {
3086
+ "epoch": 0.64882400648824,
3087
+ "grad_norm": 0.3934047222137451,
3088
+ "learning_rate": 3.5124972351249726e-06,
3089
+ "loss": 2.3412,
3090
+ "step": 8800
3091
+ },
3092
+ {
3093
+ "epoch": 0.6502986065029861,
3094
+ "grad_norm": 0.3608020842075348,
3095
+ "learning_rate": 3.497751234977513e-06,
3096
+ "loss": 2.3325,
3097
+ "step": 8820
3098
+ },
3099
+ {
3100
+ "epoch": 0.6517732065177321,
3101
+ "grad_norm": 0.3528784215450287,
3102
+ "learning_rate": 3.4830052348300526e-06,
3103
+ "loss": 2.3539,
3104
+ "step": 8840
3105
+ },
3106
+ {
3107
+ "epoch": 0.653247806532478,
3108
+ "grad_norm": 0.34913375973701477,
3109
+ "learning_rate": 3.4682592346825925e-06,
3110
+ "loss": 2.3501,
3111
+ "step": 8860
3112
+ },
3113
+ {
3114
+ "epoch": 0.654722406547224,
3115
+ "grad_norm": 0.39281442761421204,
3116
+ "learning_rate": 3.4535132345351325e-06,
3117
+ "loss": 2.3413,
3118
+ "step": 8880
3119
+ },
3120
+ {
3121
+ "epoch": 0.6561970065619701,
3122
+ "grad_norm": 0.352173775434494,
3123
+ "learning_rate": 3.4387672343876725e-06,
3124
+ "loss": 2.3636,
3125
+ "step": 8900
3126
+ },
3127
+ {
3128
+ "epoch": 0.6576716065767161,
3129
+ "grad_norm": 0.44138431549072266,
3130
+ "learning_rate": 3.4240212342402125e-06,
3131
+ "loss": 2.3365,
3132
+ "step": 8920
3133
+ },
3134
+ {
3135
+ "epoch": 0.6591462065914621,
3136
+ "grad_norm": 0.38679641485214233,
3137
+ "learning_rate": 3.4092752340927525e-06,
3138
+ "loss": 2.3473,
3139
+ "step": 8940
3140
+ },
3141
+ {
3142
+ "epoch": 0.660620806606208,
3143
+ "grad_norm": 0.37204068899154663,
3144
+ "learning_rate": 3.394529233945293e-06,
3145
+ "loss": 2.3422,
3146
+ "step": 8960
3147
+ },
3148
+ {
3149
+ "epoch": 0.6620954066209541,
3150
+ "grad_norm": 0.36612871289253235,
3151
+ "learning_rate": 3.3797832337978325e-06,
3152
+ "loss": 2.3423,
3153
+ "step": 8980
3154
+ },
3155
+ {
3156
+ "epoch": 0.6635700066357001,
3157
+ "grad_norm": 0.36784979701042175,
3158
+ "learning_rate": 3.365037233650373e-06,
3159
+ "loss": 2.3458,
3160
+ "step": 9000
3161
+ },
3162
+ {
3163
+ "epoch": 0.6650446066504461,
3164
+ "grad_norm": 0.3783581852912903,
3165
+ "learning_rate": 3.3502912335029125e-06,
3166
+ "loss": 2.351,
3167
+ "step": 9020
3168
+ },
3169
+ {
3170
+ "epoch": 0.666519206665192,
3171
+ "grad_norm": 0.4059109389781952,
3172
+ "learning_rate": 3.335545233355453e-06,
3173
+ "loss": 2.3504,
3174
+ "step": 9040
3175
+ },
3176
+ {
3177
+ "epoch": 0.667993806679938,
3178
+ "grad_norm": 0.39486610889434814,
3179
+ "learning_rate": 3.3207992332079924e-06,
3180
+ "loss": 2.3425,
3181
+ "step": 9060
3182
+ },
3183
+ {
3184
+ "epoch": 0.6694684066946841,
3185
+ "grad_norm": 0.35696399211883545,
3186
+ "learning_rate": 3.306053233060533e-06,
3187
+ "loss": 2.3668,
3188
+ "step": 9080
3189
+ },
3190
+ {
3191
+ "epoch": 0.6709430067094301,
3192
+ "grad_norm": 0.3573276102542877,
3193
+ "learning_rate": 3.2913072329130724e-06,
3194
+ "loss": 2.3503,
3195
+ "step": 9100
3196
+ },
3197
+ {
3198
+ "epoch": 0.672417606724176,
3199
+ "grad_norm": 0.4014319181442261,
3200
+ "learning_rate": 3.276561232765613e-06,
3201
+ "loss": 2.3364,
3202
+ "step": 9120
3203
+ },
3204
+ {
3205
+ "epoch": 0.673892206738922,
3206
+ "grad_norm": 0.3681239187717438,
3207
+ "learning_rate": 3.2618152326181524e-06,
3208
+ "loss": 2.3451,
3209
+ "step": 9140
3210
+ },
3211
+ {
3212
+ "epoch": 0.6753668067536681,
3213
+ "grad_norm": 0.4503883421421051,
3214
+ "learning_rate": 3.247069232470693e-06,
3215
+ "loss": 2.3408,
3216
+ "step": 9160
3217
+ },
3218
+ {
3219
+ "epoch": 0.6768414067684141,
3220
+ "grad_norm": 0.37346574664115906,
3221
+ "learning_rate": 3.232323232323233e-06,
3222
+ "loss": 2.3598,
3223
+ "step": 9180
3224
+ },
3225
+ {
3226
+ "epoch": 0.6783160067831601,
3227
+ "grad_norm": 0.3615313470363617,
3228
+ "learning_rate": 3.2175772321757724e-06,
3229
+ "loss": 2.3434,
3230
+ "step": 9200
3231
+ },
3232
+ {
3233
+ "epoch": 0.679790606797906,
3234
+ "grad_norm": 0.3929205536842346,
3235
+ "learning_rate": 3.2028312320283128e-06,
3236
+ "loss": 2.3439,
3237
+ "step": 9220
3238
+ },
3239
+ {
3240
+ "epoch": 0.681265206812652,
3241
+ "grad_norm": 0.37737640738487244,
3242
+ "learning_rate": 3.1880852318808523e-06,
3243
+ "loss": 2.349,
3244
+ "step": 9240
3245
+ },
3246
+ {
3247
+ "epoch": 0.6827398068273981,
3248
+ "grad_norm": 0.41763532161712646,
3249
+ "learning_rate": 3.1733392317333928e-06,
3250
+ "loss": 2.3422,
3251
+ "step": 9260
3252
+ },
3253
+ {
3254
+ "epoch": 0.6842144068421441,
3255
+ "grad_norm": 0.4058336615562439,
3256
+ "learning_rate": 3.1585932315859323e-06,
3257
+ "loss": 2.3459,
3258
+ "step": 9280
3259
+ },
3260
+ {
3261
+ "epoch": 0.68568900685689,
3262
+ "grad_norm": 0.3860774636268616,
3263
+ "learning_rate": 3.1438472314384727e-06,
3264
+ "loss": 2.3524,
3265
+ "step": 9300
3266
+ },
3267
+ {
3268
+ "epoch": 0.687163606871636,
3269
+ "grad_norm": 0.401663213968277,
3270
+ "learning_rate": 3.1291012312910123e-06,
3271
+ "loss": 2.3326,
3272
+ "step": 9320
3273
+ },
3274
+ {
3275
+ "epoch": 0.6886382068863821,
3276
+ "grad_norm": 0.3691912889480591,
3277
+ "learning_rate": 3.1143552311435527e-06,
3278
+ "loss": 2.337,
3279
+ "step": 9340
3280
+ },
3281
+ {
3282
+ "epoch": 0.6901128069011281,
3283
+ "grad_norm": 0.4243042767047882,
3284
+ "learning_rate": 3.0996092309960923e-06,
3285
+ "loss": 2.3432,
3286
+ "step": 9360
3287
+ },
3288
+ {
3289
+ "epoch": 0.6915874069158741,
3290
+ "grad_norm": 0.3880211412906647,
3291
+ "learning_rate": 3.0848632308486327e-06,
3292
+ "loss": 2.3664,
3293
+ "step": 9380
3294
+ },
3295
+ {
3296
+ "epoch": 0.69306200693062,
3297
+ "grad_norm": 0.3545699119567871,
3298
+ "learning_rate": 3.0701172307011723e-06,
3299
+ "loss": 2.3448,
3300
+ "step": 9400
3301
+ },
3302
+ {
3303
+ "epoch": 0.6945366069453661,
3304
+ "grad_norm": 0.38192018866539,
3305
+ "learning_rate": 3.0553712305537127e-06,
3306
+ "loss": 2.3564,
3307
+ "step": 9420
3308
+ },
3309
+ {
3310
+ "epoch": 0.6960112069601121,
3311
+ "grad_norm": 0.39317330718040466,
3312
+ "learning_rate": 3.0406252304062527e-06,
3313
+ "loss": 2.3676,
3314
+ "step": 9440
3315
+ },
3316
+ {
3317
+ "epoch": 0.6974858069748581,
3318
+ "grad_norm": 0.3686807453632355,
3319
+ "learning_rate": 3.0258792302587927e-06,
3320
+ "loss": 2.3513,
3321
+ "step": 9460
3322
+ },
3323
+ {
3324
+ "epoch": 0.698960406989604,
3325
+ "grad_norm": 0.43928787112236023,
3326
+ "learning_rate": 3.0111332301113326e-06,
3327
+ "loss": 2.3368,
3328
+ "step": 9480
3329
+ },
3330
+ {
3331
+ "epoch": 0.70043500700435,
3332
+ "grad_norm": 0.3794805705547333,
3333
+ "learning_rate": 2.9963872299638726e-06,
3334
+ "loss": 2.3504,
3335
+ "step": 9500
3336
+ },
3337
+ {
3338
+ "epoch": 0.7019096070190961,
3339
+ "grad_norm": 0.391825407743454,
3340
+ "learning_rate": 2.9816412298164126e-06,
3341
+ "loss": 2.355,
3342
+ "step": 9520
3343
+ },
3344
+ {
3345
+ "epoch": 0.7033842070338421,
3346
+ "grad_norm": 0.37879860401153564,
3347
+ "learning_rate": 2.966895229668952e-06,
3348
+ "loss": 2.3454,
3349
+ "step": 9540
3350
+ },
3351
+ {
3352
+ "epoch": 0.704858807048588,
3353
+ "grad_norm": 0.4129418134689331,
3354
+ "learning_rate": 2.9521492295214926e-06,
3355
+ "loss": 2.3496,
3356
+ "step": 9560
3357
+ },
3358
+ {
3359
+ "epoch": 0.706333407063334,
3360
+ "grad_norm": 0.4124239683151245,
3361
+ "learning_rate": 2.937403229374032e-06,
3362
+ "loss": 2.353,
3363
+ "step": 9580
3364
+ },
3365
+ {
3366
+ "epoch": 0.7078080070780801,
3367
+ "grad_norm": 0.391527384519577,
3368
+ "learning_rate": 2.9226572292265726e-06,
3369
+ "loss": 2.3397,
3370
+ "step": 9600
3371
+ },
3372
+ {
3373
+ "epoch": 0.7092826070928261,
3374
+ "grad_norm": 0.4149307906627655,
3375
+ "learning_rate": 2.907911229079112e-06,
3376
+ "loss": 2.3408,
3377
+ "step": 9620
3378
+ },
3379
+ {
3380
+ "epoch": 0.7107572071075721,
3381
+ "grad_norm": 0.3845043480396271,
3382
+ "learning_rate": 2.8931652289316526e-06,
3383
+ "loss": 2.3498,
3384
+ "step": 9640
3385
+ },
3386
+ {
3387
+ "epoch": 0.712231807122318,
3388
+ "grad_norm": 0.355175256729126,
3389
+ "learning_rate": 2.878419228784193e-06,
3390
+ "loss": 2.3558,
3391
+ "step": 9660
3392
+ },
3393
+ {
3394
+ "epoch": 0.7137064071370641,
3395
+ "grad_norm": 0.37544727325439453,
3396
+ "learning_rate": 2.8636732286367325e-06,
3397
+ "loss": 2.3629,
3398
+ "step": 9680
3399
+ },
3400
+ {
3401
+ "epoch": 0.7151810071518101,
3402
+ "grad_norm": 0.394980251789093,
3403
+ "learning_rate": 2.8489272284892725e-06,
3404
+ "loss": 2.3456,
3405
+ "step": 9700
3406
+ },
3407
+ {
3408
+ "epoch": 0.7166556071665561,
3409
+ "grad_norm": 0.39948976039886475,
3410
+ "learning_rate": 2.8341812283418125e-06,
3411
+ "loss": 2.3542,
3412
+ "step": 9720
3413
+ },
3414
+ {
3415
+ "epoch": 0.718130207181302,
3416
+ "grad_norm": 0.3865370750427246,
3417
+ "learning_rate": 2.8194352281943525e-06,
3418
+ "loss": 2.3262,
3419
+ "step": 9740
3420
+ },
3421
+ {
3422
+ "epoch": 0.719604807196048,
3423
+ "grad_norm": 0.36873430013656616,
3424
+ "learning_rate": 2.8046892280468925e-06,
3425
+ "loss": 2.343,
3426
+ "step": 9760
3427
+ },
3428
+ {
3429
+ "epoch": 0.7210794072107941,
3430
+ "grad_norm": 0.37330904603004456,
3431
+ "learning_rate": 2.7899432278994325e-06,
3432
+ "loss": 2.3629,
3433
+ "step": 9780
3434
+ },
3435
+ {
3436
+ "epoch": 0.7225540072255401,
3437
+ "grad_norm": 0.392511785030365,
3438
+ "learning_rate": 2.7751972277519725e-06,
3439
+ "loss": 2.3364,
3440
+ "step": 9800
3441
+ },
3442
+ {
3443
+ "epoch": 0.7240286072402861,
3444
+ "grad_norm": 0.4280540943145752,
3445
+ "learning_rate": 2.7604512276045125e-06,
3446
+ "loss": 2.344,
3447
+ "step": 9820
3448
+ },
3449
+ {
3450
+ "epoch": 0.725503207255032,
3451
+ "grad_norm": 0.36759504675865173,
3452
+ "learning_rate": 2.7457052274570525e-06,
3453
+ "loss": 2.3639,
3454
+ "step": 9840
3455
+ },
3456
+ {
3457
+ "epoch": 0.7269778072697781,
3458
+ "grad_norm": 0.3967907428741455,
3459
+ "learning_rate": 2.7309592273095924e-06,
3460
+ "loss": 2.3459,
3461
+ "step": 9860
3462
+ },
3463
+ {
3464
+ "epoch": 0.7284524072845241,
3465
+ "grad_norm": 0.3818022906780243,
3466
+ "learning_rate": 2.716213227162133e-06,
3467
+ "loss": 2.3471,
3468
+ "step": 9880
3469
+ },
3470
+ {
3471
+ "epoch": 0.7299270072992701,
3472
+ "grad_norm": 0.3735957443714142,
3473
+ "learning_rate": 2.7014672270146724e-06,
3474
+ "loss": 2.3324,
3475
+ "step": 9900
3476
+ },
3477
+ {
3478
+ "epoch": 0.731401607314016,
3479
+ "grad_norm": 0.3850245773792267,
3480
+ "learning_rate": 2.686721226867213e-06,
3481
+ "loss": 2.3563,
3482
+ "step": 9920
3483
+ },
3484
+ {
3485
+ "epoch": 0.732876207328762,
3486
+ "grad_norm": 0.3557223081588745,
3487
+ "learning_rate": 2.6719752267197524e-06,
3488
+ "loss": 2.3416,
3489
+ "step": 9940
3490
+ },
3491
+ {
3492
+ "epoch": 0.7343508073435081,
3493
+ "grad_norm": 0.3680027723312378,
3494
+ "learning_rate": 2.657229226572293e-06,
3495
+ "loss": 2.3486,
3496
+ "step": 9960
3497
+ },
3498
+ {
3499
+ "epoch": 0.7358254073582541,
3500
+ "grad_norm": 0.39317336678504944,
3501
+ "learning_rate": 2.6424832264248324e-06,
3502
+ "loss": 2.3489,
3503
+ "step": 9980
3504
+ },
3505
+ {
3506
+ "epoch": 0.737300007373,
3507
+ "grad_norm": 0.44913551211357117,
3508
+ "learning_rate": 2.627737226277373e-06,
3509
+ "loss": 2.3476,
3510
+ "step": 10000
3511
+ }
3512
+ ],
3513
+ "logging_steps": 20,
3514
+ "max_steps": 13563,
3515
+ "num_input_tokens_seen": 0,
3516
+ "num_train_epochs": 1,
3517
+ "save_steps": 5000,
3518
+ "stateful_callbacks": {
3519
+ "TrainerControl": {
3520
+ "args": {
3521
+ "should_epoch_stop": false,
3522
+ "should_evaluate": false,
3523
+ "should_log": false,
3524
+ "should_save": true,
3525
+ "should_training_stop": false
3526
+ },
3527
+ "attributes": {}
3528
+ }
3529
+ },
3530
+ "total_flos": 4.276403373604864e+18,
3531
+ "train_batch_size": 24,
3532
+ "trial_name": null,
3533
+ "trial_params": null
3534
+ }
checkpoint-13563/adapter_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "data4elm/Llama-400M-12L",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 32,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.1,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 16,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "embed_tokens",
28
+ "gate_proj",
29
+ "down_proj",
30
+ "v_proj",
31
+ "k_proj",
32
+ "up_proj",
33
+ "q_proj",
34
+ "lm_head",
35
+ "o_proj"
36
+ ],
37
+ "task_type": "CAUSAL_LM",
38
+ "trainable_token_indices": null,
39
+ "use_dora": true,
40
+ "use_rslora": false
41
+ }
checkpoint-13563/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad8ca1b35f4a1ea23c04e2ec7703a98e2c472464a0abe7fe6711c8af1d70dc1b
3
+ size 177776400
checkpoint-13563/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-13563/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-13563/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
checkpoint-13563/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-3391/adapter_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "data4elm/Llama-400M-12L",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 32,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.1,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 16,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "o_proj",
28
+ "v_proj",
29
+ "down_proj",
30
+ "k_proj",
31
+ "lm_head",
32
+ "q_proj",
33
+ "gate_proj",
34
+ "embed_tokens",
35
+ "up_proj"
36
+ ],
37
+ "task_type": "CAUSAL_LM",
38
+ "trainable_token_indices": null,
39
+ "use_dora": true,
40
+ "use_rslora": false
41
+ }
checkpoint-3391/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ecb81619e041f55d01eba5f63a98408e7fbd3e95b4da446d2f232504023735e
3
+ size 177776400
checkpoint-3391/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-3391/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-3391/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
checkpoint-3391/trainer_state.json ADDED
@@ -0,0 +1,1217 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 500,
7
+ "global_step": 3391,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.005897965202005308,
14
+ "grad_norm": 0.7298315763473511,
15
+ "learning_rate": 9.94396933058095e-06,
16
+ "loss": 2.4477,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.011795930404010616,
21
+ "grad_norm": 0.43002140522003174,
22
+ "learning_rate": 9.884989678560897e-06,
23
+ "loss": 2.4211,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.017693895606015926,
28
+ "grad_norm": 0.47817477583885193,
29
+ "learning_rate": 9.826010026540843e-06,
30
+ "loss": 2.4279,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.023591860808021232,
35
+ "grad_norm": 0.44502127170562744,
36
+ "learning_rate": 9.76703037452079e-06,
37
+ "loss": 2.4101,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.029489826010026542,
42
+ "grad_norm": 0.4299854040145874,
43
+ "learning_rate": 9.708050722500738e-06,
44
+ "loss": 2.4168,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.03538779121203185,
49
+ "grad_norm": 0.38239866495132446,
50
+ "learning_rate": 9.649071070480684e-06,
51
+ "loss": 2.4079,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.04128575641403716,
56
+ "grad_norm": 0.344647079706192,
57
+ "learning_rate": 9.590091418460632e-06,
58
+ "loss": 2.4105,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.047183721616042465,
63
+ "grad_norm": 0.35823318362236023,
64
+ "learning_rate": 9.531111766440579e-06,
65
+ "loss": 2.4044,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.05308168681804777,
70
+ "grad_norm": 0.37203237414360046,
71
+ "learning_rate": 9.472132114420525e-06,
72
+ "loss": 2.4006,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.058979652020053085,
77
+ "grad_norm": 0.44176626205444336,
78
+ "learning_rate": 9.413152462400472e-06,
79
+ "loss": 2.4083,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.06487761722205838,
84
+ "grad_norm": 0.3652000427246094,
85
+ "learning_rate": 9.35417281038042e-06,
86
+ "loss": 2.3931,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.0707755824240637,
91
+ "grad_norm": 0.40255945920944214,
92
+ "learning_rate": 9.295193158360366e-06,
93
+ "loss": 2.3942,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.07667354762606901,
98
+ "grad_norm": 0.4122031033039093,
99
+ "learning_rate": 9.236213506340314e-06,
100
+ "loss": 2.3895,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.08257151282807432,
105
+ "grad_norm": 0.3695772588253021,
106
+ "learning_rate": 9.177233854320259e-06,
107
+ "loss": 2.3914,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.08846947803007962,
112
+ "grad_norm": 0.35018590092658997,
113
+ "learning_rate": 9.118254202300207e-06,
114
+ "loss": 2.392,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.09436744323208493,
119
+ "grad_norm": 0.3700025975704193,
120
+ "learning_rate": 9.059274550280154e-06,
121
+ "loss": 2.4027,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.10026540843409024,
126
+ "grad_norm": 0.34140926599502563,
127
+ "learning_rate": 9.000294898260102e-06,
128
+ "loss": 2.3926,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.10616337363609554,
133
+ "grad_norm": 0.36568552255630493,
134
+ "learning_rate": 8.941315246240048e-06,
135
+ "loss": 2.393,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.11206133883810085,
140
+ "grad_norm": 0.4640246331691742,
141
+ "learning_rate": 8.882335594219995e-06,
142
+ "loss": 2.3887,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.11795930404010617,
147
+ "grad_norm": 0.35777565836906433,
148
+ "learning_rate": 8.823355942199941e-06,
149
+ "loss": 2.3829,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.12385726924211148,
154
+ "grad_norm": 0.3578435182571411,
155
+ "learning_rate": 8.764376290179889e-06,
156
+ "loss": 2.3929,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.12975523444411677,
161
+ "grad_norm": 0.4518575668334961,
162
+ "learning_rate": 8.705396638159836e-06,
163
+ "loss": 2.3943,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.13565319964612207,
168
+ "grad_norm": 0.3837476074695587,
169
+ "learning_rate": 8.646416986139784e-06,
170
+ "loss": 2.3793,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.1415511648481274,
175
+ "grad_norm": 0.35441145300865173,
176
+ "learning_rate": 8.58743733411973e-06,
177
+ "loss": 2.3861,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.14744913005013272,
182
+ "grad_norm": 0.34809452295303345,
183
+ "learning_rate": 8.528457682099676e-06,
184
+ "loss": 2.3859,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.15334709525213802,
189
+ "grad_norm": 0.32996416091918945,
190
+ "learning_rate": 8.469478030079623e-06,
191
+ "loss": 2.3788,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.15924506045414333,
196
+ "grad_norm": 0.37359902262687683,
197
+ "learning_rate": 8.410498378059571e-06,
198
+ "loss": 2.3958,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.16514302565614863,
203
+ "grad_norm": 0.33828896284103394,
204
+ "learning_rate": 8.351518726039517e-06,
205
+ "loss": 2.3684,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.17104099085815394,
210
+ "grad_norm": 0.3458431661128998,
211
+ "learning_rate": 8.292539074019464e-06,
212
+ "loss": 2.4049,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.17693895606015925,
217
+ "grad_norm": 0.5570428967475891,
218
+ "learning_rate": 8.23355942199941e-06,
219
+ "loss": 2.3813,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.18283692126216455,
224
+ "grad_norm": 0.3951890468597412,
225
+ "learning_rate": 8.174579769979358e-06,
226
+ "loss": 2.3821,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.18873488646416986,
231
+ "grad_norm": 0.3401094675064087,
232
+ "learning_rate": 8.115600117959305e-06,
233
+ "loss": 2.3746,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.19463285166617517,
238
+ "grad_norm": 0.3608352541923523,
239
+ "learning_rate": 8.056620465939251e-06,
240
+ "loss": 2.3784,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.20053081686818047,
245
+ "grad_norm": 0.3732820749282837,
246
+ "learning_rate": 7.9976408139192e-06,
247
+ "loss": 2.3792,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.20642878207018578,
252
+ "grad_norm": 0.62112957239151,
253
+ "learning_rate": 7.938661161899146e-06,
254
+ "loss": 2.3811,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.21232674727219109,
259
+ "grad_norm": 0.3565405607223511,
260
+ "learning_rate": 7.879681509879092e-06,
261
+ "loss": 2.3883,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.2182247124741964,
266
+ "grad_norm": 0.40740129351615906,
267
+ "learning_rate": 7.820701857859039e-06,
268
+ "loss": 2.3884,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.2241226776762017,
273
+ "grad_norm": 0.3595851957798004,
274
+ "learning_rate": 7.761722205838987e-06,
275
+ "loss": 2.3787,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.23002064287820703,
280
+ "grad_norm": 0.5515618324279785,
281
+ "learning_rate": 7.702742553818933e-06,
282
+ "loss": 2.3692,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.23591860808021234,
287
+ "grad_norm": 0.43797263503074646,
288
+ "learning_rate": 7.64376290179888e-06,
289
+ "loss": 2.3721,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.24181657328221765,
294
+ "grad_norm": 0.3455728590488434,
295
+ "learning_rate": 7.584783249778827e-06,
296
+ "loss": 2.3821,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.24771453848422295,
301
+ "grad_norm": 0.38383591175079346,
302
+ "learning_rate": 7.525803597758774e-06,
303
+ "loss": 2.3785,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.25361250368622823,
308
+ "grad_norm": 0.3334074318408966,
309
+ "learning_rate": 7.4668239457387205e-06,
310
+ "loss": 2.3803,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.25951046888823354,
315
+ "grad_norm": 0.32531243562698364,
316
+ "learning_rate": 7.407844293718668e-06,
317
+ "loss": 2.3726,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.26540843409023884,
322
+ "grad_norm": 0.37065383791923523,
323
+ "learning_rate": 7.348864641698614e-06,
324
+ "loss": 2.3771,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.27130639929224415,
329
+ "grad_norm": 0.9803745746612549,
330
+ "learning_rate": 7.289884989678561e-06,
331
+ "loss": 2.3925,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.2772043644942495,
336
+ "grad_norm": 0.379090815782547,
337
+ "learning_rate": 7.230905337658508e-06,
338
+ "loss": 2.3786,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.2831023296962548,
343
+ "grad_norm": 0.353801429271698,
344
+ "learning_rate": 7.171925685638456e-06,
345
+ "loss": 2.3911,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.2890002948982601,
350
+ "grad_norm": 0.38585972785949707,
351
+ "learning_rate": 7.112946033618402e-06,
352
+ "loss": 2.3785,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.29489826010026543,
357
+ "grad_norm": 0.3725855052471161,
358
+ "learning_rate": 7.05396638159835e-06,
359
+ "loss": 2.3647,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.30079622530227074,
364
+ "grad_norm": 0.3538314402103424,
365
+ "learning_rate": 6.994986729578296e-06,
366
+ "loss": 2.3734,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.30669419050427604,
371
+ "grad_norm": 0.3888925611972809,
372
+ "learning_rate": 6.936007077558243e-06,
373
+ "loss": 2.3713,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.31259215570628135,
378
+ "grad_norm": 0.38058924674987793,
379
+ "learning_rate": 6.87702742553819e-06,
380
+ "loss": 2.3845,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.31849012090828666,
385
+ "grad_norm": 0.41260024905204773,
386
+ "learning_rate": 6.818047773518137e-06,
387
+ "loss": 2.3688,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.32438808611029196,
392
+ "grad_norm": 0.37049782276153564,
393
+ "learning_rate": 6.759068121498083e-06,
394
+ "loss": 2.3776,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.33028605131229727,
399
+ "grad_norm": 0.3474882245063782,
400
+ "learning_rate": 6.700088469478031e-06,
401
+ "loss": 2.3698,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.3361840165143026,
406
+ "grad_norm": 0.387926310300827,
407
+ "learning_rate": 6.641108817457977e-06,
408
+ "loss": 2.3648,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.3420819817163079,
413
+ "grad_norm": 0.3975593149662018,
414
+ "learning_rate": 6.582129165437925e-06,
415
+ "loss": 2.3798,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.3479799469183132,
420
+ "grad_norm": 0.4951271414756775,
421
+ "learning_rate": 6.523149513417872e-06,
422
+ "loss": 2.3694,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.3538779121203185,
427
+ "grad_norm": 0.3920055329799652,
428
+ "learning_rate": 6.464169861397819e-06,
429
+ "loss": 2.3873,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.3597758773223238,
434
+ "grad_norm": 0.3636319637298584,
435
+ "learning_rate": 6.405190209377765e-06,
436
+ "loss": 2.3603,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.3656738425243291,
441
+ "grad_norm": 0.5144618153572083,
442
+ "learning_rate": 6.346210557357713e-06,
443
+ "loss": 2.3774,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.3715718077263344,
448
+ "grad_norm": 0.3716528117656708,
449
+ "learning_rate": 6.287230905337659e-06,
450
+ "loss": 2.3714,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.3774697729283397,
455
+ "grad_norm": 0.32973116636276245,
456
+ "learning_rate": 6.228251253317605e-06,
457
+ "loss": 2.3823,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.383367738130345,
462
+ "grad_norm": 0.37797972559928894,
463
+ "learning_rate": 6.169271601297553e-06,
464
+ "loss": 2.3627,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.38926570333235033,
469
+ "grad_norm": 0.33120372891426086,
470
+ "learning_rate": 6.110291949277499e-06,
471
+ "loss": 2.3788,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.39516366853435564,
476
+ "grad_norm": 0.4278649687767029,
477
+ "learning_rate": 6.051312297257446e-06,
478
+ "loss": 2.3751,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.40106163373636095,
483
+ "grad_norm": 0.4075045585632324,
484
+ "learning_rate": 5.992332645237393e-06,
485
+ "loss": 2.3658,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.40695959893836625,
490
+ "grad_norm": 0.3727574348449707,
491
+ "learning_rate": 5.933352993217341e-06,
492
+ "loss": 2.383,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.41285756414037156,
497
+ "grad_norm": 0.36080676317214966,
498
+ "learning_rate": 5.874373341197287e-06,
499
+ "loss": 2.3775,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.41875552934237686,
504
+ "grad_norm": 0.36724549531936646,
505
+ "learning_rate": 5.8153936891772346e-06,
506
+ "loss": 2.3946,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.42465349454438217,
511
+ "grad_norm": 0.35853520035743713,
512
+ "learning_rate": 5.756414037157181e-06,
513
+ "loss": 2.3764,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.4305514597463875,
518
+ "grad_norm": 0.39550721645355225,
519
+ "learning_rate": 5.697434385137128e-06,
520
+ "loss": 2.3658,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.4364494249483928,
525
+ "grad_norm": 0.37395161390304565,
526
+ "learning_rate": 5.638454733117075e-06,
527
+ "loss": 2.3731,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.4423473901503981,
532
+ "grad_norm": 0.4243367612361908,
533
+ "learning_rate": 5.579475081097022e-06,
534
+ "loss": 2.3747,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.4482453553524034,
539
+ "grad_norm": 0.4698735177516937,
540
+ "learning_rate": 5.520495429076968e-06,
541
+ "loss": 2.3779,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.4541433205544087,
546
+ "grad_norm": 0.35367387533187866,
547
+ "learning_rate": 5.461515777056916e-06,
548
+ "loss": 2.3777,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.46004128575641406,
553
+ "grad_norm": 0.38224881887435913,
554
+ "learning_rate": 5.402536125036862e-06,
555
+ "loss": 2.3611,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.46593925095841937,
560
+ "grad_norm": 0.3544207513332367,
561
+ "learning_rate": 5.34355647301681e-06,
562
+ "loss": 2.3713,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.4718372161604247,
567
+ "grad_norm": 0.45923078060150146,
568
+ "learning_rate": 5.2845768209967566e-06,
569
+ "loss": 2.3783,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.47773518136243,
574
+ "grad_norm": 0.3729366958141327,
575
+ "learning_rate": 5.225597168976704e-06,
576
+ "loss": 2.3754,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.4836331465644353,
581
+ "grad_norm": 0.4686289131641388,
582
+ "learning_rate": 5.16661751695665e-06,
583
+ "loss": 2.3642,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.4895311117664406,
588
+ "grad_norm": 0.41152337193489075,
589
+ "learning_rate": 5.1076378649365975e-06,
590
+ "loss": 2.3731,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.4954290769684459,
595
+ "grad_norm": 0.37498942017555237,
596
+ "learning_rate": 5.048658212916544e-06,
597
+ "loss": 2.3553,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.5013270421704512,
602
+ "grad_norm": 0.35418274998664856,
603
+ "learning_rate": 4.989678560896491e-06,
604
+ "loss": 2.373,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.5072250073724565,
609
+ "grad_norm": 0.385016530752182,
610
+ "learning_rate": 4.930698908876438e-06,
611
+ "loss": 2.3599,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.5131229725744618,
616
+ "grad_norm": 0.4020742177963257,
617
+ "learning_rate": 4.871719256856385e-06,
618
+ "loss": 2.3699,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.5190209377764671,
623
+ "grad_norm": 0.3605809807777405,
624
+ "learning_rate": 4.812739604836332e-06,
625
+ "loss": 2.38,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.5249189029784724,
630
+ "grad_norm": 0.45836788415908813,
631
+ "learning_rate": 4.7537599528162786e-06,
632
+ "loss": 2.3676,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.5308168681804777,
637
+ "grad_norm": 0.3600791394710541,
638
+ "learning_rate": 4.694780300796226e-06,
639
+ "loss": 2.3504,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.536714833382483,
644
+ "grad_norm": 0.34572452306747437,
645
+ "learning_rate": 4.635800648776172e-06,
646
+ "loss": 2.3756,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.5426127985844883,
651
+ "grad_norm": 0.3912692368030548,
652
+ "learning_rate": 4.5768209967561195e-06,
653
+ "loss": 2.3622,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.5485107637864937,
658
+ "grad_norm": 0.3616994023323059,
659
+ "learning_rate": 4.517841344736067e-06,
660
+ "loss": 2.3628,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.554408728988499,
665
+ "grad_norm": 0.3876311480998993,
666
+ "learning_rate": 4.458861692716013e-06,
667
+ "loss": 2.3576,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.5603066941905043,
672
+ "grad_norm": 0.41625073552131653,
673
+ "learning_rate": 4.3998820406959605e-06,
674
+ "loss": 2.3602,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.5662046593925096,
679
+ "grad_norm": 0.472136527299881,
680
+ "learning_rate": 4.340902388675907e-06,
681
+ "loss": 2.3546,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.5721026245945149,
686
+ "grad_norm": 0.3669559955596924,
687
+ "learning_rate": 4.281922736655854e-06,
688
+ "loss": 2.38,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.5780005897965202,
693
+ "grad_norm": 0.398993581533432,
694
+ "learning_rate": 4.222943084635801e-06,
695
+ "loss": 2.3722,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.5838985549985255,
700
+ "grad_norm": 0.38298431038856506,
701
+ "learning_rate": 4.163963432615748e-06,
702
+ "loss": 2.3752,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.5897965202005309,
707
+ "grad_norm": 0.3662157654762268,
708
+ "learning_rate": 4.104983780595695e-06,
709
+ "loss": 2.3558,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.5956944854025361,
714
+ "grad_norm": 0.3607335388660431,
715
+ "learning_rate": 4.0460041285756415e-06,
716
+ "loss": 2.3659,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.6015924506045415,
721
+ "grad_norm": 0.37250858545303345,
722
+ "learning_rate": 3.987024476555589e-06,
723
+ "loss": 2.3685,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.6074904158065467,
728
+ "grad_norm": 0.3815813660621643,
729
+ "learning_rate": 3.928044824535536e-06,
730
+ "loss": 2.3624,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.6133883810085521,
735
+ "grad_norm": 0.3587498962879181,
736
+ "learning_rate": 3.8690651725154825e-06,
737
+ "loss": 2.372,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.6192863462105573,
742
+ "grad_norm": 0.4261331558227539,
743
+ "learning_rate": 3.8100855204954297e-06,
744
+ "loss": 2.3695,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.6251843114125627,
749
+ "grad_norm": 0.35808065533638,
750
+ "learning_rate": 3.7511058684753766e-06,
751
+ "loss": 2.3663,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.631082276614568,
756
+ "grad_norm": 0.6072697639465332,
757
+ "learning_rate": 3.6921262164553234e-06,
758
+ "loss": 2.3722,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.6369802418165733,
763
+ "grad_norm": 0.5162146687507629,
764
+ "learning_rate": 3.6331465644352702e-06,
765
+ "loss": 2.3666,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.6428782070185786,
770
+ "grad_norm": 0.3236039876937866,
771
+ "learning_rate": 3.574166912415217e-06,
772
+ "loss": 2.3652,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.6487761722205839,
777
+ "grad_norm": 0.3428741991519928,
778
+ "learning_rate": 3.5151872603951644e-06,
779
+ "loss": 2.3563,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.6546741374225892,
784
+ "grad_norm": 0.38159841299057007,
785
+ "learning_rate": 3.4562076083751108e-06,
786
+ "loss": 2.3736,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.6605721026245945,
791
+ "grad_norm": 0.3707337975502014,
792
+ "learning_rate": 3.3972279563550576e-06,
793
+ "loss": 2.3674,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.6664700678265998,
798
+ "grad_norm": 0.4276551902294159,
799
+ "learning_rate": 3.3382483043350045e-06,
800
+ "loss": 2.3531,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.6723680330286051,
805
+ "grad_norm": 0.35105422139167786,
806
+ "learning_rate": 3.2792686523149513e-06,
807
+ "loss": 2.3679,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.6782659982306104,
812
+ "grad_norm": 0.3761979341506958,
813
+ "learning_rate": 3.220289000294898e-06,
814
+ "loss": 2.362,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.6841639634326158,
819
+ "grad_norm": 0.3713517487049103,
820
+ "learning_rate": 3.1613093482748454e-06,
821
+ "loss": 2.3618,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.690061928634621,
826
+ "grad_norm": 0.3936586380004883,
827
+ "learning_rate": 3.1023296962547922e-06,
828
+ "loss": 2.3708,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.6959598938366264,
833
+ "grad_norm": 0.4322130084037781,
834
+ "learning_rate": 3.043350044234739e-06,
835
+ "loss": 2.3754,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.7018578590386316,
840
+ "grad_norm": 0.35027167201042175,
841
+ "learning_rate": 2.984370392214686e-06,
842
+ "loss": 2.3699,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.707755824240637,
847
+ "grad_norm": 0.40044116973876953,
848
+ "learning_rate": 2.925390740194633e-06,
849
+ "loss": 2.3596,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.7136537894426422,
854
+ "grad_norm": 0.3692649006843567,
855
+ "learning_rate": 2.86641108817458e-06,
856
+ "loss": 2.3618,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.7195517546446476,
861
+ "grad_norm": 0.3762916624546051,
862
+ "learning_rate": 2.807431436154527e-06,
863
+ "loss": 2.3582,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.7254497198466529,
868
+ "grad_norm": 0.3630368113517761,
869
+ "learning_rate": 2.7484517841344737e-06,
870
+ "loss": 2.3655,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.7313476850486582,
875
+ "grad_norm": 0.37902596592903137,
876
+ "learning_rate": 2.6894721321144206e-06,
877
+ "loss": 2.3612,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.7372456502506636,
882
+ "grad_norm": 0.3508872985839844,
883
+ "learning_rate": 2.630492480094368e-06,
884
+ "loss": 2.3764,
885
+ "step": 2500
886
+ },
887
+ {
888
+ "epoch": 0.7431436154526688,
889
+ "grad_norm": 0.3574802875518799,
890
+ "learning_rate": 2.5715128280743147e-06,
891
+ "loss": 2.3673,
892
+ "step": 2520
893
+ },
894
+ {
895
+ "epoch": 0.7490415806546742,
896
+ "grad_norm": 0.41408953070640564,
897
+ "learning_rate": 2.5125331760542615e-06,
898
+ "loss": 2.3713,
899
+ "step": 2540
900
+ },
901
+ {
902
+ "epoch": 0.7549395458566794,
903
+ "grad_norm": 0.3340831995010376,
904
+ "learning_rate": 2.4535535240342083e-06,
905
+ "loss": 2.3694,
906
+ "step": 2560
907
+ },
908
+ {
909
+ "epoch": 0.7608375110586848,
910
+ "grad_norm": 0.37367573380470276,
911
+ "learning_rate": 2.394573872014155e-06,
912
+ "loss": 2.3677,
913
+ "step": 2580
914
+ },
915
+ {
916
+ "epoch": 0.76673547626069,
917
+ "grad_norm": 0.3758946657180786,
918
+ "learning_rate": 2.3355942199941025e-06,
919
+ "loss": 2.3728,
920
+ "step": 2600
921
+ },
922
+ {
923
+ "epoch": 0.7726334414626954,
924
+ "grad_norm": 0.39376673102378845,
925
+ "learning_rate": 2.2766145679740493e-06,
926
+ "loss": 2.3533,
927
+ "step": 2620
928
+ },
929
+ {
930
+ "epoch": 0.7785314066647007,
931
+ "grad_norm": 0.35029420256614685,
932
+ "learning_rate": 2.217634915953996e-06,
933
+ "loss": 2.3654,
934
+ "step": 2640
935
+ },
936
+ {
937
+ "epoch": 0.784429371866706,
938
+ "grad_norm": 0.3405188322067261,
939
+ "learning_rate": 2.158655263933943e-06,
940
+ "loss": 2.3678,
941
+ "step": 2660
942
+ },
943
+ {
944
+ "epoch": 0.7903273370687113,
945
+ "grad_norm": 0.34242868423461914,
946
+ "learning_rate": 2.09967561191389e-06,
947
+ "loss": 2.3552,
948
+ "step": 2680
949
+ },
950
+ {
951
+ "epoch": 0.7962253022707166,
952
+ "grad_norm": 0.3884572684764862,
953
+ "learning_rate": 2.0406959598938367e-06,
954
+ "loss": 2.3419,
955
+ "step": 2700
956
+ },
957
+ {
958
+ "epoch": 0.8021232674727219,
959
+ "grad_norm": 0.3567847013473511,
960
+ "learning_rate": 1.9817163078737835e-06,
961
+ "loss": 2.3751,
962
+ "step": 2720
963
+ },
964
+ {
965
+ "epoch": 0.8080212326747273,
966
+ "grad_norm": 0.37844085693359375,
967
+ "learning_rate": 1.9227366558537303e-06,
968
+ "loss": 2.3659,
969
+ "step": 2740
970
+ },
971
+ {
972
+ "epoch": 0.8139191978767325,
973
+ "grad_norm": 0.34508031606674194,
974
+ "learning_rate": 1.8637570038336774e-06,
975
+ "loss": 2.3617,
976
+ "step": 2760
977
+ },
978
+ {
979
+ "epoch": 0.8198171630787379,
980
+ "grad_norm": 0.37041574716567993,
981
+ "learning_rate": 1.8047773518136245e-06,
982
+ "loss": 2.3676,
983
+ "step": 2780
984
+ },
985
+ {
986
+ "epoch": 0.8257151282807431,
987
+ "grad_norm": 0.36701688170433044,
988
+ "learning_rate": 1.7457976997935713e-06,
989
+ "loss": 2.3542,
990
+ "step": 2800
991
+ },
992
+ {
993
+ "epoch": 0.8316130934827485,
994
+ "grad_norm": 0.3514913022518158,
995
+ "learning_rate": 1.6868180477735183e-06,
996
+ "loss": 2.3691,
997
+ "step": 2820
998
+ },
999
+ {
1000
+ "epoch": 0.8375110586847537,
1001
+ "grad_norm": 0.3825013339519501,
1002
+ "learning_rate": 1.6278383957534652e-06,
1003
+ "loss": 2.3606,
1004
+ "step": 2840
1005
+ },
1006
+ {
1007
+ "epoch": 0.8434090238867591,
1008
+ "grad_norm": 0.3623409867286682,
1009
+ "learning_rate": 1.568858743733412e-06,
1010
+ "loss": 2.3632,
1011
+ "step": 2860
1012
+ },
1013
+ {
1014
+ "epoch": 0.8493069890887643,
1015
+ "grad_norm": 0.35610345005989075,
1016
+ "learning_rate": 1.509879091713359e-06,
1017
+ "loss": 2.3578,
1018
+ "step": 2880
1019
+ },
1020
+ {
1021
+ "epoch": 0.8552049542907697,
1022
+ "grad_norm": 0.3956283628940582,
1023
+ "learning_rate": 1.450899439693306e-06,
1024
+ "loss": 2.3677,
1025
+ "step": 2900
1026
+ },
1027
+ {
1028
+ "epoch": 0.861102919492775,
1029
+ "grad_norm": 0.38382914662361145,
1030
+ "learning_rate": 1.391919787673253e-06,
1031
+ "loss": 2.3776,
1032
+ "step": 2920
1033
+ },
1034
+ {
1035
+ "epoch": 0.8670008846947803,
1036
+ "grad_norm": 0.3666318356990814,
1037
+ "learning_rate": 1.3329401356531998e-06,
1038
+ "loss": 2.3582,
1039
+ "step": 2940
1040
+ },
1041
+ {
1042
+ "epoch": 0.8728988498967856,
1043
+ "grad_norm": 0.4062568247318268,
1044
+ "learning_rate": 1.2739604836331469e-06,
1045
+ "loss": 2.379,
1046
+ "step": 2960
1047
+ },
1048
+ {
1049
+ "epoch": 0.8787968150987909,
1050
+ "grad_norm": 0.34597164392471313,
1051
+ "learning_rate": 1.2149808316130935e-06,
1052
+ "loss": 2.3638,
1053
+ "step": 2980
1054
+ },
1055
+ {
1056
+ "epoch": 0.8846947803007962,
1057
+ "grad_norm": 0.406476229429245,
1058
+ "learning_rate": 1.1560011795930406e-06,
1059
+ "loss": 2.3661,
1060
+ "step": 3000
1061
+ },
1062
+ {
1063
+ "epoch": 0.8905927455028015,
1064
+ "grad_norm": 0.4490988552570343,
1065
+ "learning_rate": 1.0970215275729874e-06,
1066
+ "loss": 2.3681,
1067
+ "step": 3020
1068
+ },
1069
+ {
1070
+ "epoch": 0.8964907107048068,
1071
+ "grad_norm": 0.34622323513031006,
1072
+ "learning_rate": 1.0380418755529344e-06,
1073
+ "loss": 2.3594,
1074
+ "step": 3040
1075
+ },
1076
+ {
1077
+ "epoch": 0.9023886759068122,
1078
+ "grad_norm": 0.4984548091888428,
1079
+ "learning_rate": 9.790622235328813e-07,
1080
+ "loss": 2.3708,
1081
+ "step": 3060
1082
+ },
1083
+ {
1084
+ "epoch": 0.9082866411088174,
1085
+ "grad_norm": 0.37141624093055725,
1086
+ "learning_rate": 9.200825715128281e-07,
1087
+ "loss": 2.3677,
1088
+ "step": 3080
1089
+ },
1090
+ {
1091
+ "epoch": 0.9141846063108228,
1092
+ "grad_norm": 0.3768744170665741,
1093
+ "learning_rate": 8.61102919492775e-07,
1094
+ "loss": 2.3686,
1095
+ "step": 3100
1096
+ },
1097
+ {
1098
+ "epoch": 0.9200825715128281,
1099
+ "grad_norm": 0.39627739787101746,
1100
+ "learning_rate": 8.021232674727219e-07,
1101
+ "loss": 2.3544,
1102
+ "step": 3120
1103
+ },
1104
+ {
1105
+ "epoch": 0.9259805367148334,
1106
+ "grad_norm": 0.3602909743785858,
1107
+ "learning_rate": 7.431436154526689e-07,
1108
+ "loss": 2.3795,
1109
+ "step": 3140
1110
+ },
1111
+ {
1112
+ "epoch": 0.9318785019168387,
1113
+ "grad_norm": 0.3863148093223572,
1114
+ "learning_rate": 6.841639634326158e-07,
1115
+ "loss": 2.359,
1116
+ "step": 3160
1117
+ },
1118
+ {
1119
+ "epoch": 0.937776467118844,
1120
+ "grad_norm": 0.3720771372318268,
1121
+ "learning_rate": 6.251843114125628e-07,
1122
+ "loss": 2.3644,
1123
+ "step": 3180
1124
+ },
1125
+ {
1126
+ "epoch": 0.9436744323208494,
1127
+ "grad_norm": 0.3867350220680237,
1128
+ "learning_rate": 5.662046593925096e-07,
1129
+ "loss": 2.3505,
1130
+ "step": 3200
1131
+ },
1132
+ {
1133
+ "epoch": 0.9495723975228546,
1134
+ "grad_norm": 0.431013822555542,
1135
+ "learning_rate": 5.072250073724565e-07,
1136
+ "loss": 2.3595,
1137
+ "step": 3220
1138
+ },
1139
+ {
1140
+ "epoch": 0.95547036272486,
1141
+ "grad_norm": 0.3696461617946625,
1142
+ "learning_rate": 4.482453553524035e-07,
1143
+ "loss": 2.354,
1144
+ "step": 3240
1145
+ },
1146
+ {
1147
+ "epoch": 0.9613683279268652,
1148
+ "grad_norm": 0.42161649465560913,
1149
+ "learning_rate": 3.8926570333235034e-07,
1150
+ "loss": 2.3627,
1151
+ "step": 3260
1152
+ },
1153
+ {
1154
+ "epoch": 0.9672662931288706,
1155
+ "grad_norm": 0.4679885506629944,
1156
+ "learning_rate": 3.302860513122973e-07,
1157
+ "loss": 2.3592,
1158
+ "step": 3280
1159
+ },
1160
+ {
1161
+ "epoch": 0.9731642583308758,
1162
+ "grad_norm": 0.37652766704559326,
1163
+ "learning_rate": 2.713063992922442e-07,
1164
+ "loss": 2.3628,
1165
+ "step": 3300
1166
+ },
1167
+ {
1168
+ "epoch": 0.9790622235328812,
1169
+ "grad_norm": 0.37805137038230896,
1170
+ "learning_rate": 2.123267472721911e-07,
1171
+ "loss": 2.3688,
1172
+ "step": 3320
1173
+ },
1174
+ {
1175
+ "epoch": 0.9849601887348864,
1176
+ "grad_norm": 0.40008819103240967,
1177
+ "learning_rate": 1.5334709525213802e-07,
1178
+ "loss": 2.3729,
1179
+ "step": 3340
1180
+ },
1181
+ {
1182
+ "epoch": 0.9908581539368918,
1183
+ "grad_norm": 0.3313311040401459,
1184
+ "learning_rate": 9.436744323208494e-08,
1185
+ "loss": 2.3651,
1186
+ "step": 3360
1187
+ },
1188
+ {
1189
+ "epoch": 0.9967561191388971,
1190
+ "grad_norm": 0.4864294230937958,
1191
+ "learning_rate": 3.538779121203185e-08,
1192
+ "loss": 2.3677,
1193
+ "step": 3380
1194
+ }
1195
+ ],
1196
+ "logging_steps": 20,
1197
+ "max_steps": 3391,
1198
+ "num_input_tokens_seen": 0,
1199
+ "num_train_epochs": 1,
1200
+ "save_steps": 5000,
1201
+ "stateful_callbacks": {
1202
+ "TrainerControl": {
1203
+ "args": {
1204
+ "should_epoch_stop": false,
1205
+ "should_evaluate": false,
1206
+ "should_log": false,
1207
+ "should_save": true,
1208
+ "should_training_stop": true
1209
+ },
1210
+ "attributes": {}
1211
+ }
1212
+ },
1213
+ "total_flos": 1.4501283838820352e+18,
1214
+ "train_batch_size": 24,
1215
+ "trial_name": null,
1216
+ "trial_params": null
1217
+ }
checkpoint-5000/adapter_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "data4elm/Llama-400M-12L",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 32,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.1,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 16,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "embed_tokens",
28
+ "gate_proj",
29
+ "down_proj",
30
+ "v_proj",
31
+ "k_proj",
32
+ "up_proj",
33
+ "q_proj",
34
+ "lm_head",
35
+ "o_proj"
36
+ ],
37
+ "task_type": "CAUSAL_LM",
38
+ "trainable_token_indices": null,
39
+ "use_dora": true,
40
+ "use_rslora": false
41
+ }
checkpoint-5000/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c290928573a582944c1933574fc2e407d7b2d044dc709b104e2ee533edf13d48
3
+ size 177776400
checkpoint-5000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-5000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-5000/tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
checkpoint-5000/trainer_state.json ADDED
@@ -0,0 +1,1784 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.3686500036865,
6
+ "eval_steps": 500,
7
+ "global_step": 5000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0014746000147460002,
14
+ "grad_norm": 0.4404066503047943,
15
+ "learning_rate": 9.985991299859913e-06,
16
+ "loss": 2.4466,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.0029492000294920003,
21
+ "grad_norm": 0.5653194785118103,
22
+ "learning_rate": 9.971245299712454e-06,
23
+ "loss": 2.4307,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004423800044238001,
28
+ "grad_norm": 0.4618721902370453,
29
+ "learning_rate": 9.956499299564993e-06,
30
+ "loss": 2.4367,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005898400058984001,
35
+ "grad_norm": 0.4391142427921295,
36
+ "learning_rate": 9.941753299417534e-06,
37
+ "loss": 2.4183,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.007373000073730001,
42
+ "grad_norm": 0.474587619304657,
43
+ "learning_rate": 9.927007299270073e-06,
44
+ "loss": 2.42,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008847600088476001,
49
+ "grad_norm": 0.420256644487381,
50
+ "learning_rate": 9.912261299122614e-06,
51
+ "loss": 2.4164,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.010322200103222,
56
+ "grad_norm": 0.40069517493247986,
57
+ "learning_rate": 9.897515298975153e-06,
58
+ "loss": 2.3962,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.011796800117968001,
63
+ "grad_norm": 0.36043497920036316,
64
+ "learning_rate": 9.882769298827694e-06,
65
+ "loss": 2.4026,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.013271400132714002,
70
+ "grad_norm": 0.354769229888916,
71
+ "learning_rate": 9.868023298680233e-06,
72
+ "loss": 2.4013,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.014746000147460001,
77
+ "grad_norm": 0.40864863991737366,
78
+ "learning_rate": 9.853277298532774e-06,
79
+ "loss": 2.4077,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.016220600162206,
84
+ "grad_norm": 0.37636885046958923,
85
+ "learning_rate": 9.838531298385315e-06,
86
+ "loss": 2.4058,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.017695200176952003,
91
+ "grad_norm": 0.37770289182662964,
92
+ "learning_rate": 9.823785298237854e-06,
93
+ "loss": 2.3789,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.019169800191698002,
98
+ "grad_norm": 0.4997946321964264,
99
+ "learning_rate": 9.809039298090393e-06,
100
+ "loss": 2.3921,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.020644400206444,
105
+ "grad_norm": 0.42146745324134827,
106
+ "learning_rate": 9.794293297942934e-06,
107
+ "loss": 2.389,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.022119000221190004,
112
+ "grad_norm": 0.31817182898521423,
113
+ "learning_rate": 9.779547297795475e-06,
114
+ "loss": 2.3969,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.023593600235936003,
119
+ "grad_norm": 0.41462913155555725,
120
+ "learning_rate": 9.764801297648014e-06,
121
+ "loss": 2.3903,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.025068200250682002,
126
+ "grad_norm": 0.35277777910232544,
127
+ "learning_rate": 9.750055297500553e-06,
128
+ "loss": 2.3886,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.026542800265428004,
133
+ "grad_norm": 0.39531123638153076,
134
+ "learning_rate": 9.735309297353094e-06,
135
+ "loss": 2.3847,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.028017400280174003,
140
+ "grad_norm": 0.3687122166156769,
141
+ "learning_rate": 9.720563297205633e-06,
142
+ "loss": 2.3824,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.029492000294920002,
147
+ "grad_norm": 0.5362212061882019,
148
+ "learning_rate": 9.705817297058172e-06,
149
+ "loss": 2.3861,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.030966600309666,
154
+ "grad_norm": 0.5793041586875916,
155
+ "learning_rate": 9.691071296910713e-06,
156
+ "loss": 2.4069,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.032441200324412,
161
+ "grad_norm": 0.3618432581424713,
162
+ "learning_rate": 9.676325296763254e-06,
163
+ "loss": 2.379,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.033915800339158,
168
+ "grad_norm": 0.39441928267478943,
169
+ "learning_rate": 9.661579296615793e-06,
170
+ "loss": 2.3878,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.035390400353904006,
175
+ "grad_norm": 0.3998846113681793,
176
+ "learning_rate": 9.646833296468334e-06,
177
+ "loss": 2.4028,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03686500036865,
182
+ "grad_norm": 0.354809045791626,
183
+ "learning_rate": 9.632087296320873e-06,
184
+ "loss": 2.384,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.038339600383396004,
189
+ "grad_norm": 0.6007148623466492,
190
+ "learning_rate": 9.617341296173414e-06,
191
+ "loss": 2.3974,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.039814200398142006,
196
+ "grad_norm": 0.37758493423461914,
197
+ "learning_rate": 9.602595296025953e-06,
198
+ "loss": 2.3762,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.041288800412888,
203
+ "grad_norm": 0.4141014516353607,
204
+ "learning_rate": 9.587849295878494e-06,
205
+ "loss": 2.3831,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.042763400427634005,
210
+ "grad_norm": 0.48960772156715393,
211
+ "learning_rate": 9.573103295731033e-06,
212
+ "loss": 2.3961,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04423800044238001,
217
+ "grad_norm": 0.3811470568180084,
218
+ "learning_rate": 9.558357295583574e-06,
219
+ "loss": 2.3751,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.045712600457126,
224
+ "grad_norm": 0.35424861311912537,
225
+ "learning_rate": 9.543611295436113e-06,
226
+ "loss": 2.3725,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.047187200471872005,
231
+ "grad_norm": 0.36190420389175415,
232
+ "learning_rate": 9.528865295288654e-06,
233
+ "loss": 2.3944,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.04866180048661801,
238
+ "grad_norm": 0.42110538482666016,
239
+ "learning_rate": 9.514119295141195e-06,
240
+ "loss": 2.3909,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.050136400501364004,
245
+ "grad_norm": 0.4361230134963989,
246
+ "learning_rate": 9.499373294993734e-06,
247
+ "loss": 2.3752,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.051611000516110006,
252
+ "grad_norm": 0.3302260935306549,
253
+ "learning_rate": 9.484627294846273e-06,
254
+ "loss": 2.3699,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.05308560053085601,
259
+ "grad_norm": 0.3521312177181244,
260
+ "learning_rate": 9.469881294698814e-06,
261
+ "loss": 2.3772,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.054560200545602004,
266
+ "grad_norm": 0.3700699806213379,
267
+ "learning_rate": 9.455135294551355e-06,
268
+ "loss": 2.3751,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05603480056034801,
273
+ "grad_norm": 0.3583256006240845,
274
+ "learning_rate": 9.440389294403894e-06,
275
+ "loss": 2.3675,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.057509400575094,
280
+ "grad_norm": 0.33290043473243713,
281
+ "learning_rate": 9.425643294256433e-06,
282
+ "loss": 2.3784,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.058984000589840005,
287
+ "grad_norm": 0.3490481674671173,
288
+ "learning_rate": 9.410897294108974e-06,
289
+ "loss": 2.3597,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.06045860060458601,
294
+ "grad_norm": 0.479775071144104,
295
+ "learning_rate": 9.396151293961515e-06,
296
+ "loss": 2.3852,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.061933200619332,
301
+ "grad_norm": 0.36794590950012207,
302
+ "learning_rate": 9.381405293814054e-06,
303
+ "loss": 2.3748,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.06340780063407801,
308
+ "grad_norm": 0.38288605213165283,
309
+ "learning_rate": 9.366659293666593e-06,
310
+ "loss": 2.3673,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.064882400648824,
315
+ "grad_norm": 0.40629106760025024,
316
+ "learning_rate": 9.351913293519134e-06,
317
+ "loss": 2.3906,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06635700066357,
322
+ "grad_norm": 0.3594074249267578,
323
+ "learning_rate": 9.337167293371675e-06,
324
+ "loss": 2.3773,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.067831600678316,
329
+ "grad_norm": 0.463144451379776,
330
+ "learning_rate": 9.322421293224214e-06,
331
+ "loss": 2.3846,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06930620069306201,
336
+ "grad_norm": 0.35561177134513855,
337
+ "learning_rate": 9.307675293076753e-06,
338
+ "loss": 2.3723,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.07078080070780801,
343
+ "grad_norm": 0.3446957767009735,
344
+ "learning_rate": 9.292929292929294e-06,
345
+ "loss": 2.4011,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.07225540072255401,
350
+ "grad_norm": 0.5307037830352783,
351
+ "learning_rate": 9.278183292781835e-06,
352
+ "loss": 2.3747,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.0737300007373,
357
+ "grad_norm": 0.3605501055717468,
358
+ "learning_rate": 9.263437292634374e-06,
359
+ "loss": 2.3523,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.075204600752046,
364
+ "grad_norm": 0.3705900013446808,
365
+ "learning_rate": 9.248691292486913e-06,
366
+ "loss": 2.3693,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07667920076679201,
371
+ "grad_norm": 0.3397590219974518,
372
+ "learning_rate": 9.233945292339454e-06,
373
+ "loss": 2.362,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07815380078153801,
378
+ "grad_norm": 0.32325974106788635,
379
+ "learning_rate": 9.219199292191993e-06,
380
+ "loss": 2.3863,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07962840079628401,
385
+ "grad_norm": 0.3661843240261078,
386
+ "learning_rate": 9.204453292044534e-06,
387
+ "loss": 2.3769,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.08110300081103,
392
+ "grad_norm": 0.3777139186859131,
393
+ "learning_rate": 9.189707291897075e-06,
394
+ "loss": 2.3626,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.082577600825776,
399
+ "grad_norm": 0.37038654088974,
400
+ "learning_rate": 9.174961291749614e-06,
401
+ "loss": 2.3895,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.084052200840522,
406
+ "grad_norm": 0.3628358542919159,
407
+ "learning_rate": 9.160215291602153e-06,
408
+ "loss": 2.3649,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.08552680085526801,
413
+ "grad_norm": 0.3579985201358795,
414
+ "learning_rate": 9.145469291454694e-06,
415
+ "loss": 2.3696,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.08700140087001401,
420
+ "grad_norm": 0.42436483502388,
421
+ "learning_rate": 9.130723291307235e-06,
422
+ "loss": 2.3845,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08847600088476001,
427
+ "grad_norm": 0.3851209580898285,
428
+ "learning_rate": 9.115977291159774e-06,
429
+ "loss": 2.36,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.089950600899506,
434
+ "grad_norm": 0.5885825157165527,
435
+ "learning_rate": 9.101231291012313e-06,
436
+ "loss": 2.3776,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.091425200914252,
441
+ "grad_norm": 0.42304036021232605,
442
+ "learning_rate": 9.086485290864854e-06,
443
+ "loss": 2.3632,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.09289980092899801,
448
+ "grad_norm": 0.4534706473350525,
449
+ "learning_rate": 9.071739290717394e-06,
450
+ "loss": 2.3531,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.09437440094374401,
455
+ "grad_norm": 0.38540026545524597,
456
+ "learning_rate": 9.056993290569934e-06,
457
+ "loss": 2.3755,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.09584900095849001,
462
+ "grad_norm": 0.38401320576667786,
463
+ "learning_rate": 9.042247290422473e-06,
464
+ "loss": 2.3614,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.09732360097323602,
469
+ "grad_norm": 0.35363873839378357,
470
+ "learning_rate": 9.027501290275014e-06,
471
+ "loss": 2.3655,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.098798200987982,
476
+ "grad_norm": 0.36643800139427185,
477
+ "learning_rate": 9.012755290127554e-06,
478
+ "loss": 2.369,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.10027280100272801,
483
+ "grad_norm": 0.3596145212650299,
484
+ "learning_rate": 8.998009289980094e-06,
485
+ "loss": 2.3703,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.10174740101747401,
490
+ "grad_norm": 0.554706871509552,
491
+ "learning_rate": 8.983263289832633e-06,
492
+ "loss": 2.3631,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.10322200103222001,
497
+ "grad_norm": 0.3995848298072815,
498
+ "learning_rate": 8.968517289685174e-06,
499
+ "loss": 2.3671,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.10469660104696601,
504
+ "grad_norm": 0.3653299808502197,
505
+ "learning_rate": 8.953771289537714e-06,
506
+ "loss": 2.3783,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.10617120106171202,
511
+ "grad_norm": 0.39819827675819397,
512
+ "learning_rate": 8.939025289390254e-06,
513
+ "loss": 2.3617,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.107645801076458,
518
+ "grad_norm": 0.3512992262840271,
519
+ "learning_rate": 8.924279289242793e-06,
520
+ "loss": 2.3649,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.10912040109120401,
525
+ "grad_norm": 0.43283718824386597,
526
+ "learning_rate": 8.909533289095333e-06,
527
+ "loss": 2.3624,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.11059500110595001,
532
+ "grad_norm": 0.3857913613319397,
533
+ "learning_rate": 8.894787288947874e-06,
534
+ "loss": 2.3686,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.11206960112069601,
539
+ "grad_norm": 0.3893970251083374,
540
+ "learning_rate": 8.880041288800413e-06,
541
+ "loss": 2.3722,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.11354420113544202,
546
+ "grad_norm": 0.6659526228904724,
547
+ "learning_rate": 8.865295288652953e-06,
548
+ "loss": 2.347,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.115018801150188,
553
+ "grad_norm": 0.3801191449165344,
554
+ "learning_rate": 8.850549288505493e-06,
555
+ "loss": 2.3595,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.11649340116493401,
560
+ "grad_norm": 0.3556024432182312,
561
+ "learning_rate": 8.835803288358034e-06,
562
+ "loss": 2.3597,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.11796800117968001,
567
+ "grad_norm": 0.45088592171669006,
568
+ "learning_rate": 8.821057288210573e-06,
569
+ "loss": 2.3623,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.11944260119442601,
574
+ "grad_norm": 0.3952132761478424,
575
+ "learning_rate": 8.806311288063114e-06,
576
+ "loss": 2.3616,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.12091720120917201,
581
+ "grad_norm": 0.4083469808101654,
582
+ "learning_rate": 8.791565287915653e-06,
583
+ "loss": 2.3597,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.12239180122391802,
588
+ "grad_norm": 0.374976247549057,
589
+ "learning_rate": 8.776819287768194e-06,
590
+ "loss": 2.3656,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.123866401238664,
595
+ "grad_norm": 0.37194111943244934,
596
+ "learning_rate": 8.762073287620733e-06,
597
+ "loss": 2.368,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.12534100125341002,
602
+ "grad_norm": 0.35863035917282104,
603
+ "learning_rate": 8.747327287473274e-06,
604
+ "loss": 2.3648,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.12681560126815603,
609
+ "grad_norm": 0.43522682785987854,
610
+ "learning_rate": 8.732581287325813e-06,
611
+ "loss": 2.3663,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.128290201282902,
616
+ "grad_norm": 0.3582712411880493,
617
+ "learning_rate": 8.717835287178354e-06,
618
+ "loss": 2.3684,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.129764801297648,
623
+ "grad_norm": 0.40035149455070496,
624
+ "learning_rate": 8.703089287030893e-06,
625
+ "loss": 2.3679,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.131239401312394,
630
+ "grad_norm": 0.36125797033309937,
631
+ "learning_rate": 8.688343286883434e-06,
632
+ "loss": 2.3501,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.13271400132714,
637
+ "grad_norm": 0.3568665087223053,
638
+ "learning_rate": 8.673597286735973e-06,
639
+ "loss": 2.3594,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.134188601341886,
644
+ "grad_norm": 0.4135202169418335,
645
+ "learning_rate": 8.658851286588512e-06,
646
+ "loss": 2.387,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.135663201356632,
651
+ "grad_norm": 0.3586069345474243,
652
+ "learning_rate": 8.644105286441053e-06,
653
+ "loss": 2.3702,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.13713780137137802,
658
+ "grad_norm": 0.47354263067245483,
659
+ "learning_rate": 8.629359286293594e-06,
660
+ "loss": 2.3709,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.13861240138612402,
665
+ "grad_norm": 0.3735561668872833,
666
+ "learning_rate": 8.614613286146133e-06,
667
+ "loss": 2.3639,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.14008700140087002,
672
+ "grad_norm": 0.351646363735199,
673
+ "learning_rate": 8.599867285998672e-06,
674
+ "loss": 2.3678,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.14156160141561602,
679
+ "grad_norm": 0.3624926507472992,
680
+ "learning_rate": 8.585121285851213e-06,
681
+ "loss": 2.3725,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.14303620143036203,
686
+ "grad_norm": 0.41470956802368164,
687
+ "learning_rate": 8.570375285703754e-06,
688
+ "loss": 2.364,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.14451080144510803,
693
+ "grad_norm": 0.3892461657524109,
694
+ "learning_rate": 8.555629285556293e-06,
695
+ "loss": 2.3623,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.145985401459854,
700
+ "grad_norm": 0.37670454382896423,
701
+ "learning_rate": 8.540883285408832e-06,
702
+ "loss": 2.3649,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.1474600014746,
707
+ "grad_norm": 0.36948758363723755,
708
+ "learning_rate": 8.526137285261373e-06,
709
+ "loss": 2.3689,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.148934601489346,
714
+ "grad_norm": 0.3593231737613678,
715
+ "learning_rate": 8.511391285113914e-06,
716
+ "loss": 2.3693,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.150409201504092,
721
+ "grad_norm": 0.3645029366016388,
722
+ "learning_rate": 8.496645284966453e-06,
723
+ "loss": 2.3607,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.151883801518838,
728
+ "grad_norm": 0.761593759059906,
729
+ "learning_rate": 8.481899284818994e-06,
730
+ "loss": 2.362,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.15335840153358402,
735
+ "grad_norm": 0.3621225953102112,
736
+ "learning_rate": 8.467153284671533e-06,
737
+ "loss": 2.3566,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.15483300154833002,
742
+ "grad_norm": 0.353289395570755,
743
+ "learning_rate": 8.452407284524074e-06,
744
+ "loss": 2.3565,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.15630760156307602,
749
+ "grad_norm": 0.4106830060482025,
750
+ "learning_rate": 8.437661284376613e-06,
751
+ "loss": 2.3536,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.15778220157782202,
756
+ "grad_norm": 0.41519981622695923,
757
+ "learning_rate": 8.422915284229154e-06,
758
+ "loss": 2.3532,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.15925680159256803,
763
+ "grad_norm": 0.3569793701171875,
764
+ "learning_rate": 8.408169284081693e-06,
765
+ "loss": 2.3648,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.16073140160731403,
770
+ "grad_norm": 0.3916598856449127,
771
+ "learning_rate": 8.393423283934234e-06,
772
+ "loss": 2.3675,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.16220600162206,
777
+ "grad_norm": 0.3435116708278656,
778
+ "learning_rate": 8.378677283786773e-06,
779
+ "loss": 2.3685,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.163680601636806,
784
+ "grad_norm": 0.47191643714904785,
785
+ "learning_rate": 8.363931283639314e-06,
786
+ "loss": 2.3596,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.165155201651552,
791
+ "grad_norm": 0.54694002866745,
792
+ "learning_rate": 8.349185283491853e-06,
793
+ "loss": 2.3551,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.166629801666298,
798
+ "grad_norm": 0.9548392295837402,
799
+ "learning_rate": 8.334439283344394e-06,
800
+ "loss": 2.3784,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.168104401681044,
805
+ "grad_norm": 0.3765574097633362,
806
+ "learning_rate": 8.319693283196933e-06,
807
+ "loss": 2.3668,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.16957900169579002,
812
+ "grad_norm": 0.748131275177002,
813
+ "learning_rate": 8.304947283049474e-06,
814
+ "loss": 2.3492,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.17105360171053602,
819
+ "grad_norm": 0.3873535096645355,
820
+ "learning_rate": 8.290201282902015e-06,
821
+ "loss": 2.3592,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.17252820172528202,
826
+ "grad_norm": 0.3429403305053711,
827
+ "learning_rate": 8.275455282754554e-06,
828
+ "loss": 2.3688,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.17400280174002802,
833
+ "grad_norm": 0.3973939120769501,
834
+ "learning_rate": 8.260709282607093e-06,
835
+ "loss": 2.3666,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.17547740175477403,
840
+ "grad_norm": 0.35062074661254883,
841
+ "learning_rate": 8.245963282459634e-06,
842
+ "loss": 2.3534,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.17695200176952003,
847
+ "grad_norm": 0.3548290431499481,
848
+ "learning_rate": 8.231217282312175e-06,
849
+ "loss": 2.3477,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.17842660178426603,
854
+ "grad_norm": 0.4035143256187439,
855
+ "learning_rate": 8.216471282164714e-06,
856
+ "loss": 2.3601,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.179901201799012,
861
+ "grad_norm": 0.37481680512428284,
862
+ "learning_rate": 8.201725282017253e-06,
863
+ "loss": 2.3579,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.181375801813758,
868
+ "grad_norm": 0.5035015344619751,
869
+ "learning_rate": 8.186979281869794e-06,
870
+ "loss": 2.366,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.182850401828504,
875
+ "grad_norm": 0.3541349768638611,
876
+ "learning_rate": 8.172233281722333e-06,
877
+ "loss": 2.368,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.18432500184325,
882
+ "grad_norm": 0.3855954110622406,
883
+ "learning_rate": 8.157487281574874e-06,
884
+ "loss": 2.3542,
885
+ "step": 2500
886
+ },
887
+ {
888
+ "epoch": 0.18579960185799602,
889
+ "grad_norm": 0.4045879542827606,
890
+ "learning_rate": 8.142741281427413e-06,
891
+ "loss": 2.3579,
892
+ "step": 2520
893
+ },
894
+ {
895
+ "epoch": 0.18727420187274202,
896
+ "grad_norm": 0.3711334764957428,
897
+ "learning_rate": 8.127995281279954e-06,
898
+ "loss": 2.3711,
899
+ "step": 2540
900
+ },
901
+ {
902
+ "epoch": 0.18874880188748802,
903
+ "grad_norm": 0.3625940978527069,
904
+ "learning_rate": 8.113249281132493e-06,
905
+ "loss": 2.3573,
906
+ "step": 2560
907
+ },
908
+ {
909
+ "epoch": 0.19022340190223402,
910
+ "grad_norm": 0.6012184619903564,
911
+ "learning_rate": 8.098503280985034e-06,
912
+ "loss": 2.3705,
913
+ "step": 2580
914
+ },
915
+ {
916
+ "epoch": 0.19169800191698003,
917
+ "grad_norm": 0.40935376286506653,
918
+ "learning_rate": 8.083757280837573e-06,
919
+ "loss": 2.3615,
920
+ "step": 2600
921
+ },
922
+ {
923
+ "epoch": 0.19317260193172603,
924
+ "grad_norm": 0.34611454606056213,
925
+ "learning_rate": 8.069011280690114e-06,
926
+ "loss": 2.358,
927
+ "step": 2620
928
+ },
929
+ {
930
+ "epoch": 0.19464720194647203,
931
+ "grad_norm": 0.36542612314224243,
932
+ "learning_rate": 8.054265280542653e-06,
933
+ "loss": 2.3656,
934
+ "step": 2640
935
+ },
936
+ {
937
+ "epoch": 0.196121801961218,
938
+ "grad_norm": 0.38626089692115784,
939
+ "learning_rate": 8.039519280395194e-06,
940
+ "loss": 2.3643,
941
+ "step": 2660
942
+ },
943
+ {
944
+ "epoch": 0.197596401975964,
945
+ "grad_norm": 0.39764684438705444,
946
+ "learning_rate": 8.024773280247733e-06,
947
+ "loss": 2.3556,
948
+ "step": 2680
949
+ },
950
+ {
951
+ "epoch": 0.19907100199071,
952
+ "grad_norm": 0.3800354301929474,
953
+ "learning_rate": 8.010027280100274e-06,
954
+ "loss": 2.3696,
955
+ "step": 2700
956
+ },
957
+ {
958
+ "epoch": 0.20054560200545601,
959
+ "grad_norm": 0.37549829483032227,
960
+ "learning_rate": 7.995281279952813e-06,
961
+ "loss": 2.3623,
962
+ "step": 2720
963
+ },
964
+ {
965
+ "epoch": 0.20202020202020202,
966
+ "grad_norm": 0.3357870280742645,
967
+ "learning_rate": 7.980535279805354e-06,
968
+ "loss": 2.3609,
969
+ "step": 2740
970
+ },
971
+ {
972
+ "epoch": 0.20349480203494802,
973
+ "grad_norm": 0.38587677478790283,
974
+ "learning_rate": 7.965789279657895e-06,
975
+ "loss": 2.3648,
976
+ "step": 2760
977
+ },
978
+ {
979
+ "epoch": 0.20496940204969402,
980
+ "grad_norm": 0.3734722137451172,
981
+ "learning_rate": 7.951043279510434e-06,
982
+ "loss": 2.3637,
983
+ "step": 2780
984
+ },
985
+ {
986
+ "epoch": 0.20644400206444002,
987
+ "grad_norm": 0.37882205843925476,
988
+ "learning_rate": 7.936297279362973e-06,
989
+ "loss": 2.3638,
990
+ "step": 2800
991
+ },
992
+ {
993
+ "epoch": 0.20791860207918603,
994
+ "grad_norm": 0.3540538549423218,
995
+ "learning_rate": 7.921551279215514e-06,
996
+ "loss": 2.3483,
997
+ "step": 2820
998
+ },
999
+ {
1000
+ "epoch": 0.20939320209393203,
1001
+ "grad_norm": 0.3712068796157837,
1002
+ "learning_rate": 7.906805279068054e-06,
1003
+ "loss": 2.3553,
1004
+ "step": 2840
1005
+ },
1006
+ {
1007
+ "epoch": 0.21086780210867803,
1008
+ "grad_norm": 0.3830094039440155,
1009
+ "learning_rate": 7.892059278920594e-06,
1010
+ "loss": 2.3526,
1011
+ "step": 2860
1012
+ },
1013
+ {
1014
+ "epoch": 0.21234240212342403,
1015
+ "grad_norm": 0.37301984429359436,
1016
+ "learning_rate": 7.877313278773133e-06,
1017
+ "loss": 2.3597,
1018
+ "step": 2880
1019
+ },
1020
+ {
1021
+ "epoch": 0.21381700213817,
1022
+ "grad_norm": 0.3589264452457428,
1023
+ "learning_rate": 7.862567278625674e-06,
1024
+ "loss": 2.3526,
1025
+ "step": 2900
1026
+ },
1027
+ {
1028
+ "epoch": 0.215291602152916,
1029
+ "grad_norm": 0.33692067861557007,
1030
+ "learning_rate": 7.847821278478214e-06,
1031
+ "loss": 2.3529,
1032
+ "step": 2920
1033
+ },
1034
+ {
1035
+ "epoch": 0.21676620216766201,
1036
+ "grad_norm": 0.3636477589607239,
1037
+ "learning_rate": 7.833075278330754e-06,
1038
+ "loss": 2.3583,
1039
+ "step": 2940
1040
+ },
1041
+ {
1042
+ "epoch": 0.21824080218240802,
1043
+ "grad_norm": 0.39444780349731445,
1044
+ "learning_rate": 7.818329278183293e-06,
1045
+ "loss": 2.3448,
1046
+ "step": 2960
1047
+ },
1048
+ {
1049
+ "epoch": 0.21971540219715402,
1050
+ "grad_norm": 0.3866218626499176,
1051
+ "learning_rate": 7.803583278035834e-06,
1052
+ "loss": 2.3626,
1053
+ "step": 2980
1054
+ },
1055
+ {
1056
+ "epoch": 0.22119000221190002,
1057
+ "grad_norm": 0.37035319209098816,
1058
+ "learning_rate": 7.788837277888374e-06,
1059
+ "loss": 2.3672,
1060
+ "step": 3000
1061
+ },
1062
+ {
1063
+ "epoch": 0.22266460222664602,
1064
+ "grad_norm": 0.3661469519138336,
1065
+ "learning_rate": 7.774091277740914e-06,
1066
+ "loss": 2.3658,
1067
+ "step": 3020
1068
+ },
1069
+ {
1070
+ "epoch": 0.22413920224139203,
1071
+ "grad_norm": 0.391837477684021,
1072
+ "learning_rate": 7.759345277593453e-06,
1073
+ "loss": 2.3672,
1074
+ "step": 3040
1075
+ },
1076
+ {
1077
+ "epoch": 0.22561380225613803,
1078
+ "grad_norm": 0.36069273948669434,
1079
+ "learning_rate": 7.744599277445994e-06,
1080
+ "loss": 2.3685,
1081
+ "step": 3060
1082
+ },
1083
+ {
1084
+ "epoch": 0.22708840227088403,
1085
+ "grad_norm": 0.3551023602485657,
1086
+ "learning_rate": 7.729853277298534e-06,
1087
+ "loss": 2.3697,
1088
+ "step": 3080
1089
+ },
1090
+ {
1091
+ "epoch": 0.22856300228563003,
1092
+ "grad_norm": 0.3549511730670929,
1093
+ "learning_rate": 7.715107277151073e-06,
1094
+ "loss": 2.3708,
1095
+ "step": 3100
1096
+ },
1097
+ {
1098
+ "epoch": 0.230037602300376,
1099
+ "grad_norm": 0.3584568202495575,
1100
+ "learning_rate": 7.700361277003613e-06,
1101
+ "loss": 2.3491,
1102
+ "step": 3120
1103
+ },
1104
+ {
1105
+ "epoch": 0.231512202315122,
1106
+ "grad_norm": 0.3881671726703644,
1107
+ "learning_rate": 7.685615276856153e-06,
1108
+ "loss": 2.3603,
1109
+ "step": 3140
1110
+ },
1111
+ {
1112
+ "epoch": 0.23298680232986801,
1113
+ "grad_norm": 0.37565991282463074,
1114
+ "learning_rate": 7.670869276708693e-06,
1115
+ "loss": 2.349,
1116
+ "step": 3160
1117
+ },
1118
+ {
1119
+ "epoch": 0.23446140234461402,
1120
+ "grad_norm": 0.3800000548362732,
1121
+ "learning_rate": 7.656123276561233e-06,
1122
+ "loss": 2.3634,
1123
+ "step": 3180
1124
+ },
1125
+ {
1126
+ "epoch": 0.23593600235936002,
1127
+ "grad_norm": 0.3795914053916931,
1128
+ "learning_rate": 7.641377276413774e-06,
1129
+ "loss": 2.3541,
1130
+ "step": 3200
1131
+ },
1132
+ {
1133
+ "epoch": 0.23741060237410602,
1134
+ "grad_norm": 0.34739038348197937,
1135
+ "learning_rate": 7.626631276266313e-06,
1136
+ "loss": 2.364,
1137
+ "step": 3220
1138
+ },
1139
+ {
1140
+ "epoch": 0.23888520238885202,
1141
+ "grad_norm": 0.3631921112537384,
1142
+ "learning_rate": 7.611885276118853e-06,
1143
+ "loss": 2.3472,
1144
+ "step": 3240
1145
+ },
1146
+ {
1147
+ "epoch": 0.24035980240359803,
1148
+ "grad_norm": 0.34708309173583984,
1149
+ "learning_rate": 7.597139275971393e-06,
1150
+ "loss": 2.3745,
1151
+ "step": 3260
1152
+ },
1153
+ {
1154
+ "epoch": 0.24183440241834403,
1155
+ "grad_norm": 0.39002010226249695,
1156
+ "learning_rate": 7.582393275823933e-06,
1157
+ "loss": 2.3496,
1158
+ "step": 3280
1159
+ },
1160
+ {
1161
+ "epoch": 0.24330900243309003,
1162
+ "grad_norm": 0.3856920003890991,
1163
+ "learning_rate": 7.5676472756764725e-06,
1164
+ "loss": 2.3556,
1165
+ "step": 3300
1166
+ },
1167
+ {
1168
+ "epoch": 0.24478360244783604,
1169
+ "grad_norm": 0.3604213297367096,
1170
+ "learning_rate": 7.552901275529013e-06,
1171
+ "loss": 2.3588,
1172
+ "step": 3320
1173
+ },
1174
+ {
1175
+ "epoch": 0.24625820246258204,
1176
+ "grad_norm": 0.4610792100429535,
1177
+ "learning_rate": 7.538155275381553e-06,
1178
+ "loss": 2.3495,
1179
+ "step": 3340
1180
+ },
1181
+ {
1182
+ "epoch": 0.247732802477328,
1183
+ "grad_norm": 0.37130406498908997,
1184
+ "learning_rate": 7.523409275234093e-06,
1185
+ "loss": 2.3654,
1186
+ "step": 3360
1187
+ },
1188
+ {
1189
+ "epoch": 0.24920740249207401,
1190
+ "grad_norm": 0.4634559154510498,
1191
+ "learning_rate": 7.5086632750866325e-06,
1192
+ "loss": 2.3472,
1193
+ "step": 3380
1194
+ },
1195
+ {
1196
+ "epoch": 0.25068200250682005,
1197
+ "grad_norm": 0.3728873133659363,
1198
+ "learning_rate": 7.493917274939173e-06,
1199
+ "loss": 2.3601,
1200
+ "step": 3400
1201
+ },
1202
+ {
1203
+ "epoch": 0.252156602521566,
1204
+ "grad_norm": 0.35525408387184143,
1205
+ "learning_rate": 7.479171274791713e-06,
1206
+ "loss": 2.3483,
1207
+ "step": 3420
1208
+ },
1209
+ {
1210
+ "epoch": 0.25363120253631205,
1211
+ "grad_norm": 0.35072821378707886,
1212
+ "learning_rate": 7.464425274644253e-06,
1213
+ "loss": 2.3575,
1214
+ "step": 3440
1215
+ },
1216
+ {
1217
+ "epoch": 0.255105802551058,
1218
+ "grad_norm": 0.33984318375587463,
1219
+ "learning_rate": 7.449679274496794e-06,
1220
+ "loss": 2.3562,
1221
+ "step": 3460
1222
+ },
1223
+ {
1224
+ "epoch": 0.256580402565804,
1225
+ "grad_norm": 0.3893386423587799,
1226
+ "learning_rate": 7.434933274349333e-06,
1227
+ "loss": 2.3417,
1228
+ "step": 3480
1229
+ },
1230
+ {
1231
+ "epoch": 0.25805500258055003,
1232
+ "grad_norm": 0.3702589273452759,
1233
+ "learning_rate": 7.420187274201873e-06,
1234
+ "loss": 2.3575,
1235
+ "step": 3500
1236
+ },
1237
+ {
1238
+ "epoch": 0.259529602595296,
1239
+ "grad_norm": 0.37328028678894043,
1240
+ "learning_rate": 7.405441274054413e-06,
1241
+ "loss": 2.3691,
1242
+ "step": 3520
1243
+ },
1244
+ {
1245
+ "epoch": 0.26100420261004204,
1246
+ "grad_norm": 0.362596720457077,
1247
+ "learning_rate": 7.390695273906954e-06,
1248
+ "loss": 2.3516,
1249
+ "step": 3540
1250
+ },
1251
+ {
1252
+ "epoch": 0.262478802624788,
1253
+ "grad_norm": 0.35985130071640015,
1254
+ "learning_rate": 7.375949273759493e-06,
1255
+ "loss": 2.3579,
1256
+ "step": 3560
1257
+ },
1258
+ {
1259
+ "epoch": 0.26395340263953404,
1260
+ "grad_norm": 0.39259010553359985,
1261
+ "learning_rate": 7.361203273612033e-06,
1262
+ "loss": 2.3674,
1263
+ "step": 3580
1264
+ },
1265
+ {
1266
+ "epoch": 0.26542800265428,
1267
+ "grad_norm": 0.3379274904727936,
1268
+ "learning_rate": 7.346457273464573e-06,
1269
+ "loss": 2.3409,
1270
+ "step": 3600
1271
+ },
1272
+ {
1273
+ "epoch": 0.26690260266902605,
1274
+ "grad_norm": 0.38404515385627747,
1275
+ "learning_rate": 7.331711273317113e-06,
1276
+ "loss": 2.3493,
1277
+ "step": 3620
1278
+ },
1279
+ {
1280
+ "epoch": 0.268377202683772,
1281
+ "grad_norm": 0.3580029606819153,
1282
+ "learning_rate": 7.316965273169654e-06,
1283
+ "loss": 2.3426,
1284
+ "step": 3640
1285
+ },
1286
+ {
1287
+ "epoch": 0.26985180269851805,
1288
+ "grad_norm": 0.36124977469444275,
1289
+ "learning_rate": 7.302219273022193e-06,
1290
+ "loss": 2.3563,
1291
+ "step": 3660
1292
+ },
1293
+ {
1294
+ "epoch": 0.271326402713264,
1295
+ "grad_norm": 0.37032002210617065,
1296
+ "learning_rate": 7.287473272874733e-06,
1297
+ "loss": 2.3558,
1298
+ "step": 3680
1299
+ },
1300
+ {
1301
+ "epoch": 0.27280100272801,
1302
+ "grad_norm": 0.37544354796409607,
1303
+ "learning_rate": 7.272727272727273e-06,
1304
+ "loss": 2.353,
1305
+ "step": 3700
1306
+ },
1307
+ {
1308
+ "epoch": 0.27427560274275603,
1309
+ "grad_norm": 0.38063186407089233,
1310
+ "learning_rate": 7.257981272579814e-06,
1311
+ "loss": 2.355,
1312
+ "step": 3720
1313
+ },
1314
+ {
1315
+ "epoch": 0.275750202757502,
1316
+ "grad_norm": 0.34436681866645813,
1317
+ "learning_rate": 7.243235272432353e-06,
1318
+ "loss": 2.3671,
1319
+ "step": 3740
1320
+ },
1321
+ {
1322
+ "epoch": 0.27722480277224804,
1323
+ "grad_norm": 0.3745051622390747,
1324
+ "learning_rate": 7.228489272284893e-06,
1325
+ "loss": 2.343,
1326
+ "step": 3760
1327
+ },
1328
+ {
1329
+ "epoch": 0.278699402786994,
1330
+ "grad_norm": 0.36901938915252686,
1331
+ "learning_rate": 7.213743272137433e-06,
1332
+ "loss": 2.3367,
1333
+ "step": 3780
1334
+ },
1335
+ {
1336
+ "epoch": 0.28017400280174004,
1337
+ "grad_norm": 0.38288217782974243,
1338
+ "learning_rate": 7.198997271989974e-06,
1339
+ "loss": 2.3598,
1340
+ "step": 3800
1341
+ },
1342
+ {
1343
+ "epoch": 0.281648602816486,
1344
+ "grad_norm": 0.37386777997016907,
1345
+ "learning_rate": 7.184251271842513e-06,
1346
+ "loss": 2.3618,
1347
+ "step": 3820
1348
+ },
1349
+ {
1350
+ "epoch": 0.28312320283123205,
1351
+ "grad_norm": 0.35673725605010986,
1352
+ "learning_rate": 7.169505271695053e-06,
1353
+ "loss": 2.3514,
1354
+ "step": 3840
1355
+ },
1356
+ {
1357
+ "epoch": 0.284597802845978,
1358
+ "grad_norm": 0.38071954250335693,
1359
+ "learning_rate": 7.154759271547593e-06,
1360
+ "loss": 2.3564,
1361
+ "step": 3860
1362
+ },
1363
+ {
1364
+ "epoch": 0.28607240286072405,
1365
+ "grad_norm": 0.3805045187473297,
1366
+ "learning_rate": 7.140013271400134e-06,
1367
+ "loss": 2.3599,
1368
+ "step": 3880
1369
+ },
1370
+ {
1371
+ "epoch": 0.28754700287547,
1372
+ "grad_norm": 0.3432954251766205,
1373
+ "learning_rate": 7.125267271252674e-06,
1374
+ "loss": 2.3568,
1375
+ "step": 3900
1376
+ },
1377
+ {
1378
+ "epoch": 0.28902160289021606,
1379
+ "grad_norm": 0.35117459297180176,
1380
+ "learning_rate": 7.110521271105213e-06,
1381
+ "loss": 2.3566,
1382
+ "step": 3920
1383
+ },
1384
+ {
1385
+ "epoch": 0.29049620290496203,
1386
+ "grad_norm": 0.39861348271369934,
1387
+ "learning_rate": 7.095775270957753e-06,
1388
+ "loss": 2.3553,
1389
+ "step": 3940
1390
+ },
1391
+ {
1392
+ "epoch": 0.291970802919708,
1393
+ "grad_norm": 0.3653263747692108,
1394
+ "learning_rate": 7.081029270810294e-06,
1395
+ "loss": 2.3781,
1396
+ "step": 3960
1397
+ },
1398
+ {
1399
+ "epoch": 0.29344540293445404,
1400
+ "grad_norm": 0.3614581823348999,
1401
+ "learning_rate": 7.066283270662834e-06,
1402
+ "loss": 2.3407,
1403
+ "step": 3980
1404
+ },
1405
+ {
1406
+ "epoch": 0.2949200029492,
1407
+ "grad_norm": 0.39455854892730713,
1408
+ "learning_rate": 7.051537270515373e-06,
1409
+ "loss": 2.3541,
1410
+ "step": 4000
1411
+ },
1412
+ {
1413
+ "epoch": 0.29639460296394604,
1414
+ "grad_norm": 0.33767521381378174,
1415
+ "learning_rate": 7.036791270367913e-06,
1416
+ "loss": 2.3332,
1417
+ "step": 4020
1418
+ },
1419
+ {
1420
+ "epoch": 0.297869202978692,
1421
+ "grad_norm": 0.36079707741737366,
1422
+ "learning_rate": 7.022045270220453e-06,
1423
+ "loss": 2.3546,
1424
+ "step": 4040
1425
+ },
1426
+ {
1427
+ "epoch": 0.29934380299343805,
1428
+ "grad_norm": 0.395107626914978,
1429
+ "learning_rate": 7.007299270072994e-06,
1430
+ "loss": 2.3483,
1431
+ "step": 4060
1432
+ },
1433
+ {
1434
+ "epoch": 0.300818403008184,
1435
+ "grad_norm": 0.3456408381462097,
1436
+ "learning_rate": 6.992553269925533e-06,
1437
+ "loss": 2.3657,
1438
+ "step": 4080
1439
+ },
1440
+ {
1441
+ "epoch": 0.30229300302293005,
1442
+ "grad_norm": 0.4138198792934418,
1443
+ "learning_rate": 6.977807269778073e-06,
1444
+ "loss": 2.3616,
1445
+ "step": 4100
1446
+ },
1447
+ {
1448
+ "epoch": 0.303767603037676,
1449
+ "grad_norm": 0.382722944021225,
1450
+ "learning_rate": 6.963061269630613e-06,
1451
+ "loss": 2.3607,
1452
+ "step": 4120
1453
+ },
1454
+ {
1455
+ "epoch": 0.30524220305242206,
1456
+ "grad_norm": 0.4129588007926941,
1457
+ "learning_rate": 6.948315269483154e-06,
1458
+ "loss": 2.3582,
1459
+ "step": 4140
1460
+ },
1461
+ {
1462
+ "epoch": 0.30671680306716803,
1463
+ "grad_norm": 0.35810399055480957,
1464
+ "learning_rate": 6.933569269335694e-06,
1465
+ "loss": 2.345,
1466
+ "step": 4160
1467
+ },
1468
+ {
1469
+ "epoch": 0.308191403081914,
1470
+ "grad_norm": 0.3707892596721649,
1471
+ "learning_rate": 6.918823269188233e-06,
1472
+ "loss": 2.3389,
1473
+ "step": 4180
1474
+ },
1475
+ {
1476
+ "epoch": 0.30966600309666004,
1477
+ "grad_norm": 0.43102672696113586,
1478
+ "learning_rate": 6.904077269040773e-06,
1479
+ "loss": 2.3636,
1480
+ "step": 4200
1481
+ },
1482
+ {
1483
+ "epoch": 0.311140603111406,
1484
+ "grad_norm": 0.37081730365753174,
1485
+ "learning_rate": 6.889331268893314e-06,
1486
+ "loss": 2.3589,
1487
+ "step": 4220
1488
+ },
1489
+ {
1490
+ "epoch": 0.31261520312615204,
1491
+ "grad_norm": 0.3554447591304779,
1492
+ "learning_rate": 6.874585268745854e-06,
1493
+ "loss": 2.3639,
1494
+ "step": 4240
1495
+ },
1496
+ {
1497
+ "epoch": 0.314089803140898,
1498
+ "grad_norm": 0.3365320563316345,
1499
+ "learning_rate": 6.859839268598393e-06,
1500
+ "loss": 2.3604,
1501
+ "step": 4260
1502
+ },
1503
+ {
1504
+ "epoch": 0.31556440315564405,
1505
+ "grad_norm": 0.40808168053627014,
1506
+ "learning_rate": 6.845093268450933e-06,
1507
+ "loss": 2.3496,
1508
+ "step": 4280
1509
+ },
1510
+ {
1511
+ "epoch": 0.31703900317039,
1512
+ "grad_norm": 0.380753755569458,
1513
+ "learning_rate": 6.830347268303474e-06,
1514
+ "loss": 2.3478,
1515
+ "step": 4300
1516
+ },
1517
+ {
1518
+ "epoch": 0.31851360318513605,
1519
+ "grad_norm": 0.3863361179828644,
1520
+ "learning_rate": 6.815601268156014e-06,
1521
+ "loss": 2.3381,
1522
+ "step": 4320
1523
+ },
1524
+ {
1525
+ "epoch": 0.319988203199882,
1526
+ "grad_norm": 0.40554317831993103,
1527
+ "learning_rate": 6.800855268008554e-06,
1528
+ "loss": 2.3518,
1529
+ "step": 4340
1530
+ },
1531
+ {
1532
+ "epoch": 0.32146280321462806,
1533
+ "grad_norm": 0.39253705739974976,
1534
+ "learning_rate": 6.786109267861093e-06,
1535
+ "loss": 2.3567,
1536
+ "step": 4360
1537
+ },
1538
+ {
1539
+ "epoch": 0.32293740322937403,
1540
+ "grad_norm": 0.37108898162841797,
1541
+ "learning_rate": 6.771363267713633e-06,
1542
+ "loss": 2.3509,
1543
+ "step": 4380
1544
+ },
1545
+ {
1546
+ "epoch": 0.32441200324412,
1547
+ "grad_norm": 0.36466971039772034,
1548
+ "learning_rate": 6.756617267566174e-06,
1549
+ "loss": 2.3627,
1550
+ "step": 4400
1551
+ },
1552
+ {
1553
+ "epoch": 0.32588660325886604,
1554
+ "grad_norm": 0.37712493538856506,
1555
+ "learning_rate": 6.741871267418714e-06,
1556
+ "loss": 2.3754,
1557
+ "step": 4420
1558
+ },
1559
+ {
1560
+ "epoch": 0.327361203273612,
1561
+ "grad_norm": 0.3712570071220398,
1562
+ "learning_rate": 6.727125267271253e-06,
1563
+ "loss": 2.3619,
1564
+ "step": 4440
1565
+ },
1566
+ {
1567
+ "epoch": 0.32883580328835804,
1568
+ "grad_norm": 0.34880587458610535,
1569
+ "learning_rate": 6.712379267123793e-06,
1570
+ "loss": 2.3574,
1571
+ "step": 4460
1572
+ },
1573
+ {
1574
+ "epoch": 0.330310403303104,
1575
+ "grad_norm": 0.4109377861022949,
1576
+ "learning_rate": 6.697633266976334e-06,
1577
+ "loss": 2.3574,
1578
+ "step": 4480
1579
+ },
1580
+ {
1581
+ "epoch": 0.33178500331785005,
1582
+ "grad_norm": 0.3626454770565033,
1583
+ "learning_rate": 6.682887266828874e-06,
1584
+ "loss": 2.3527,
1585
+ "step": 4500
1586
+ },
1587
+ {
1588
+ "epoch": 0.333259603332596,
1589
+ "grad_norm": 0.3580588400363922,
1590
+ "learning_rate": 6.668141266681413e-06,
1591
+ "loss": 2.348,
1592
+ "step": 4520
1593
+ },
1594
+ {
1595
+ "epoch": 0.33473420334734205,
1596
+ "grad_norm": 0.3460945785045624,
1597
+ "learning_rate": 6.653395266533953e-06,
1598
+ "loss": 2.3461,
1599
+ "step": 4540
1600
+ },
1601
+ {
1602
+ "epoch": 0.336208803362088,
1603
+ "grad_norm": 0.39259958267211914,
1604
+ "learning_rate": 6.6386492663864936e-06,
1605
+ "loss": 2.3818,
1606
+ "step": 4560
1607
+ },
1608
+ {
1609
+ "epoch": 0.33768340337683406,
1610
+ "grad_norm": 0.44788721203804016,
1611
+ "learning_rate": 6.6239032662390335e-06,
1612
+ "loss": 2.3636,
1613
+ "step": 4580
1614
+ },
1615
+ {
1616
+ "epoch": 0.33915800339158003,
1617
+ "grad_norm": 0.40668484568595886,
1618
+ "learning_rate": 6.6091572660915735e-06,
1619
+ "loss": 2.3542,
1620
+ "step": 4600
1621
+ },
1622
+ {
1623
+ "epoch": 0.340632603406326,
1624
+ "grad_norm": 0.3672851026058197,
1625
+ "learning_rate": 6.594411265944113e-06,
1626
+ "loss": 2.3497,
1627
+ "step": 4620
1628
+ },
1629
+ {
1630
+ "epoch": 0.34210720342107204,
1631
+ "grad_norm": 0.3685692846775055,
1632
+ "learning_rate": 6.5796652657966535e-06,
1633
+ "loss": 2.361,
1634
+ "step": 4640
1635
+ },
1636
+ {
1637
+ "epoch": 0.343581803435818,
1638
+ "grad_norm": 0.4522005617618561,
1639
+ "learning_rate": 6.5649192656491935e-06,
1640
+ "loss": 2.3498,
1641
+ "step": 4660
1642
+ },
1643
+ {
1644
+ "epoch": 0.34505640345056404,
1645
+ "grad_norm": 0.4089388847351074,
1646
+ "learning_rate": 6.5501732655017335e-06,
1647
+ "loss": 2.3479,
1648
+ "step": 4680
1649
+ },
1650
+ {
1651
+ "epoch": 0.34653100346531,
1652
+ "grad_norm": 0.3874075710773468,
1653
+ "learning_rate": 6.535427265354273e-06,
1654
+ "loss": 2.3437,
1655
+ "step": 4700
1656
+ },
1657
+ {
1658
+ "epoch": 0.34800560348005605,
1659
+ "grad_norm": 0.3733789622783661,
1660
+ "learning_rate": 6.520681265206813e-06,
1661
+ "loss": 2.3524,
1662
+ "step": 4720
1663
+ },
1664
+ {
1665
+ "epoch": 0.349480203494802,
1666
+ "grad_norm": 0.3892102539539337,
1667
+ "learning_rate": 6.5059352650593535e-06,
1668
+ "loss": 2.346,
1669
+ "step": 4740
1670
+ },
1671
+ {
1672
+ "epoch": 0.35095480350954805,
1673
+ "grad_norm": 0.36850956082344055,
1674
+ "learning_rate": 6.4911892649118935e-06,
1675
+ "loss": 2.3532,
1676
+ "step": 4760
1677
+ },
1678
+ {
1679
+ "epoch": 0.352429403524294,
1680
+ "grad_norm": 0.3519984185695648,
1681
+ "learning_rate": 6.476443264764433e-06,
1682
+ "loss": 2.3429,
1683
+ "step": 4780
1684
+ },
1685
+ {
1686
+ "epoch": 0.35390400353904006,
1687
+ "grad_norm": 0.36120033264160156,
1688
+ "learning_rate": 6.461697264616973e-06,
1689
+ "loss": 2.3397,
1690
+ "step": 4800
1691
+ },
1692
+ {
1693
+ "epoch": 0.35537860355378603,
1694
+ "grad_norm": 0.40259799361228943,
1695
+ "learning_rate": 6.446951264469513e-06,
1696
+ "loss": 2.3443,
1697
+ "step": 4820
1698
+ },
1699
+ {
1700
+ "epoch": 0.35685320356853206,
1701
+ "grad_norm": 0.370540976524353,
1702
+ "learning_rate": 6.432205264322053e-06,
1703
+ "loss": 2.3494,
1704
+ "step": 4840
1705
+ },
1706
+ {
1707
+ "epoch": 0.35832780358327804,
1708
+ "grad_norm": 0.7460556626319885,
1709
+ "learning_rate": 6.417459264174593e-06,
1710
+ "loss": 2.3452,
1711
+ "step": 4860
1712
+ },
1713
+ {
1714
+ "epoch": 0.359802403598024,
1715
+ "grad_norm": 0.3665451109409332,
1716
+ "learning_rate": 6.4027132640271325e-06,
1717
+ "loss": 2.3693,
1718
+ "step": 4880
1719
+ },
1720
+ {
1721
+ "epoch": 0.36127700361277004,
1722
+ "grad_norm": 0.38104239106178284,
1723
+ "learning_rate": 6.387967263879673e-06,
1724
+ "loss": 2.3568,
1725
+ "step": 4900
1726
+ },
1727
+ {
1728
+ "epoch": 0.362751603627516,
1729
+ "grad_norm": 0.36739110946655273,
1730
+ "learning_rate": 6.373221263732213e-06,
1731
+ "loss": 2.361,
1732
+ "step": 4920
1733
+ },
1734
+ {
1735
+ "epoch": 0.36422620364226205,
1736
+ "grad_norm": 0.405453085899353,
1737
+ "learning_rate": 6.358475263584753e-06,
1738
+ "loss": 2.3448,
1739
+ "step": 4940
1740
+ },
1741
+ {
1742
+ "epoch": 0.365700803657008,
1743
+ "grad_norm": 0.349317342042923,
1744
+ "learning_rate": 6.3437292634372925e-06,
1745
+ "loss": 2.3569,
1746
+ "step": 4960
1747
+ },
1748
+ {
1749
+ "epoch": 0.36717540367175405,
1750
+ "grad_norm": 0.3882797360420227,
1751
+ "learning_rate": 6.328983263289833e-06,
1752
+ "loss": 2.3457,
1753
+ "step": 4980
1754
+ },
1755
+ {
1756
+ "epoch": 0.3686500036865,
1757
+ "grad_norm": 0.42195364832878113,
1758
+ "learning_rate": 6.314237263142373e-06,
1759
+ "loss": 2.364,
1760
+ "step": 5000
1761
+ }
1762
+ ],
1763
+ "logging_steps": 20,
1764
+ "max_steps": 13563,
1765
+ "num_input_tokens_seen": 0,
1766
+ "num_train_epochs": 1,
1767
+ "save_steps": 5000,
1768
+ "stateful_callbacks": {
1769
+ "TrainerControl": {
1770
+ "args": {
1771
+ "should_epoch_stop": false,
1772
+ "should_evaluate": false,
1773
+ "should_log": false,
1774
+ "should_save": true,
1775
+ "should_training_stop": false
1776
+ },
1777
+ "attributes": {}
1778
+ }
1779
+ },
1780
+ "total_flos": 2.138201686802432e+18,
1781
+ "train_batch_size": 24,
1782
+ "trial_name": null,
1783
+ "trial_params": null
1784
+ }
dataset_info.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "description": "High-quality text dataset for edge language model training",
3
+ "size": "8.1 GB",
4
+ "instances": "2,604,072",
5
+ "estimated_tokens": "~0.8B",
6
+ "format": "JSONL format"
7
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "padding_side": "right",
39
+ "sp_model_kwargs": {},
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false
43
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "total_flos": 5.80008589556659e+18,
4
+ "train_loss": 2.354225961330374,
5
+ "train_runtime": 12020.9549,
6
+ "train_samples": 2604072,
7
+ "train_samples_per_second": 216.628,
8
+ "train_steps_per_second": 1.128
9
+ }
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff