Hugging Face's logo

bitext
/
OpenELM-450M_Banking

Text Generation
Transformers
Safetensors
openelm
conversational
custom_code
Model card Files Community
malmarjeh commited on
Commit
4109443
•
1 Parent(s): fde37e0

Upload folder using huggingface_hub

Browse files
Files changed (12) hide show
  1. added_tokens.json +4 -0
  2. config.json +102 -0
  3. generation_config.json +7 -0
  4. model.safetensors +3 -0
  5. optimizer.pt +3 -0
  6. rng_state.pth +3 -0
  7. scheduler.pt +3 -0
  8. special_tokens_map.json +28 -0
  9. tokenizer.model +3 -0
  10. tokenizer_config.json +64 -0
  11. trainer_state.json +3213 -0
  12. training_args.bin +3 -0
added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "<|im_end|>": 32001,
3
+ "<|im_start|>": 32000
4
+ }
config.json ADDED
@@ -0,0 +1,102 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "apple/OpenELM-450M",
3
+ "activation_fn_name": "swish",
4
+ "architectures": [
5
+ "OpenELMForCausalLM"
6
+ ],
7
+ "auto_map": {
8
+ "AutoConfig": "apple/OpenELM-450M--configuration_openelm.OpenELMConfig",
9
+ "AutoModelForCausalLM": "apple/OpenELM-450M--modeling_openelm.OpenELMForCausalLM"
10
+ },
11
+ "bos_token_id": 32000,
12
+ "eos_token_id": 32001,
13
+ "ffn_dim_divisor": 256,
14
+ "ffn_multipliers": [
15
+ 0.5,
16
+ 0.68,
17
+ 0.87,
18
+ 1.05,
19
+ 1.24,
20
+ 1.42,
21
+ 1.61,
22
+ 1.79,
23
+ 1.97,
24
+ 2.16,
25
+ 2.34,
26
+ 2.53,
27
+ 2.71,
28
+ 2.89,
29
+ 3.08,
30
+ 3.26,
31
+ 3.45,
32
+ 3.63,
33
+ 3.82,
34
+ 4.0
35
+ ],
36
+ "ffn_with_glu": true,
37
+ "head_dim": 64,
38
+ "initializer_range": 0.02,
39
+ "max_context_length": 2048,
40
+ "model_dim": 1536,
41
+ "model_type": "openelm",
42
+ "normalization_layer_name": "rms_norm",
43
+ "normalize_qk_projections": true,
44
+ "num_gqa_groups": 4,
45
+ "num_kv_heads": [
46
+ 3,
47
+ 3,
48
+ 3,
49
+ 4,
50
+ 4,
51
+ 4,
52
+ 4,
53
+ 4,
54
+ 4,
55
+ 4,
56
+ 5,
57
+ 5,
58
+ 5,
59
+ 5,
60
+ 5,
61
+ 5,
62
+ 6,
63
+ 6,
64
+ 6,
65
+ 6
66
+ ],
67
+ "num_query_heads": [
68
+ 12,
69
+ 12,
70
+ 12,
71
+ 16,
72
+ 16,
73
+ 16,
74
+ 16,
75
+ 16,
76
+ 16,
77
+ 16,
78
+ 20,
79
+ 20,
80
+ 20,
81
+ 20,
82
+ 20,
83
+ 20,
84
+ 24,
85
+ 24,
86
+ 24,
87
+ 24
88
+ ],
89
+ "num_transformer_layers": 20,
90
+ "pad_token_id": 32001,
91
+ "qkv_multipliers": [
92
+ 0.5,
93
+ 1.0
94
+ ],
95
+ "rope_freq_constant": 10000,
96
+ "rope_max_length": 4096,
97
+ "share_input_output_layers": true,
98
+ "torch_dtype": "bfloat16",
99
+ "transformers_version": "4.42.4",
100
+ "use_cache": true,
101
+ "vocab_size": 32002
102
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 32000,
4
+ "eos_token_id": 32001,
5
+ "pad_token_id": 32001,
6
+ "transformers_version": "4.42.4"
7
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3528a6c5a49dcf652ebeaa860d25e66dc1f398d20f489412366682650ec1a6b8
3
+ size 914382688
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:510cf3e13114b6b75ea49f229f564cb57e1c7a5bdb97c04bc6c82b51cfdb2baa
3
+ size 916689914
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d138cfe3a4adf21f048848ee35837c9a757a0a3616ff7adbb45b69aac247435
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d208a65231ad5d02cea5427d3b30b1572b39ecf33b694c1047d9f0f88f8d120
3
+ size 1064
special_tokens_map.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ {
4
+ "content": "<|im_start|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ {
11
+ "content": "<|im_end|>",
12
+ "lstrip": false,
13
+ "normalized": false,
14
+ "rstrip": false,
15
+ "single_word": false
16
+ }
17
+ ],
18
+ "bos_token": "<|im_start|>",
19
+ "eos_token": "<|im_end|>",
20
+ "pad_token": "<unk>",
21
+ "unk_token": {
22
+ "content": "<unk>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ }
28
+ }
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
tokenizer_config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "<|im_start|>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ },
38
+ "32001": {
39
+ "content": "<|im_end|>",
40
+ "lstrip": false,
41
+ "normalized": false,
42
+ "rstrip": false,
43
+ "single_word": false,
44
+ "special": true
45
+ }
46
+ },
47
+ "additional_special_tokens": [
48
+ "<|im_start|>",
49
+ "<|im_end|>"
50
+ ],
51
+ "bos_token": "<|im_start|>",
52
+ "chat_template": "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
53
+ "clean_up_tokenization_spaces": false,
54
+ "eos_token": "<|im_end|>",
55
+ "legacy": false,
56
+ "model_max_length": 1000000000000000019884624838656,
57
+ "pad_token": "<unk>",
58
+ "padding_side": "right",
59
+ "sp_model_kwargs": {},
60
+ "spaces_between_special_tokens": false,
61
+ "tokenizer_class": "LlamaTokenizer",
62
+ "unk_token": "<unk>",
63
+ "use_default_system_prompt": false
64
+ }
trainer_state.json ADDED
@@ -0,0 +1,3213 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
+ "eval_steps": 500,
6
+ "global_step": 452,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.004424778761061947,
13
+ "grad_norm": 1624.0,
14
+ "learning_rate": 9.523809523809523e-06,
15
+ "loss": 18.8236,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.008849557522123894,
20
+ "grad_norm": 1632.0,
21
+ "learning_rate": 1.9047619047619046e-05,
22
+ "loss": 18.845,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.01327433628318584,
27
+ "grad_norm": 1104.0,
28
+ "learning_rate": 2.857142857142857e-05,
29
+ "loss": 13.897,
30
+ "step": 3
31
+ },
32
+ {
33
+ "epoch": 0.017699115044247787,
34
+ "grad_norm": 556.0,
35
+ "learning_rate": 3.809523809523809e-05,
36
+ "loss": 12.7695,
37
+ "step": 4
38
+ },
39
+ {
40
+ "epoch": 0.022123893805309734,
41
+ "grad_norm": 352.0,
42
+ "learning_rate": 4.761904761904762e-05,
43
+ "loss": 12.0053,
44
+ "step": 5
45
+ },
46
+ {
47
+ "epoch": 0.02654867256637168,
48
+ "grad_norm": 121.5,
49
+ "learning_rate": 5.714285714285714e-05,
50
+ "loss": 9.8363,
51
+ "step": 6
52
+ },
53
+ {
54
+ "epoch": 0.030973451327433628,
55
+ "grad_norm": 93.0,
56
+ "learning_rate": 6.666666666666667e-05,
57
+ "loss": 8.3813,
58
+ "step": 7
59
+ },
60
+ {
61
+ "epoch": 0.035398230088495575,
62
+ "grad_norm": 32.75,
63
+ "learning_rate": 7.619047619047618e-05,
64
+ "loss": 7.5272,
65
+ "step": 8
66
+ },
67
+ {
68
+ "epoch": 0.03982300884955752,
69
+ "grad_norm": 54.75,
70
+ "learning_rate": 8.571428571428571e-05,
71
+ "loss": 7.1281,
72
+ "step": 9
73
+ },
74
+ {
75
+ "epoch": 0.04424778761061947,
76
+ "grad_norm": 20.125,
77
+ "learning_rate": 9.523809523809524e-05,
78
+ "loss": 6.8148,
79
+ "step": 10
80
+ },
81
+ {
82
+ "epoch": 0.048672566371681415,
83
+ "grad_norm": 43.25,
84
+ "learning_rate": 0.00010476190476190477,
85
+ "loss": 6.8122,
86
+ "step": 11
87
+ },
88
+ {
89
+ "epoch": 0.05309734513274336,
90
+ "grad_norm": 7.46875,
91
+ "learning_rate": 0.00011428571428571428,
92
+ "loss": 6.4345,
93
+ "step": 12
94
+ },
95
+ {
96
+ "epoch": 0.05752212389380531,
97
+ "grad_norm": 17.75,
98
+ "learning_rate": 0.0001238095238095238,
99
+ "loss": 6.4061,
100
+ "step": 13
101
+ },
102
+ {
103
+ "epoch": 0.061946902654867256,
104
+ "grad_norm": 34.25,
105
+ "learning_rate": 0.00013333333333333334,
106
+ "loss": 6.4773,
107
+ "step": 14
108
+ },
109
+ {
110
+ "epoch": 0.06637168141592921,
111
+ "grad_norm": 11.9375,
112
+ "learning_rate": 0.00014285714285714287,
113
+ "loss": 6.2721,
114
+ "step": 15
115
+ },
116
+ {
117
+ "epoch": 0.07079646017699115,
118
+ "grad_norm": 13.25,
119
+ "learning_rate": 0.00015238095238095237,
120
+ "loss": 6.1657,
121
+ "step": 16
122
+ },
123
+ {
124
+ "epoch": 0.0752212389380531,
125
+ "grad_norm": 10.125,
126
+ "learning_rate": 0.00016190476190476192,
127
+ "loss": 5.9112,
128
+ "step": 17
129
+ },
130
+ {
131
+ "epoch": 0.07964601769911504,
132
+ "grad_norm": 13.875,
133
+ "learning_rate": 0.00017142857142857143,
134
+ "loss": 5.3251,
135
+ "step": 18
136
+ },
137
+ {
138
+ "epoch": 0.084070796460177,
139
+ "grad_norm": 179.0,
140
+ "learning_rate": 0.00018095238095238095,
141
+ "loss": 7.1923,
142
+ "step": 19
143
+ },
144
+ {
145
+ "epoch": 0.08849557522123894,
146
+ "grad_norm": 26.875,
147
+ "learning_rate": 0.00019047619047619048,
148
+ "loss": 4.1209,
149
+ "step": 20
150
+ },
151
+ {
152
+ "epoch": 0.09292035398230089,
153
+ "grad_norm": 43.25,
154
+ "learning_rate": 0.0002,
155
+ "loss": 3.2117,
156
+ "step": 21
157
+ },
158
+ {
159
+ "epoch": 0.09734513274336283,
160
+ "grad_norm": 11.25,
161
+ "learning_rate": 0.00019999885675796823,
162
+ "loss": 2.2957,
163
+ "step": 22
164
+ },
165
+ {
166
+ "epoch": 0.10176991150442478,
167
+ "grad_norm": 9.25,
168
+ "learning_rate": 0.00019999542705801296,
169
+ "loss": 1.8052,
170
+ "step": 23
171
+ },
172
+ {
173
+ "epoch": 0.10619469026548672,
174
+ "grad_norm": 5.75,
175
+ "learning_rate": 0.0001999897109785537,
176
+ "loss": 1.5344,
177
+ "step": 24
178
+ },
179
+ {
180
+ "epoch": 0.11061946902654868,
181
+ "grad_norm": 2.234375,
182
+ "learning_rate": 0.00019998170865028774,
183
+ "loss": 1.3085,
184
+ "step": 25
185
+ },
186
+ {
187
+ "epoch": 0.11504424778761062,
188
+ "grad_norm": 2.71875,
189
+ "learning_rate": 0.00019997142025618701,
190
+ "loss": 1.2024,
191
+ "step": 26
192
+ },
193
+ {
194
+ "epoch": 0.11946902654867257,
195
+ "grad_norm": 2.09375,
196
+ "learning_rate": 0.00019995884603149402,
197
+ "loss": 1.1342,
198
+ "step": 27
199
+ },
200
+ {
201
+ "epoch": 0.12389380530973451,
202
+ "grad_norm": 1.5234375,
203
+ "learning_rate": 0.00019994398626371643,
204
+ "loss": 1.073,
205
+ "step": 28
206
+ },
207
+ {
208
+ "epoch": 0.12831858407079647,
209
+ "grad_norm": 1.6953125,
210
+ "learning_rate": 0.00019992684129262038,
211
+ "loss": 0.9815,
212
+ "step": 29
213
+ },
214
+ {
215
+ "epoch": 0.13274336283185842,
216
+ "grad_norm": 1.28125,
217
+ "learning_rate": 0.00019990741151022301,
218
+ "loss": 0.9502,
219
+ "step": 30
220
+ },
221
+ {
222
+ "epoch": 0.13716814159292035,
223
+ "grad_norm": 0.953125,
224
+ "learning_rate": 0.00019988569736078312,
225
+ "loss": 0.8713,
226
+ "step": 31
227
+ },
228
+ {
229
+ "epoch": 0.1415929203539823,
230
+ "grad_norm": 1.0546875,
231
+ "learning_rate": 0.00019986169934079135,
232
+ "loss": 0.8431,
233
+ "step": 32
234
+ },
235
+ {
236
+ "epoch": 0.14601769911504425,
237
+ "grad_norm": 0.734375,
238
+ "learning_rate": 0.0001998354179989585,
239
+ "loss": 0.7957,
240
+ "step": 33
241
+ },
242
+ {
243
+ "epoch": 0.1504424778761062,
244
+ "grad_norm": 0.875,
245
+ "learning_rate": 0.00019980685393620337,
246
+ "loss": 0.8088,
247
+ "step": 34
248
+ },
249
+ {
250
+ "epoch": 0.15486725663716813,
251
+ "grad_norm": 0.8046875,
252
+ "learning_rate": 0.00019977600780563863,
253
+ "loss": 0.7827,
254
+ "step": 35
255
+ },
256
+ {
257
+ "epoch": 0.1592920353982301,
258
+ "grad_norm": 0.75390625,
259
+ "learning_rate": 0.00019974288031255618,
260
+ "loss": 0.7153,
261
+ "step": 36
262
+ },
263
+ {
264
+ "epoch": 0.16371681415929204,
265
+ "grad_norm": 0.74609375,
266
+ "learning_rate": 0.00019970747221441083,
267
+ "loss": 0.7449,
268
+ "step": 37
269
+ },
270
+ {
271
+ "epoch": 0.168141592920354,
272
+ "grad_norm": 0.69921875,
273
+ "learning_rate": 0.00019966978432080316,
274
+ "loss": 0.734,
275
+ "step": 38
276
+ },
277
+ {
278
+ "epoch": 0.17256637168141592,
279
+ "grad_norm": 0.671875,
280
+ "learning_rate": 0.00019962981749346078,
281
+ "loss": 0.7157,
282
+ "step": 39
283
+ },
284
+ {
285
+ "epoch": 0.17699115044247787,
286
+ "grad_norm": 0.640625,
287
+ "learning_rate": 0.0001995875726462189,
288
+ "loss": 0.733,
289
+ "step": 40
290
+ },
291
+ {
292
+ "epoch": 0.18141592920353983,
293
+ "grad_norm": 0.65234375,
294
+ "learning_rate": 0.00019954305074499916,
295
+ "loss": 0.6717,
296
+ "step": 41
297
+ },
298
+ {
299
+ "epoch": 0.18584070796460178,
300
+ "grad_norm": 0.57421875,
301
+ "learning_rate": 0.00019949625280778777,
302
+ "loss": 0.7242,
303
+ "step": 42
304
+ },
305
+ {
306
+ "epoch": 0.1902654867256637,
307
+ "grad_norm": 0.625,
308
+ "learning_rate": 0.00019944717990461207,
309
+ "loss": 0.6803,
310
+ "step": 43
311
+ },
312
+ {
313
+ "epoch": 0.19469026548672566,
314
+ "grad_norm": 0.578125,
315
+ "learning_rate": 0.00019939583315751623,
316
+ "loss": 0.6824,
317
+ "step": 44
318
+ },
319
+ {
320
+ "epoch": 0.19911504424778761,
321
+ "grad_norm": 0.578125,
322
+ "learning_rate": 0.0001993422137405354,
323
+ "loss": 0.669,
324
+ "step": 45
325
+ },
326
+ {
327
+ "epoch": 0.20353982300884957,
328
+ "grad_norm": 0.56640625,
329
+ "learning_rate": 0.000199286322879669,
330
+ "loss": 0.6621,
331
+ "step": 46
332
+ },
333
+ {
334
+ "epoch": 0.2079646017699115,
335
+ "grad_norm": 0.609375,
336
+ "learning_rate": 0.00019922816185285265,
337
+ "loss": 0.6433,
338
+ "step": 47
339
+ },
340
+ {
341
+ "epoch": 0.21238938053097345,
342
+ "grad_norm": 0.53515625,
343
+ "learning_rate": 0.000199167731989929,
344
+ "loss": 0.647,
345
+ "step": 48
346
+ },
347
+ {
348
+ "epoch": 0.2168141592920354,
349
+ "grad_norm": 0.6171875,
350
+ "learning_rate": 0.00019910503467261722,
351
+ "loss": 0.642,
352
+ "step": 49
353
+ },
354
+ {
355
+ "epoch": 0.22123893805309736,
356
+ "grad_norm": 0.53515625,
357
+ "learning_rate": 0.00019904007133448147,
358
+ "loss": 0.6283,
359
+ "step": 50
360
+ },
361
+ {
362
+ "epoch": 0.22566371681415928,
363
+ "grad_norm": 0.6171875,
364
+ "learning_rate": 0.0001989728434608981,
365
+ "loss": 0.6374,
366
+ "step": 51
367
+ },
368
+ {
369
+ "epoch": 0.23008849557522124,
370
+ "grad_norm": 0.60546875,
371
+ "learning_rate": 0.00019890335258902178,
372
+ "loss": 0.6671,
373
+ "step": 52
374
+ },
375
+ {
376
+ "epoch": 0.2345132743362832,
377
+ "grad_norm": 0.515625,
378
+ "learning_rate": 0.00019883160030775016,
379
+ "loss": 0.6303,
380
+ "step": 53
381
+ },
382
+ {
383
+ "epoch": 0.23893805309734514,
384
+ "grad_norm": 0.62890625,
385
+ "learning_rate": 0.0001987575882576878,
386
+ "loss": 0.6423,
387
+ "step": 54
388
+ },
389
+ {
390
+ "epoch": 0.24336283185840707,
391
+ "grad_norm": 0.54296875,
392
+ "learning_rate": 0.00019868131813110832,
393
+ "loss": 0.6428,
394
+ "step": 55
395
+ },
396
+ {
397
+ "epoch": 0.24778761061946902,
398
+ "grad_norm": 0.515625,
399
+ "learning_rate": 0.0001986027916719161,
400
+ "loss": 0.6105,
401
+ "step": 56
402
+ },
403
+ {
404
+ "epoch": 0.252212389380531,
405
+ "grad_norm": 0.5390625,
406
+ "learning_rate": 0.00019852201067560606,
407
+ "loss": 0.6109,
408
+ "step": 57
409
+ },
410
+ {
411
+ "epoch": 0.25663716814159293,
412
+ "grad_norm": 0.58984375,
413
+ "learning_rate": 0.00019843897698922284,
414
+ "loss": 0.5753,
415
+ "step": 58
416
+ },
417
+ {
418
+ "epoch": 0.2610619469026549,
419
+ "grad_norm": 0.52734375,
420
+ "learning_rate": 0.00019835369251131846,
421
+ "loss": 0.607,
422
+ "step": 59
423
+ },
424
+ {
425
+ "epoch": 0.26548672566371684,
426
+ "grad_norm": 0.515625,
427
+ "learning_rate": 0.00019826615919190887,
428
+ "loss": 0.6029,
429
+ "step": 60
430
+ },
431
+ {
432
+ "epoch": 0.26991150442477874,
433
+ "grad_norm": 0.51171875,
434
+ "learning_rate": 0.0001981763790324295,
435
+ "loss": 0.5668,
436
+ "step": 61
437
+ },
438
+ {
439
+ "epoch": 0.2743362831858407,
440
+ "grad_norm": 0.515625,
441
+ "learning_rate": 0.00019808435408568938,
442
+ "loss": 0.6282,
443
+ "step": 62
444
+ },
445
+ {
446
+ "epoch": 0.27876106194690264,
447
+ "grad_norm": 0.515625,
448
+ "learning_rate": 0.0001979900864558242,
449
+ "loss": 0.5816,
450
+ "step": 63
451
+ },
452
+ {
453
+ "epoch": 0.2831858407079646,
454
+ "grad_norm": 0.53125,
455
+ "learning_rate": 0.0001978935782982484,
456
+ "loss": 0.6071,
457
+ "step": 64
458
+ },
459
+ {
460
+ "epoch": 0.28761061946902655,
461
+ "grad_norm": 0.48046875,
462
+ "learning_rate": 0.00019779483181960557,
463
+ "loss": 0.56,
464
+ "step": 65
465
+ },
466
+ {
467
+ "epoch": 0.2920353982300885,
468
+ "grad_norm": 0.498046875,
469
+ "learning_rate": 0.0001976938492777182,
470
+ "loss": 0.6138,
471
+ "step": 66
472
+ },
473
+ {
474
+ "epoch": 0.29646017699115046,
475
+ "grad_norm": 0.486328125,
476
+ "learning_rate": 0.00019759063298153598,
477
+ "loss": 0.5735,
478
+ "step": 67
479
+ },
480
+ {
481
+ "epoch": 0.3008849557522124,
482
+ "grad_norm": 0.5234375,
483
+ "learning_rate": 0.00019748518529108316,
484
+ "loss": 0.5871,
485
+ "step": 68
486
+ },
487
+ {
488
+ "epoch": 0.3053097345132743,
489
+ "grad_norm": 0.5234375,
490
+ "learning_rate": 0.00019737750861740431,
491
+ "loss": 0.6027,
492
+ "step": 69
493
+ },
494
+ {
495
+ "epoch": 0.30973451327433627,
496
+ "grad_norm": 0.5,
497
+ "learning_rate": 0.00019726760542250946,
498
+ "loss": 0.6089,
499
+ "step": 70
500
+ },
501
+ {
502
+ "epoch": 0.3141592920353982,
503
+ "grad_norm": 0.474609375,
504
+ "learning_rate": 0.0001971554782193176,
505
+ "loss": 0.5682,
506
+ "step": 71
507
+ },
508
+ {
509
+ "epoch": 0.3185840707964602,
510
+ "grad_norm": 0.474609375,
511
+ "learning_rate": 0.0001970411295715994,
512
+ "loss": 0.5707,
513
+ "step": 72
514
+ },
515
+ {
516
+ "epoch": 0.3230088495575221,
517
+ "grad_norm": 0.515625,
518
+ "learning_rate": 0.00019692456209391846,
519
+ "loss": 0.5949,
520
+ "step": 73
521
+ },
522
+ {
523
+ "epoch": 0.3274336283185841,
524
+ "grad_norm": 0.52734375,
525
+ "learning_rate": 0.00019680577845157155,
526
+ "loss": 0.577,
527
+ "step": 74
528
+ },
529
+ {
530
+ "epoch": 0.33185840707964603,
531
+ "grad_norm": 0.69140625,
532
+ "learning_rate": 0.00019668478136052774,
533
+ "loss": 0.5871,
534
+ "step": 75
535
+ },
536
+ {
537
+ "epoch": 0.336283185840708,
538
+ "grad_norm": 0.5078125,
539
+ "learning_rate": 0.00019656157358736626,
540
+ "loss": 0.5763,
541
+ "step": 76
542
+ },
543
+ {
544
+ "epoch": 0.3407079646017699,
545
+ "grad_norm": 0.498046875,
546
+ "learning_rate": 0.0001964361579492132,
547
+ "loss": 0.5875,
548
+ "step": 77
549
+ },
550
+ {
551
+ "epoch": 0.34513274336283184,
552
+ "grad_norm": 0.45703125,
553
+ "learning_rate": 0.00019630853731367713,
554
+ "loss": 0.5811,
555
+ "step": 78
556
+ },
557
+ {
558
+ "epoch": 0.3495575221238938,
559
+ "grad_norm": 0.5625,
560
+ "learning_rate": 0.0001961787145987835,
561
+ "loss": 0.6122,
562
+ "step": 79
563
+ },
564
+ {
565
+ "epoch": 0.35398230088495575,
566
+ "grad_norm": 0.4765625,
567
+ "learning_rate": 0.00019604669277290806,
568
+ "loss": 0.5584,
569
+ "step": 80
570
+ },
571
+ {
572
+ "epoch": 0.3584070796460177,
573
+ "grad_norm": 0.46875,
574
+ "learning_rate": 0.0001959124748547088,
575
+ "loss": 0.5629,
576
+ "step": 81
577
+ },
578
+ {
579
+ "epoch": 0.36283185840707965,
580
+ "grad_norm": 0.50390625,
581
+ "learning_rate": 0.00019577606391305702,
582
+ "loss": 0.5758,
583
+ "step": 82
584
+ },
585
+ {
586
+ "epoch": 0.3672566371681416,
587
+ "grad_norm": 0.478515625,
588
+ "learning_rate": 0.0001956374630669672,
589
+ "loss": 0.5383,
590
+ "step": 83
591
+ },
592
+ {
593
+ "epoch": 0.37168141592920356,
594
+ "grad_norm": 0.46484375,
595
+ "learning_rate": 0.00019549667548552556,
596
+ "loss": 0.5869,
597
+ "step": 84
598
+ },
599
+ {
600
+ "epoch": 0.37610619469026546,
601
+ "grad_norm": 0.51953125,
602
+ "learning_rate": 0.00019535370438781766,
603
+ "loss": 0.5642,
604
+ "step": 85
605
+ },
606
+ {
607
+ "epoch": 0.3805309734513274,
608
+ "grad_norm": 0.498046875,
609
+ "learning_rate": 0.000195208553042855,
610
+ "loss": 0.5895,
611
+ "step": 86
612
+ },
613
+ {
614
+ "epoch": 0.38495575221238937,
615
+ "grad_norm": 0.48828125,
616
+ "learning_rate": 0.00019506122476949981,
617
+ "loss": 0.5744,
618
+ "step": 87
619
+ },
620
+ {
621
+ "epoch": 0.3893805309734513,
622
+ "grad_norm": 0.5078125,
623
+ "learning_rate": 0.00019491172293638968,
624
+ "loss": 0.5537,
625
+ "step": 88
626
+ },
627
+ {
628
+ "epoch": 0.3938053097345133,
629
+ "grad_norm": 0.50390625,
630
+ "learning_rate": 0.00019476005096186015,
631
+ "loss": 0.5748,
632
+ "step": 89
633
+ },
634
+ {
635
+ "epoch": 0.39823008849557523,
636
+ "grad_norm": 0.486328125,
637
+ "learning_rate": 0.00019460621231386676,
638
+ "loss": 0.5511,
639
+ "step": 90
640
+ },
641
+ {
642
+ "epoch": 0.4026548672566372,
643
+ "grad_norm": 0.470703125,
644
+ "learning_rate": 0.00019445021050990571,
645
+ "loss": 0.5636,
646
+ "step": 91
647
+ },
648
+ {
649
+ "epoch": 0.40707964601769914,
650
+ "grad_norm": 0.482421875,
651
+ "learning_rate": 0.00019429204911693333,
652
+ "loss": 0.5649,
653
+ "step": 92
654
+ },
655
+ {
656
+ "epoch": 0.41150442477876104,
657
+ "grad_norm": 0.49609375,
658
+ "learning_rate": 0.00019413173175128473,
659
+ "loss": 0.5468,
660
+ "step": 93
661
+ },
662
+ {
663
+ "epoch": 0.415929203539823,
664
+ "grad_norm": 0.45703125,
665
+ "learning_rate": 0.00019396926207859084,
666
+ "loss": 0.5347,
667
+ "step": 94
668
+ },
669
+ {
670
+ "epoch": 0.42035398230088494,
671
+ "grad_norm": 0.5078125,
672
+ "learning_rate": 0.00019380464381369493,
673
+ "loss": 0.5667,
674
+ "step": 95
675
+ },
676
+ {
677
+ "epoch": 0.4247787610619469,
678
+ "grad_norm": 0.490234375,
679
+ "learning_rate": 0.0001936378807205673,
680
+ "loss": 0.5644,
681
+ "step": 96
682
+ },
683
+ {
684
+ "epoch": 0.42920353982300885,
685
+ "grad_norm": 0.4765625,
686
+ "learning_rate": 0.00019346897661221956,
687
+ "loss": 0.5384,
688
+ "step": 97
689
+ },
690
+ {
691
+ "epoch": 0.4336283185840708,
692
+ "grad_norm": 0.47265625,
693
+ "learning_rate": 0.00019329793535061723,
694
+ "loss": 0.5569,
695
+ "step": 98
696
+ },
697
+ {
698
+ "epoch": 0.43805309734513276,
699
+ "grad_norm": 0.48046875,
700
+ "learning_rate": 0.0001931247608465915,
701
+ "loss": 0.5494,
702
+ "step": 99
703
+ },
704
+ {
705
+ "epoch": 0.4424778761061947,
706
+ "grad_norm": 0.427734375,
707
+ "learning_rate": 0.00019294945705974973,
708
+ "loss": 0.536,
709
+ "step": 100
710
+ },
711
+ {
712
+ "epoch": 0.4469026548672566,
713
+ "grad_norm": 0.447265625,
714
+ "learning_rate": 0.00019277202799838518,
715
+ "loss": 0.5319,
716
+ "step": 101
717
+ },
718
+ {
719
+ "epoch": 0.45132743362831856,
720
+ "grad_norm": 0.4921875,
721
+ "learning_rate": 0.000192592477719385,
722
+ "loss": 0.5649,
723
+ "step": 102
724
+ },
725
+ {
726
+ "epoch": 0.4557522123893805,
727
+ "grad_norm": 0.474609375,
728
+ "learning_rate": 0.00019241081032813772,
729
+ "loss": 0.5652,
730
+ "step": 103
731
+ },
732
+ {
733
+ "epoch": 0.46017699115044247,
734
+ "grad_norm": 0.462890625,
735
+ "learning_rate": 0.00019222702997843927,
736
+ "loss": 0.5222,
737
+ "step": 104
738
+ },
739
+ {
740
+ "epoch": 0.4646017699115044,
741
+ "grad_norm": 0.4765625,
742
+ "learning_rate": 0.00019204114087239806,
743
+ "loss": 0.551,
744
+ "step": 105
745
+ },
746
+ {
747
+ "epoch": 0.4690265486725664,
748
+ "grad_norm": 0.4921875,
749
+ "learning_rate": 0.00019185314726033893,
750
+ "loss": 0.555,
751
+ "step": 106
752
+ },
753
+ {
754
+ "epoch": 0.47345132743362833,
755
+ "grad_norm": 0.470703125,
756
+ "learning_rate": 0.00019166305344070578,
757
+ "loss": 0.5404,
758
+ "step": 107
759
+ },
760
+ {
761
+ "epoch": 0.4778761061946903,
762
+ "grad_norm": 0.494140625,
763
+ "learning_rate": 0.0001914708637599636,
764
+ "loss": 0.5462,
765
+ "step": 108
766
+ },
767
+ {
768
+ "epoch": 0.4823008849557522,
769
+ "grad_norm": 0.79296875,
770
+ "learning_rate": 0.0001912765826124987,
771
+ "loss": 0.5532,
772
+ "step": 109
773
+ },
774
+ {
775
+ "epoch": 0.48672566371681414,
776
+ "grad_norm": 0.55078125,
777
+ "learning_rate": 0.0001910802144405186,
778
+ "loss": 0.5417,
779
+ "step": 110
780
+ },
781
+ {
782
+ "epoch": 0.4911504424778761,
783
+ "grad_norm": 0.5,
784
+ "learning_rate": 0.0001908817637339503,
785
+ "loss": 0.5418,
786
+ "step": 111
787
+ },
788
+ {
789
+ "epoch": 0.49557522123893805,
790
+ "grad_norm": 0.515625,
791
+ "learning_rate": 0.00019068123503033753,
792
+ "loss": 0.5362,
793
+ "step": 112
794
+ },
795
+ {
796
+ "epoch": 0.5,
797
+ "grad_norm": 0.5,
798
+ "learning_rate": 0.00019047863291473717,
799
+ "loss": 0.5512,
800
+ "step": 113
801
+ },
802
+ {
803
+ "epoch": 0.504424778761062,
804
+ "grad_norm": 0.51171875,
805
+ "learning_rate": 0.0001902739620196143,
806
+ "loss": 0.5541,
807
+ "step": 114
808
+ },
809
+ {
810
+ "epoch": 0.5088495575221239,
811
+ "grad_norm": 0.5390625,
812
+ "learning_rate": 0.00019006722702473629,
813
+ "loss": 0.5138,
814
+ "step": 115
815
+ },
816
+ {
817
+ "epoch": 0.5132743362831859,
818
+ "grad_norm": 0.490234375,
819
+ "learning_rate": 0.00018985843265706588,
820
+ "loss": 0.5457,
821
+ "step": 116
822
+ },
823
+ {
824
+ "epoch": 0.5176991150442478,
825
+ "grad_norm": 0.5390625,
826
+ "learning_rate": 0.000189647583690653,
827
+ "loss": 0.5161,
828
+ "step": 117
829
+ },
830
+ {
831
+ "epoch": 0.5221238938053098,
832
+ "grad_norm": 0.48046875,
833
+ "learning_rate": 0.0001894346849465257,
834
+ "loss": 0.5442,
835
+ "step": 118
836
+ },
837
+ {
838
+ "epoch": 0.5265486725663717,
839
+ "grad_norm": 0.4375,
840
+ "learning_rate": 0.0001892197412925798,
841
+ "loss": 0.5079,
842
+ "step": 119
843
+ },
844
+ {
845
+ "epoch": 0.5309734513274337,
846
+ "grad_norm": 0.48828125,
847
+ "learning_rate": 0.00018900275764346768,
848
+ "loss": 0.4962,
849
+ "step": 120
850
+ },
851
+ {
852
+ "epoch": 0.5353982300884956,
853
+ "grad_norm": 0.5078125,
854
+ "learning_rate": 0.00018878373896048594,
855
+ "loss": 0.5248,
856
+ "step": 121
857
+ },
858
+ {
859
+ "epoch": 0.5398230088495575,
860
+ "grad_norm": 0.470703125,
861
+ "learning_rate": 0.0001885626902514618,
862
+ "loss": 0.5247,
863
+ "step": 122
864
+ },
865
+ {
866
+ "epoch": 0.5442477876106194,
867
+ "grad_norm": 0.498046875,
868
+ "learning_rate": 0.00018833961657063885,
869
+ "loss": 0.5397,
870
+ "step": 123
871
+ },
872
+ {
873
+ "epoch": 0.5486725663716814,
874
+ "grad_norm": 0.48046875,
875
+ "learning_rate": 0.0001881145230185612,
876
+ "loss": 0.5146,
877
+ "step": 124
878
+ },
879
+ {
880
+ "epoch": 0.5530973451327433,
881
+ "grad_norm": 0.46875,
882
+ "learning_rate": 0.00018788741474195706,
883
+ "loss": 0.5654,
884
+ "step": 125
885
+ },
886
+ {
887
+ "epoch": 0.5575221238938053,
888
+ "grad_norm": 0.45703125,
889
+ "learning_rate": 0.00018765829693362095,
890
+ "loss": 0.52,
891
+ "step": 126
892
+ },
893
+ {
894
+ "epoch": 0.5619469026548672,
895
+ "grad_norm": 0.423828125,
896
+ "learning_rate": 0.0001874271748322951,
897
+ "loss": 0.5543,
898
+ "step": 127
899
+ },
900
+ {
901
+ "epoch": 0.5663716814159292,
902
+ "grad_norm": 0.44140625,
903
+ "learning_rate": 0.00018719405372254948,
904
+ "loss": 0.5199,
905
+ "step": 128
906
+ },
907
+ {
908
+ "epoch": 0.5707964601769911,
909
+ "grad_norm": 0.4765625,
910
+ "learning_rate": 0.0001869589389346611,
911
+ "loss": 0.5064,
912
+ "step": 129
913
+ },
914
+ {
915
+ "epoch": 0.5752212389380531,
916
+ "grad_norm": 0.4921875,
917
+ "learning_rate": 0.00018672183584449216,
918
+ "loss": 0.5137,
919
+ "step": 130
920
+ },
921
+ {
922
+ "epoch": 0.5796460176991151,
923
+ "grad_norm": 0.4921875,
924
+ "learning_rate": 0.00018648274987336704,
925
+ "loss": 0.5058,
926
+ "step": 131
927
+ },
928
+ {
929
+ "epoch": 0.584070796460177,
930
+ "grad_norm": 0.458984375,
931
+ "learning_rate": 0.00018624168648794832,
932
+ "loss": 0.5264,
933
+ "step": 132
934
+ },
935
+ {
936
+ "epoch": 0.588495575221239,
937
+ "grad_norm": 0.4296875,
938
+ "learning_rate": 0.00018599865120011192,
939
+ "loss": 0.5229,
940
+ "step": 133
941
+ },
942
+ {
943
+ "epoch": 0.5929203539823009,
944
+ "grad_norm": 0.470703125,
945
+ "learning_rate": 0.00018575364956682094,
946
+ "loss": 0.546,
947
+ "step": 134
948
+ },
949
+ {
950
+ "epoch": 0.5973451327433629,
951
+ "grad_norm": 0.48046875,
952
+ "learning_rate": 0.00018550668718999872,
953
+ "loss": 0.5311,
954
+ "step": 135
955
+ },
956
+ {
957
+ "epoch": 0.6017699115044248,
958
+ "grad_norm": 0.47265625,
959
+ "learning_rate": 0.00018525776971640062,
960
+ "loss": 0.5043,
961
+ "step": 136
962
+ },
963
+ {
964
+ "epoch": 0.6061946902654868,
965
+ "grad_norm": 0.46875,
966
+ "learning_rate": 0.00018500690283748504,
967
+ "loss": 0.4815,
968
+ "step": 137
969
+ },
970
+ {
971
+ "epoch": 0.6106194690265486,
972
+ "grad_norm": 0.474609375,
973
+ "learning_rate": 0.00018475409228928312,
974
+ "loss": 0.517,
975
+ "step": 138
976
+ },
977
+ {
978
+ "epoch": 0.6150442477876106,
979
+ "grad_norm": 0.5859375,
980
+ "learning_rate": 0.0001844993438522678,
981
+ "loss": 0.513,
982
+ "step": 139
983
+ },
984
+ {
985
+ "epoch": 0.6194690265486725,
986
+ "grad_norm": 0.53125,
987
+ "learning_rate": 0.00018424266335122152,
988
+ "loss": 0.5296,
989
+ "step": 140
990
+ },
991
+ {
992
+ "epoch": 0.6238938053097345,
993
+ "grad_norm": 0.5,
994
+ "learning_rate": 0.000183984056655103,
995
+ "loss": 0.5186,
996
+ "step": 141
997
+ },
998
+ {
999
+ "epoch": 0.6283185840707964,
1000
+ "grad_norm": 0.4921875,
1001
+ "learning_rate": 0.0001837235296769131,
1002
+ "loss": 0.5239,
1003
+ "step": 142
1004
+ },
1005
+ {
1006
+ "epoch": 0.6327433628318584,
1007
+ "grad_norm": 0.4765625,
1008
+ "learning_rate": 0.00018346108837355972,
1009
+ "loss": 0.5024,
1010
+ "step": 143
1011
+ },
1012
+ {
1013
+ "epoch": 0.6371681415929203,
1014
+ "grad_norm": 0.45703125,
1015
+ "learning_rate": 0.0001831967387457214,
1016
+ "loss": 0.5144,
1017
+ "step": 144
1018
+ },
1019
+ {
1020
+ "epoch": 0.6415929203539823,
1021
+ "grad_norm": 0.484375,
1022
+ "learning_rate": 0.00018293048683771024,
1023
+ "loss": 0.5342,
1024
+ "step": 145
1025
+ },
1026
+ {
1027
+ "epoch": 0.6460176991150443,
1028
+ "grad_norm": 0.462890625,
1029
+ "learning_rate": 0.00018266233873733375,
1030
+ "loss": 0.5143,
1031
+ "step": 146
1032
+ },
1033
+ {
1034
+ "epoch": 0.6504424778761062,
1035
+ "grad_norm": 0.439453125,
1036
+ "learning_rate": 0.00018239230057575542,
1037
+ "loss": 0.5271,
1038
+ "step": 147
1039
+ },
1040
+ {
1041
+ "epoch": 0.6548672566371682,
1042
+ "grad_norm": 0.46484375,
1043
+ "learning_rate": 0.00018212037852735486,
1044
+ "loss": 0.5185,
1045
+ "step": 148
1046
+ },
1047
+ {
1048
+ "epoch": 0.6592920353982301,
1049
+ "grad_norm": 0.435546875,
1050
+ "learning_rate": 0.00018184657880958635,
1051
+ "loss": 0.5237,
1052
+ "step": 149
1053
+ },
1054
+ {
1055
+ "epoch": 0.6637168141592921,
1056
+ "grad_norm": 0.494140625,
1057
+ "learning_rate": 0.00018157090768283678,
1058
+ "loss": 0.5137,
1059
+ "step": 150
1060
+ },
1061
+ {
1062
+ "epoch": 0.668141592920354,
1063
+ "grad_norm": 0.482421875,
1064
+ "learning_rate": 0.00018129337145028255,
1065
+ "loss": 0.5284,
1066
+ "step": 151
1067
+ },
1068
+ {
1069
+ "epoch": 0.672566371681416,
1070
+ "grad_norm": 0.431640625,
1071
+ "learning_rate": 0.00018101397645774539,
1072
+ "loss": 0.5252,
1073
+ "step": 152
1074
+ },
1075
+ {
1076
+ "epoch": 0.6769911504424779,
1077
+ "grad_norm": 0.435546875,
1078
+ "learning_rate": 0.00018073272909354727,
1079
+ "loss": 0.5105,
1080
+ "step": 153
1081
+ },
1082
+ {
1083
+ "epoch": 0.6814159292035398,
1084
+ "grad_norm": 0.47265625,
1085
+ "learning_rate": 0.00018044963578836435,
1086
+ "loss": 0.5384,
1087
+ "step": 154
1088
+ },
1089
+ {
1090
+ "epoch": 0.6858407079646017,
1091
+ "grad_norm": 0.4140625,
1092
+ "learning_rate": 0.00018016470301507995,
1093
+ "loss": 0.5086,
1094
+ "step": 155
1095
+ },
1096
+ {
1097
+ "epoch": 0.6902654867256637,
1098
+ "grad_norm": 0.435546875,
1099
+ "learning_rate": 0.00017987793728863651,
1100
+ "loss": 0.5227,
1101
+ "step": 156
1102
+ },
1103
+ {
1104
+ "epoch": 0.6946902654867256,
1105
+ "grad_norm": 0.44921875,
1106
+ "learning_rate": 0.00017958934516588667,
1107
+ "loss": 0.4982,
1108
+ "step": 157
1109
+ },
1110
+ {
1111
+ "epoch": 0.6991150442477876,
1112
+ "grad_norm": 0.44921875,
1113
+ "learning_rate": 0.00017929893324544332,
1114
+ "loss": 0.5109,
1115
+ "step": 158
1116
+ },
1117
+ {
1118
+ "epoch": 0.7035398230088495,
1119
+ "grad_norm": 0.439453125,
1120
+ "learning_rate": 0.00017900670816752874,
1121
+ "loss": 0.5242,
1122
+ "step": 159
1123
+ },
1124
+ {
1125
+ "epoch": 0.7079646017699115,
1126
+ "grad_norm": 0.4453125,
1127
+ "learning_rate": 0.00017871267661382278,
1128
+ "loss": 0.5019,
1129
+ "step": 160
1130
+ },
1131
+ {
1132
+ "epoch": 0.7123893805309734,
1133
+ "grad_norm": 0.40234375,
1134
+ "learning_rate": 0.00017841684530731005,
1135
+ "loss": 0.5048,
1136
+ "step": 161
1137
+ },
1138
+ {
1139
+ "epoch": 0.7168141592920354,
1140
+ "grad_norm": 0.392578125,
1141
+ "learning_rate": 0.0001781192210121262,
1142
+ "loss": 0.4957,
1143
+ "step": 162
1144
+ },
1145
+ {
1146
+ "epoch": 0.7212389380530974,
1147
+ "grad_norm": 0.455078125,
1148
+ "learning_rate": 0.00017781981053340337,
1149
+ "loss": 0.5035,
1150
+ "step": 163
1151
+ },
1152
+ {
1153
+ "epoch": 0.7256637168141593,
1154
+ "grad_norm": 0.451171875,
1155
+ "learning_rate": 0.00017751862071711442,
1156
+ "loss": 0.5161,
1157
+ "step": 164
1158
+ },
1159
+ {
1160
+ "epoch": 0.7300884955752213,
1161
+ "grad_norm": 0.42578125,
1162
+ "learning_rate": 0.00017721565844991643,
1163
+ "loss": 0.4838,
1164
+ "step": 165
1165
+ },
1166
+ {
1167
+ "epoch": 0.7345132743362832,
1168
+ "grad_norm": 0.46484375,
1169
+ "learning_rate": 0.00017691093065899346,
1170
+ "loss": 0.5135,
1171
+ "step": 166
1172
+ },
1173
+ {
1174
+ "epoch": 0.7389380530973452,
1175
+ "grad_norm": 0.462890625,
1176
+ "learning_rate": 0.0001766044443118978,
1177
+ "loss": 0.5174,
1178
+ "step": 167
1179
+ },
1180
+ {
1181
+ "epoch": 0.7433628318584071,
1182
+ "grad_norm": 0.41796875,
1183
+ "learning_rate": 0.00017629620641639103,
1184
+ "loss": 0.4982,
1185
+ "step": 168
1186
+ },
1187
+ {
1188
+ "epoch": 0.7477876106194691,
1189
+ "grad_norm": 0.494140625,
1190
+ "learning_rate": 0.0001759862240202834,
1191
+ "loss": 0.4902,
1192
+ "step": 169
1193
+ },
1194
+ {
1195
+ "epoch": 0.7522123893805309,
1196
+ "grad_norm": 0.451171875,
1197
+ "learning_rate": 0.00017567450421127306,
1198
+ "loss": 0.4858,
1199
+ "step": 170
1200
+ },
1201
+ {
1202
+ "epoch": 0.7566371681415929,
1203
+ "grad_norm": 0.4453125,
1204
+ "learning_rate": 0.0001753610541167838,
1205
+ "loss": 0.5178,
1206
+ "step": 171
1207
+ },
1208
+ {
1209
+ "epoch": 0.7610619469026548,
1210
+ "grad_norm": 0.453125,
1211
+ "learning_rate": 0.00017504588090380197,
1212
+ "loss": 0.5209,
1213
+ "step": 172
1214
+ },
1215
+ {
1216
+ "epoch": 0.7654867256637168,
1217
+ "grad_norm": 0.435546875,
1218
+ "learning_rate": 0.00017472899177871297,
1219
+ "loss": 0.5042,
1220
+ "step": 173
1221
+ },
1222
+ {
1223
+ "epoch": 0.7699115044247787,
1224
+ "grad_norm": 0.4609375,
1225
+ "learning_rate": 0.00017441039398713608,
1226
+ "loss": 0.539,
1227
+ "step": 174
1228
+ },
1229
+ {
1230
+ "epoch": 0.7743362831858407,
1231
+ "grad_norm": 0.423828125,
1232
+ "learning_rate": 0.00017409009481375904,
1233
+ "loss": 0.5053,
1234
+ "step": 175
1235
+ },
1236
+ {
1237
+ "epoch": 0.7787610619469026,
1238
+ "grad_norm": 0.439453125,
1239
+ "learning_rate": 0.0001737681015821714,
1240
+ "loss": 0.5357,
1241
+ "step": 176
1242
+ },
1243
+ {
1244
+ "epoch": 0.7831858407079646,
1245
+ "grad_norm": 0.44921875,
1246
+ "learning_rate": 0.00017344442165469714,
1247
+ "loss": 0.5108,
1248
+ "step": 177
1249
+ },
1250
+ {
1251
+ "epoch": 0.7876106194690266,
1252
+ "grad_norm": 0.435546875,
1253
+ "learning_rate": 0.00017311906243222614,
1254
+ "loss": 0.4909,
1255
+ "step": 178
1256
+ },
1257
+ {
1258
+ "epoch": 0.7920353982300885,
1259
+ "grad_norm": 0.443359375,
1260
+ "learning_rate": 0.0001727920313540452,
1261
+ "loss": 0.506,
1262
+ "step": 179
1263
+ },
1264
+ {
1265
+ "epoch": 0.7964601769911505,
1266
+ "grad_norm": 0.455078125,
1267
+ "learning_rate": 0.00017246333589766787,
1268
+ "loss": 0.5186,
1269
+ "step": 180
1270
+ },
1271
+ {
1272
+ "epoch": 0.8008849557522124,
1273
+ "grad_norm": 0.439453125,
1274
+ "learning_rate": 0.00017213298357866326,
1275
+ "loss": 0.4941,
1276
+ "step": 181
1277
+ },
1278
+ {
1279
+ "epoch": 0.8053097345132744,
1280
+ "grad_norm": 0.455078125,
1281
+ "learning_rate": 0.00017180098195048458,
1282
+ "loss": 0.4811,
1283
+ "step": 182
1284
+ },
1285
+ {
1286
+ "epoch": 0.8097345132743363,
1287
+ "grad_norm": 0.435546875,
1288
+ "learning_rate": 0.00017146733860429612,
1289
+ "loss": 0.5137,
1290
+ "step": 183
1291
+ },
1292
+ {
1293
+ "epoch": 0.8141592920353983,
1294
+ "grad_norm": 0.455078125,
1295
+ "learning_rate": 0.00017113206116879982,
1296
+ "loss": 0.5152,
1297
+ "step": 184
1298
+ },
1299
+ {
1300
+ "epoch": 0.8185840707964602,
1301
+ "grad_norm": 0.427734375,
1302
+ "learning_rate": 0.00017079515731006085,
1303
+ "loss": 0.5052,
1304
+ "step": 185
1305
+ },
1306
+ {
1307
+ "epoch": 0.8230088495575221,
1308
+ "grad_norm": 0.431640625,
1309
+ "learning_rate": 0.00017045663473133215,
1310
+ "loss": 0.5018,
1311
+ "step": 186
1312
+ },
1313
+ {
1314
+ "epoch": 0.827433628318584,
1315
+ "grad_norm": 0.4765625,
1316
+ "learning_rate": 0.00017011650117287866,
1317
+ "loss": 0.5055,
1318
+ "step": 187
1319
+ },
1320
+ {
1321
+ "epoch": 0.831858407079646,
1322
+ "grad_norm": 0.416015625,
1323
+ "learning_rate": 0.00016977476441179992,
1324
+ "loss": 0.5109,
1325
+ "step": 188
1326
+ },
1327
+ {
1328
+ "epoch": 0.8362831858407079,
1329
+ "grad_norm": 0.4140625,
1330
+ "learning_rate": 0.00016943143226185253,
1331
+ "loss": 0.4591,
1332
+ "step": 189
1333
+ },
1334
+ {
1335
+ "epoch": 0.8407079646017699,
1336
+ "grad_norm": 0.498046875,
1337
+ "learning_rate": 0.00016908651257327138,
1338
+ "loss": 0.5182,
1339
+ "step": 190
1340
+ },
1341
+ {
1342
+ "epoch": 0.8451327433628318,
1343
+ "grad_norm": 0.431640625,
1344
+ "learning_rate": 0.00016874001323259011,
1345
+ "loss": 0.4982,
1346
+ "step": 191
1347
+ },
1348
+ {
1349
+ "epoch": 0.8495575221238938,
1350
+ "grad_norm": 0.4296875,
1351
+ "learning_rate": 0.00016839194216246108,
1352
+ "loss": 0.4869,
1353
+ "step": 192
1354
+ },
1355
+ {
1356
+ "epoch": 0.8539823008849557,
1357
+ "grad_norm": 0.451171875,
1358
+ "learning_rate": 0.0001680423073214737,
1359
+ "loss": 0.4933,
1360
+ "step": 193
1361
+ },
1362
+ {
1363
+ "epoch": 0.8584070796460177,
1364
+ "grad_norm": 0.408203125,
1365
+ "learning_rate": 0.00016769111670397296,
1366
+ "loss": 0.5087,
1367
+ "step": 194
1368
+ },
1369
+ {
1370
+ "epoch": 0.8628318584070797,
1371
+ "grad_norm": 0.455078125,
1372
+ "learning_rate": 0.00016733837833987633,
1373
+ "loss": 0.5372,
1374
+ "step": 195
1375
+ },
1376
+ {
1377
+ "epoch": 0.8672566371681416,
1378
+ "grad_norm": 0.43359375,
1379
+ "learning_rate": 0.0001669841002944903,
1380
+ "loss": 0.5232,
1381
+ "step": 196
1382
+ },
1383
+ {
1384
+ "epoch": 0.8716814159292036,
1385
+ "grad_norm": 0.390625,
1386
+ "learning_rate": 0.00016662829066832596,
1387
+ "loss": 0.5069,
1388
+ "step": 197
1389
+ },
1390
+ {
1391
+ "epoch": 0.8761061946902655,
1392
+ "grad_norm": 0.42578125,
1393
+ "learning_rate": 0.00016627095759691362,
1394
+ "loss": 0.5029,
1395
+ "step": 198
1396
+ },
1397
+ {
1398
+ "epoch": 0.8805309734513275,
1399
+ "grad_norm": 0.42578125,
1400
+ "learning_rate": 0.0001659121092506171,
1401
+ "loss": 0.5079,
1402
+ "step": 199
1403
+ },
1404
+ {
1405
+ "epoch": 0.8849557522123894,
1406
+ "grad_norm": 0.419921875,
1407
+ "learning_rate": 0.00016555175383444655,
1408
+ "loss": 0.4895,
1409
+ "step": 200
1410
+ },
1411
+ {
1412
+ "epoch": 0.8893805309734514,
1413
+ "grad_norm": 0.443359375,
1414
+ "learning_rate": 0.00016518989958787126,
1415
+ "loss": 0.5173,
1416
+ "step": 201
1417
+ },
1418
+ {
1419
+ "epoch": 0.8938053097345132,
1420
+ "grad_norm": 0.42578125,
1421
+ "learning_rate": 0.00016482655478463083,
1422
+ "loss": 0.4811,
1423
+ "step": 202
1424
+ },
1425
+ {
1426
+ "epoch": 0.8982300884955752,
1427
+ "grad_norm": 0.412109375,
1428
+ "learning_rate": 0.00016446172773254629,
1429
+ "loss": 0.5187,
1430
+ "step": 203
1431
+ },
1432
+ {
1433
+ "epoch": 0.9026548672566371,
1434
+ "grad_norm": 0.4296875,
1435
+ "learning_rate": 0.00016409542677333006,
1436
+ "loss": 0.5053,
1437
+ "step": 204
1438
+ },
1439
+ {
1440
+ "epoch": 0.9070796460176991,
1441
+ "grad_norm": 0.431640625,
1442
+ "learning_rate": 0.0001637276602823952,
1443
+ "loss": 0.5353,
1444
+ "step": 205
1445
+ },
1446
+ {
1447
+ "epoch": 0.911504424778761,
1448
+ "grad_norm": 0.4453125,
1449
+ "learning_rate": 0.00016335843666866388,
1450
+ "loss": 0.4827,
1451
+ "step": 206
1452
+ },
1453
+ {
1454
+ "epoch": 0.915929203539823,
1455
+ "grad_norm": 0.4140625,
1456
+ "learning_rate": 0.00016298776437437523,
1457
+ "loss": 0.4936,
1458
+ "step": 207
1459
+ },
1460
+ {
1461
+ "epoch": 0.9203539823008849,
1462
+ "grad_norm": 0.443359375,
1463
+ "learning_rate": 0.0001626156518748922,
1464
+ "loss": 0.5207,
1465
+ "step": 208
1466
+ },
1467
+ {
1468
+ "epoch": 0.9247787610619469,
1469
+ "grad_norm": 0.423828125,
1470
+ "learning_rate": 0.0001622421076785077,
1471
+ "loss": 0.4725,
1472
+ "step": 209
1473
+ },
1474
+ {
1475
+ "epoch": 0.9292035398230089,
1476
+ "grad_norm": 0.408203125,
1477
+ "learning_rate": 0.00016186714032625035,
1478
+ "loss": 0.4816,
1479
+ "step": 210
1480
+ },
1481
+ {
1482
+ "epoch": 0.9336283185840708,
1483
+ "grad_norm": 0.419921875,
1484
+ "learning_rate": 0.00016149075839168886,
1485
+ "loss": 0.5515,
1486
+ "step": 211
1487
+ },
1488
+ {
1489
+ "epoch": 0.9380530973451328,
1490
+ "grad_norm": 0.41796875,
1491
+ "learning_rate": 0.0001611129704807362,
1492
+ "loss": 0.5059,
1493
+ "step": 212
1494
+ },
1495
+ {
1496
+ "epoch": 0.9424778761061947,
1497
+ "grad_norm": 0.443359375,
1498
+ "learning_rate": 0.0001607337852314527,
1499
+ "loss": 0.496,
1500
+ "step": 213
1501
+ },
1502
+ {
1503
+ "epoch": 0.9469026548672567,
1504
+ "grad_norm": 0.408203125,
1505
+ "learning_rate": 0.0001603532113138487,
1506
+ "loss": 0.484,
1507
+ "step": 214
1508
+ },
1509
+ {
1510
+ "epoch": 0.9513274336283186,
1511
+ "grad_norm": 0.41015625,
1512
+ "learning_rate": 0.00015997125742968617,
1513
+ "loss": 0.5204,
1514
+ "step": 215
1515
+ },
1516
+ {
1517
+ "epoch": 0.9557522123893806,
1518
+ "grad_norm": 0.40625,
1519
+ "learning_rate": 0.0001595879323122798,
1520
+ "loss": 0.4906,
1521
+ "step": 216
1522
+ },
1523
+ {
1524
+ "epoch": 0.9601769911504425,
1525
+ "grad_norm": 0.416015625,
1526
+ "learning_rate": 0.00015920324472629732,
1527
+ "loss": 0.492,
1528
+ "step": 217
1529
+ },
1530
+ {
1531
+ "epoch": 0.9646017699115044,
1532
+ "grad_norm": 0.412109375,
1533
+ "learning_rate": 0.00015881720346755905,
1534
+ "loss": 0.4574,
1535
+ "step": 218
1536
+ },
1537
+ {
1538
+ "epoch": 0.9690265486725663,
1539
+ "grad_norm": 0.4296875,
1540
+ "learning_rate": 0.00015842981736283686,
1541
+ "loss": 0.5018,
1542
+ "step": 219
1543
+ },
1544
+ {
1545
+ "epoch": 0.9734513274336283,
1546
+ "grad_norm": 0.419921875,
1547
+ "learning_rate": 0.00015804109526965232,
1548
+ "loss": 0.4766,
1549
+ "step": 220
1550
+ },
1551
+ {
1552
+ "epoch": 0.9778761061946902,
1553
+ "grad_norm": 0.384765625,
1554
+ "learning_rate": 0.0001576510460760741,
1555
+ "loss": 0.4928,
1556
+ "step": 221
1557
+ },
1558
+ {
1559
+ "epoch": 0.9823008849557522,
1560
+ "grad_norm": 0.400390625,
1561
+ "learning_rate": 0.0001572596787005149,
1562
+ "loss": 0.4919,
1563
+ "step": 222
1564
+ },
1565
+ {
1566
+ "epoch": 0.9867256637168141,
1567
+ "grad_norm": 0.41796875,
1568
+ "learning_rate": 0.00015686700209152738,
1569
+ "loss": 0.4737,
1570
+ "step": 223
1571
+ },
1572
+ {
1573
+ "epoch": 0.9911504424778761,
1574
+ "grad_norm": 0.423828125,
1575
+ "learning_rate": 0.00015647302522759962,
1576
+ "loss": 0.5073,
1577
+ "step": 224
1578
+ },
1579
+ {
1580
+ "epoch": 0.995575221238938,
1581
+ "grad_norm": 0.44140625,
1582
+ "learning_rate": 0.00015607775711694977,
1583
+ "loss": 0.4684,
1584
+ "step": 225
1585
+ },
1586
+ {
1587
+ "epoch": 1.0,
1588
+ "grad_norm": 0.466796875,
1589
+ "learning_rate": 0.0001556812067973203,
1590
+ "loss": 0.4789,
1591
+ "step": 226
1592
+ },
1593
+ {
1594
+ "epoch": 1.0,
1595
+ "eval_loss": 0.49126777052879333,
1596
+ "eval_runtime": 5.6505,
1597
+ "eval_samples_per_second": 148.306,
1598
+ "eval_steps_per_second": 18.583,
1599
+ "step": 226
1600
+ },
1601
+ {
1602
+ "epoch": 1.0044247787610618,
1603
+ "grad_norm": 0.4609375,
1604
+ "learning_rate": 0.00015528338333577101,
1605
+ "loss": 0.4303,
1606
+ "step": 227
1607
+ },
1608
+ {
1609
+ "epoch": 1.008849557522124,
1610
+ "grad_norm": 0.474609375,
1611
+ "learning_rate": 0.00015488429582847192,
1612
+ "loss": 0.4457,
1613
+ "step": 228
1614
+ },
1615
+ {
1616
+ "epoch": 1.0132743362831858,
1617
+ "grad_norm": 0.4921875,
1618
+ "learning_rate": 0.00015448395340049537,
1619
+ "loss": 0.4697,
1620
+ "step": 229
1621
+ },
1622
+ {
1623
+ "epoch": 1.0176991150442478,
1624
+ "grad_norm": 0.470703125,
1625
+ "learning_rate": 0.0001540823652056071,
1626
+ "loss": 0.4294,
1627
+ "step": 230
1628
+ },
1629
+ {
1630
+ "epoch": 1.0221238938053097,
1631
+ "grad_norm": 0.400390625,
1632
+ "learning_rate": 0.0001536795404260572,
1633
+ "loss": 0.4192,
1634
+ "step": 231
1635
+ },
1636
+ {
1637
+ "epoch": 1.0265486725663717,
1638
+ "grad_norm": 0.435546875,
1639
+ "learning_rate": 0.00015327548827237007,
1640
+ "loss": 0.4449,
1641
+ "step": 232
1642
+ },
1643
+ {
1644
+ "epoch": 1.0309734513274336,
1645
+ "grad_norm": 0.408203125,
1646
+ "learning_rate": 0.0001528702179831338,
1647
+ "loss": 0.4394,
1648
+ "step": 233
1649
+ },
1650
+ {
1651
+ "epoch": 1.0353982300884956,
1652
+ "grad_norm": 0.41015625,
1653
+ "learning_rate": 0.00015246373882478898,
1654
+ "loss": 0.435,
1655
+ "step": 234
1656
+ },
1657
+ {
1658
+ "epoch": 1.0398230088495575,
1659
+ "grad_norm": 0.42578125,
1660
+ "learning_rate": 0.0001520560600914168,
1661
+ "loss": 0.4531,
1662
+ "step": 235
1663
+ },
1664
+ {
1665
+ "epoch": 1.0442477876106195,
1666
+ "grad_norm": 0.4140625,
1667
+ "learning_rate": 0.00015164719110452652,
1668
+ "loss": 0.4443,
1669
+ "step": 236
1670
+ },
1671
+ {
1672
+ "epoch": 1.0486725663716814,
1673
+ "grad_norm": 0.419921875,
1674
+ "learning_rate": 0.0001512371412128424,
1675
+ "loss": 0.4459,
1676
+ "step": 237
1677
+ },
1678
+ {
1679
+ "epoch": 1.0530973451327434,
1680
+ "grad_norm": 0.431640625,
1681
+ "learning_rate": 0.00015082591979208976,
1682
+ "loss": 0.4355,
1683
+ "step": 238
1684
+ },
1685
+ {
1686
+ "epoch": 1.0575221238938053,
1687
+ "grad_norm": 0.474609375,
1688
+ "learning_rate": 0.00015041353624478093,
1689
+ "loss": 0.443,
1690
+ "step": 239
1691
+ },
1692
+ {
1693
+ "epoch": 1.0619469026548674,
1694
+ "grad_norm": 0.4140625,
1695
+ "learning_rate": 0.00015000000000000001,
1696
+ "loss": 0.4295,
1697
+ "step": 240
1698
+ },
1699
+ {
1700
+ "epoch": 1.0663716814159292,
1701
+ "grad_norm": 0.384765625,
1702
+ "learning_rate": 0.0001495853205131873,
1703
+ "loss": 0.4249,
1704
+ "step": 241
1705
+ },
1706
+ {
1707
+ "epoch": 1.0707964601769913,
1708
+ "grad_norm": 0.431640625,
1709
+ "learning_rate": 0.00014916950726592322,
1710
+ "loss": 0.4464,
1711
+ "step": 242
1712
+ },
1713
+ {
1714
+ "epoch": 1.075221238938053,
1715
+ "grad_norm": 0.453125,
1716
+ "learning_rate": 0.00014875256976571135,
1717
+ "loss": 0.4404,
1718
+ "step": 243
1719
+ },
1720
+ {
1721
+ "epoch": 1.079646017699115,
1722
+ "grad_norm": 0.42578125,
1723
+ "learning_rate": 0.00014833451754576123,
1724
+ "loss": 0.4294,
1725
+ "step": 244
1726
+ },
1727
+ {
1728
+ "epoch": 1.084070796460177,
1729
+ "grad_norm": 0.427734375,
1730
+ "learning_rate": 0.00014791536016477022,
1731
+ "loss": 0.4172,
1732
+ "step": 245
1733
+ },
1734
+ {
1735
+ "epoch": 1.0884955752212389,
1736
+ "grad_norm": 0.43359375,
1737
+ "learning_rate": 0.00014749510720670506,
1738
+ "loss": 0.4333,
1739
+ "step": 246
1740
+ },
1741
+ {
1742
+ "epoch": 1.092920353982301,
1743
+ "grad_norm": 0.416015625,
1744
+ "learning_rate": 0.00014707376828058263,
1745
+ "loss": 0.4522,
1746
+ "step": 247
1747
+ },
1748
+ {
1749
+ "epoch": 1.0973451327433628,
1750
+ "grad_norm": 0.427734375,
1751
+ "learning_rate": 0.00014665135302025035,
1752
+ "loss": 0.46,
1753
+ "step": 248
1754
+ },
1755
+ {
1756
+ "epoch": 1.1017699115044248,
1757
+ "grad_norm": 0.40234375,
1758
+ "learning_rate": 0.00014622787108416584,
1759
+ "loss": 0.42,
1760
+ "step": 249
1761
+ },
1762
+ {
1763
+ "epoch": 1.1061946902654867,
1764
+ "grad_norm": 0.43359375,
1765
+ "learning_rate": 0.00014580333215517607,
1766
+ "loss": 0.4367,
1767
+ "step": 250
1768
+ },
1769
+ {
1770
+ "epoch": 1.1106194690265487,
1771
+ "grad_norm": 0.40625,
1772
+ "learning_rate": 0.000145377745940296,
1773
+ "loss": 0.4305,
1774
+ "step": 251
1775
+ },
1776
+ {
1777
+ "epoch": 1.1150442477876106,
1778
+ "grad_norm": 0.384765625,
1779
+ "learning_rate": 0.00014495112217048658,
1780
+ "loss": 0.4518,
1781
+ "step": 252
1782
+ },
1783
+ {
1784
+ "epoch": 1.1194690265486726,
1785
+ "grad_norm": 0.44140625,
1786
+ "learning_rate": 0.00014452347060043237,
1787
+ "loss": 0.4227,
1788
+ "step": 253
1789
+ },
1790
+ {
1791
+ "epoch": 1.1238938053097345,
1792
+ "grad_norm": 0.431640625,
1793
+ "learning_rate": 0.00014409480100831834,
1794
+ "loss": 0.4336,
1795
+ "step": 254
1796
+ },
1797
+ {
1798
+ "epoch": 1.1283185840707965,
1799
+ "grad_norm": 0.40625,
1800
+ "learning_rate": 0.0001436651231956064,
1801
+ "loss": 0.4077,
1802
+ "step": 255
1803
+ },
1804
+ {
1805
+ "epoch": 1.1327433628318584,
1806
+ "grad_norm": 0.423828125,
1807
+ "learning_rate": 0.00014323444698681126,
1808
+ "loss": 0.426,
1809
+ "step": 256
1810
+ },
1811
+ {
1812
+ "epoch": 1.1371681415929205,
1813
+ "grad_norm": 0.421875,
1814
+ "learning_rate": 0.0001428027822292758,
1815
+ "loss": 0.4403,
1816
+ "step": 257
1817
+ },
1818
+ {
1819
+ "epoch": 1.1415929203539823,
1820
+ "grad_norm": 0.4609375,
1821
+ "learning_rate": 0.0001423701387929459,
1822
+ "loss": 0.443,
1823
+ "step": 258
1824
+ },
1825
+ {
1826
+ "epoch": 1.1460176991150441,
1827
+ "grad_norm": 0.4296875,
1828
+ "learning_rate": 0.0001419365265701448,
1829
+ "loss": 0.438,
1830
+ "step": 259
1831
+ },
1832
+ {
1833
+ "epoch": 1.1504424778761062,
1834
+ "grad_norm": 0.419921875,
1835
+ "learning_rate": 0.00014150195547534686,
1836
+ "loss": 0.4318,
1837
+ "step": 260
1838
+ },
1839
+ {
1840
+ "epoch": 1.154867256637168,
1841
+ "grad_norm": 0.408203125,
1842
+ "learning_rate": 0.0001410664354449509,
1843
+ "loss": 0.4413,
1844
+ "step": 261
1845
+ },
1846
+ {
1847
+ "epoch": 1.1592920353982301,
1848
+ "grad_norm": 0.392578125,
1849
+ "learning_rate": 0.00014062997643705306,
1850
+ "loss": 0.4325,
1851
+ "step": 262
1852
+ },
1853
+ {
1854
+ "epoch": 1.163716814159292,
1855
+ "grad_norm": 0.39453125,
1856
+ "learning_rate": 0.00014019258843121893,
1857
+ "loss": 0.4164,
1858
+ "step": 263
1859
+ },
1860
+ {
1861
+ "epoch": 1.168141592920354,
1862
+ "grad_norm": 0.431640625,
1863
+ "learning_rate": 0.0001397542814282556,
1864
+ "loss": 0.4393,
1865
+ "step": 264
1866
+ },
1867
+ {
1868
+ "epoch": 1.1725663716814159,
1869
+ "grad_norm": 0.423828125,
1870
+ "learning_rate": 0.00013931506544998283,
1871
+ "loss": 0.4346,
1872
+ "step": 265
1873
+ },
1874
+ {
1875
+ "epoch": 1.176991150442478,
1876
+ "grad_norm": 0.439453125,
1877
+ "learning_rate": 0.00013887495053900397,
1878
+ "loss": 0.4294,
1879
+ "step": 266
1880
+ },
1881
+ {
1882
+ "epoch": 1.1814159292035398,
1883
+ "grad_norm": 0.416015625,
1884
+ "learning_rate": 0.00013843394675847634,
1885
+ "loss": 0.4404,
1886
+ "step": 267
1887
+ },
1888
+ {
1889
+ "epoch": 1.1858407079646018,
1890
+ "grad_norm": 0.388671875,
1891
+ "learning_rate": 0.00013799206419188103,
1892
+ "loss": 0.4198,
1893
+ "step": 268
1894
+ },
1895
+ {
1896
+ "epoch": 1.1902654867256637,
1897
+ "grad_norm": 0.40234375,
1898
+ "learning_rate": 0.00013754931294279263,
1899
+ "loss": 0.4158,
1900
+ "step": 269
1901
+ },
1902
+ {
1903
+ "epoch": 1.1946902654867257,
1904
+ "grad_norm": 0.3984375,
1905
+ "learning_rate": 0.00013710570313464778,
1906
+ "loss": 0.4234,
1907
+ "step": 270
1908
+ },
1909
+ {
1910
+ "epoch": 1.1991150442477876,
1911
+ "grad_norm": 0.4140625,
1912
+ "learning_rate": 0.0001366612449105141,
1913
+ "loss": 0.4233,
1914
+ "step": 271
1915
+ },
1916
+ {
1917
+ "epoch": 1.2035398230088497,
1918
+ "grad_norm": 0.439453125,
1919
+ "learning_rate": 0.00013621594843285802,
1920
+ "loss": 0.4518,
1921
+ "step": 272
1922
+ },
1923
+ {
1924
+ "epoch": 1.2079646017699115,
1925
+ "grad_norm": 0.412109375,
1926
+ "learning_rate": 0.0001357698238833126,
1927
+ "loss": 0.4465,
1928
+ "step": 273
1929
+ },
1930
+ {
1931
+ "epoch": 1.2123893805309733,
1932
+ "grad_norm": 0.412109375,
1933
+ "learning_rate": 0.00013532288146244446,
1934
+ "loss": 0.4446,
1935
+ "step": 274
1936
+ },
1937
+ {
1938
+ "epoch": 1.2168141592920354,
1939
+ "grad_norm": 0.4140625,
1940
+ "learning_rate": 0.00013487513138952094,
1941
+ "loss": 0.4384,
1942
+ "step": 275
1943
+ },
1944
+ {
1945
+ "epoch": 1.2212389380530975,
1946
+ "grad_norm": 0.41015625,
1947
+ "learning_rate": 0.00013442658390227602,
1948
+ "loss": 0.4286,
1949
+ "step": 276
1950
+ },
1951
+ {
1952
+ "epoch": 1.2256637168141593,
1953
+ "grad_norm": 0.390625,
1954
+ "learning_rate": 0.00013397724925667657,
1955
+ "loss": 0.43,
1956
+ "step": 277
1957
+ },
1958
+ {
1959
+ "epoch": 1.2300884955752212,
1960
+ "grad_norm": 0.38671875,
1961
+ "learning_rate": 0.00013352713772668765,
1962
+ "loss": 0.4293,
1963
+ "step": 278
1964
+ },
1965
+ {
1966
+ "epoch": 1.2345132743362832,
1967
+ "grad_norm": 0.39453125,
1968
+ "learning_rate": 0.00013307625960403763,
1969
+ "loss": 0.4326,
1970
+ "step": 279
1971
+ },
1972
+ {
1973
+ "epoch": 1.238938053097345,
1974
+ "grad_norm": 0.3671875,
1975
+ "learning_rate": 0.00013262462519798293,
1976
+ "loss": 0.4241,
1977
+ "step": 280
1978
+ },
1979
+ {
1980
+ "epoch": 1.2433628318584071,
1981
+ "grad_norm": 0.3984375,
1982
+ "learning_rate": 0.00013217224483507228,
1983
+ "loss": 0.4336,
1984
+ "step": 281
1985
+ },
1986
+ {
1987
+ "epoch": 1.247787610619469,
1988
+ "grad_norm": 0.427734375,
1989
+ "learning_rate": 0.00013171912885891063,
1990
+ "loss": 0.4418,
1991
+ "step": 282
1992
+ },
1993
+ {
1994
+ "epoch": 1.252212389380531,
1995
+ "grad_norm": 0.416015625,
1996
+ "learning_rate": 0.00013126528762992247,
1997
+ "loss": 0.4258,
1998
+ "step": 283
1999
+ },
2000
+ {
2001
+ "epoch": 1.2566371681415929,
2002
+ "grad_norm": 0.37109375,
2003
+ "learning_rate": 0.00013081073152511525,
2004
+ "loss": 0.438,
2005
+ "step": 284
2006
+ },
2007
+ {
2008
+ "epoch": 1.261061946902655,
2009
+ "grad_norm": 0.40234375,
2010
+ "learning_rate": 0.00013035547093784186,
2011
+ "loss": 0.4364,
2012
+ "step": 285
2013
+ },
2014
+ {
2015
+ "epoch": 1.2654867256637168,
2016
+ "grad_norm": 0.392578125,
2017
+ "learning_rate": 0.00012989951627756304,
2018
+ "loss": 0.409,
2019
+ "step": 286
2020
+ },
2021
+ {
2022
+ "epoch": 1.2699115044247788,
2023
+ "grad_norm": 0.416015625,
2024
+ "learning_rate": 0.0001294428779696095,
2025
+ "loss": 0.4448,
2026
+ "step": 287
2027
+ },
2028
+ {
2029
+ "epoch": 1.2743362831858407,
2030
+ "grad_norm": 0.41015625,
2031
+ "learning_rate": 0.00012898556645494325,
2032
+ "loss": 0.4497,
2033
+ "step": 288
2034
+ },
2035
+ {
2036
+ "epoch": 1.2787610619469025,
2037
+ "grad_norm": 0.404296875,
2038
+ "learning_rate": 0.00012852759218991933,
2039
+ "loss": 0.4188,
2040
+ "step": 289
2041
+ },
2042
+ {
2043
+ "epoch": 1.2831858407079646,
2044
+ "grad_norm": 0.388671875,
2045
+ "learning_rate": 0.00012806896564604626,
2046
+ "loss": 0.4226,
2047
+ "step": 290
2048
+ },
2049
+ {
2050
+ "epoch": 1.2876106194690267,
2051
+ "grad_norm": 0.427734375,
2052
+ "learning_rate": 0.00012760969730974694,
2053
+ "loss": 0.4281,
2054
+ "step": 291
2055
+ },
2056
+ {
2057
+ "epoch": 1.2920353982300885,
2058
+ "grad_norm": 0.4375,
2059
+ "learning_rate": 0.00012714979768211853,
2060
+ "loss": 0.4626,
2061
+ "step": 292
2062
+ },
2063
+ {
2064
+ "epoch": 1.2964601769911503,
2065
+ "grad_norm": 0.416015625,
2066
+ "learning_rate": 0.0001266892772786929,
2067
+ "loss": 0.4292,
2068
+ "step": 293
2069
+ },
2070
+ {
2071
+ "epoch": 1.3008849557522124,
2072
+ "grad_norm": 0.412109375,
2073
+ "learning_rate": 0.00012622814662919561,
2074
+ "loss": 0.4309,
2075
+ "step": 294
2076
+ },
2077
+ {
2078
+ "epoch": 1.3053097345132743,
2079
+ "grad_norm": 0.404296875,
2080
+ "learning_rate": 0.0001257664162773055,
2081
+ "loss": 0.4283,
2082
+ "step": 295
2083
+ },
2084
+ {
2085
+ "epoch": 1.3097345132743363,
2086
+ "grad_norm": 0.408203125,
2087
+ "learning_rate": 0.00012530409678041343,
2088
+ "loss": 0.4431,
2089
+ "step": 296
2090
+ },
2091
+ {
2092
+ "epoch": 1.3141592920353982,
2093
+ "grad_norm": 0.400390625,
2094
+ "learning_rate": 0.00012484119870938103,
2095
+ "loss": 0.4419,
2096
+ "step": 297
2097
+ },
2098
+ {
2099
+ "epoch": 1.3185840707964602,
2100
+ "grad_norm": 0.380859375,
2101
+ "learning_rate": 0.00012437773264829897,
2102
+ "loss": 0.4513,
2103
+ "step": 298
2104
+ },
2105
+ {
2106
+ "epoch": 1.323008849557522,
2107
+ "grad_norm": 0.39453125,
2108
+ "learning_rate": 0.00012391370919424485,
2109
+ "loss": 0.4389,
2110
+ "step": 299
2111
+ },
2112
+ {
2113
+ "epoch": 1.3274336283185841,
2114
+ "grad_norm": 0.390625,
2115
+ "learning_rate": 0.00012344913895704097,
2116
+ "loss": 0.4495,
2117
+ "step": 300
2118
+ },
2119
+ {
2120
+ "epoch": 1.331858407079646,
2121
+ "grad_norm": 0.40234375,
2122
+ "learning_rate": 0.00012298403255901186,
2123
+ "loss": 0.4169,
2124
+ "step": 301
2125
+ },
2126
+ {
2127
+ "epoch": 1.336283185840708,
2128
+ "grad_norm": 0.388671875,
2129
+ "learning_rate": 0.00012251840063474108,
2130
+ "loss": 0.4368,
2131
+ "step": 302
2132
+ },
2133
+ {
2134
+ "epoch": 1.3407079646017699,
2135
+ "grad_norm": 0.38671875,
2136
+ "learning_rate": 0.00012205225383082843,
2137
+ "loss": 0.4198,
2138
+ "step": 303
2139
+ },
2140
+ {
2141
+ "epoch": 1.3451327433628317,
2142
+ "grad_norm": 0.419921875,
2143
+ "learning_rate": 0.00012158560280564626,
2144
+ "loss": 0.4386,
2145
+ "step": 304
2146
+ },
2147
+ {
2148
+ "epoch": 1.3495575221238938,
2149
+ "grad_norm": 0.412109375,
2150
+ "learning_rate": 0.00012111845822909596,
2151
+ "loss": 0.4164,
2152
+ "step": 305
2153
+ },
2154
+ {
2155
+ "epoch": 1.3539823008849559,
2156
+ "grad_norm": 0.408203125,
2157
+ "learning_rate": 0.00012065083078236374,
2158
+ "loss": 0.4568,
2159
+ "step": 306
2160
+ },
2161
+ {
2162
+ "epoch": 1.3584070796460177,
2163
+ "grad_norm": 0.40625,
2164
+ "learning_rate": 0.00012018273115767673,
2165
+ "loss": 0.4283,
2166
+ "step": 307
2167
+ },
2168
+ {
2169
+ "epoch": 1.3628318584070795,
2170
+ "grad_norm": 0.416015625,
2171
+ "learning_rate": 0.00011971417005805818,
2172
+ "loss": 0.4254,
2173
+ "step": 308
2174
+ },
2175
+ {
2176
+ "epoch": 1.3672566371681416,
2177
+ "grad_norm": 0.4140625,
2178
+ "learning_rate": 0.000119245158197083,
2179
+ "loss": 0.434,
2180
+ "step": 309
2181
+ },
2182
+ {
2183
+ "epoch": 1.3716814159292037,
2184
+ "grad_norm": 0.408203125,
2185
+ "learning_rate": 0.00011877570629863266,
2186
+ "loss": 0.4203,
2187
+ "step": 310
2188
+ },
2189
+ {
2190
+ "epoch": 1.3761061946902655,
2191
+ "grad_norm": 0.408203125,
2192
+ "learning_rate": 0.00011830582509664995,
2193
+ "loss": 0.4373,
2194
+ "step": 311
2195
+ },
2196
+ {
2197
+ "epoch": 1.3805309734513274,
2198
+ "grad_norm": 0.41796875,
2199
+ "learning_rate": 0.00011783552533489372,
2200
+ "loss": 0.4303,
2201
+ "step": 312
2202
+ },
2203
+ {
2204
+ "epoch": 1.3849557522123894,
2205
+ "grad_norm": 0.390625,
2206
+ "learning_rate": 0.00011736481776669306,
2207
+ "loss": 0.429,
2208
+ "step": 313
2209
+ },
2210
+ {
2211
+ "epoch": 1.3893805309734513,
2212
+ "grad_norm": 0.408203125,
2213
+ "learning_rate": 0.0001168937131547015,
2214
+ "loss": 0.4122,
2215
+ "step": 314
2216
+ },
2217
+ {
2218
+ "epoch": 1.3938053097345133,
2219
+ "grad_norm": 0.412109375,
2220
+ "learning_rate": 0.00011642222227065089,
2221
+ "loss": 0.4519,
2222
+ "step": 315
2223
+ },
2224
+ {
2225
+ "epoch": 1.3982300884955752,
2226
+ "grad_norm": 0.43359375,
2227
+ "learning_rate": 0.00011595035589510522,
2228
+ "loss": 0.4303,
2229
+ "step": 316
2230
+ },
2231
+ {
2232
+ "epoch": 1.4026548672566372,
2233
+ "grad_norm": 0.38671875,
2234
+ "learning_rate": 0.00011547812481721388,
2235
+ "loss": 0.4215,
2236
+ "step": 317
2237
+ },
2238
+ {
2239
+ "epoch": 1.407079646017699,
2240
+ "grad_norm": 0.408203125,
2241
+ "learning_rate": 0.00011500553983446527,
2242
+ "loss": 0.4276,
2243
+ "step": 318
2244
+ },
2245
+ {
2246
+ "epoch": 1.411504424778761,
2247
+ "grad_norm": 0.427734375,
2248
+ "learning_rate": 0.00011453261175243973,
2249
+ "loss": 0.4249,
2250
+ "step": 319
2251
+ },
2252
+ {
2253
+ "epoch": 1.415929203539823,
2254
+ "grad_norm": 0.392578125,
2255
+ "learning_rate": 0.00011405935138456241,
2256
+ "loss": 0.4168,
2257
+ "step": 320
2258
+ },
2259
+ {
2260
+ "epoch": 1.420353982300885,
2261
+ "grad_norm": 0.41015625,
2262
+ "learning_rate": 0.0001135857695518563,
2263
+ "loss": 0.4194,
2264
+ "step": 321
2265
+ },
2266
+ {
2267
+ "epoch": 1.424778761061947,
2268
+ "grad_norm": 0.4296875,
2269
+ "learning_rate": 0.00011311187708269442,
2270
+ "loss": 0.4154,
2271
+ "step": 322
2272
+ },
2273
+ {
2274
+ "epoch": 1.4292035398230087,
2275
+ "grad_norm": 0.412109375,
2276
+ "learning_rate": 0.00011263768481255264,
2277
+ "loss": 0.426,
2278
+ "step": 323
2279
+ },
2280
+ {
2281
+ "epoch": 1.4336283185840708,
2282
+ "grad_norm": 0.4140625,
2283
+ "learning_rate": 0.00011216320358376157,
2284
+ "loss": 0.4359,
2285
+ "step": 324
2286
+ },
2287
+ {
2288
+ "epoch": 1.4380530973451329,
2289
+ "grad_norm": 0.400390625,
2290
+ "learning_rate": 0.00011168844424525902,
2291
+ "loss": 0.412,
2292
+ "step": 325
2293
+ },
2294
+ {
2295
+ "epoch": 1.4424778761061947,
2296
+ "grad_norm": 0.3984375,
2297
+ "learning_rate": 0.00011121341765234146,
2298
+ "loss": 0.4418,
2299
+ "step": 326
2300
+ },
2301
+ {
2302
+ "epoch": 1.4469026548672566,
2303
+ "grad_norm": 0.390625,
2304
+ "learning_rate": 0.00011073813466641632,
2305
+ "loss": 0.4193,
2306
+ "step": 327
2307
+ },
2308
+ {
2309
+ "epoch": 1.4513274336283186,
2310
+ "grad_norm": 0.416015625,
2311
+ "learning_rate": 0.00011026260615475333,
2312
+ "loss": 0.4161,
2313
+ "step": 328
2314
+ },
2315
+ {
2316
+ "epoch": 1.4557522123893805,
2317
+ "grad_norm": 0.384765625,
2318
+ "learning_rate": 0.00010978684299023607,
2319
+ "loss": 0.4367,
2320
+ "step": 329
2321
+ },
2322
+ {
2323
+ "epoch": 1.4601769911504425,
2324
+ "grad_norm": 0.400390625,
2325
+ "learning_rate": 0.00010931085605111354,
2326
+ "loss": 0.4334,
2327
+ "step": 330
2328
+ },
2329
+ {
2330
+ "epoch": 1.4646017699115044,
2331
+ "grad_norm": 0.384765625,
2332
+ "learning_rate": 0.0001088346562207512,
2333
+ "loss": 0.4069,
2334
+ "step": 331
2335
+ },
2336
+ {
2337
+ "epoch": 1.4690265486725664,
2338
+ "grad_norm": 0.40234375,
2339
+ "learning_rate": 0.00010835825438738232,
2340
+ "loss": 0.4183,
2341
+ "step": 332
2342
+ },
2343
+ {
2344
+ "epoch": 1.4734513274336283,
2345
+ "grad_norm": 0.40234375,
2346
+ "learning_rate": 0.00010788166144385888,
2347
+ "loss": 0.4275,
2348
+ "step": 333
2349
+ },
2350
+ {
2351
+ "epoch": 1.4778761061946903,
2352
+ "grad_norm": 0.42578125,
2353
+ "learning_rate": 0.00010740488828740258,
2354
+ "loss": 0.4342,
2355
+ "step": 334
2356
+ },
2357
+ {
2358
+ "epoch": 1.4823008849557522,
2359
+ "grad_norm": 0.375,
2360
+ "learning_rate": 0.00010692794581935566,
2361
+ "loss": 0.4199,
2362
+ "step": 335
2363
+ },
2364
+ {
2365
+ "epoch": 1.4867256637168142,
2366
+ "grad_norm": 0.3984375,
2367
+ "learning_rate": 0.00010645084494493165,
2368
+ "loss": 0.4134,
2369
+ "step": 336
2370
+ },
2371
+ {
2372
+ "epoch": 1.491150442477876,
2373
+ "grad_norm": 0.404296875,
2374
+ "learning_rate": 0.00010597359657296602,
2375
+ "loss": 0.4057,
2376
+ "step": 337
2377
+ },
2378
+ {
2379
+ "epoch": 1.495575221238938,
2380
+ "grad_norm": 0.38671875,
2381
+ "learning_rate": 0.0001054962116156667,
2382
+ "loss": 0.4129,
2383
+ "step": 338
2384
+ },
2385
+ {
2386
+ "epoch": 1.5,
2387
+ "grad_norm": 0.392578125,
2388
+ "learning_rate": 0.00010501870098836473,
2389
+ "loss": 0.4337,
2390
+ "step": 339
2391
+ },
2392
+ {
2393
+ "epoch": 1.504424778761062,
2394
+ "grad_norm": 0.40234375,
2395
+ "learning_rate": 0.00010454107560926443,
2396
+ "loss": 0.4292,
2397
+ "step": 340
2398
+ },
2399
+ {
2400
+ "epoch": 1.508849557522124,
2401
+ "grad_norm": 0.3671875,
2402
+ "learning_rate": 0.00010406334639919403,
2403
+ "loss": 0.4074,
2404
+ "step": 341
2405
+ },
2406
+ {
2407
+ "epoch": 1.5132743362831858,
2408
+ "grad_norm": 0.427734375,
2409
+ "learning_rate": 0.00010358552428135575,
2410
+ "loss": 0.4187,
2411
+ "step": 342
2412
+ },
2413
+ {
2414
+ "epoch": 1.5176991150442478,
2415
+ "grad_norm": 0.427734375,
2416
+ "learning_rate": 0.0001031076201810762,
2417
+ "loss": 0.4427,
2418
+ "step": 343
2419
+ },
2420
+ {
2421
+ "epoch": 1.5221238938053099,
2422
+ "grad_norm": 0.3828125,
2423
+ "learning_rate": 0.00010262964502555643,
2424
+ "loss": 0.4249,
2425
+ "step": 344
2426
+ },
2427
+ {
2428
+ "epoch": 1.5265486725663717,
2429
+ "grad_norm": 0.37890625,
2430
+ "learning_rate": 0.00010215160974362223,
2431
+ "loss": 0.4132,
2432
+ "step": 345
2433
+ },
2434
+ {
2435
+ "epoch": 1.5309734513274336,
2436
+ "grad_norm": 0.4140625,
2437
+ "learning_rate": 0.00010167352526547416,
2438
+ "loss": 0.4181,
2439
+ "step": 346
2440
+ },
2441
+ {
2442
+ "epoch": 1.5353982300884956,
2443
+ "grad_norm": 0.408203125,
2444
+ "learning_rate": 0.00010119540252243755,
2445
+ "loss": 0.4282,
2446
+ "step": 347
2447
+ },
2448
+ {
2449
+ "epoch": 1.5398230088495575,
2450
+ "grad_norm": 0.38671875,
2451
+ "learning_rate": 0.00010071725244671282,
2452
+ "loss": 0.4314,
2453
+ "step": 348
2454
+ },
2455
+ {
2456
+ "epoch": 1.5442477876106193,
2457
+ "grad_norm": 0.396484375,
2458
+ "learning_rate": 0.00010023908597112514,
2459
+ "loss": 0.4404,
2460
+ "step": 349
2461
+ },
2462
+ {
2463
+ "epoch": 1.5486725663716814,
2464
+ "grad_norm": 0.4296875,
2465
+ "learning_rate": 9.976091402887487e-05,
2466
+ "loss": 0.4394,
2467
+ "step": 350
2468
+ },
2469
+ {
2470
+ "epoch": 1.5530973451327434,
2471
+ "grad_norm": 0.419921875,
2472
+ "learning_rate": 9.928274755328723e-05,
2473
+ "loss": 0.4282,
2474
+ "step": 351
2475
+ },
2476
+ {
2477
+ "epoch": 1.5575221238938053,
2478
+ "grad_norm": 0.392578125,
2479
+ "learning_rate": 9.880459747756247e-05,
2480
+ "loss": 0.4234,
2481
+ "step": 352
2482
+ },
2483
+ {
2484
+ "epoch": 1.5619469026548671,
2485
+ "grad_norm": 0.388671875,
2486
+ "learning_rate": 9.83264747345259e-05,
2487
+ "loss": 0.412,
2488
+ "step": 353
2489
+ },
2490
+ {
2491
+ "epoch": 1.5663716814159292,
2492
+ "grad_norm": 0.40234375,
2493
+ "learning_rate": 9.784839025637778e-05,
2494
+ "loss": 0.4258,
2495
+ "step": 354
2496
+ },
2497
+ {
2498
+ "epoch": 1.5707964601769913,
2499
+ "grad_norm": 0.453125,
2500
+ "learning_rate": 9.737035497444361e-05,
2501
+ "loss": 0.4149,
2502
+ "step": 355
2503
+ },
2504
+ {
2505
+ "epoch": 1.575221238938053,
2506
+ "grad_norm": 0.404296875,
2507
+ "learning_rate": 9.689237981892382e-05,
2508
+ "loss": 0.4095,
2509
+ "step": 356
2510
+ },
2511
+ {
2512
+ "epoch": 1.579646017699115,
2513
+ "grad_norm": 0.41015625,
2514
+ "learning_rate": 9.641447571864429e-05,
2515
+ "loss": 0.3994,
2516
+ "step": 357
2517
+ },
2518
+ {
2519
+ "epoch": 1.584070796460177,
2520
+ "grad_norm": 0.37109375,
2521
+ "learning_rate": 9.593665360080599e-05,
2522
+ "loss": 0.417,
2523
+ "step": 358
2524
+ },
2525
+ {
2526
+ "epoch": 1.588495575221239,
2527
+ "grad_norm": 0.408203125,
2528
+ "learning_rate": 9.545892439073562e-05,
2529
+ "loss": 0.4328,
2530
+ "step": 359
2531
+ },
2532
+ {
2533
+ "epoch": 1.592920353982301,
2534
+ "grad_norm": 0.4296875,
2535
+ "learning_rate": 9.49812990116353e-05,
2536
+ "loss": 0.4364,
2537
+ "step": 360
2538
+ },
2539
+ {
2540
+ "epoch": 1.5973451327433628,
2541
+ "grad_norm": 0.39453125,
2542
+ "learning_rate": 9.450378838433331e-05,
2543
+ "loss": 0.4098,
2544
+ "step": 361
2545
+ },
2546
+ {
2547
+ "epoch": 1.6017699115044248,
2548
+ "grad_norm": 0.408203125,
2549
+ "learning_rate": 9.4026403427034e-05,
2550
+ "loss": 0.4155,
2551
+ "step": 362
2552
+ },
2553
+ {
2554
+ "epoch": 1.606194690265487,
2555
+ "grad_norm": 0.375,
2556
+ "learning_rate": 9.354915505506839e-05,
2557
+ "loss": 0.4338,
2558
+ "step": 363
2559
+ },
2560
+ {
2561
+ "epoch": 1.6106194690265485,
2562
+ "grad_norm": 0.408203125,
2563
+ "learning_rate": 9.307205418064437e-05,
2564
+ "loss": 0.4326,
2565
+ "step": 364
2566
+ },
2567
+ {
2568
+ "epoch": 1.6150442477876106,
2569
+ "grad_norm": 0.396484375,
2570
+ "learning_rate": 9.259511171259746e-05,
2571
+ "loss": 0.4167,
2572
+ "step": 365
2573
+ },
2574
+ {
2575
+ "epoch": 1.6194690265486726,
2576
+ "grad_norm": 0.40625,
2577
+ "learning_rate": 9.211833855614114e-05,
2578
+ "loss": 0.4413,
2579
+ "step": 366
2580
+ },
2581
+ {
2582
+ "epoch": 1.6238938053097345,
2583
+ "grad_norm": 0.376953125,
2584
+ "learning_rate": 9.164174561261771e-05,
2585
+ "loss": 0.4223,
2586
+ "step": 367
2587
+ },
2588
+ {
2589
+ "epoch": 1.6283185840707963,
2590
+ "grad_norm": 0.388671875,
2591
+ "learning_rate": 9.116534377924883e-05,
2592
+ "loss": 0.4171,
2593
+ "step": 368
2594
+ },
2595
+ {
2596
+ "epoch": 1.6327433628318584,
2597
+ "grad_norm": 0.384765625,
2598
+ "learning_rate": 9.06891439488865e-05,
2599
+ "loss": 0.4042,
2600
+ "step": 369
2601
+ },
2602
+ {
2603
+ "epoch": 1.6371681415929205,
2604
+ "grad_norm": 0.39453125,
2605
+ "learning_rate": 9.021315700976395e-05,
2606
+ "loss": 0.4168,
2607
+ "step": 370
2608
+ },
2609
+ {
2610
+ "epoch": 1.6415929203539823,
2611
+ "grad_norm": 0.3828125,
2612
+ "learning_rate": 8.973739384524674e-05,
2613
+ "loss": 0.4121,
2614
+ "step": 371
2615
+ },
2616
+ {
2617
+ "epoch": 1.6460176991150441,
2618
+ "grad_norm": 0.357421875,
2619
+ "learning_rate": 8.92618653335837e-05,
2620
+ "loss": 0.4084,
2621
+ "step": 372
2622
+ },
2623
+ {
2624
+ "epoch": 1.6504424778761062,
2625
+ "grad_norm": 0.390625,
2626
+ "learning_rate": 8.878658234765858e-05,
2627
+ "loss": 0.41,
2628
+ "step": 373
2629
+ },
2630
+ {
2631
+ "epoch": 1.6548672566371683,
2632
+ "grad_norm": 0.369140625,
2633
+ "learning_rate": 8.831155575474102e-05,
2634
+ "loss": 0.4131,
2635
+ "step": 374
2636
+ },
2637
+ {
2638
+ "epoch": 1.6592920353982301,
2639
+ "grad_norm": 0.37109375,
2640
+ "learning_rate": 8.783679641623845e-05,
2641
+ "loss": 0.4277,
2642
+ "step": 375
2643
+ },
2644
+ {
2645
+ "epoch": 1.663716814159292,
2646
+ "grad_norm": 0.37890625,
2647
+ "learning_rate": 8.73623151874474e-05,
2648
+ "loss": 0.4201,
2649
+ "step": 376
2650
+ },
2651
+ {
2652
+ "epoch": 1.668141592920354,
2653
+ "grad_norm": 0.39453125,
2654
+ "learning_rate": 8.688812291730563e-05,
2655
+ "loss": 0.417,
2656
+ "step": 377
2657
+ },
2658
+ {
2659
+ "epoch": 1.672566371681416,
2660
+ "grad_norm": 0.37109375,
2661
+ "learning_rate": 8.641423044814374e-05,
2662
+ "loss": 0.4111,
2663
+ "step": 378
2664
+ },
2665
+ {
2666
+ "epoch": 1.676991150442478,
2667
+ "grad_norm": 0.37109375,
2668
+ "learning_rate": 8.59406486154376e-05,
2669
+ "loss": 0.404,
2670
+ "step": 379
2671
+ },
2672
+ {
2673
+ "epoch": 1.6814159292035398,
2674
+ "grad_norm": 0.40625,
2675
+ "learning_rate": 8.54673882475603e-05,
2676
+ "loss": 0.428,
2677
+ "step": 380
2678
+ },
2679
+ {
2680
+ "epoch": 1.6858407079646018,
2681
+ "grad_norm": 0.396484375,
2682
+ "learning_rate": 8.499446016553474e-05,
2683
+ "loss": 0.4051,
2684
+ "step": 381
2685
+ },
2686
+ {
2687
+ "epoch": 1.6902654867256637,
2688
+ "grad_norm": 0.384765625,
2689
+ "learning_rate": 8.452187518278613e-05,
2690
+ "loss": 0.4164,
2691
+ "step": 382
2692
+ },
2693
+ {
2694
+ "epoch": 1.6946902654867255,
2695
+ "grad_norm": 0.375,
2696
+ "learning_rate": 8.404964410489485e-05,
2697
+ "loss": 0.3898,
2698
+ "step": 383
2699
+ },
2700
+ {
2701
+ "epoch": 1.6991150442477876,
2702
+ "grad_norm": 0.40234375,
2703
+ "learning_rate": 8.357777772934913e-05,
2704
+ "loss": 0.4264,
2705
+ "step": 384
2706
+ },
2707
+ {
2708
+ "epoch": 1.7035398230088497,
2709
+ "grad_norm": 0.357421875,
2710
+ "learning_rate": 8.310628684529856e-05,
2711
+ "loss": 0.405,
2712
+ "step": 385
2713
+ },
2714
+ {
2715
+ "epoch": 1.7079646017699115,
2716
+ "grad_norm": 0.380859375,
2717
+ "learning_rate": 8.263518223330697e-05,
2718
+ "loss": 0.4261,
2719
+ "step": 386
2720
+ },
2721
+ {
2722
+ "epoch": 1.7123893805309733,
2723
+ "grad_norm": 0.376953125,
2724
+ "learning_rate": 8.216447466510631e-05,
2725
+ "loss": 0.4001,
2726
+ "step": 387
2727
+ },
2728
+ {
2729
+ "epoch": 1.7168141592920354,
2730
+ "grad_norm": 0.404296875,
2731
+ "learning_rate": 8.169417490335007e-05,
2732
+ "loss": 0.4212,
2733
+ "step": 388
2734
+ },
2735
+ {
2736
+ "epoch": 1.7212389380530975,
2737
+ "grad_norm": 0.37109375,
2738
+ "learning_rate": 8.122429370136739e-05,
2739
+ "loss": 0.4097,
2740
+ "step": 389
2741
+ },
2742
+ {
2743
+ "epoch": 1.7256637168141593,
2744
+ "grad_norm": 0.375,
2745
+ "learning_rate": 8.075484180291701e-05,
2746
+ "loss": 0.4133,
2747
+ "step": 390
2748
+ },
2749
+ {
2750
+ "epoch": 1.7300884955752212,
2751
+ "grad_norm": 0.3984375,
2752
+ "learning_rate": 8.028582994194185e-05,
2753
+ "loss": 0.4104,
2754
+ "step": 391
2755
+ },
2756
+ {
2757
+ "epoch": 1.7345132743362832,
2758
+ "grad_norm": 0.390625,
2759
+ "learning_rate": 7.981726884232328e-05,
2760
+ "loss": 0.4147,
2761
+ "step": 392
2762
+ },
2763
+ {
2764
+ "epoch": 1.7389380530973453,
2765
+ "grad_norm": 0.408203125,
2766
+ "learning_rate": 7.934916921763628e-05,
2767
+ "loss": 0.4097,
2768
+ "step": 393
2769
+ },
2770
+ {
2771
+ "epoch": 1.7433628318584071,
2772
+ "grad_norm": 0.39453125,
2773
+ "learning_rate": 7.888154177090405e-05,
2774
+ "loss": 0.4126,
2775
+ "step": 394
2776
+ },
2777
+ {
2778
+ "epoch": 1.747787610619469,
2779
+ "grad_norm": 0.390625,
2780
+ "learning_rate": 7.841439719435377e-05,
2781
+ "loss": 0.414,
2782
+ "step": 395
2783
+ },
2784
+ {
2785
+ "epoch": 1.752212389380531,
2786
+ "grad_norm": 0.388671875,
2787
+ "learning_rate": 7.79477461691716e-05,
2788
+ "loss": 0.4348,
2789
+ "step": 396
2790
+ },
2791
+ {
2792
+ "epoch": 1.7566371681415929,
2793
+ "grad_norm": 0.38671875,
2794
+ "learning_rate": 7.748159936525896e-05,
2795
+ "loss": 0.4122,
2796
+ "step": 397
2797
+ },
2798
+ {
2799
+ "epoch": 1.7610619469026547,
2800
+ "grad_norm": 0.380859375,
2801
+ "learning_rate": 7.701596744098818e-05,
2802
+ "loss": 0.4134,
2803
+ "step": 398
2804
+ },
2805
+ {
2806
+ "epoch": 1.7654867256637168,
2807
+ "grad_norm": 0.375,
2808
+ "learning_rate": 7.655086104295904e-05,
2809
+ "loss": 0.406,
2810
+ "step": 399
2811
+ },
2812
+ {
2813
+ "epoch": 1.7699115044247788,
2814
+ "grad_norm": 0.39453125,
2815
+ "learning_rate": 7.608629080575518e-05,
2816
+ "loss": 0.4179,
2817
+ "step": 400
2818
+ },
2819
+ {
2820
+ "epoch": 1.7743362831858407,
2821
+ "grad_norm": 0.3828125,
2822
+ "learning_rate": 7.562226735170106e-05,
2823
+ "loss": 0.427,
2824
+ "step": 401
2825
+ },
2826
+ {
2827
+ "epoch": 1.7787610619469025,
2828
+ "grad_norm": 0.408203125,
2829
+ "learning_rate": 7.5158801290619e-05,
2830
+ "loss": 0.4255,
2831
+ "step": 402
2832
+ },
2833
+ {
2834
+ "epoch": 1.7831858407079646,
2835
+ "grad_norm": 0.384765625,
2836
+ "learning_rate": 7.469590321958662e-05,
2837
+ "loss": 0.407,
2838
+ "step": 403
2839
+ },
2840
+ {
2841
+ "epoch": 1.7876106194690267,
2842
+ "grad_norm": 0.388671875,
2843
+ "learning_rate": 7.423358372269455e-05,
2844
+ "loss": 0.4208,
2845
+ "step": 404
2846
+ },
2847
+ {
2848
+ "epoch": 1.7920353982300885,
2849
+ "grad_norm": 0.37109375,
2850
+ "learning_rate": 7.377185337080442e-05,
2851
+ "loss": 0.4167,
2852
+ "step": 405
2853
+ },
2854
+ {
2855
+ "epoch": 1.7964601769911503,
2856
+ "grad_norm": 0.37890625,
2857
+ "learning_rate": 7.331072272130712e-05,
2858
+ "loss": 0.4207,
2859
+ "step": 406
2860
+ },
2861
+ {
2862
+ "epoch": 1.8008849557522124,
2863
+ "grad_norm": 0.384765625,
2864
+ "learning_rate": 7.285020231788149e-05,
2865
+ "loss": 0.4195,
2866
+ "step": 407
2867
+ },
2868
+ {
2869
+ "epoch": 1.8053097345132745,
2870
+ "grad_norm": 0.3828125,
2871
+ "learning_rate": 7.239030269025311e-05,
2872
+ "loss": 0.4033,
2873
+ "step": 408
2874
+ },
2875
+ {
2876
+ "epoch": 1.8097345132743363,
2877
+ "grad_norm": 0.36328125,
2878
+ "learning_rate": 7.193103435395378e-05,
2879
+ "loss": 0.4012,
2880
+ "step": 409
2881
+ },
2882
+ {
2883
+ "epoch": 1.8141592920353982,
2884
+ "grad_norm": 0.384765625,
2885
+ "learning_rate": 7.147240781008068e-05,
2886
+ "loss": 0.4151,
2887
+ "step": 410
2888
+ },
2889
+ {
2890
+ "epoch": 1.8185840707964602,
2891
+ "grad_norm": 0.37890625,
2892
+ "learning_rate": 7.101443354505678e-05,
2893
+ "loss": 0.4091,
2894
+ "step": 411
2895
+ },
2896
+ {
2897
+ "epoch": 1.823008849557522,
2898
+ "grad_norm": 0.392578125,
2899
+ "learning_rate": 7.055712203039055e-05,
2900
+ "loss": 0.453,
2901
+ "step": 412
2902
+ },
2903
+ {
2904
+ "epoch": 1.827433628318584,
2905
+ "grad_norm": 0.400390625,
2906
+ "learning_rate": 7.010048372243698e-05,
2907
+ "loss": 0.4153,
2908
+ "step": 413
2909
+ },
2910
+ {
2911
+ "epoch": 1.831858407079646,
2912
+ "grad_norm": 0.3671875,
2913
+ "learning_rate": 6.964452906215815e-05,
2914
+ "loss": 0.3963,
2915
+ "step": 414
2916
+ },
2917
+ {
2918
+ "epoch": 1.836283185840708,
2919
+ "grad_norm": 0.373046875,
2920
+ "learning_rate": 6.918926847488476e-05,
2921
+ "loss": 0.4217,
2922
+ "step": 415
2923
+ },
2924
+ {
2925
+ "epoch": 1.8407079646017699,
2926
+ "grad_norm": 0.37109375,
2927
+ "learning_rate": 6.873471237007754e-05,
2928
+ "loss": 0.408,
2929
+ "step": 416
2930
+ },
2931
+ {
2932
+ "epoch": 1.8451327433628317,
2933
+ "grad_norm": 0.384765625,
2934
+ "learning_rate": 6.82808711410894e-05,
2935
+ "loss": 0.4029,
2936
+ "step": 417
2937
+ },
2938
+ {
2939
+ "epoch": 1.8495575221238938,
2940
+ "grad_norm": 0.408203125,
2941
+ "learning_rate": 6.782775516492771e-05,
2942
+ "loss": 0.4386,
2943
+ "step": 418
2944
+ },
2945
+ {
2946
+ "epoch": 1.8539823008849559,
2947
+ "grad_norm": 0.373046875,
2948
+ "learning_rate": 6.73753748020171e-05,
2949
+ "loss": 0.4093,
2950
+ "step": 419
2951
+ },
2952
+ {
2953
+ "epoch": 1.8584070796460177,
2954
+ "grad_norm": 0.3828125,
2955
+ "learning_rate": 6.69237403959624e-05,
2956
+ "loss": 0.4261,
2957
+ "step": 420
2958
+ },
2959
+ {
2960
+ "epoch": 1.8628318584070795,
2961
+ "grad_norm": 0.37109375,
2962
+ "learning_rate": 6.64728622733124e-05,
2963
+ "loss": 0.3903,
2964
+ "step": 421
2965
+ },
2966
+ {
2967
+ "epoch": 1.8672566371681416,
2968
+ "grad_norm": 0.388671875,
2969
+ "learning_rate": 6.602275074332345e-05,
2970
+ "loss": 0.4108,
2971
+ "step": 422
2972
+ },
2973
+ {
2974
+ "epoch": 1.8716814159292037,
2975
+ "grad_norm": 0.3828125,
2976
+ "learning_rate": 6.5573416097724e-05,
2977
+ "loss": 0.4265,
2978
+ "step": 423
2979
+ },
2980
+ {
2981
+ "epoch": 1.8761061946902655,
2982
+ "grad_norm": 0.361328125,
2983
+ "learning_rate": 6.512486861047911e-05,
2984
+ "loss": 0.4117,
2985
+ "step": 424
2986
+ },
2987
+ {
2988
+ "epoch": 1.8805309734513274,
2989
+ "grad_norm": 0.384765625,
2990
+ "learning_rate": 6.467711853755558e-05,
2991
+ "loss": 0.4202,
2992
+ "step": 425
2993
+ },
2994
+ {
2995
+ "epoch": 1.8849557522123894,
2996
+ "grad_norm": 0.388671875,
2997
+ "learning_rate": 6.423017611668745e-05,
2998
+ "loss": 0.4054,
2999
+ "step": 426
3000
+ },
3001
+ {
3002
+ "epoch": 1.8893805309734515,
3003
+ "grad_norm": 0.376953125,
3004
+ "learning_rate": 6.378405156714202e-05,
3005
+ "loss": 0.4023,
3006
+ "step": 427
3007
+ },
3008
+ {
3009
+ "epoch": 1.893805309734513,
3010
+ "grad_norm": 0.37109375,
3011
+ "learning_rate": 6.333875508948593e-05,
3012
+ "loss": 0.4138,
3013
+ "step": 428
3014
+ },
3015
+ {
3016
+ "epoch": 1.8982300884955752,
3017
+ "grad_norm": 0.384765625,
3018
+ "learning_rate": 6.289429686535226e-05,
3019
+ "loss": 0.4308,
3020
+ "step": 429
3021
+ },
3022
+ {
3023
+ "epoch": 1.9026548672566372,
3024
+ "grad_norm": 0.404296875,
3025
+ "learning_rate": 6.245068705720739e-05,
3026
+ "loss": 0.4343,
3027
+ "step": 430
3028
+ },
3029
+ {
3030
+ "epoch": 1.907079646017699,
3031
+ "grad_norm": 0.37890625,
3032
+ "learning_rate": 6.200793580811896e-05,
3033
+ "loss": 0.4085,
3034
+ "step": 431
3035
+ },
3036
+ {
3037
+ "epoch": 1.911504424778761,
3038
+ "grad_norm": 0.41796875,
3039
+ "learning_rate": 6.15660532415237e-05,
3040
+ "loss": 0.4365,
3041
+ "step": 432
3042
+ },
3043
+ {
3044
+ "epoch": 1.915929203539823,
3045
+ "grad_norm": 0.400390625,
3046
+ "learning_rate": 6.112504946099604e-05,
3047
+ "loss": 0.4068,
3048
+ "step": 433
3049
+ },
3050
+ {
3051
+ "epoch": 1.920353982300885,
3052
+ "grad_norm": 0.38671875,
3053
+ "learning_rate": 6.0684934550017184e-05,
3054
+ "loss": 0.4157,
3055
+ "step": 434
3056
+ },
3057
+ {
3058
+ "epoch": 1.924778761061947,
3059
+ "grad_norm": 0.376953125,
3060
+ "learning_rate": 6.024571857174443e-05,
3061
+ "loss": 0.438,
3062
+ "step": 435
3063
+ },
3064
+ {
3065
+ "epoch": 1.9292035398230087,
3066
+ "grad_norm": 0.375,
3067
+ "learning_rate": 5.980741156878109e-05,
3068
+ "loss": 0.41,
3069
+ "step": 436
3070
+ },
3071
+ {
3072
+ "epoch": 1.9336283185840708,
3073
+ "grad_norm": 0.365234375,
3074
+ "learning_rate": 5.9370023562946986e-05,
3075
+ "loss": 0.4104,
3076
+ "step": 437
3077
+ },
3078
+ {
3079
+ "epoch": 1.9380530973451329,
3080
+ "grad_norm": 0.37890625,
3081
+ "learning_rate": 5.8933564555049105e-05,
3082
+ "loss": 0.4245,
3083
+ "step": 438
3084
+ },
3085
+ {
3086
+ "epoch": 1.9424778761061947,
3087
+ "grad_norm": 0.3671875,
3088
+ "learning_rate": 5.849804452465317e-05,
3089
+ "loss": 0.4005,
3090
+ "step": 439
3091
+ },
3092
+ {
3093
+ "epoch": 1.9469026548672566,
3094
+ "grad_norm": 0.375,
3095
+ "learning_rate": 5.806347342985521e-05,
3096
+ "loss": 0.3941,
3097
+ "step": 440
3098
+ },
3099
+ {
3100
+ "epoch": 1.9513274336283186,
3101
+ "grad_norm": 0.3671875,
3102
+ "learning_rate": 5.7629861207054136e-05,
3103
+ "loss": 0.4167,
3104
+ "step": 441
3105
+ },
3106
+ {
3107
+ "epoch": 1.9557522123893807,
3108
+ "grad_norm": 0.3671875,
3109
+ "learning_rate": 5.7197217770724245e-05,
3110
+ "loss": 0.4032,
3111
+ "step": 442
3112
+ },
3113
+ {
3114
+ "epoch": 1.9601769911504425,
3115
+ "grad_norm": 0.376953125,
3116
+ "learning_rate": 5.6765553013188766e-05,
3117
+ "loss": 0.408,
3118
+ "step": 443
3119
+ },
3120
+ {
3121
+ "epoch": 1.9646017699115044,
3122
+ "grad_norm": 0.3828125,
3123
+ "learning_rate": 5.633487680439361e-05,
3124
+ "loss": 0.4142,
3125
+ "step": 444
3126
+ },
3127
+ {
3128
+ "epoch": 1.9690265486725664,
3129
+ "grad_norm": 0.419921875,
3130
+ "learning_rate": 5.5905198991681695e-05,
3131
+ "loss": 0.4044,
3132
+ "step": 445
3133
+ },
3134
+ {
3135
+ "epoch": 1.9734513274336283,
3136
+ "grad_norm": 0.38671875,
3137
+ "learning_rate": 5.547652939956764e-05,
3138
+ "loss": 0.3855,
3139
+ "step": 446
3140
+ },
3141
+ {
3142
+ "epoch": 1.9778761061946901,
3143
+ "grad_norm": 0.365234375,
3144
+ "learning_rate": 5.5048877829513424e-05,
3145
+ "loss": 0.3921,
3146
+ "step": 447
3147
+ },
3148
+ {
3149
+ "epoch": 1.9823008849557522,
3150
+ "grad_norm": 0.39453125,
3151
+ "learning_rate": 5.462225405970401e-05,
3152
+ "loss": 0.4108,
3153
+ "step": 448
3154
+ },
3155
+ {
3156
+ "epoch": 1.9867256637168142,
3157
+ "grad_norm": 0.39453125,
3158
+ "learning_rate": 5.419666784482398e-05,
3159
+ "loss": 0.427,
3160
+ "step": 449
3161
+ },
3162
+ {
3163
+ "epoch": 1.991150442477876,
3164
+ "grad_norm": 0.396484375,
3165
+ "learning_rate": 5.3772128915834184e-05,
3166
+ "loss": 0.404,
3167
+ "step": 450
3168
+ },
3169
+ {
3170
+ "epoch": 1.995575221238938,
3171
+ "grad_norm": 0.408203125,
3172
+ "learning_rate": 5.3348646979749687e-05,
3173
+ "loss": 0.4006,
3174
+ "step": 451
3175
+ },
3176
+ {
3177
+ "epoch": 2.0,
3178
+ "grad_norm": 0.47265625,
3179
+ "learning_rate": 5.29262317194174e-05,
3180
+ "loss": 0.4206,
3181
+ "step": 452
3182
+ },
3183
+ {
3184
+ "epoch": 2.0,
3185
+ "eval_loss": 0.4563676416873932,
3186
+ "eval_runtime": 5.7045,
3187
+ "eval_samples_per_second": 146.901,
3188
+ "eval_steps_per_second": 18.406,
3189
+ "step": 452
3190
+ }
3191
+ ],
3192
+ "logging_steps": 1,
3193
+ "max_steps": 678,
3194
+ "num_input_tokens_seen": 0,
3195
+ "num_train_epochs": 3,
3196
+ "save_steps": 500,
3197
+ "stateful_callbacks": {
3198
+ "TrainerControl": {
3199
+ "args": {
3200
+ "should_epoch_stop": false,
3201
+ "should_evaluate": false,
3202
+ "should_log": false,
3203
+ "should_save": true,
3204
+ "should_training_stop": false
3205
+ },
3206
+ "attributes": {}
3207
+ }
3208
+ },
3209
+ "total_flos": 6.948804377680282e+16,
3210
+ "train_batch_size": 60,
3211
+ "trial_name": null,
3212
+ "trial_params": null
3213
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:db2c3f970b2251a9f1e03ac99629a4b7a340c83e86517d4ecc0b4995a8d47b70
3
+ size 5368