shivank21 commited on
Commit
65b5d02
·
verified ·
1 Parent(s): cbb3657

Upload folder using huggingface_hub

Browse files
config.json ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 4096,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
14
+ "max_position_embeddings": 4096,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 32,
18
+ "num_hidden_layers": 30,
19
+ "num_key_value_heads": 32,
20
+ "pretraining_tp": 1,
21
+ "rms_norm_eps": 1e-06,
22
+ "rope_scaling": null,
23
+ "rope_theta": 10000.0,
24
+ "tie_word_embeddings": false,
25
+ "torch_dtype": "bfloat16",
26
+ "transformers_version": "4.51.3",
27
+ "use_cache": false,
28
+ "vocab_size": 102400
29
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 100000,
4
+ "eos_token_id": 100001,
5
+ "transformers_version": "4.51.3"
6
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<|begin▁of▁sentence|>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<|end▁of▁sentence|>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<|end▁of▁sentence|>",
18
+ "lstrip": false,
19
+ "normalized": true,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ }
23
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,138 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "100000": {
7
+ "content": "<|begin▁of▁sentence|>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "100001": {
15
+ "content": "<|end▁of▁sentence|>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "100002": {
23
+ "content": "ø",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": false
29
+ },
30
+ "100003": {
31
+ "content": "ö",
32
+ "lstrip": false,
33
+ "normalized": true,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": false
37
+ },
38
+ "100004": {
39
+ "content": "ú",
40
+ "lstrip": false,
41
+ "normalized": true,
42
+ "rstrip": false,
43
+ "single_word": false,
44
+ "special": false
45
+ },
46
+ "100005": {
47
+ "content": "ÿ",
48
+ "lstrip": false,
49
+ "normalized": true,
50
+ "rstrip": false,
51
+ "single_word": false,
52
+ "special": false
53
+ },
54
+ "100006": {
55
+ "content": "õ",
56
+ "lstrip": false,
57
+ "normalized": true,
58
+ "rstrip": false,
59
+ "single_word": false,
60
+ "special": false
61
+ },
62
+ "100007": {
63
+ "content": "÷",
64
+ "lstrip": false,
65
+ "normalized": true,
66
+ "rstrip": false,
67
+ "single_word": false,
68
+ "special": false
69
+ },
70
+ "100008": {
71
+ "content": "û",
72
+ "lstrip": false,
73
+ "normalized": true,
74
+ "rstrip": false,
75
+ "single_word": false,
76
+ "special": false
77
+ },
78
+ "100009": {
79
+ "content": "ý",
80
+ "lstrip": false,
81
+ "normalized": true,
82
+ "rstrip": false,
83
+ "single_word": false,
84
+ "special": false
85
+ },
86
+ "100010": {
87
+ "content": "À",
88
+ "lstrip": false,
89
+ "normalized": true,
90
+ "rstrip": false,
91
+ "single_word": false,
92
+ "special": false
93
+ },
94
+ "100011": {
95
+ "content": "ù",
96
+ "lstrip": false,
97
+ "normalized": true,
98
+ "rstrip": false,
99
+ "single_word": false,
100
+ "special": false
101
+ },
102
+ "100012": {
103
+ "content": "Á",
104
+ "lstrip": false,
105
+ "normalized": true,
106
+ "rstrip": false,
107
+ "single_word": false,
108
+ "special": false
109
+ },
110
+ "100013": {
111
+ "content": "þ",
112
+ "lstrip": false,
113
+ "normalized": true,
114
+ "rstrip": false,
115
+ "single_word": false,
116
+ "special": false
117
+ },
118
+ "100014": {
119
+ "content": "ü",
120
+ "lstrip": false,
121
+ "normalized": true,
122
+ "rstrip": false,
123
+ "single_word": false,
124
+ "special": false
125
+ }
126
+ },
127
+ "bos_token": "<|begin▁of▁sentence|>",
128
+ "clean_up_tokenization_spaces": false,
129
+ "eos_token": "<|end▁of▁sentence|>",
130
+ "extra_special_tokens": {},
131
+ "legacy": true,
132
+ "model_max_length": 4096,
133
+ "pad_token": "<|end▁of▁sentence|>",
134
+ "sp_model_kwargs": {},
135
+ "tokenizer_class": "LlamaTokenizerFast",
136
+ "unk_token": null,
137
+ "use_default_system_prompt": false
138
+ }
trainer_state.json ADDED
@@ -0,0 +1,2031 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 3750,
3
+ "best_metric": 0.5079798102378845,
4
+ "best_model_checkpoint": "./results/checkpoint-3500",
5
+ "epoch": 4.997752808988764,
6
+ "eval_steps": 250,
7
+ "global_step": 9455,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.026437541308658295,
14
+ "grad_norm": 0.2475886046886444,
15
+ "learning_rate": 0.00034625958983852136,
16
+ "loss": 0.8762,
17
+ "mean_token_accuracy": 0.8128470236063003,
18
+ "num_tokens": 1638400.0,
19
+ "step": 50
20
+ },
21
+ {
22
+ "epoch": 0.05287508261731659,
23
+ "grad_norm": 0.25265493988990784,
24
+ "learning_rate": 0.000407611186724682,
25
+ "loss": 0.6461,
26
+ "mean_token_accuracy": 0.8552572166919709,
27
+ "num_tokens": 3276800.0,
28
+ "step": 100
29
+ },
30
+ {
31
+ "epoch": 0.07931262392597488,
32
+ "grad_norm": 0.21018819510936737,
33
+ "learning_rate": 0.0004434995702624468,
34
+ "loss": 0.635,
35
+ "mean_token_accuracy": 0.8568304070830345,
36
+ "num_tokens": 4915200.0,
37
+ "step": 150
38
+ },
39
+ {
40
+ "epoch": 0.10575016523463318,
41
+ "grad_norm": 0.23957596719264984,
42
+ "learning_rate": 0.0004689627836108426,
43
+ "loss": 0.6235,
44
+ "mean_token_accuracy": 0.8593035244941711,
45
+ "num_tokens": 6553600.0,
46
+ "step": 200
47
+ },
48
+ {
49
+ "epoch": 0.13218770654329148,
50
+ "grad_norm": 0.192116379737854,
51
+ "learning_rate": 0.0004887135863147016,
52
+ "loss": 0.6217,
53
+ "step": 250
54
+ },
55
+ {
56
+ "epoch": 0.13218770654329148,
57
+ "eval_loss": 0.630805492401123,
58
+ "eval_mean_token_accuracy": 0.8574222893971973,
59
+ "eval_num_tokens": 8192000.0,
60
+ "eval_runtime": 1444.1199,
61
+ "eval_samples_per_second": 5.238,
62
+ "eval_steps_per_second": 0.655,
63
+ "step": 250
64
+ },
65
+ {
66
+ "epoch": 0.15862524785194976,
67
+ "grad_norm": 0.2664429545402527,
68
+ "learning_rate": 0.0004991822047759241,
69
+ "loss": 0.6239,
70
+ "mean_token_accuracy": 0.859162701368332,
71
+ "num_tokens": 9830400.0,
72
+ "step": 300
73
+ },
74
+ {
75
+ "epoch": 0.18506278916060806,
76
+ "grad_norm": 0.5154497027397156,
77
+ "learning_rate": 0.0004964562206956711,
78
+ "loss": 0.6973,
79
+ "mean_token_accuracy": 0.8490096586942673,
80
+ "num_tokens": 11468800.0,
81
+ "step": 350
82
+ },
83
+ {
84
+ "epoch": 0.21150033046926636,
85
+ "grad_norm": 0.2535574436187744,
86
+ "learning_rate": 0.0004937302366154182,
87
+ "loss": 0.6079,
88
+ "mean_token_accuracy": 0.8621737584471703,
89
+ "num_tokens": 13107200.0,
90
+ "step": 400
91
+ },
92
+ {
93
+ "epoch": 0.23793787177792466,
94
+ "grad_norm": 0.16894356906414032,
95
+ "learning_rate": 0.0004910042525351653,
96
+ "loss": 0.6082,
97
+ "mean_token_accuracy": 0.8616935542225838,
98
+ "num_tokens": 14745600.0,
99
+ "step": 450
100
+ },
101
+ {
102
+ "epoch": 0.26437541308658297,
103
+ "grad_norm": 0.1515253186225891,
104
+ "learning_rate": 0.00048827826845491225,
105
+ "loss": 0.6043,
106
+ "step": 500
107
+ },
108
+ {
109
+ "epoch": 0.26437541308658297,
110
+ "eval_loss": 0.6044251918792725,
111
+ "eval_mean_token_accuracy": 0.8618915110637472,
112
+ "eval_num_tokens": 16384000.0,
113
+ "eval_runtime": 1444.8043,
114
+ "eval_samples_per_second": 5.236,
115
+ "eval_steps_per_second": 0.655,
116
+ "step": 500
117
+ },
118
+ {
119
+ "epoch": 0.29081295439524124,
120
+ "grad_norm": 0.15358857810497284,
121
+ "learning_rate": 0.0004855522843746593,
122
+ "loss": 0.6137,
123
+ "mean_token_accuracy": 0.8617021070420742,
124
+ "num_tokens": 18022400.0,
125
+ "step": 550
126
+ },
127
+ {
128
+ "epoch": 0.3172504957038995,
129
+ "grad_norm": 0.34029558300971985,
130
+ "learning_rate": 0.00048282630029440626,
131
+ "loss": 0.5973,
132
+ "mean_token_accuracy": 0.8634732463955879,
133
+ "num_tokens": 19660800.0,
134
+ "step": 600
135
+ },
136
+ {
137
+ "epoch": 0.34368803701255785,
138
+ "grad_norm": 0.22880446910858154,
139
+ "learning_rate": 0.00048010031621415335,
140
+ "loss": 0.5984,
141
+ "mean_token_accuracy": 0.8633871012926102,
142
+ "num_tokens": 21299200.0,
143
+ "step": 650
144
+ },
145
+ {
146
+ "epoch": 0.3701255783212161,
147
+ "grad_norm": 0.17699383199214935,
148
+ "learning_rate": 0.0004773743321339004,
149
+ "loss": 0.5871,
150
+ "mean_token_accuracy": 0.8653806248307228,
151
+ "num_tokens": 22937600.0,
152
+ "step": 700
153
+ },
154
+ {
155
+ "epoch": 0.3965631196298744,
156
+ "grad_norm": 0.18763592839241028,
157
+ "learning_rate": 0.00047464834805364736,
158
+ "loss": 0.5889,
159
+ "step": 750
160
+ },
161
+ {
162
+ "epoch": 0.3965631196298744,
163
+ "eval_loss": 0.5848008394241333,
164
+ "eval_mean_token_accuracy": 0.865356788007704,
165
+ "eval_num_tokens": 24576000.0,
166
+ "eval_runtime": 1443.8073,
167
+ "eval_samples_per_second": 5.24,
168
+ "eval_steps_per_second": 0.655,
169
+ "step": 750
170
+ },
171
+ {
172
+ "epoch": 0.4230006609385327,
173
+ "grad_norm": 0.1546117663383484,
174
+ "learning_rate": 0.0004719223639733944,
175
+ "loss": 0.5821,
176
+ "mean_token_accuracy": 0.8651845099031925,
177
+ "num_tokens": 26214400.0,
178
+ "step": 800
179
+ },
180
+ {
181
+ "epoch": 0.449438202247191,
182
+ "grad_norm": 0.13760164380073547,
183
+ "learning_rate": 0.0004691963798931415,
184
+ "loss": 0.5824,
185
+ "mean_token_accuracy": 0.8658082884550095,
186
+ "num_tokens": 27852800.0,
187
+ "step": 850
188
+ },
189
+ {
190
+ "epoch": 0.47587574355584933,
191
+ "grad_norm": 0.1396537572145462,
192
+ "learning_rate": 0.00046647039581288846,
193
+ "loss": 0.5707,
194
+ "mean_token_accuracy": 0.8684304764866829,
195
+ "num_tokens": 29491200.0,
196
+ "step": 900
197
+ },
198
+ {
199
+ "epoch": 0.5023132848645075,
200
+ "grad_norm": 0.13558807969093323,
201
+ "learning_rate": 0.0004637444117326355,
202
+ "loss": 0.58,
203
+ "mean_token_accuracy": 0.8662952136993408,
204
+ "num_tokens": 31129600.0,
205
+ "step": 950
206
+ },
207
+ {
208
+ "epoch": 0.5287508261731659,
209
+ "grad_norm": 0.12448519468307495,
210
+ "learning_rate": 0.0004610184276523825,
211
+ "loss": 0.5769,
212
+ "step": 1000
213
+ },
214
+ {
215
+ "epoch": 0.5287508261731659,
216
+ "eval_loss": 0.5692248940467834,
217
+ "eval_mean_token_accuracy": 0.8682624975546218,
218
+ "eval_num_tokens": 32768000.0,
219
+ "eval_runtime": 1443.7648,
220
+ "eval_samples_per_second": 5.24,
221
+ "eval_steps_per_second": 0.655,
222
+ "step": 1000
223
+ },
224
+ {
225
+ "epoch": 0.5551883674818242,
226
+ "grad_norm": 0.12668457627296448,
227
+ "learning_rate": 0.00045829244357212956,
228
+ "loss": 0.5683,
229
+ "mean_token_accuracy": 0.8675177201628685,
230
+ "num_tokens": 34406400.0,
231
+ "step": 1050
232
+ },
233
+ {
234
+ "epoch": 0.5816259087904825,
235
+ "grad_norm": 0.1283566802740097,
236
+ "learning_rate": 0.0004555664594918766,
237
+ "loss": 0.5686,
238
+ "mean_token_accuracy": 0.8686253690719604,
239
+ "num_tokens": 36044800.0,
240
+ "step": 1100
241
+ },
242
+ {
243
+ "epoch": 0.6080634500991408,
244
+ "grad_norm": 0.15960244834423065,
245
+ "learning_rate": 0.0004528404754116236,
246
+ "loss": 0.5644,
247
+ "mean_token_accuracy": 0.8688807463645936,
248
+ "num_tokens": 37683200.0,
249
+ "step": 1150
250
+ },
251
+ {
252
+ "epoch": 0.634500991407799,
253
+ "grad_norm": 0.11330343037843704,
254
+ "learning_rate": 0.0004501144913313706,
255
+ "loss": 0.5649,
256
+ "mean_token_accuracy": 0.8686217039823532,
257
+ "num_tokens": 39321600.0,
258
+ "step": 1200
259
+ },
260
+ {
261
+ "epoch": 0.6609385327164574,
262
+ "grad_norm": 0.12131626158952713,
263
+ "learning_rate": 0.0004473885072511177,
264
+ "loss": 0.5638,
265
+ "step": 1250
266
+ },
267
+ {
268
+ "epoch": 0.6609385327164574,
269
+ "eval_loss": 0.5589728951454163,
270
+ "eval_mean_token_accuracy": 0.8698779833493223,
271
+ "eval_num_tokens": 40960000.0,
272
+ "eval_runtime": 1443.9339,
273
+ "eval_samples_per_second": 5.239,
274
+ "eval_steps_per_second": 0.655,
275
+ "step": 1250
276
+ },
277
+ {
278
+ "epoch": 0.6873760740251157,
279
+ "grad_norm": 0.135975182056427,
280
+ "learning_rate": 0.0004446625231708647,
281
+ "loss": 0.5606,
282
+ "mean_token_accuracy": 0.8690710559487342,
283
+ "num_tokens": 42598400.0,
284
+ "step": 1300
285
+ },
286
+ {
287
+ "epoch": 0.713813615333774,
288
+ "grad_norm": 0.15456999838352203,
289
+ "learning_rate": 0.0004419365390906117,
290
+ "loss": 0.5495,
291
+ "mean_token_accuracy": 0.8715310388803482,
292
+ "num_tokens": 44236800.0,
293
+ "step": 1350
294
+ },
295
+ {
296
+ "epoch": 0.7402511566424322,
297
+ "grad_norm": 0.17465010285377502,
298
+ "learning_rate": 0.00043921055501035873,
299
+ "loss": 0.5424,
300
+ "mean_token_accuracy": 0.873205646276474,
301
+ "num_tokens": 45875200.0,
302
+ "step": 1400
303
+ },
304
+ {
305
+ "epoch": 0.7666886979510905,
306
+ "grad_norm": 0.11691431701183319,
307
+ "learning_rate": 0.0004364845709301058,
308
+ "loss": 0.5407,
309
+ "mean_token_accuracy": 0.8731714335083961,
310
+ "num_tokens": 47513600.0,
311
+ "step": 1450
312
+ },
313
+ {
314
+ "epoch": 0.7931262392597488,
315
+ "grad_norm": 0.1413394808769226,
316
+ "learning_rate": 0.0004337585868498528,
317
+ "loss": 0.5477,
318
+ "step": 1500
319
+ },
320
+ {
321
+ "epoch": 0.7931262392597488,
322
+ "eval_loss": 0.5501642227172852,
323
+ "eval_mean_token_accuracy": 0.8714489584619348,
324
+ "eval_num_tokens": 49152000.0,
325
+ "eval_runtime": 1443.5916,
326
+ "eval_samples_per_second": 5.24,
327
+ "eval_steps_per_second": 0.655,
328
+ "step": 1500
329
+ },
330
+ {
331
+ "epoch": 0.8195637805684072,
332
+ "grad_norm": 0.10578064620494843,
333
+ "learning_rate": 0.00043103260276959983,
334
+ "loss": 0.5508,
335
+ "mean_token_accuracy": 0.8713572224974633,
336
+ "num_tokens": 50790400.0,
337
+ "step": 1550
338
+ },
339
+ {
340
+ "epoch": 0.8460013218770654,
341
+ "grad_norm": 0.10736281424760818,
342
+ "learning_rate": 0.0004283066186893468,
343
+ "loss": 0.5481,
344
+ "mean_token_accuracy": 0.8717375376820564,
345
+ "num_tokens": 52428800.0,
346
+ "step": 1600
347
+ },
348
+ {
349
+ "epoch": 0.8724388631857237,
350
+ "grad_norm": 0.12125309556722641,
351
+ "learning_rate": 0.0004255806346090939,
352
+ "loss": 0.5537,
353
+ "mean_token_accuracy": 0.86990652769804,
354
+ "num_tokens": 54067200.0,
355
+ "step": 1650
356
+ },
357
+ {
358
+ "epoch": 0.898876404494382,
359
+ "grad_norm": 0.10367817431688309,
360
+ "learning_rate": 0.00042285465052884093,
361
+ "loss": 0.5446,
362
+ "mean_token_accuracy": 0.8718084079027176,
363
+ "num_tokens": 55705600.0,
364
+ "step": 1700
365
+ },
366
+ {
367
+ "epoch": 0.9253139458030403,
368
+ "grad_norm": 0.19373266398906708,
369
+ "learning_rate": 0.0004201286664485879,
370
+ "loss": 0.5486,
371
+ "step": 1750
372
+ },
373
+ {
374
+ "epoch": 0.9253139458030403,
375
+ "eval_loss": 0.5393346548080444,
376
+ "eval_mean_token_accuracy": 0.8730530525236755,
377
+ "eval_num_tokens": 57344000.0,
378
+ "eval_runtime": 1444.4931,
379
+ "eval_samples_per_second": 5.237,
380
+ "eval_steps_per_second": 0.655,
381
+ "step": 1750
382
+ },
383
+ {
384
+ "epoch": 0.9517514871116987,
385
+ "grad_norm": 0.10371921956539154,
386
+ "learning_rate": 0.00041740268236833495,
387
+ "loss": 0.5381,
388
+ "mean_token_accuracy": 0.8720436228811741,
389
+ "num_tokens": 58982400.0,
390
+ "step": 1800
391
+ },
392
+ {
393
+ "epoch": 0.9781890284203569,
394
+ "grad_norm": 0.09864738583564758,
395
+ "learning_rate": 0.00041467669828808203,
396
+ "loss": 0.5321,
397
+ "mean_token_accuracy": 0.8747250729799271,
398
+ "num_tokens": 60620800.0,
399
+ "step": 1850
400
+ },
401
+ {
402
+ "epoch": 1.0042300066093852,
403
+ "grad_norm": 0.12164577841758728,
404
+ "learning_rate": 0.00041200523388943414,
405
+ "loss": 0.519,
406
+ "mean_token_accuracy": 0.8759235533966025,
407
+ "num_tokens": 62234624.0,
408
+ "step": 1900
409
+ },
410
+ {
411
+ "epoch": 1.0306675479180436,
412
+ "grad_norm": 0.16176024079322815,
413
+ "learning_rate": 0.0004092792498091811,
414
+ "loss": 0.4182,
415
+ "mean_token_accuracy": 0.8925769734382629,
416
+ "num_tokens": 63873024.0,
417
+ "step": 1950
418
+ },
419
+ {
420
+ "epoch": 1.057105089226702,
421
+ "grad_norm": 0.09831411391496658,
422
+ "learning_rate": 0.00040655326572892816,
423
+ "loss": 0.426,
424
+ "step": 2000
425
+ },
426
+ {
427
+ "epoch": 1.057105089226702,
428
+ "eval_loss": 0.5495159029960632,
429
+ "eval_mean_token_accuracy": 0.8728643428321621,
430
+ "eval_num_tokens": 65511424.0,
431
+ "eval_runtime": 1443.7886,
432
+ "eval_samples_per_second": 5.24,
433
+ "eval_steps_per_second": 0.655,
434
+ "step": 2000
435
+ },
436
+ {
437
+ "epoch": 1.0835426305353602,
438
+ "grad_norm": 0.11064837127923965,
439
+ "learning_rate": 0.00040382728164867513,
440
+ "loss": 0.4203,
441
+ "mean_token_accuracy": 0.891697817593813,
442
+ "num_tokens": 67149824.0,
443
+ "step": 2050
444
+ },
445
+ {
446
+ "epoch": 1.1099801718440185,
447
+ "grad_norm": 0.12444479018449783,
448
+ "learning_rate": 0.0004011012975684222,
449
+ "loss": 0.4324,
450
+ "mean_token_accuracy": 0.8894703006744384,
451
+ "num_tokens": 68788224.0,
452
+ "step": 2100
453
+ },
454
+ {
455
+ "epoch": 1.1364177131526767,
456
+ "grad_norm": 0.10909954458475113,
457
+ "learning_rate": 0.00039837531348816925,
458
+ "loss": 0.4243,
459
+ "mean_token_accuracy": 0.8913825675845146,
460
+ "num_tokens": 70426624.0,
461
+ "step": 2150
462
+ },
463
+ {
464
+ "epoch": 1.162855254461335,
465
+ "grad_norm": 0.09556654095649719,
466
+ "learning_rate": 0.00039564932940791623,
467
+ "loss": 0.426,
468
+ "mean_token_accuracy": 0.8912683191895485,
469
+ "num_tokens": 72065024.0,
470
+ "step": 2200
471
+ },
472
+ {
473
+ "epoch": 1.1892927957699935,
474
+ "grad_norm": 0.11516989767551422,
475
+ "learning_rate": 0.00039292334532766327,
476
+ "loss": 0.4313,
477
+ "step": 2250
478
+ },
479
+ {
480
+ "epoch": 1.1892927957699935,
481
+ "eval_loss": 0.5482946634292603,
482
+ "eval_mean_token_accuracy": 0.873437058408971,
483
+ "eval_num_tokens": 73703424.0,
484
+ "eval_runtime": 1445.2243,
485
+ "eval_samples_per_second": 5.234,
486
+ "eval_steps_per_second": 0.655,
487
+ "step": 2250
488
+ },
489
+ {
490
+ "epoch": 1.2157303370786516,
491
+ "grad_norm": 0.11027960479259491,
492
+ "learning_rate": 0.00039019736124741035,
493
+ "loss": 0.4393,
494
+ "mean_token_accuracy": 0.889146803766489,
495
+ "num_tokens": 75341824.0,
496
+ "step": 2300
497
+ },
498
+ {
499
+ "epoch": 1.24216787838731,
500
+ "grad_norm": 0.118904247879982,
501
+ "learning_rate": 0.00038747137716715733,
502
+ "loss": 0.4408,
503
+ "mean_token_accuracy": 0.8877504804730415,
504
+ "num_tokens": 76980224.0,
505
+ "step": 2350
506
+ },
507
+ {
508
+ "epoch": 1.2686054196959682,
509
+ "grad_norm": 0.10859134048223495,
510
+ "learning_rate": 0.00038474539308690437,
511
+ "loss": 0.4349,
512
+ "mean_token_accuracy": 0.8895466673374176,
513
+ "num_tokens": 78618624.0,
514
+ "step": 2400
515
+ },
516
+ {
517
+ "epoch": 1.2950429610046266,
518
+ "grad_norm": 0.09974760562181473,
519
+ "learning_rate": 0.0003820194090066514,
520
+ "loss": 0.4375,
521
+ "mean_token_accuracy": 0.8887909257411957,
522
+ "num_tokens": 80257024.0,
523
+ "step": 2450
524
+ },
525
+ {
526
+ "epoch": 1.321480502313285,
527
+ "grad_norm": 0.10627176612615585,
528
+ "learning_rate": 0.00037929342492639843,
529
+ "loss": 0.4354,
530
+ "step": 2500
531
+ },
532
+ {
533
+ "epoch": 1.321480502313285,
534
+ "eval_loss": 0.5409240126609802,
535
+ "eval_mean_token_accuracy": 0.8742006784639701,
536
+ "eval_num_tokens": 81895424.0,
537
+ "eval_runtime": 1443.9561,
538
+ "eval_samples_per_second": 5.239,
539
+ "eval_steps_per_second": 0.655,
540
+ "step": 2500
541
+ },
542
+ {
543
+ "epoch": 1.3479180436219431,
544
+ "grad_norm": 0.10294341295957565,
545
+ "learning_rate": 0.00037656744084614547,
546
+ "loss": 0.4372,
547
+ "mean_token_accuracy": 0.8892436383664608,
548
+ "num_tokens": 83533824.0,
549
+ "step": 2550
550
+ },
551
+ {
552
+ "epoch": 1.3743555849306015,
553
+ "grad_norm": 0.12393230944871902,
554
+ "learning_rate": 0.0003738414567658925,
555
+ "loss": 0.4386,
556
+ "mean_token_accuracy": 0.8886467415094376,
557
+ "num_tokens": 85172224.0,
558
+ "step": 2600
559
+ },
560
+ {
561
+ "epoch": 1.4007931262392597,
562
+ "grad_norm": 0.09495224803686142,
563
+ "learning_rate": 0.0003711154726856395,
564
+ "loss": 0.4302,
565
+ "mean_token_accuracy": 0.8906915840506554,
566
+ "num_tokens": 86810624.0,
567
+ "step": 2650
568
+ },
569
+ {
570
+ "epoch": 1.427230667547918,
571
+ "grad_norm": 0.10494447499513626,
572
+ "learning_rate": 0.00036838948860538656,
573
+ "loss": 0.4312,
574
+ "mean_token_accuracy": 0.8900836914777756,
575
+ "num_tokens": 88449024.0,
576
+ "step": 2700
577
+ },
578
+ {
579
+ "epoch": 1.4536682088565764,
580
+ "grad_norm": 0.12757033109664917,
581
+ "learning_rate": 0.0003656635045251336,
582
+ "loss": 0.437,
583
+ "step": 2750
584
+ },
585
+ {
586
+ "epoch": 1.4536682088565764,
587
+ "eval_loss": 0.53493332862854,
588
+ "eval_mean_token_accuracy": 0.8759168862162382,
589
+ "eval_num_tokens": 90087424.0,
590
+ "eval_runtime": 1443.9928,
591
+ "eval_samples_per_second": 5.239,
592
+ "eval_steps_per_second": 0.655,
593
+ "step": 2750
594
+ },
595
+ {
596
+ "epoch": 1.4801057501652346,
597
+ "grad_norm": 0.12719060480594635,
598
+ "learning_rate": 0.0003629375204448806,
599
+ "loss": 0.4384,
600
+ "mean_token_accuracy": 0.8888401076197624,
601
+ "num_tokens": 91725824.0,
602
+ "step": 2800
603
+ },
604
+ {
605
+ "epoch": 1.5065432914738928,
606
+ "grad_norm": 0.1001153439283371,
607
+ "learning_rate": 0.0003602115363646276,
608
+ "loss": 0.443,
609
+ "mean_token_accuracy": 0.8874511170387268,
610
+ "num_tokens": 93364224.0,
611
+ "step": 2850
612
+ },
613
+ {
614
+ "epoch": 1.5329808327825512,
615
+ "grad_norm": 0.10144038498401642,
616
+ "learning_rate": 0.0003574855522843747,
617
+ "loss": 0.4439,
618
+ "mean_token_accuracy": 0.8875879681110382,
619
+ "num_tokens": 95002624.0,
620
+ "step": 2900
621
+ },
622
+ {
623
+ "epoch": 1.5594183740912095,
624
+ "grad_norm": 0.10192844271659851,
625
+ "learning_rate": 0.0003547595682041217,
626
+ "loss": 0.4345,
627
+ "mean_token_accuracy": 0.8896059301495552,
628
+ "num_tokens": 96641024.0,
629
+ "step": 2950
630
+ },
631
+ {
632
+ "epoch": 1.585855915399868,
633
+ "grad_norm": 0.11319652944803238,
634
+ "learning_rate": 0.0003520335841238687,
635
+ "loss": 0.4378,
636
+ "step": 3000
637
+ },
638
+ {
639
+ "epoch": 1.585855915399868,
640
+ "eval_loss": 0.5288907289505005,
641
+ "eval_mean_token_accuracy": 0.8770441555951665,
642
+ "eval_num_tokens": 98279424.0,
643
+ "eval_runtime": 1443.9606,
644
+ "eval_samples_per_second": 5.239,
645
+ "eval_steps_per_second": 0.655,
646
+ "step": 3000
647
+ },
648
+ {
649
+ "epoch": 1.612293456708526,
650
+ "grad_norm": 0.10589715093374252,
651
+ "learning_rate": 0.00034930760004361574,
652
+ "loss": 0.4334,
653
+ "mean_token_accuracy": 0.8895524734258652,
654
+ "num_tokens": 99917824.0,
655
+ "step": 3050
656
+ },
657
+ {
658
+ "epoch": 1.6387309980171842,
659
+ "grad_norm": 0.09860669076442719,
660
+ "learning_rate": 0.0003465816159633628,
661
+ "loss": 0.4377,
662
+ "mean_token_accuracy": 0.8887499925494194,
663
+ "num_tokens": 101556224.0,
664
+ "step": 3100
665
+ },
666
+ {
667
+ "epoch": 1.6651685393258426,
668
+ "grad_norm": 0.11345923691987991,
669
+ "learning_rate": 0.0003438556318831098,
670
+ "loss": 0.4413,
671
+ "mean_token_accuracy": 0.8883473780751229,
672
+ "num_tokens": 103194624.0,
673
+ "step": 3150
674
+ },
675
+ {
676
+ "epoch": 1.691606080634501,
677
+ "grad_norm": 0.10729037970304489,
678
+ "learning_rate": 0.00034112964780285684,
679
+ "loss": 0.4365,
680
+ "mean_token_accuracy": 0.8893462771177292,
681
+ "num_tokens": 104833024.0,
682
+ "step": 3200
683
+ },
684
+ {
685
+ "epoch": 1.7180436219431594,
686
+ "grad_norm": 0.0929078608751297,
687
+ "learning_rate": 0.0003384036637226039,
688
+ "loss": 0.4339,
689
+ "step": 3250
690
+ },
691
+ {
692
+ "epoch": 1.7180436219431594,
693
+ "eval_loss": 0.522659182548523,
694
+ "eval_mean_token_accuracy": 0.8778351331740051,
695
+ "eval_num_tokens": 106471424.0,
696
+ "eval_runtime": 1444.1286,
697
+ "eval_samples_per_second": 5.238,
698
+ "eval_steps_per_second": 0.655,
699
+ "step": 3250
700
+ },
701
+ {
702
+ "epoch": 1.7444811632518176,
703
+ "grad_norm": 0.09583564847707748,
704
+ "learning_rate": 0.0003356776796423509,
705
+ "loss": 0.4292,
706
+ "mean_token_accuracy": 0.8906918911635876,
707
+ "num_tokens": 108109824.0,
708
+ "step": 3300
709
+ },
710
+ {
711
+ "epoch": 1.7709187045604757,
712
+ "grad_norm": 0.09396171569824219,
713
+ "learning_rate": 0.00033295169556209794,
714
+ "loss": 0.4259,
715
+ "mean_token_accuracy": 0.8916892632842064,
716
+ "num_tokens": 109748224.0,
717
+ "step": 3350
718
+ },
719
+ {
720
+ "epoch": 1.7973562458691341,
721
+ "grad_norm": 0.08548162132501602,
722
+ "learning_rate": 0.0003302257114818449,
723
+ "loss": 0.4327,
724
+ "mean_token_accuracy": 0.8902602550387383,
725
+ "num_tokens": 111386624.0,
726
+ "step": 3400
727
+ },
728
+ {
729
+ "epoch": 1.8237937871777925,
730
+ "grad_norm": 0.1594037264585495,
731
+ "learning_rate": 0.000327499727401592,
732
+ "loss": 0.4279,
733
+ "mean_token_accuracy": 0.8916428324580192,
734
+ "num_tokens": 113025024.0,
735
+ "step": 3450
736
+ },
737
+ {
738
+ "epoch": 1.8502313284864509,
739
+ "grad_norm": 0.09137943387031555,
740
+ "learning_rate": 0.00032477374332133904,
741
+ "loss": 0.4259,
742
+ "step": 3500
743
+ },
744
+ {
745
+ "epoch": 1.8502313284864509,
746
+ "eval_loss": 0.5174182653427124,
747
+ "eval_mean_token_accuracy": 0.878870385834833,
748
+ "eval_num_tokens": 114663424.0,
749
+ "eval_runtime": 1444.5837,
750
+ "eval_samples_per_second": 5.237,
751
+ "eval_steps_per_second": 0.655,
752
+ "step": 3500
753
+ },
754
+ {
755
+ "epoch": 1.876668869795109,
756
+ "grad_norm": 0.13611243665218353,
757
+ "learning_rate": 0.000322047759241086,
758
+ "loss": 0.4334,
759
+ "mean_token_accuracy": 0.8908357685804367,
760
+ "num_tokens": 116301824.0,
761
+ "step": 3550
762
+ },
763
+ {
764
+ "epoch": 1.9031064111037672,
765
+ "grad_norm": 0.10409994423389435,
766
+ "learning_rate": 0.00031932177516083305,
767
+ "loss": 0.4396,
768
+ "mean_token_accuracy": 0.8884219127893448,
769
+ "num_tokens": 117940224.0,
770
+ "step": 3600
771
+ },
772
+ {
773
+ "epoch": 1.9295439524124256,
774
+ "grad_norm": 0.08799432218074799,
775
+ "learning_rate": 0.00031659579108058014,
776
+ "loss": 0.4238,
777
+ "mean_token_accuracy": 0.892235452234745,
778
+ "num_tokens": 119578624.0,
779
+ "step": 3650
780
+ },
781
+ {
782
+ "epoch": 1.955981493721084,
783
+ "grad_norm": 0.0893888846039772,
784
+ "learning_rate": 0.0003138698070003271,
785
+ "loss": 0.4292,
786
+ "mean_token_accuracy": 0.8910862573981285,
787
+ "num_tokens": 121217024.0,
788
+ "step": 3700
789
+ },
790
+ {
791
+ "epoch": 1.9824190350297424,
792
+ "grad_norm": 0.08708135783672333,
793
+ "learning_rate": 0.00031114382292007415,
794
+ "loss": 0.4319,
795
+ "step": 3750
796
+ },
797
+ {
798
+ "epoch": 1.9824190350297424,
799
+ "eval_loss": 0.5079798102378845,
800
+ "eval_mean_token_accuracy": 0.880161566348711,
801
+ "eval_num_tokens": 122855424.0,
802
+ "eval_runtime": 1444.3228,
803
+ "eval_samples_per_second": 5.238,
804
+ "eval_steps_per_second": 0.655,
805
+ "step": 3750
806
+ },
807
+ {
808
+ "epoch": 2.0084600132187704,
809
+ "grad_norm": 0.10443489998579025,
810
+ "learning_rate": 0.00030847235852142626,
811
+ "loss": 0.3692,
812
+ "mean_token_accuracy": 0.89680361027081,
813
+ "num_tokens": 124469248.0,
814
+ "step": 3800
815
+ },
816
+ {
817
+ "epoch": 2.034897554527429,
818
+ "grad_norm": 0.09454915672540665,
819
+ "learning_rate": 0.00030574637444117324,
820
+ "loss": 0.2389,
821
+ "mean_token_accuracy": 0.9304979285597801,
822
+ "num_tokens": 126107648.0,
823
+ "step": 3850
824
+ },
825
+ {
826
+ "epoch": 2.061335095836087,
827
+ "grad_norm": 0.10566703230142593,
828
+ "learning_rate": 0.0003030203903609203,
829
+ "loss": 0.2421,
830
+ "mean_token_accuracy": 0.9295185798406601,
831
+ "num_tokens": 127746048.0,
832
+ "step": 3900
833
+ },
834
+ {
835
+ "epoch": 2.0877726371447456,
836
+ "grad_norm": 0.10303833335638046,
837
+ "learning_rate": 0.00030029440628066736,
838
+ "loss": 0.2398,
839
+ "mean_token_accuracy": 0.9298442155122757,
840
+ "num_tokens": 129384448.0,
841
+ "step": 3950
842
+ },
843
+ {
844
+ "epoch": 2.114210178453404,
845
+ "grad_norm": 0.0934186801314354,
846
+ "learning_rate": 0.00029756842220041434,
847
+ "loss": 0.2444,
848
+ "step": 4000
849
+ },
850
+ {
851
+ "epoch": 2.114210178453404,
852
+ "eval_loss": 0.5857627391815186,
853
+ "eval_mean_token_accuracy": 0.8763663001342757,
854
+ "eval_num_tokens": 131022848.0,
855
+ "eval_runtime": 1444.4937,
856
+ "eval_samples_per_second": 5.237,
857
+ "eval_steps_per_second": 0.655,
858
+ "step": 4000
859
+ },
860
+ {
861
+ "epoch": 2.140647719762062,
862
+ "grad_norm": 0.11742925643920898,
863
+ "learning_rate": 0.0002948424381201614,
864
+ "loss": 0.2433,
865
+ "mean_token_accuracy": 0.9287463417649269,
866
+ "num_tokens": 132661248.0,
867
+ "step": 4050
868
+ },
869
+ {
870
+ "epoch": 2.1670852610707203,
871
+ "grad_norm": 0.09310949593782425,
872
+ "learning_rate": 0.0002921164540399084,
873
+ "loss": 0.2502,
874
+ "mean_token_accuracy": 0.9270796775817871,
875
+ "num_tokens": 134299648.0,
876
+ "step": 4100
877
+ },
878
+ {
879
+ "epoch": 2.1935228023793787,
880
+ "grad_norm": 0.10311378538608551,
881
+ "learning_rate": 0.00028939046995965544,
882
+ "loss": 0.2487,
883
+ "mean_token_accuracy": 0.9274065324664116,
884
+ "num_tokens": 135938048.0,
885
+ "step": 4150
886
+ },
887
+ {
888
+ "epoch": 2.219960343688037,
889
+ "grad_norm": 0.10651703178882599,
890
+ "learning_rate": 0.00028666448587940247,
891
+ "loss": 0.2481,
892
+ "mean_token_accuracy": 0.9273692658543586,
893
+ "num_tokens": 137576448.0,
894
+ "step": 4200
895
+ },
896
+ {
897
+ "epoch": 2.2463978849966955,
898
+ "grad_norm": 0.11140276491641998,
899
+ "learning_rate": 0.0002839385017991495,
900
+ "loss": 0.2513,
901
+ "step": 4250
902
+ },
903
+ {
904
+ "epoch": 2.2463978849966955,
905
+ "eval_loss": 0.5914037823677063,
906
+ "eval_mean_token_accuracy": 0.8769333329815693,
907
+ "eval_num_tokens": 139214848.0,
908
+ "eval_runtime": 1445.0132,
909
+ "eval_samples_per_second": 5.235,
910
+ "eval_steps_per_second": 0.655,
911
+ "step": 4250
912
+ },
913
+ {
914
+ "epoch": 2.2728354263053534,
915
+ "grad_norm": 0.10259566456079483,
916
+ "learning_rate": 0.00028121251771889654,
917
+ "loss": 0.253,
918
+ "mean_token_accuracy": 0.9264763659238815,
919
+ "num_tokens": 140853248.0,
920
+ "step": 4300
921
+ },
922
+ {
923
+ "epoch": 2.299272967614012,
924
+ "grad_norm": 0.10440666228532791,
925
+ "learning_rate": 0.00027848653363864357,
926
+ "loss": 0.25,
927
+ "mean_token_accuracy": 0.9270870080590248,
928
+ "num_tokens": 142491648.0,
929
+ "step": 4350
930
+ },
931
+ {
932
+ "epoch": 2.32571050892267,
933
+ "grad_norm": 0.11085245013237,
934
+ "learning_rate": 0.0002757605495583906,
935
+ "loss": 0.2507,
936
+ "mean_token_accuracy": 0.9270876184105873,
937
+ "num_tokens": 144130048.0,
938
+ "step": 4400
939
+ },
940
+ {
941
+ "epoch": 2.3521480502313286,
942
+ "grad_norm": 0.11133650690317154,
943
+ "learning_rate": 0.0002730345654781376,
944
+ "loss": 0.2509,
945
+ "mean_token_accuracy": 0.9270026966929436,
946
+ "num_tokens": 145768448.0,
947
+ "step": 4450
948
+ },
949
+ {
950
+ "epoch": 2.378585591539987,
951
+ "grad_norm": 0.09678333252668381,
952
+ "learning_rate": 0.00027030858139788467,
953
+ "loss": 0.252,
954
+ "step": 4500
955
+ },
956
+ {
957
+ "epoch": 2.378585591539987,
958
+ "eval_loss": 0.588071882724762,
959
+ "eval_mean_token_accuracy": 0.8775507902242164,
960
+ "eval_num_tokens": 147406848.0,
961
+ "eval_runtime": 1444.9835,
962
+ "eval_samples_per_second": 5.235,
963
+ "eval_steps_per_second": 0.655,
964
+ "step": 4500
965
+ },
966
+ {
967
+ "epoch": 2.405023132848645,
968
+ "grad_norm": 0.09919407218694687,
969
+ "learning_rate": 0.0002675825973176317,
970
+ "loss": 0.2597,
971
+ "mean_token_accuracy": 0.9257050442695618,
972
+ "num_tokens": 149045248.0,
973
+ "step": 4550
974
+ },
975
+ {
976
+ "epoch": 2.4314606741573033,
977
+ "grad_norm": 0.09900713711977005,
978
+ "learning_rate": 0.0002648566132373787,
979
+ "loss": 0.2594,
980
+ "mean_token_accuracy": 0.9249798482656479,
981
+ "num_tokens": 150683648.0,
982
+ "step": 4600
983
+ },
984
+ {
985
+ "epoch": 2.4578982154659617,
986
+ "grad_norm": 0.11208795011043549,
987
+ "learning_rate": 0.0002621306291571257,
988
+ "loss": 0.2512,
989
+ "mean_token_accuracy": 0.926915942132473,
990
+ "num_tokens": 152322048.0,
991
+ "step": 4650
992
+ },
993
+ {
994
+ "epoch": 2.48433575677462,
995
+ "grad_norm": 0.1047317236661911,
996
+ "learning_rate": 0.0002594046450768728,
997
+ "loss": 0.2575,
998
+ "mean_token_accuracy": 0.9254233974218369,
999
+ "num_tokens": 153960448.0,
1000
+ "step": 4700
1001
+ },
1002
+ {
1003
+ "epoch": 2.5107732980832784,
1004
+ "grad_norm": 0.10790830850601196,
1005
+ "learning_rate": 0.0002566786609966198,
1006
+ "loss": 0.2597,
1007
+ "step": 4750
1008
+ },
1009
+ {
1010
+ "epoch": 2.5107732980832784,
1011
+ "eval_loss": 0.5828577876091003,
1012
+ "eval_mean_token_accuracy": 0.8786447106153717,
1013
+ "eval_num_tokens": 155598848.0,
1014
+ "eval_runtime": 1445.085,
1015
+ "eval_samples_per_second": 5.235,
1016
+ "eval_steps_per_second": 0.655,
1017
+ "step": 4750
1018
+ },
1019
+ {
1020
+ "epoch": 2.5372108393919364,
1021
+ "grad_norm": 0.1063418760895729,
1022
+ "learning_rate": 0.0002539526769163668,
1023
+ "loss": 0.2532,
1024
+ "mean_token_accuracy": 0.9256470040977001,
1025
+ "num_tokens": 157237248.0,
1026
+ "step": 4800
1027
+ },
1028
+ {
1029
+ "epoch": 2.5636483807005948,
1030
+ "grad_norm": 0.10517441481351852,
1031
+ "learning_rate": 0.00025122669283611385,
1032
+ "loss": 0.257,
1033
+ "mean_token_accuracy": 0.92532870054245,
1034
+ "num_tokens": 158875648.0,
1035
+ "step": 4850
1036
+ },
1037
+ {
1038
+ "epoch": 2.590085922009253,
1039
+ "grad_norm": 0.10175232589244843,
1040
+ "learning_rate": 0.0002485007087558609,
1041
+ "loss": 0.2578,
1042
+ "mean_token_accuracy": 0.9254624971747398,
1043
+ "num_tokens": 160514048.0,
1044
+ "step": 4900
1045
+ },
1046
+ {
1047
+ "epoch": 2.6165234633179115,
1048
+ "grad_norm": 0.10122060030698776,
1049
+ "learning_rate": 0.0002457747246756079,
1050
+ "loss": 0.2585,
1051
+ "mean_token_accuracy": 0.925160077214241,
1052
+ "num_tokens": 162152448.0,
1053
+ "step": 4950
1054
+ },
1055
+ {
1056
+ "epoch": 2.64296100462657,
1057
+ "grad_norm": 0.17676186561584473,
1058
+ "learning_rate": 0.00024304874059535492,
1059
+ "loss": 0.2549,
1060
+ "step": 5000
1061
+ },
1062
+ {
1063
+ "epoch": 2.64296100462657,
1064
+ "eval_loss": 0.5834685564041138,
1065
+ "eval_mean_token_accuracy": 0.8788562039618261,
1066
+ "eval_num_tokens": 163790848.0,
1067
+ "eval_runtime": 1444.3995,
1068
+ "eval_samples_per_second": 5.237,
1069
+ "eval_steps_per_second": 0.655,
1070
+ "step": 5000
1071
+ },
1072
+ {
1073
+ "epoch": 2.669398545935228,
1074
+ "grad_norm": 0.09294166415929794,
1075
+ "learning_rate": 0.00024032275651510195,
1076
+ "loss": 0.2578,
1077
+ "mean_token_accuracy": 0.9256011824309825,
1078
+ "num_tokens": 165429248.0,
1079
+ "step": 5050
1080
+ },
1081
+ {
1082
+ "epoch": 2.6958360872438862,
1083
+ "grad_norm": 0.10612129420042038,
1084
+ "learning_rate": 0.000237596772434849,
1085
+ "loss": 0.2525,
1086
+ "mean_token_accuracy": 0.9266947796940803,
1087
+ "num_tokens": 167067648.0,
1088
+ "step": 5100
1089
+ },
1090
+ {
1091
+ "epoch": 2.7222736285525446,
1092
+ "grad_norm": 0.09773480892181396,
1093
+ "learning_rate": 0.00023487078835459602,
1094
+ "loss": 0.2574,
1095
+ "mean_token_accuracy": 0.9255730789899826,
1096
+ "num_tokens": 168706048.0,
1097
+ "step": 5150
1098
+ },
1099
+ {
1100
+ "epoch": 2.748711169861203,
1101
+ "grad_norm": 0.10722821205854416,
1102
+ "learning_rate": 0.00023214480427434303,
1103
+ "loss": 0.255,
1104
+ "mean_token_accuracy": 0.9261156010627747,
1105
+ "num_tokens": 170344448.0,
1106
+ "step": 5200
1107
+ },
1108
+ {
1109
+ "epoch": 2.7751487111698614,
1110
+ "grad_norm": 0.10466789454221725,
1111
+ "learning_rate": 0.00022941882019409009,
1112
+ "loss": 0.2579,
1113
+ "step": 5250
1114
+ },
1115
+ {
1116
+ "epoch": 2.7751487111698614,
1117
+ "eval_loss": 0.5760974287986755,
1118
+ "eval_mean_token_accuracy": 0.8801544366628876,
1119
+ "eval_num_tokens": 171982848.0,
1120
+ "eval_runtime": 1445.5403,
1121
+ "eval_samples_per_second": 5.233,
1122
+ "eval_steps_per_second": 0.654,
1123
+ "step": 5250
1124
+ },
1125
+ {
1126
+ "epoch": 2.8015862524785193,
1127
+ "grad_norm": 0.1087755486369133,
1128
+ "learning_rate": 0.0002266928361138371,
1129
+ "loss": 0.2609,
1130
+ "mean_token_accuracy": 0.9251267817616463,
1131
+ "num_tokens": 173621248.0,
1132
+ "step": 5300
1133
+ },
1134
+ {
1135
+ "epoch": 2.8280237937871777,
1136
+ "grad_norm": 0.09775590896606445,
1137
+ "learning_rate": 0.00022396685203358413,
1138
+ "loss": 0.2557,
1139
+ "mean_token_accuracy": 0.9260593932867051,
1140
+ "num_tokens": 175259648.0,
1141
+ "step": 5350
1142
+ },
1143
+ {
1144
+ "epoch": 2.854461335095836,
1145
+ "grad_norm": 0.09694032371044159,
1146
+ "learning_rate": 0.00022124086795333116,
1147
+ "loss": 0.2575,
1148
+ "mean_token_accuracy": 0.9255865186452865,
1149
+ "num_tokens": 176898048.0,
1150
+ "step": 5400
1151
+ },
1152
+ {
1153
+ "epoch": 2.8808988764044945,
1154
+ "grad_norm": 0.09860485792160034,
1155
+ "learning_rate": 0.0002185148838730782,
1156
+ "loss": 0.2567,
1157
+ "mean_token_accuracy": 0.9257099312543869,
1158
+ "num_tokens": 178536448.0,
1159
+ "step": 5450
1160
+ },
1161
+ {
1162
+ "epoch": 2.907336417713153,
1163
+ "grad_norm": 0.10467279702425003,
1164
+ "learning_rate": 0.0002157888997928252,
1165
+ "loss": 0.2537,
1166
+ "step": 5500
1167
+ },
1168
+ {
1169
+ "epoch": 2.907336417713153,
1170
+ "eval_loss": 0.5691426396369934,
1171
+ "eval_mean_token_accuracy": 0.8808775775402343,
1172
+ "eval_num_tokens": 180174848.0,
1173
+ "eval_runtime": 1445.1515,
1174
+ "eval_samples_per_second": 5.235,
1175
+ "eval_steps_per_second": 0.655,
1176
+ "step": 5500
1177
+ },
1178
+ {
1179
+ "epoch": 2.933773959021811,
1180
+ "grad_norm": 0.13350491225719452,
1181
+ "learning_rate": 0.00021306291571257226,
1182
+ "loss": 0.2548,
1183
+ "mean_token_accuracy": 0.9265344049036502,
1184
+ "num_tokens": 181813248.0,
1185
+ "step": 5550
1186
+ },
1187
+ {
1188
+ "epoch": 2.960211500330469,
1189
+ "grad_norm": 0.10815941542387009,
1190
+ "learning_rate": 0.00021033693163231926,
1191
+ "loss": 0.2529,
1192
+ "mean_token_accuracy": 0.9266226869821549,
1193
+ "num_tokens": 183451648.0,
1194
+ "step": 5600
1195
+ },
1196
+ {
1197
+ "epoch": 2.9866490416391276,
1198
+ "grad_norm": 0.09711950272321701,
1199
+ "learning_rate": 0.0002076109475520663,
1200
+ "loss": 0.2477,
1201
+ "mean_token_accuracy": 0.928189767897129,
1202
+ "num_tokens": 185090048.0,
1203
+ "step": 5650
1204
+ },
1205
+ {
1206
+ "epoch": 3.012690019828156,
1207
+ "grad_norm": 0.07202441990375519,
1208
+ "learning_rate": 0.0002049394831534184,
1209
+ "loss": 0.1737,
1210
+ "mean_token_accuracy": 0.9494730455016122,
1211
+ "num_tokens": 186703872.0,
1212
+ "step": 5700
1213
+ },
1214
+ {
1215
+ "epoch": 3.0391275611368145,
1216
+ "grad_norm": 0.09105918556451797,
1217
+ "learning_rate": 0.0002022134990731654,
1218
+ "loss": 0.0857,
1219
+ "step": 5750
1220
+ },
1221
+ {
1222
+ "epoch": 3.0391275611368145,
1223
+ "eval_loss": 0.7057962417602539,
1224
+ "eval_mean_token_accuracy": 0.8797050480736739,
1225
+ "eval_num_tokens": 188342272.0,
1226
+ "eval_runtime": 1445.2714,
1227
+ "eval_samples_per_second": 5.234,
1228
+ "eval_steps_per_second": 0.655,
1229
+ "step": 5750
1230
+ },
1231
+ {
1232
+ "epoch": 3.0655651024454724,
1233
+ "grad_norm": 0.08339951187372208,
1234
+ "learning_rate": 0.00019948751499291245,
1235
+ "loss": 0.0836,
1236
+ "mean_token_accuracy": 0.9740020345151424,
1237
+ "num_tokens": 189980672.0,
1238
+ "step": 5800
1239
+ },
1240
+ {
1241
+ "epoch": 3.092002643754131,
1242
+ "grad_norm": 0.07610207796096802,
1243
+ "learning_rate": 0.00019676153091265948,
1244
+ "loss": 0.0835,
1245
+ "mean_token_accuracy": 0.9742461076378822,
1246
+ "num_tokens": 191619072.0,
1247
+ "step": 5850
1248
+ },
1249
+ {
1250
+ "epoch": 3.118440185062789,
1251
+ "grad_norm": 0.08562301844358444,
1252
+ "learning_rate": 0.0001940355468324065,
1253
+ "loss": 0.0841,
1254
+ "mean_token_accuracy": 0.9739308601617813,
1255
+ "num_tokens": 193257472.0,
1256
+ "step": 5900
1257
+ },
1258
+ {
1259
+ "epoch": 3.1448777263714476,
1260
+ "grad_norm": 0.08178895711898804,
1261
+ "learning_rate": 0.00019130956275215352,
1262
+ "loss": 0.0824,
1263
+ "mean_token_accuracy": 0.9746059566736222,
1264
+ "num_tokens": 194895872.0,
1265
+ "step": 5950
1266
+ },
1267
+ {
1268
+ "epoch": 3.1713152676801055,
1269
+ "grad_norm": 0.08008107542991638,
1270
+ "learning_rate": 0.00018858357867190058,
1271
+ "loss": 0.0826,
1272
+ "step": 6000
1273
+ },
1274
+ {
1275
+ "epoch": 3.1713152676801055,
1276
+ "eval_loss": 0.718048095703125,
1277
+ "eval_mean_token_accuracy": 0.8796352478842907,
1278
+ "eval_num_tokens": 196534272.0,
1279
+ "eval_runtime": 1444.6019,
1280
+ "eval_samples_per_second": 5.237,
1281
+ "eval_steps_per_second": 0.655,
1282
+ "step": 6000
1283
+ },
1284
+ {
1285
+ "epoch": 3.197752808988764,
1286
+ "grad_norm": 0.08383160084486008,
1287
+ "learning_rate": 0.00018585759459164758,
1288
+ "loss": 0.0833,
1289
+ "mean_token_accuracy": 0.9743710477650166,
1290
+ "num_tokens": 198172672.0,
1291
+ "step": 6050
1292
+ },
1293
+ {
1294
+ "epoch": 3.2241903502974223,
1295
+ "grad_norm": 0.08407159149646759,
1296
+ "learning_rate": 0.00018313161051139462,
1297
+ "loss": 0.0843,
1298
+ "mean_token_accuracy": 0.9739204734563828,
1299
+ "num_tokens": 199811072.0,
1300
+ "step": 6100
1301
+ },
1302
+ {
1303
+ "epoch": 3.2506278916060807,
1304
+ "grad_norm": 0.0758429542183876,
1305
+ "learning_rate": 0.00018040562643114165,
1306
+ "loss": 0.0825,
1307
+ "mean_token_accuracy": 0.9741593536734581,
1308
+ "num_tokens": 201449472.0,
1309
+ "step": 6150
1310
+ },
1311
+ {
1312
+ "epoch": 3.277065432914739,
1313
+ "grad_norm": 0.08588337153196335,
1314
+ "learning_rate": 0.00017767964235088868,
1315
+ "loss": 0.0841,
1316
+ "mean_token_accuracy": 0.9740646561980247,
1317
+ "num_tokens": 203087872.0,
1318
+ "step": 6200
1319
+ },
1320
+ {
1321
+ "epoch": 3.303502974223397,
1322
+ "grad_norm": 0.08413270860910416,
1323
+ "learning_rate": 0.0001749536582706357,
1324
+ "loss": 0.0859,
1325
+ "step": 6250
1326
+ },
1327
+ {
1328
+ "epoch": 3.303502974223397,
1329
+ "eval_loss": 0.7220771908760071,
1330
+ "eval_mean_token_accuracy": 0.8795976611700925,
1331
+ "eval_num_tokens": 204726272.0,
1332
+ "eval_runtime": 1445.7274,
1333
+ "eval_samples_per_second": 5.233,
1334
+ "eval_steps_per_second": 0.654,
1335
+ "step": 6250
1336
+ },
1337
+ {
1338
+ "epoch": 3.3299405155320554,
1339
+ "grad_norm": 0.0801534354686737,
1340
+ "learning_rate": 0.00017222767419038275,
1341
+ "loss": 0.0856,
1342
+ "mean_token_accuracy": 0.9733324344456196,
1343
+ "num_tokens": 206364672.0,
1344
+ "step": 6300
1345
+ },
1346
+ {
1347
+ "epoch": 3.3563780568407138,
1348
+ "grad_norm": 0.08935518562793732,
1349
+ "learning_rate": 0.00016950169011012976,
1350
+ "loss": 0.0851,
1351
+ "mean_token_accuracy": 0.9735490143299103,
1352
+ "num_tokens": 208003072.0,
1353
+ "step": 6350
1354
+ },
1355
+ {
1356
+ "epoch": 3.382815598149372,
1357
+ "grad_norm": 0.08593747019767761,
1358
+ "learning_rate": 0.0001667757060298768,
1359
+ "loss": 0.0898,
1360
+ "mean_token_accuracy": 0.9727987706661224,
1361
+ "num_tokens": 209641472.0,
1362
+ "step": 6400
1363
+ },
1364
+ {
1365
+ "epoch": 3.4092531394580305,
1366
+ "grad_norm": 0.08515259623527527,
1367
+ "learning_rate": 0.00016404972194962382,
1368
+ "loss": 0.0862,
1369
+ "mean_token_accuracy": 0.9732246008515358,
1370
+ "num_tokens": 211279872.0,
1371
+ "step": 6450
1372
+ },
1373
+ {
1374
+ "epoch": 3.4356906807666885,
1375
+ "grad_norm": 0.08400440216064453,
1376
+ "learning_rate": 0.00016132373786937086,
1377
+ "loss": 0.0848,
1378
+ "step": 6500
1379
+ },
1380
+ {
1381
+ "epoch": 3.4356906807666885,
1382
+ "eval_loss": 0.7208439111709595,
1383
+ "eval_mean_token_accuracy": 0.8801698074154319,
1384
+ "eval_num_tokens": 212918272.0,
1385
+ "eval_runtime": 1445.2121,
1386
+ "eval_samples_per_second": 5.235,
1387
+ "eval_steps_per_second": 0.655,
1388
+ "step": 6500
1389
+ },
1390
+ {
1391
+ "epoch": 3.462128222075347,
1392
+ "grad_norm": 0.09465966373682022,
1393
+ "learning_rate": 0.00015859775378911786,
1394
+ "loss": 0.0845,
1395
+ "mean_token_accuracy": 0.9737170253694057,
1396
+ "num_tokens": 214556672.0,
1397
+ "step": 6550
1398
+ },
1399
+ {
1400
+ "epoch": 3.4885657633840053,
1401
+ "grad_norm": 0.08350127190351486,
1402
+ "learning_rate": 0.0001558717697088649,
1403
+ "loss": 0.0845,
1404
+ "mean_token_accuracy": 0.9738056141138077,
1405
+ "num_tokens": 216195072.0,
1406
+ "step": 6600
1407
+ },
1408
+ {
1409
+ "epoch": 3.5150033046926636,
1410
+ "grad_norm": 0.09258560091257095,
1411
+ "learning_rate": 0.00015314578562861193,
1412
+ "loss": 0.0867,
1413
+ "mean_token_accuracy": 0.973144568502903,
1414
+ "num_tokens": 217833472.0,
1415
+ "step": 6650
1416
+ },
1417
+ {
1418
+ "epoch": 3.541440846001322,
1419
+ "grad_norm": 0.08615857362747192,
1420
+ "learning_rate": 0.00015041980154835896,
1421
+ "loss": 0.0836,
1422
+ "mean_token_accuracy": 0.9740035626292228,
1423
+ "num_tokens": 219471872.0,
1424
+ "step": 6700
1425
+ },
1426
+ {
1427
+ "epoch": 3.56787838730998,
1428
+ "grad_norm": 0.08243568986654282,
1429
+ "learning_rate": 0.00014769381746810597,
1430
+ "loss": 0.0847,
1431
+ "step": 6750
1432
+ },
1433
+ {
1434
+ "epoch": 3.56787838730998,
1435
+ "eval_loss": 0.7248117923736572,
1436
+ "eval_mean_token_accuracy": 0.8803664739489807,
1437
+ "eval_num_tokens": 221110272.0,
1438
+ "eval_runtime": 1445.5737,
1439
+ "eval_samples_per_second": 5.233,
1440
+ "eval_steps_per_second": 0.654,
1441
+ "step": 6750
1442
+ },
1443
+ {
1444
+ "epoch": 3.5943159286186384,
1445
+ "grad_norm": 0.08806952089071274,
1446
+ "learning_rate": 0.00014496783338785303,
1447
+ "loss": 0.0861,
1448
+ "mean_token_accuracy": 0.9733803936839104,
1449
+ "num_tokens": 222748672.0,
1450
+ "step": 6800
1451
+ },
1452
+ {
1453
+ "epoch": 3.6207534699272967,
1454
+ "grad_norm": 0.08953431993722916,
1455
+ "learning_rate": 0.00014224184930760003,
1456
+ "loss": 0.0856,
1457
+ "mean_token_accuracy": 0.9733651205897331,
1458
+ "num_tokens": 224387072.0,
1459
+ "step": 6850
1460
+ },
1461
+ {
1462
+ "epoch": 3.647191011235955,
1463
+ "grad_norm": 0.08528242260217667,
1464
+ "learning_rate": 0.00013951586522734707,
1465
+ "loss": 0.0852,
1466
+ "mean_token_accuracy": 0.9733651199936867,
1467
+ "num_tokens": 226025472.0,
1468
+ "step": 6900
1469
+ },
1470
+ {
1471
+ "epoch": 3.6736285525446135,
1472
+ "grad_norm": 0.08363524079322815,
1473
+ "learning_rate": 0.0001367898811470941,
1474
+ "loss": 0.0845,
1475
+ "mean_token_accuracy": 0.9736650982499122,
1476
+ "num_tokens": 227663872.0,
1477
+ "step": 6950
1478
+ },
1479
+ {
1480
+ "epoch": 3.7000660938532715,
1481
+ "grad_norm": 0.07602488994598389,
1482
+ "learning_rate": 0.00013406389706684113,
1483
+ "loss": 0.0845,
1484
+ "step": 7000
1485
+ },
1486
+ {
1487
+ "epoch": 3.7000660938532715,
1488
+ "eval_loss": 0.7213518023490906,
1489
+ "eval_mean_token_accuracy": 0.8806469935695415,
1490
+ "eval_num_tokens": 229302272.0,
1491
+ "eval_runtime": 1444.415,
1492
+ "eval_samples_per_second": 5.237,
1493
+ "eval_steps_per_second": 0.655,
1494
+ "step": 7000
1495
+ },
1496
+ {
1497
+ "epoch": 3.72650363516193,
1498
+ "grad_norm": 0.08517055958509445,
1499
+ "learning_rate": 0.00013133791298658814,
1500
+ "loss": 0.0847,
1501
+ "mean_token_accuracy": 0.9737151943147182,
1502
+ "num_tokens": 230940672.0,
1503
+ "step": 7050
1504
+ },
1505
+ {
1506
+ "epoch": 3.7529411764705882,
1507
+ "grad_norm": 0.0767749696969986,
1508
+ "learning_rate": 0.0001286119289063352,
1509
+ "loss": 0.0828,
1510
+ "mean_token_accuracy": 0.9742509955167771,
1511
+ "num_tokens": 232579072.0,
1512
+ "step": 7100
1513
+ },
1514
+ {
1515
+ "epoch": 3.7793787177792466,
1516
+ "grad_norm": 0.08359658718109131,
1517
+ "learning_rate": 0.0001258859448260822,
1518
+ "loss": 0.0849,
1519
+ "mean_token_accuracy": 0.973515414595604,
1520
+ "num_tokens": 234217472.0,
1521
+ "step": 7150
1522
+ },
1523
+ {
1524
+ "epoch": 3.805816259087905,
1525
+ "grad_norm": 0.07397591322660446,
1526
+ "learning_rate": 0.00012315996074582924,
1527
+ "loss": 0.0835,
1528
+ "mean_token_accuracy": 0.9739962309598923,
1529
+ "num_tokens": 235855872.0,
1530
+ "step": 7200
1531
+ },
1532
+ {
1533
+ "epoch": 3.832253800396563,
1534
+ "grad_norm": 0.08941724896430969,
1535
+ "learning_rate": 0.00012043397666557627,
1536
+ "loss": 0.0821,
1537
+ "step": 7250
1538
+ },
1539
+ {
1540
+ "epoch": 3.832253800396563,
1541
+ "eval_loss": 0.727186381816864,
1542
+ "eval_mean_token_accuracy": 0.881018961017782,
1543
+ "eval_num_tokens": 237494272.0,
1544
+ "eval_runtime": 1445.0501,
1545
+ "eval_samples_per_second": 5.235,
1546
+ "eval_steps_per_second": 0.655,
1547
+ "step": 7250
1548
+ },
1549
+ {
1550
+ "epoch": 3.8586913417052213,
1551
+ "grad_norm": 0.08002132922410965,
1552
+ "learning_rate": 0.0001177079925853233,
1553
+ "loss": 0.0824,
1554
+ "mean_token_accuracy": 0.9744101479649544,
1555
+ "num_tokens": 239132672.0,
1556
+ "step": 7300
1557
+ },
1558
+ {
1559
+ "epoch": 3.8851288830138797,
1560
+ "grad_norm": 0.08109039068222046,
1561
+ "learning_rate": 0.00011498200850507034,
1562
+ "loss": 0.0842,
1563
+ "mean_token_accuracy": 0.9738893151283264,
1564
+ "num_tokens": 240771072.0,
1565
+ "step": 7350
1566
+ },
1567
+ {
1568
+ "epoch": 3.911566424322538,
1569
+ "grad_norm": 0.08812084048986435,
1570
+ "learning_rate": 0.00011225602442481736,
1571
+ "loss": 0.0808,
1572
+ "mean_token_accuracy": 0.9748161256313324,
1573
+ "num_tokens": 242409472.0,
1574
+ "step": 7400
1575
+ },
1576
+ {
1577
+ "epoch": 3.9380039656311965,
1578
+ "grad_norm": 0.08132945001125336,
1579
+ "learning_rate": 0.00010953004034456439,
1580
+ "loss": 0.0824,
1581
+ "mean_token_accuracy": 0.9742913186550141,
1582
+ "num_tokens": 244047872.0,
1583
+ "step": 7450
1584
+ },
1585
+ {
1586
+ "epoch": 3.9644415069398544,
1587
+ "grad_norm": 0.08432092517614365,
1588
+ "learning_rate": 0.00010680405626431142,
1589
+ "loss": 0.0815,
1590
+ "step": 7500
1591
+ },
1592
+ {
1593
+ "epoch": 3.9644415069398544,
1594
+ "eval_loss": 0.727695643901825,
1595
+ "eval_mean_token_accuracy": 0.8814008755109275,
1596
+ "eval_num_tokens": 245686272.0,
1597
+ "eval_runtime": 1444.3382,
1598
+ "eval_samples_per_second": 5.238,
1599
+ "eval_steps_per_second": 0.655,
1600
+ "step": 7500
1601
+ },
1602
+ {
1603
+ "epoch": 3.990879048248513,
1604
+ "grad_norm": 0.08465278148651123,
1605
+ "learning_rate": 0.00010407807218405844,
1606
+ "loss": 0.0823,
1607
+ "mean_token_accuracy": 0.9746123726665974,
1608
+ "num_tokens": 247324672.0,
1609
+ "step": 7550
1610
+ },
1611
+ {
1612
+ "epoch": 4.016920026437541,
1613
+ "grad_norm": 0.04739232733845711,
1614
+ "learning_rate": 0.00010135208810380548,
1615
+ "loss": 0.0478,
1616
+ "mean_token_accuracy": 0.9862868610372398,
1617
+ "num_tokens": 248938496.0,
1618
+ "step": 7600
1619
+ },
1620
+ {
1621
+ "epoch": 4.0433575677462,
1622
+ "grad_norm": 0.05006248131394386,
1623
+ "learning_rate": 9.86261040235525e-05,
1624
+ "loss": 0.0278,
1625
+ "mean_token_accuracy": 0.9926484674215317,
1626
+ "num_tokens": 250576896.0,
1627
+ "step": 7650
1628
+ },
1629
+ {
1630
+ "epoch": 4.069795109054858,
1631
+ "grad_norm": 0.048755794763565063,
1632
+ "learning_rate": 9.590011994329953e-05,
1633
+ "loss": 0.0275,
1634
+ "mean_token_accuracy": 0.9927639365196228,
1635
+ "num_tokens": 252215296.0,
1636
+ "step": 7700
1637
+ },
1638
+ {
1639
+ "epoch": 4.0962326503635165,
1640
+ "grad_norm": 0.04973224923014641,
1641
+ "learning_rate": 9.317413586304656e-05,
1642
+ "loss": 0.0269,
1643
+ "step": 7750
1644
+ },
1645
+ {
1646
+ "epoch": 4.0962326503635165,
1647
+ "eval_loss": 0.8192626237869263,
1648
+ "eval_mean_token_accuracy": 0.88220195217062,
1649
+ "eval_num_tokens": 253853696.0,
1650
+ "eval_runtime": 1444.8032,
1651
+ "eval_samples_per_second": 5.236,
1652
+ "eval_steps_per_second": 0.655,
1653
+ "step": 7750
1654
+ },
1655
+ {
1656
+ "epoch": 4.122670191672174,
1657
+ "grad_norm": 0.05098351836204529,
1658
+ "learning_rate": 9.044815178279358e-05,
1659
+ "loss": 0.0268,
1660
+ "mean_token_accuracy": 0.9928580224514008,
1661
+ "num_tokens": 255492096.0,
1662
+ "step": 7800
1663
+ },
1664
+ {
1665
+ "epoch": 4.149107732980832,
1666
+ "grad_norm": 0.04829578474164009,
1667
+ "learning_rate": 8.772216770254061e-05,
1668
+ "loss": 0.0272,
1669
+ "mean_token_accuracy": 0.9927712675929069,
1670
+ "num_tokens": 257130496.0,
1671
+ "step": 7850
1672
+ },
1673
+ {
1674
+ "epoch": 4.175545274289491,
1675
+ "grad_norm": 0.04151781275868416,
1676
+ "learning_rate": 8.499618362228765e-05,
1677
+ "loss": 0.0269,
1678
+ "mean_token_accuracy": 0.9928335842490196,
1679
+ "num_tokens": 258768896.0,
1680
+ "step": 7900
1681
+ },
1682
+ {
1683
+ "epoch": 4.201982815598149,
1684
+ "grad_norm": 0.05445162579417229,
1685
+ "learning_rate": 8.227019954203467e-05,
1686
+ "loss": 0.0265,
1687
+ "mean_token_accuracy": 0.9929741024971008,
1688
+ "num_tokens": 260407296.0,
1689
+ "step": 7950
1690
+ },
1691
+ {
1692
+ "epoch": 4.228420356906808,
1693
+ "grad_norm": 0.03888251632452011,
1694
+ "learning_rate": 7.95442154617817e-05,
1695
+ "loss": 0.0273,
1696
+ "step": 8000
1697
+ },
1698
+ {
1699
+ "epoch": 4.228420356906808,
1700
+ "eval_loss": 0.8266236782073975,
1701
+ "eval_mean_token_accuracy": 0.8823872005712659,
1702
+ "eval_num_tokens": 262045696.0,
1703
+ "eval_runtime": 1444.6622,
1704
+ "eval_samples_per_second": 5.237,
1705
+ "eval_steps_per_second": 0.655,
1706
+ "step": 8000
1707
+ },
1708
+ {
1709
+ "epoch": 4.254857898215466,
1710
+ "grad_norm": 0.049200285226106644,
1711
+ "learning_rate": 7.681823138152873e-05,
1712
+ "loss": 0.0265,
1713
+ "mean_token_accuracy": 0.9928293077647686,
1714
+ "num_tokens": 263684096.0,
1715
+ "step": 8050
1716
+ },
1717
+ {
1718
+ "epoch": 4.281295439524124,
1719
+ "grad_norm": 0.04635064676403999,
1720
+ "learning_rate": 7.409224730127575e-05,
1721
+ "loss": 0.0259,
1722
+ "mean_token_accuracy": 0.9931225627660751,
1723
+ "num_tokens": 265322496.0,
1724
+ "step": 8100
1725
+ },
1726
+ {
1727
+ "epoch": 4.307732980832783,
1728
+ "grad_norm": 0.047765497118234634,
1729
+ "learning_rate": 7.136626322102279e-05,
1730
+ "loss": 0.0269,
1731
+ "mean_token_accuracy": 0.992902010679245,
1732
+ "num_tokens": 266960896.0,
1733
+ "step": 8150
1734
+ },
1735
+ {
1736
+ "epoch": 4.334170522141441,
1737
+ "grad_norm": 0.040082089602947235,
1738
+ "learning_rate": 6.864027914076983e-05,
1739
+ "loss": 0.0327,
1740
+ "mean_token_accuracy": 0.9925085598230362,
1741
+ "num_tokens": 268599296.0,
1742
+ "step": 8200
1743
+ },
1744
+ {
1745
+ "epoch": 4.360608063450099,
1746
+ "grad_norm": 0.040357593446969986,
1747
+ "learning_rate": 6.591429506051685e-05,
1748
+ "loss": 0.0262,
1749
+ "step": 8250
1750
+ },
1751
+ {
1752
+ "epoch": 4.360608063450099,
1753
+ "eval_loss": 0.8343076109886169,
1754
+ "eval_mean_token_accuracy": 0.8827618814169983,
1755
+ "eval_num_tokens": 270237696.0,
1756
+ "eval_runtime": 1444.2593,
1757
+ "eval_samples_per_second": 5.238,
1758
+ "eval_steps_per_second": 0.655,
1759
+ "step": 8250
1760
+ },
1761
+ {
1762
+ "epoch": 4.387045604758757,
1763
+ "grad_norm": 0.04214048385620117,
1764
+ "learning_rate": 6.318831098026388e-05,
1765
+ "loss": 0.0258,
1766
+ "mean_token_accuracy": 0.9930578023195267,
1767
+ "num_tokens": 271876096.0,
1768
+ "step": 8300
1769
+ },
1770
+ {
1771
+ "epoch": 4.413483146067415,
1772
+ "grad_norm": 0.0495334193110466,
1773
+ "learning_rate": 6.0462326900010904e-05,
1774
+ "loss": 0.0259,
1775
+ "mean_token_accuracy": 0.9930883496999741,
1776
+ "num_tokens": 273514496.0,
1777
+ "step": 8350
1778
+ },
1779
+ {
1780
+ "epoch": 4.439920687376074,
1781
+ "grad_norm": 0.04878053814172745,
1782
+ "learning_rate": 5.773634281975793e-05,
1783
+ "loss": 0.0256,
1784
+ "mean_token_accuracy": 0.9932361990213394,
1785
+ "num_tokens": 275152896.0,
1786
+ "step": 8400
1787
+ },
1788
+ {
1789
+ "epoch": 4.466358228684732,
1790
+ "grad_norm": 0.04534803330898285,
1791
+ "learning_rate": 5.5010358739504963e-05,
1792
+ "loss": 0.0257,
1793
+ "mean_token_accuracy": 0.9931818246841431,
1794
+ "num_tokens": 276791296.0,
1795
+ "step": 8450
1796
+ },
1797
+ {
1798
+ "epoch": 4.492795769993391,
1799
+ "grad_norm": 0.04720517247915268,
1800
+ "learning_rate": 5.228437465925199e-05,
1801
+ "loss": 0.0255,
1802
+ "step": 8500
1803
+ },
1804
+ {
1805
+ "epoch": 4.492795769993391,
1806
+ "eval_loss": 0.8357202410697937,
1807
+ "eval_mean_token_accuracy": 0.8828653151338751,
1808
+ "eval_num_tokens": 278429696.0,
1809
+ "eval_runtime": 1445.1121,
1810
+ "eval_samples_per_second": 5.235,
1811
+ "eval_steps_per_second": 0.655,
1812
+ "step": 8500
1813
+ },
1814
+ {
1815
+ "epoch": 4.519233311302049,
1816
+ "grad_norm": 0.04248838871717453,
1817
+ "learning_rate": 4.9558390578999016e-05,
1818
+ "loss": 0.0252,
1819
+ "mean_token_accuracy": 0.9932594150304794,
1820
+ "num_tokens": 280068096.0,
1821
+ "step": 8550
1822
+ },
1823
+ {
1824
+ "epoch": 4.545670852610707,
1825
+ "grad_norm": 0.04879326745867729,
1826
+ "learning_rate": 4.683240649874604e-05,
1827
+ "loss": 0.0256,
1828
+ "mean_token_accuracy": 0.9932215362787247,
1829
+ "num_tokens": 281706496.0,
1830
+ "step": 8600
1831
+ },
1832
+ {
1833
+ "epoch": 4.572108393919366,
1834
+ "grad_norm": 0.04033661261200905,
1835
+ "learning_rate": 4.4106422418493076e-05,
1836
+ "loss": 0.0256,
1837
+ "mean_token_accuracy": 0.993228867650032,
1838
+ "num_tokens": 283344896.0,
1839
+ "step": 8650
1840
+ },
1841
+ {
1842
+ "epoch": 4.598545935228024,
1843
+ "grad_norm": 0.043814148753881454,
1844
+ "learning_rate": 4.138043833824011e-05,
1845
+ "loss": 0.0251,
1846
+ "mean_token_accuracy": 0.9933846592903137,
1847
+ "num_tokens": 284983296.0,
1848
+ "step": 8700
1849
+ },
1850
+ {
1851
+ "epoch": 4.624983476536682,
1852
+ "grad_norm": 0.04637971520423889,
1853
+ "learning_rate": 3.8654454257987135e-05,
1854
+ "loss": 0.0256,
1855
+ "step": 8750
1856
+ },
1857
+ {
1858
+ "epoch": 4.624983476536682,
1859
+ "eval_loss": 0.8386735916137695,
1860
+ "eval_mean_token_accuracy": 0.8830419348485878,
1861
+ "eval_num_tokens": 286621696.0,
1862
+ "eval_runtime": 1444.7416,
1863
+ "eval_samples_per_second": 5.236,
1864
+ "eval_steps_per_second": 0.655,
1865
+ "step": 8750
1866
+ },
1867
+ {
1868
+ "epoch": 4.65142101784534,
1869
+ "grad_norm": 0.04898098483681679,
1870
+ "learning_rate": 3.592847017773417e-05,
1871
+ "loss": 0.025,
1872
+ "mean_token_accuracy": 0.9933046251535416,
1873
+ "num_tokens": 288260096.0,
1874
+ "step": 8800
1875
+ },
1876
+ {
1877
+ "epoch": 4.677858559153998,
1878
+ "grad_norm": 0.04658009856939316,
1879
+ "learning_rate": 3.3202486097481194e-05,
1880
+ "loss": 0.025,
1881
+ "mean_token_accuracy": 0.9934017658233643,
1882
+ "num_tokens": 289898496.0,
1883
+ "step": 8850
1884
+ },
1885
+ {
1886
+ "epoch": 4.704296100462657,
1887
+ "grad_norm": 0.04260571300983429,
1888
+ "learning_rate": 3.0476502017228217e-05,
1889
+ "loss": 0.0246,
1890
+ "mean_token_accuracy": 0.9935636669397354,
1891
+ "num_tokens": 291536896.0,
1892
+ "step": 8900
1893
+ },
1894
+ {
1895
+ "epoch": 4.730733641771315,
1896
+ "grad_norm": 0.04221141338348389,
1897
+ "learning_rate": 2.775051793697525e-05,
1898
+ "loss": 0.0249,
1899
+ "mean_token_accuracy": 0.9934304803609848,
1900
+ "num_tokens": 293175296.0,
1901
+ "step": 8950
1902
+ },
1903
+ {
1904
+ "epoch": 4.757171183079974,
1905
+ "grad_norm": 0.046503596007823944,
1906
+ "learning_rate": 2.502453385672228e-05,
1907
+ "loss": 0.0247,
1908
+ "step": 9000
1909
+ },
1910
+ {
1911
+ "epoch": 4.757171183079974,
1912
+ "eval_loss": 0.8432579636573792,
1913
+ "eval_mean_token_accuracy": 0.8832672491008081,
1914
+ "eval_num_tokens": 294813696.0,
1915
+ "eval_runtime": 1444.873,
1916
+ "eval_samples_per_second": 5.236,
1917
+ "eval_steps_per_second": 0.655,
1918
+ "step": 9000
1919
+ },
1920
+ {
1921
+ "epoch": 4.783608724388632,
1922
+ "grad_norm": 0.049112509936094284,
1923
+ "learning_rate": 2.2298549776469306e-05,
1924
+ "loss": 0.0244,
1925
+ "mean_token_accuracy": 0.9935187622904778,
1926
+ "num_tokens": 296452096.0,
1927
+ "step": 9050
1928
+ },
1929
+ {
1930
+ "epoch": 4.81004626569729,
1931
+ "grad_norm": 0.049860984086990356,
1932
+ "learning_rate": 1.9572565696216336e-05,
1933
+ "loss": 0.0248,
1934
+ "mean_token_accuracy": 0.9935349524021149,
1935
+ "num_tokens": 298090496.0,
1936
+ "step": 9100
1937
+ },
1938
+ {
1939
+ "epoch": 4.836483807005949,
1940
+ "grad_norm": 0.041883405297994614,
1941
+ "learning_rate": 1.6846581615963362e-05,
1942
+ "loss": 0.0243,
1943
+ "mean_token_accuracy": 0.9936204850673676,
1944
+ "num_tokens": 299728896.0,
1945
+ "step": 9150
1946
+ },
1947
+ {
1948
+ "epoch": 4.8629213483146065,
1949
+ "grad_norm": 0.043799083679914474,
1950
+ "learning_rate": 1.4120597535710392e-05,
1951
+ "loss": 0.024,
1952
+ "mean_token_accuracy": 0.9937127381563187,
1953
+ "num_tokens": 301367296.0,
1954
+ "step": 9200
1955
+ },
1956
+ {
1957
+ "epoch": 4.889358889623265,
1958
+ "grad_norm": 0.04291240870952606,
1959
+ "learning_rate": 1.139461345545742e-05,
1960
+ "loss": 0.0238,
1961
+ "step": 9250
1962
+ },
1963
+ {
1964
+ "epoch": 4.889358889623265,
1965
+ "eval_loss": 0.8444077372550964,
1966
+ "eval_mean_token_accuracy": 0.88355980766248,
1967
+ "eval_num_tokens": 303005696.0,
1968
+ "eval_runtime": 1444.2008,
1969
+ "eval_samples_per_second": 5.238,
1970
+ "eval_steps_per_second": 0.655,
1971
+ "step": 9250
1972
+ },
1973
+ {
1974
+ "epoch": 4.915796430931923,
1975
+ "grad_norm": 0.04179864004254341,
1976
+ "learning_rate": 8.668629375204448e-06,
1977
+ "loss": 0.0237,
1978
+ "mean_token_accuracy": 0.9937710836529732,
1979
+ "num_tokens": 304644096.0,
1980
+ "step": 9300
1981
+ },
1982
+ {
1983
+ "epoch": 4.942233972240581,
1984
+ "grad_norm": 0.04021856561303139,
1985
+ "learning_rate": 5.942645294951477e-06,
1986
+ "loss": 0.0239,
1987
+ "mean_token_accuracy": 0.993884414434433,
1988
+ "num_tokens": 306282496.0,
1989
+ "step": 9350
1990
+ },
1991
+ {
1992
+ "epoch": 4.96867151354924,
1993
+ "grad_norm": 0.04214942827820778,
1994
+ "learning_rate": 3.2166612146985063e-06,
1995
+ "loss": 0.0243,
1996
+ "mean_token_accuracy": 0.9936669170856476,
1997
+ "num_tokens": 307920896.0,
1998
+ "step": 9400
1999
+ },
2000
+ {
2001
+ "epoch": 4.995109054857898,
2002
+ "grad_norm": 0.04387414827942848,
2003
+ "learning_rate": 4.906771344455349e-07,
2004
+ "loss": 0.0242,
2005
+ "mean_token_accuracy": 0.9936302602291107,
2006
+ "num_tokens": 309559296.0,
2007
+ "step": 9450
2008
+ }
2009
+ ],
2010
+ "logging_steps": 50,
2011
+ "max_steps": 9455,
2012
+ "num_input_tokens_seen": 0,
2013
+ "num_train_epochs": 5,
2014
+ "save_steps": 500,
2015
+ "stateful_callbacks": {
2016
+ "TrainerControl": {
2017
+ "args": {
2018
+ "should_epoch_stop": false,
2019
+ "should_evaluate": false,
2020
+ "should_log": false,
2021
+ "should_save": true,
2022
+ "should_training_stop": true
2023
+ },
2024
+ "attributes": {}
2025
+ }
2026
+ },
2027
+ "total_flos": 464304822288384.0,
2028
+ "train_batch_size": 1,
2029
+ "trial_name": null,
2030
+ "trial_params": null
2031
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6d1965a048417448deba65ffbd2c16e88f9d5681c449f9faee78883d4cc9800
3
+ size 8120