rootxhacker commited on
Commit
fc728b7
·
verified ·
1 Parent(s): a1c71c7

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. checkpoint-1000/config.json +30 -0
  2. checkpoint-1000/generation_config.json +7 -0
  3. checkpoint-1000/model.safetensors +3 -0
  4. checkpoint-1000/optimizer.pt +3 -0
  5. checkpoint-1000/rng_state.pth +3 -0
  6. checkpoint-1000/scheduler.pt +3 -0
  7. checkpoint-1000/special_tokens_map.json +24 -0
  8. checkpoint-1000/tokenizer.json +0 -0
  9. checkpoint-1000/tokenizer.model +3 -0
  10. checkpoint-1000/tokenizer_config.json +42 -0
  11. checkpoint-1000/trainer_state.json +934 -0
  12. checkpoint-1000/training_args.bin +3 -0
  13. checkpoint-1500/config.json +30 -0
  14. checkpoint-1500/generation_config.json +7 -0
  15. checkpoint-1500/model.safetensors +3 -0
  16. checkpoint-1500/optimizer.pt +3 -0
  17. checkpoint-1500/rng_state.pth +3 -0
  18. checkpoint-1500/scheduler.pt +3 -0
  19. checkpoint-1500/special_tokens_map.json +24 -0
  20. checkpoint-1500/tokenizer.json +0 -0
  21. checkpoint-1500/tokenizer.model +3 -0
  22. checkpoint-1500/tokenizer_config.json +42 -0
  23. checkpoint-1500/trainer_state.json +1384 -0
  24. checkpoint-1500/training_args.bin +3 -0
  25. checkpoint-1920/config.json +30 -0
  26. checkpoint-1920/generation_config.json +7 -0
  27. checkpoint-1920/model.safetensors +3 -0
  28. checkpoint-1920/optimizer.pt +3 -0
  29. checkpoint-1920/rng_state.pth +3 -0
  30. checkpoint-1920/scheduler.pt +3 -0
  31. checkpoint-1920/special_tokens_map.json +24 -0
  32. checkpoint-1920/tokenizer.json +0 -0
  33. checkpoint-1920/tokenizer.model +3 -0
  34. checkpoint-1920/tokenizer_config.json +42 -0
  35. checkpoint-1920/trainer_state.json +1762 -0
  36. checkpoint-1920/training_args.bin +3 -0
  37. checkpoint-500/config.json +30 -0
  38. checkpoint-500/generation_config.json +7 -0
  39. checkpoint-500/model.safetensors +3 -0
  40. checkpoint-500/optimizer.pt +3 -0
  41. checkpoint-500/rng_state.pth +3 -0
  42. checkpoint-500/scheduler.pt +3 -0
  43. checkpoint-500/special_tokens_map.json +24 -0
  44. checkpoint-500/tokenizer.json +0 -0
  45. checkpoint-500/tokenizer.model +3 -0
  46. checkpoint-500/tokenizer_config.json +42 -0
  47. checkpoint-500/trainer_state.json +484 -0
  48. checkpoint-500/training_args.bin +3 -0
  49. config.json +30 -0
  50. generation_config.json +7 -0
checkpoint-1000/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 512,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-1000/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-1000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2edcf40abf042530fc3cebdd36ed881b05a377f2f6f15dbee2cb706389764199
3
+ size 309900448
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a0172215d31a1b74ed7c7027aa734edf4672e58b00a00bbd3b43cc35dad9a5e
3
+ size 619836730
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95b6047bd8cc6f4cdf7c46dea47edb8e542435510070c6cd1e0a7d9ccf5fd7da
3
+ size 14244
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ec29a0c7c00a01f3112690b5202d037b7b01ca67bd030b14a4afe3e89e0aa4b
3
+ size 1064
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-1000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1000/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1000/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,934 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 5.1820546163849155,
6
+ "eval_steps": 500,
7
+ "global_step": 1000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.05201560468140442,
14
+ "grad_norm": 3.158724784851074,
15
+ "learning_rate": 1.9906250000000003e-05,
16
+ "loss": 6.2302,
17
+ "mean_token_accuracy": 0.16946521885693072,
18
+ "num_tokens": 513852.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.10403120936280884,
23
+ "grad_norm": 10.176986694335938,
24
+ "learning_rate": 1.9802083333333335e-05,
25
+ "loss": 6.0745,
26
+ "mean_token_accuracy": 0.17949572652578355,
27
+ "num_tokens": 1032088.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.15604681404421328,
32
+ "grad_norm": 3.010871410369873,
33
+ "learning_rate": 1.9697916666666667e-05,
34
+ "loss": 5.8317,
35
+ "mean_token_accuracy": 0.1901865139603615,
36
+ "num_tokens": 1539389.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.20806241872561768,
41
+ "grad_norm": 3.298682451248169,
42
+ "learning_rate": 1.9593750000000002e-05,
43
+ "loss": 5.5987,
44
+ "mean_token_accuracy": 0.20738236755132675,
45
+ "num_tokens": 2057784.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.26007802340702213,
50
+ "grad_norm": 2.472743272781372,
51
+ "learning_rate": 1.9489583333333334e-05,
52
+ "loss": 5.4572,
53
+ "mean_token_accuracy": 0.2219402514398098,
54
+ "num_tokens": 2574094.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.31209362808842656,
59
+ "grad_norm": 2.3076207637786865,
60
+ "learning_rate": 1.938541666666667e-05,
61
+ "loss": 5.2817,
62
+ "mean_token_accuracy": 0.23026303350925445,
63
+ "num_tokens": 3076289.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.3641092327698309,
68
+ "grad_norm": 2.1481237411499023,
69
+ "learning_rate": 1.928125e-05,
70
+ "loss": 5.0596,
71
+ "mean_token_accuracy": 0.24168218560516835,
72
+ "num_tokens": 3586260.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.41612483745123535,
77
+ "grad_norm": 2.2509524822235107,
78
+ "learning_rate": 1.9177083333333336e-05,
79
+ "loss": 4.9117,
80
+ "mean_token_accuracy": 0.2510342214256525,
81
+ "num_tokens": 4094922.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.4681404421326398,
86
+ "grad_norm": 3.319535732269287,
87
+ "learning_rate": 1.907291666666667e-05,
88
+ "loss": 4.7624,
89
+ "mean_token_accuracy": 0.26219444051384927,
90
+ "num_tokens": 4606336.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.5201560468140443,
95
+ "grad_norm": 2.0256783962249756,
96
+ "learning_rate": 1.8968750000000004e-05,
97
+ "loss": 4.6138,
98
+ "mean_token_accuracy": 0.2705389905720949,
99
+ "num_tokens": 5110749.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.5721716514954487,
104
+ "grad_norm": 5.284979820251465,
105
+ "learning_rate": 1.8864583333333336e-05,
106
+ "loss": 4.5011,
107
+ "mean_token_accuracy": 0.27774968966841695,
108
+ "num_tokens": 5611872.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.6241872561768531,
113
+ "grad_norm": 15.897697448730469,
114
+ "learning_rate": 1.8760416666666668e-05,
115
+ "loss": 4.3486,
116
+ "mean_token_accuracy": 0.29455167055130005,
117
+ "num_tokens": 6123716.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.6762028608582574,
122
+ "grad_norm": 3.0891332626342773,
123
+ "learning_rate": 1.8656250000000003e-05,
124
+ "loss": 4.2372,
125
+ "mean_token_accuracy": 0.3053073361515999,
126
+ "num_tokens": 6630449.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.7282184655396619,
131
+ "grad_norm": 2.85463285446167,
132
+ "learning_rate": 1.8552083333333335e-05,
133
+ "loss": 4.161,
134
+ "mean_token_accuracy": 0.3164319314062595,
135
+ "num_tokens": 7145717.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.7802340702210663,
140
+ "grad_norm": 1.2611316442489624,
141
+ "learning_rate": 1.8447916666666667e-05,
142
+ "loss": 4.1002,
143
+ "mean_token_accuracy": 0.32258255779743195,
144
+ "num_tokens": 7655884.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.8322496749024707,
149
+ "grad_norm": 1.4083240032196045,
150
+ "learning_rate": 1.8343750000000002e-05,
151
+ "loss": 4.0138,
152
+ "mean_token_accuracy": 0.3296029016375542,
153
+ "num_tokens": 8159725.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.8842652795838751,
158
+ "grad_norm": 1.0340592861175537,
159
+ "learning_rate": 1.8239583333333334e-05,
160
+ "loss": 3.9163,
161
+ "mean_token_accuracy": 0.3385446682572365,
162
+ "num_tokens": 8668116.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.9362808842652796,
167
+ "grad_norm": 0.7984447479248047,
168
+ "learning_rate": 1.8135416666666666e-05,
169
+ "loss": 3.8628,
170
+ "mean_token_accuracy": 0.3455250822007656,
171
+ "num_tokens": 9182489.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.988296488946684,
176
+ "grad_norm": 1.244081735610962,
177
+ "learning_rate": 1.803125e-05,
178
+ "loss": 3.8025,
179
+ "mean_token_accuracy": 0.35223659574985505,
180
+ "num_tokens": 9676078.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 1.0364109232769831,
185
+ "grad_norm": 2.412661552429199,
186
+ "learning_rate": 1.7927083333333337e-05,
187
+ "loss": 3.7325,
188
+ "mean_token_accuracy": 0.35705321382831884,
189
+ "num_tokens": 10138114.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 1.0884265279583876,
194
+ "grad_norm": 1.7736923694610596,
195
+ "learning_rate": 1.782291666666667e-05,
196
+ "loss": 3.7052,
197
+ "mean_token_accuracy": 0.3590236373245716,
198
+ "num_tokens": 10656504.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 1.140442132639792,
203
+ "grad_norm": 0.7769147157669067,
204
+ "learning_rate": 1.7718750000000004e-05,
205
+ "loss": 3.6433,
206
+ "mean_token_accuracy": 0.36575346291065214,
207
+ "num_tokens": 11161374.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 1.1924577373211964,
212
+ "grad_norm": 0.9089329838752747,
213
+ "learning_rate": 1.7614583333333336e-05,
214
+ "loss": 3.653,
215
+ "mean_token_accuracy": 0.3651699565351009,
216
+ "num_tokens": 11670197.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 1.2444733420026008,
221
+ "grad_norm": 0.7330015897750854,
222
+ "learning_rate": 1.7510416666666668e-05,
223
+ "loss": 3.5843,
224
+ "mean_token_accuracy": 0.3718643538653851,
225
+ "num_tokens": 12174931.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 1.2964889466840053,
230
+ "grad_norm": 1.534440040588379,
231
+ "learning_rate": 1.7406250000000003e-05,
232
+ "loss": 3.5468,
233
+ "mean_token_accuracy": 0.3751932092010975,
234
+ "num_tokens": 12685023.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 1.3485045513654097,
239
+ "grad_norm": 1.9603444337844849,
240
+ "learning_rate": 1.7302083333333335e-05,
241
+ "loss": 3.4935,
242
+ "mean_token_accuracy": 0.3808542750775814,
243
+ "num_tokens": 13197338.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 1.4005201560468141,
248
+ "grad_norm": 0.5988335013389587,
249
+ "learning_rate": 1.7197916666666667e-05,
250
+ "loss": 3.4988,
251
+ "mean_token_accuracy": 0.3815221585333347,
252
+ "num_tokens": 13704900.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 1.4525357607282183,
257
+ "grad_norm": 2.7422847747802734,
258
+ "learning_rate": 1.7093750000000002e-05,
259
+ "loss": 3.4438,
260
+ "mean_token_accuracy": 0.3865826904773712,
261
+ "num_tokens": 14210835.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 1.5045513654096228,
266
+ "grad_norm": 1.9575722217559814,
267
+ "learning_rate": 1.6989583333333334e-05,
268
+ "loss": 3.417,
269
+ "mean_token_accuracy": 0.38725101202726364,
270
+ "num_tokens": 14726584.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 1.5565669700910272,
275
+ "grad_norm": 0.6955925226211548,
276
+ "learning_rate": 1.6885416666666666e-05,
277
+ "loss": 3.3895,
278
+ "mean_token_accuracy": 0.39159547463059424,
279
+ "num_tokens": 15238009.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 1.6085825747724316,
284
+ "grad_norm": 0.769548237323761,
285
+ "learning_rate": 1.678125e-05,
286
+ "loss": 3.3979,
287
+ "mean_token_accuracy": 0.39134200662374496,
288
+ "num_tokens": 15736398.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 1.660598179453836,
293
+ "grad_norm": 0.8247058391571045,
294
+ "learning_rate": 1.6677083333333333e-05,
295
+ "loss": 3.3884,
296
+ "mean_token_accuracy": 0.3924022614955902,
297
+ "num_tokens": 16235540.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 1.7126137841352405,
302
+ "grad_norm": 0.9121173024177551,
303
+ "learning_rate": 1.657291666666667e-05,
304
+ "loss": 3.3541,
305
+ "mean_token_accuracy": 0.3951391130685806,
306
+ "num_tokens": 16747733.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 1.7646293888166449,
311
+ "grad_norm": 1.3171864748001099,
312
+ "learning_rate": 1.6468750000000004e-05,
313
+ "loss": 3.3486,
314
+ "mean_token_accuracy": 0.39698211625218394,
315
+ "num_tokens": 17265239.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 1.8166449934980493,
320
+ "grad_norm": 4.172916412353516,
321
+ "learning_rate": 1.6364583333333336e-05,
322
+ "loss": 3.325,
323
+ "mean_token_accuracy": 0.39922175258398057,
324
+ "num_tokens": 17780508.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 1.8686605981794537,
329
+ "grad_norm": 0.4353245794773102,
330
+ "learning_rate": 1.6260416666666668e-05,
331
+ "loss": 3.2807,
332
+ "mean_token_accuracy": 0.4035400986671448,
333
+ "num_tokens": 18297602.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 1.9206762028608582,
338
+ "grad_norm": 3.6861395835876465,
339
+ "learning_rate": 1.6156250000000003e-05,
340
+ "loss": 3.2832,
341
+ "mean_token_accuracy": 0.4044442743062973,
342
+ "num_tokens": 18798625.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 1.9726918075422626,
347
+ "grad_norm": 1.1718441247940063,
348
+ "learning_rate": 1.6052083333333335e-05,
349
+ "loss": 3.2664,
350
+ "mean_token_accuracy": 0.40756076350808146,
351
+ "num_tokens": 19294344.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 2.020806241872562,
356
+ "grad_norm": 0.6570408940315247,
357
+ "learning_rate": 1.5947916666666667e-05,
358
+ "loss": 3.2056,
359
+ "mean_token_accuracy": 0.4098896440621969,
360
+ "num_tokens": 19761637.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 2.0728218465539663,
365
+ "grad_norm": 0.46285247802734375,
366
+ "learning_rate": 1.5843750000000002e-05,
367
+ "loss": 3.2396,
368
+ "mean_token_accuracy": 0.41089860200881956,
369
+ "num_tokens": 20265138.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 2.1248374512353707,
374
+ "grad_norm": 3.5917506217956543,
375
+ "learning_rate": 1.5739583333333334e-05,
376
+ "loss": 3.2205,
377
+ "mean_token_accuracy": 0.41081195920705793,
378
+ "num_tokens": 20777818.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 2.176853055916775,
383
+ "grad_norm": 0.6580970287322998,
384
+ "learning_rate": 1.5635416666666666e-05,
385
+ "loss": 3.225,
386
+ "mean_token_accuracy": 0.41244959980249407,
387
+ "num_tokens": 21282514.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 2.2288686605981796,
392
+ "grad_norm": 1.2673513889312744,
393
+ "learning_rate": 1.553125e-05,
394
+ "loss": 3.2143,
395
+ "mean_token_accuracy": 0.41195737645030023,
396
+ "num_tokens": 21796246.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 2.280884265279584,
401
+ "grad_norm": 1.9437355995178223,
402
+ "learning_rate": 1.5427083333333333e-05,
403
+ "loss": 3.2027,
404
+ "mean_token_accuracy": 0.41485874727368355,
405
+ "num_tokens": 22310915.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 2.3328998699609884,
410
+ "grad_norm": 0.4803381860256195,
411
+ "learning_rate": 1.532291666666667e-05,
412
+ "loss": 3.1743,
413
+ "mean_token_accuracy": 0.417393134534359,
414
+ "num_tokens": 22823271.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 2.384915474642393,
419
+ "grad_norm": 1.0227100849151611,
420
+ "learning_rate": 1.5218750000000002e-05,
421
+ "loss": 3.1751,
422
+ "mean_token_accuracy": 0.4173254922032356,
423
+ "num_tokens": 23326555.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 2.4369310793237973,
428
+ "grad_norm": 0.9217405915260315,
429
+ "learning_rate": 1.5114583333333334e-05,
430
+ "loss": 3.1336,
431
+ "mean_token_accuracy": 0.42112250030040743,
432
+ "num_tokens": 23832549.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 2.4889466840052017,
437
+ "grad_norm": 0.7527967095375061,
438
+ "learning_rate": 1.5010416666666668e-05,
439
+ "loss": 3.1469,
440
+ "mean_token_accuracy": 0.42071041762828826,
441
+ "num_tokens": 24338061.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 2.540962288686606,
446
+ "grad_norm": 0.370236873626709,
447
+ "learning_rate": 1.4906250000000001e-05,
448
+ "loss": 3.1461,
449
+ "mean_token_accuracy": 0.41948254629969595,
450
+ "num_tokens": 24854140.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 2.5929778933680105,
455
+ "grad_norm": 0.4003697335720062,
456
+ "learning_rate": 1.4802083333333335e-05,
457
+ "loss": 3.1162,
458
+ "mean_token_accuracy": 0.4244517408311367,
459
+ "num_tokens": 25367383.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 2.644993498049415,
464
+ "grad_norm": 2.082859516143799,
465
+ "learning_rate": 1.4697916666666667e-05,
466
+ "loss": 3.1256,
467
+ "mean_token_accuracy": 0.42284995764493943,
468
+ "num_tokens": 25872016.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 2.6970091027308194,
473
+ "grad_norm": 2.390314817428589,
474
+ "learning_rate": 1.4593750000000002e-05,
475
+ "loss": 3.0869,
476
+ "mean_token_accuracy": 0.42750845700502393,
477
+ "num_tokens": 26374448.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 2.749024707412224,
482
+ "grad_norm": 0.46641266345977783,
483
+ "learning_rate": 1.4489583333333334e-05,
484
+ "loss": 3.0982,
485
+ "mean_token_accuracy": 0.42524692341685294,
486
+ "num_tokens": 26892842.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 2.8010403120936282,
491
+ "grad_norm": 2.4539034366607666,
492
+ "learning_rate": 1.4385416666666668e-05,
493
+ "loss": 3.0921,
494
+ "mean_token_accuracy": 0.425856501609087,
495
+ "num_tokens": 27399620.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 2.8530559167750327,
500
+ "grad_norm": 0.5952461957931519,
501
+ "learning_rate": 1.4281250000000002e-05,
502
+ "loss": 3.0888,
503
+ "mean_token_accuracy": 0.4283280909061432,
504
+ "num_tokens": 27901792.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 2.9050715214564367,
509
+ "grad_norm": 0.8322259783744812,
510
+ "learning_rate": 1.4177083333333335e-05,
511
+ "loss": 3.0828,
512
+ "mean_token_accuracy": 0.4277593605220318,
513
+ "num_tokens": 28412070.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 2.957087126137841,
518
+ "grad_norm": 0.499178409576416,
519
+ "learning_rate": 1.4072916666666667e-05,
520
+ "loss": 3.0776,
521
+ "mean_token_accuracy": 0.42896960899233816,
522
+ "num_tokens": 28925255.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 3.0052015604681404,
527
+ "grad_norm": 0.5208362340927124,
528
+ "learning_rate": 1.3968750000000002e-05,
529
+ "loss": 3.0475,
530
+ "mean_token_accuracy": 0.4297164734956381,
531
+ "num_tokens": 29380955.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 3.057217165149545,
536
+ "grad_norm": 1.1485410928726196,
537
+ "learning_rate": 1.3864583333333334e-05,
538
+ "loss": 3.0694,
539
+ "mean_token_accuracy": 0.42906737625598906,
540
+ "num_tokens": 29891475.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 3.109232769830949,
545
+ "grad_norm": 0.44713732600212097,
546
+ "learning_rate": 1.3760416666666668e-05,
547
+ "loss": 3.0237,
548
+ "mean_token_accuracy": 0.43474498465657235,
549
+ "num_tokens": 30397348.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 3.1612483745123536,
554
+ "grad_norm": 0.6080917716026306,
555
+ "learning_rate": 1.3656250000000002e-05,
556
+ "loss": 3.0392,
557
+ "mean_token_accuracy": 0.4334587998688221,
558
+ "num_tokens": 30901701.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 3.213263979193758,
563
+ "grad_norm": 1.6787530183792114,
564
+ "learning_rate": 1.3552083333333335e-05,
565
+ "loss": 3.0626,
566
+ "mean_token_accuracy": 0.4313603535294533,
567
+ "num_tokens": 31412858.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 3.2652795838751625,
572
+ "grad_norm": 3.046581745147705,
573
+ "learning_rate": 1.3447916666666667e-05,
574
+ "loss": 3.0271,
575
+ "mean_token_accuracy": 0.43369641304016116,
576
+ "num_tokens": 31911006.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 3.317295188556567,
581
+ "grad_norm": 1.0174599885940552,
582
+ "learning_rate": 1.3343750000000002e-05,
583
+ "loss": 3.0229,
584
+ "mean_token_accuracy": 0.43341642022132876,
585
+ "num_tokens": 32422873.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 3.3693107932379713,
590
+ "grad_norm": 0.4790953993797302,
591
+ "learning_rate": 1.3239583333333334e-05,
592
+ "loss": 3.0186,
593
+ "mean_token_accuracy": 0.43539193645119667,
594
+ "num_tokens": 32937766.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 3.4213263979193758,
599
+ "grad_norm": 4.097532272338867,
600
+ "learning_rate": 1.3135416666666666e-05,
601
+ "loss": 2.998,
602
+ "mean_token_accuracy": 0.4385197140276432,
603
+ "num_tokens": 33440445.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 3.47334200260078,
608
+ "grad_norm": 0.36602917313575745,
609
+ "learning_rate": 1.3031250000000002e-05,
610
+ "loss": 3.0098,
611
+ "mean_token_accuracy": 0.4368562549352646,
612
+ "num_tokens": 33939105.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 3.5253576072821846,
617
+ "grad_norm": 1.6092300415039062,
618
+ "learning_rate": 1.2927083333333335e-05,
619
+ "loss": 3.0193,
620
+ "mean_token_accuracy": 0.43651641234755517,
621
+ "num_tokens": 34450716.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 3.577373211963589,
626
+ "grad_norm": 3.6081929206848145,
627
+ "learning_rate": 1.2822916666666667e-05,
628
+ "loss": 2.9884,
629
+ "mean_token_accuracy": 0.4398868255317211,
630
+ "num_tokens": 34958114.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 3.6293888166449935,
635
+ "grad_norm": 2.0380210876464844,
636
+ "learning_rate": 1.2718750000000003e-05,
637
+ "loss": 3.0131,
638
+ "mean_token_accuracy": 0.43571230471134187,
639
+ "num_tokens": 35468951.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 3.681404421326398,
644
+ "grad_norm": 0.5758602023124695,
645
+ "learning_rate": 1.2614583333333334e-05,
646
+ "loss": 2.9752,
647
+ "mean_token_accuracy": 0.4403461940586567,
648
+ "num_tokens": 35977001.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 3.7334200260078023,
653
+ "grad_norm": 0.914013147354126,
654
+ "learning_rate": 1.2510416666666666e-05,
655
+ "loss": 3.0002,
656
+ "mean_token_accuracy": 0.4371410049498081,
657
+ "num_tokens": 36491206.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 3.7854356306892067,
662
+ "grad_norm": 0.8788081407546997,
663
+ "learning_rate": 1.2406250000000002e-05,
664
+ "loss": 3.0013,
665
+ "mean_token_accuracy": 0.43753816187381744,
666
+ "num_tokens": 36996454.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 3.837451235370611,
671
+ "grad_norm": 0.35317686200141907,
672
+ "learning_rate": 1.2302083333333335e-05,
673
+ "loss": 2.9874,
674
+ "mean_token_accuracy": 0.44055561572313306,
675
+ "num_tokens": 37493079.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 3.8894668400520156,
680
+ "grad_norm": 0.40355926752090454,
681
+ "learning_rate": 1.2197916666666667e-05,
682
+ "loss": 2.9743,
683
+ "mean_token_accuracy": 0.4413572125136852,
684
+ "num_tokens": 38004972.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 3.94148244473342,
689
+ "grad_norm": 0.3518731892108917,
690
+ "learning_rate": 1.2093750000000003e-05,
691
+ "loss": 2.9602,
692
+ "mean_token_accuracy": 0.4420863904058933,
693
+ "num_tokens": 38534943.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 3.9934980494148244,
698
+ "grad_norm": 0.7066172361373901,
699
+ "learning_rate": 1.1989583333333335e-05,
700
+ "loss": 2.9588,
701
+ "mean_token_accuracy": 0.44266087040305135,
702
+ "num_tokens": 39052563.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 4.041612483745124,
707
+ "grad_norm": 2.8007071018218994,
708
+ "learning_rate": 1.1885416666666666e-05,
709
+ "loss": 2.9561,
710
+ "mean_token_accuracy": 0.44085964157774643,
711
+ "num_tokens": 39521519.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 4.093628088426528,
716
+ "grad_norm": 0.5129738450050354,
717
+ "learning_rate": 1.1781250000000002e-05,
718
+ "loss": 2.942,
719
+ "mean_token_accuracy": 0.4447008825838566,
720
+ "num_tokens": 40026574.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 4.145643693107933,
725
+ "grad_norm": 0.750156044960022,
726
+ "learning_rate": 1.1677083333333334e-05,
727
+ "loss": 2.9617,
728
+ "mean_token_accuracy": 0.4419982649385929,
729
+ "num_tokens": 40546106.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 4.197659297789337,
734
+ "grad_norm": 0.567778468132019,
735
+ "learning_rate": 1.1572916666666667e-05,
736
+ "loss": 2.9514,
737
+ "mean_token_accuracy": 0.44232149720191954,
738
+ "num_tokens": 41064511.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 4.249674902470741,
743
+ "grad_norm": 0.3298749029636383,
744
+ "learning_rate": 1.1468750000000003e-05,
745
+ "loss": 2.9526,
746
+ "mean_token_accuracy": 0.4424647718667984,
747
+ "num_tokens": 41568948.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 4.301690507152146,
752
+ "grad_norm": 0.36816585063934326,
753
+ "learning_rate": 1.1364583333333335e-05,
754
+ "loss": 2.9254,
755
+ "mean_token_accuracy": 0.44615685641765596,
756
+ "num_tokens": 42075864.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 4.35370611183355,
761
+ "grad_norm": 2.7220842838287354,
762
+ "learning_rate": 1.1260416666666666e-05,
763
+ "loss": 2.9571,
764
+ "mean_token_accuracy": 0.44221931844949725,
765
+ "num_tokens": 42594515.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 4.405721716514955,
770
+ "grad_norm": 0.37389975786209106,
771
+ "learning_rate": 1.1156250000000002e-05,
772
+ "loss": 2.9441,
773
+ "mean_token_accuracy": 0.442495983093977,
774
+ "num_tokens": 43111918.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 4.457737321196359,
779
+ "grad_norm": 0.40067073702812195,
780
+ "learning_rate": 1.1052083333333334e-05,
781
+ "loss": 2.9252,
782
+ "mean_token_accuracy": 0.4479030266404152,
783
+ "num_tokens": 43613796.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 4.5097529258777636,
788
+ "grad_norm": 0.9721180200576782,
789
+ "learning_rate": 1.0947916666666667e-05,
790
+ "loss": 2.9317,
791
+ "mean_token_accuracy": 0.4480686657130718,
792
+ "num_tokens": 44116168.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 4.561768530559168,
797
+ "grad_norm": 0.6541546583175659,
798
+ "learning_rate": 1.0843750000000001e-05,
799
+ "loss": 2.9164,
800
+ "mean_token_accuracy": 0.4477278359234333,
801
+ "num_tokens": 44619073.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 4.613784135240572,
806
+ "grad_norm": 0.9818729758262634,
807
+ "learning_rate": 1.0739583333333335e-05,
808
+ "loss": 2.9236,
809
+ "mean_token_accuracy": 0.44781008884310725,
810
+ "num_tokens": 45122295.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 4.665799739921977,
815
+ "grad_norm": 0.5139995813369751,
816
+ "learning_rate": 1.0635416666666667e-05,
817
+ "loss": 2.9245,
818
+ "mean_token_accuracy": 0.4462748847901821,
819
+ "num_tokens": 45640893.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 4.717815344603381,
824
+ "grad_norm": 0.5460729002952576,
825
+ "learning_rate": 1.0531250000000002e-05,
826
+ "loss": 2.8997,
827
+ "mean_token_accuracy": 0.45041476264595987,
828
+ "num_tokens": 46149628.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 4.769830949284786,
833
+ "grad_norm": 0.4323829412460327,
834
+ "learning_rate": 1.0427083333333334e-05,
835
+ "loss": 2.9241,
836
+ "mean_token_accuracy": 0.4480301558971405,
837
+ "num_tokens": 46658455.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 4.82184655396619,
842
+ "grad_norm": 0.368254154920578,
843
+ "learning_rate": 1.0322916666666667e-05,
844
+ "loss": 2.9198,
845
+ "mean_token_accuracy": 0.4473501190543175,
846
+ "num_tokens": 47152789.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 4.8738621586475945,
851
+ "grad_norm": 1.6135244369506836,
852
+ "learning_rate": 1.0218750000000001e-05,
853
+ "loss": 2.9327,
854
+ "mean_token_accuracy": 0.44697768464684484,
855
+ "num_tokens": 47664450.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 4.925877763328999,
860
+ "grad_norm": 0.5351427793502808,
861
+ "learning_rate": 1.0114583333333335e-05,
862
+ "loss": 2.9031,
863
+ "mean_token_accuracy": 0.4496559798717499,
864
+ "num_tokens": 48166797.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 4.977893368010403,
869
+ "grad_norm": 0.7081937193870544,
870
+ "learning_rate": 1.0010416666666667e-05,
871
+ "loss": 2.8935,
872
+ "mean_token_accuracy": 0.45092770233750346,
873
+ "num_tokens": 48674756.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 5.026007802340702,
878
+ "grad_norm": 35.31280517578125,
879
+ "learning_rate": 9.90625e-06,
880
+ "loss": 2.9106,
881
+ "mean_token_accuracy": 0.4502824319375528,
882
+ "num_tokens": 49132955.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 5.078023407022107,
887
+ "grad_norm": 0.49765506386756897,
888
+ "learning_rate": 9.802083333333334e-06,
889
+ "loss": 2.9015,
890
+ "mean_token_accuracy": 0.44894649609923365,
891
+ "num_tokens": 49642865.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 5.130039011703511,
896
+ "grad_norm": 0.4912125766277313,
897
+ "learning_rate": 9.697916666666667e-06,
898
+ "loss": 2.8704,
899
+ "mean_token_accuracy": 0.4528684295713902,
900
+ "num_tokens": 50145710.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 5.1820546163849155,
905
+ "grad_norm": 0.8951883912086487,
906
+ "learning_rate": 9.593750000000001e-06,
907
+ "loss": 2.8903,
908
+ "mean_token_accuracy": 0.45256839543581007,
909
+ "num_tokens": 50646447.0,
910
+ "step": 1000
911
+ }
912
+ ],
913
+ "logging_steps": 10,
914
+ "max_steps": 1920,
915
+ "num_input_tokens_seen": 0,
916
+ "num_train_epochs": 10,
917
+ "save_steps": 500,
918
+ "stateful_callbacks": {
919
+ "TrainerControl": {
920
+ "args": {
921
+ "should_epoch_stop": false,
922
+ "should_evaluate": false,
923
+ "should_log": false,
924
+ "should_save": true,
925
+ "should_training_stop": false
926
+ },
927
+ "attributes": {}
928
+ }
929
+ },
930
+ "total_flos": 4.121866184839373e+16,
931
+ "train_batch_size": 64,
932
+ "trial_name": null,
933
+ "trial_params": null
934
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea990651e2aecabd57f309f323439c138096e78f77c92b4e54f46fe372d50d1d
3
+ size 5688
checkpoint-1500/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 512,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-1500/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-1500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef80293cac93eb28f42e3299347ff9eb37c441b4630506fc96cc667b2313a969
3
+ size 309900448
checkpoint-1500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7912b61e428816ff0a59a62aadca49f43bbea732e0282e5bfa220fb84b37760d
3
+ size 619836730
checkpoint-1500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2768285b45b2a0c05f6f50bbb8c0287fca6f62a8cde6d1b1f02151ac72ee8dc
3
+ size 14244
checkpoint-1500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c10b3ee8840c289f63142a61aedbca61a8366ca19e42a8da9d40908b5bc46815
3
+ size 1064
checkpoint-1500/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-1500/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1500/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1500/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-1500/trainer_state.json ADDED
@@ -0,0 +1,1384 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 7.775032509752926,
6
+ "eval_steps": 500,
7
+ "global_step": 1500,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.05201560468140442,
14
+ "grad_norm": 3.158724784851074,
15
+ "learning_rate": 1.9906250000000003e-05,
16
+ "loss": 6.2302,
17
+ "mean_token_accuracy": 0.16946521885693072,
18
+ "num_tokens": 513852.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.10403120936280884,
23
+ "grad_norm": 10.176986694335938,
24
+ "learning_rate": 1.9802083333333335e-05,
25
+ "loss": 6.0745,
26
+ "mean_token_accuracy": 0.17949572652578355,
27
+ "num_tokens": 1032088.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.15604681404421328,
32
+ "grad_norm": 3.010871410369873,
33
+ "learning_rate": 1.9697916666666667e-05,
34
+ "loss": 5.8317,
35
+ "mean_token_accuracy": 0.1901865139603615,
36
+ "num_tokens": 1539389.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.20806241872561768,
41
+ "grad_norm": 3.298682451248169,
42
+ "learning_rate": 1.9593750000000002e-05,
43
+ "loss": 5.5987,
44
+ "mean_token_accuracy": 0.20738236755132675,
45
+ "num_tokens": 2057784.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.26007802340702213,
50
+ "grad_norm": 2.472743272781372,
51
+ "learning_rate": 1.9489583333333334e-05,
52
+ "loss": 5.4572,
53
+ "mean_token_accuracy": 0.2219402514398098,
54
+ "num_tokens": 2574094.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.31209362808842656,
59
+ "grad_norm": 2.3076207637786865,
60
+ "learning_rate": 1.938541666666667e-05,
61
+ "loss": 5.2817,
62
+ "mean_token_accuracy": 0.23026303350925445,
63
+ "num_tokens": 3076289.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.3641092327698309,
68
+ "grad_norm": 2.1481237411499023,
69
+ "learning_rate": 1.928125e-05,
70
+ "loss": 5.0596,
71
+ "mean_token_accuracy": 0.24168218560516835,
72
+ "num_tokens": 3586260.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.41612483745123535,
77
+ "grad_norm": 2.2509524822235107,
78
+ "learning_rate": 1.9177083333333336e-05,
79
+ "loss": 4.9117,
80
+ "mean_token_accuracy": 0.2510342214256525,
81
+ "num_tokens": 4094922.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.4681404421326398,
86
+ "grad_norm": 3.319535732269287,
87
+ "learning_rate": 1.907291666666667e-05,
88
+ "loss": 4.7624,
89
+ "mean_token_accuracy": 0.26219444051384927,
90
+ "num_tokens": 4606336.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.5201560468140443,
95
+ "grad_norm": 2.0256783962249756,
96
+ "learning_rate": 1.8968750000000004e-05,
97
+ "loss": 4.6138,
98
+ "mean_token_accuracy": 0.2705389905720949,
99
+ "num_tokens": 5110749.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.5721716514954487,
104
+ "grad_norm": 5.284979820251465,
105
+ "learning_rate": 1.8864583333333336e-05,
106
+ "loss": 4.5011,
107
+ "mean_token_accuracy": 0.27774968966841695,
108
+ "num_tokens": 5611872.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.6241872561768531,
113
+ "grad_norm": 15.897697448730469,
114
+ "learning_rate": 1.8760416666666668e-05,
115
+ "loss": 4.3486,
116
+ "mean_token_accuracy": 0.29455167055130005,
117
+ "num_tokens": 6123716.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.6762028608582574,
122
+ "grad_norm": 3.0891332626342773,
123
+ "learning_rate": 1.8656250000000003e-05,
124
+ "loss": 4.2372,
125
+ "mean_token_accuracy": 0.3053073361515999,
126
+ "num_tokens": 6630449.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.7282184655396619,
131
+ "grad_norm": 2.85463285446167,
132
+ "learning_rate": 1.8552083333333335e-05,
133
+ "loss": 4.161,
134
+ "mean_token_accuracy": 0.3164319314062595,
135
+ "num_tokens": 7145717.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.7802340702210663,
140
+ "grad_norm": 1.2611316442489624,
141
+ "learning_rate": 1.8447916666666667e-05,
142
+ "loss": 4.1002,
143
+ "mean_token_accuracy": 0.32258255779743195,
144
+ "num_tokens": 7655884.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.8322496749024707,
149
+ "grad_norm": 1.4083240032196045,
150
+ "learning_rate": 1.8343750000000002e-05,
151
+ "loss": 4.0138,
152
+ "mean_token_accuracy": 0.3296029016375542,
153
+ "num_tokens": 8159725.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.8842652795838751,
158
+ "grad_norm": 1.0340592861175537,
159
+ "learning_rate": 1.8239583333333334e-05,
160
+ "loss": 3.9163,
161
+ "mean_token_accuracy": 0.3385446682572365,
162
+ "num_tokens": 8668116.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.9362808842652796,
167
+ "grad_norm": 0.7984447479248047,
168
+ "learning_rate": 1.8135416666666666e-05,
169
+ "loss": 3.8628,
170
+ "mean_token_accuracy": 0.3455250822007656,
171
+ "num_tokens": 9182489.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.988296488946684,
176
+ "grad_norm": 1.244081735610962,
177
+ "learning_rate": 1.803125e-05,
178
+ "loss": 3.8025,
179
+ "mean_token_accuracy": 0.35223659574985505,
180
+ "num_tokens": 9676078.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 1.0364109232769831,
185
+ "grad_norm": 2.412661552429199,
186
+ "learning_rate": 1.7927083333333337e-05,
187
+ "loss": 3.7325,
188
+ "mean_token_accuracy": 0.35705321382831884,
189
+ "num_tokens": 10138114.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 1.0884265279583876,
194
+ "grad_norm": 1.7736923694610596,
195
+ "learning_rate": 1.782291666666667e-05,
196
+ "loss": 3.7052,
197
+ "mean_token_accuracy": 0.3590236373245716,
198
+ "num_tokens": 10656504.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 1.140442132639792,
203
+ "grad_norm": 0.7769147157669067,
204
+ "learning_rate": 1.7718750000000004e-05,
205
+ "loss": 3.6433,
206
+ "mean_token_accuracy": 0.36575346291065214,
207
+ "num_tokens": 11161374.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 1.1924577373211964,
212
+ "grad_norm": 0.9089329838752747,
213
+ "learning_rate": 1.7614583333333336e-05,
214
+ "loss": 3.653,
215
+ "mean_token_accuracy": 0.3651699565351009,
216
+ "num_tokens": 11670197.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 1.2444733420026008,
221
+ "grad_norm": 0.7330015897750854,
222
+ "learning_rate": 1.7510416666666668e-05,
223
+ "loss": 3.5843,
224
+ "mean_token_accuracy": 0.3718643538653851,
225
+ "num_tokens": 12174931.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 1.2964889466840053,
230
+ "grad_norm": 1.534440040588379,
231
+ "learning_rate": 1.7406250000000003e-05,
232
+ "loss": 3.5468,
233
+ "mean_token_accuracy": 0.3751932092010975,
234
+ "num_tokens": 12685023.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 1.3485045513654097,
239
+ "grad_norm": 1.9603444337844849,
240
+ "learning_rate": 1.7302083333333335e-05,
241
+ "loss": 3.4935,
242
+ "mean_token_accuracy": 0.3808542750775814,
243
+ "num_tokens": 13197338.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 1.4005201560468141,
248
+ "grad_norm": 0.5988335013389587,
249
+ "learning_rate": 1.7197916666666667e-05,
250
+ "loss": 3.4988,
251
+ "mean_token_accuracy": 0.3815221585333347,
252
+ "num_tokens": 13704900.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 1.4525357607282183,
257
+ "grad_norm": 2.7422847747802734,
258
+ "learning_rate": 1.7093750000000002e-05,
259
+ "loss": 3.4438,
260
+ "mean_token_accuracy": 0.3865826904773712,
261
+ "num_tokens": 14210835.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 1.5045513654096228,
266
+ "grad_norm": 1.9575722217559814,
267
+ "learning_rate": 1.6989583333333334e-05,
268
+ "loss": 3.417,
269
+ "mean_token_accuracy": 0.38725101202726364,
270
+ "num_tokens": 14726584.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 1.5565669700910272,
275
+ "grad_norm": 0.6955925226211548,
276
+ "learning_rate": 1.6885416666666666e-05,
277
+ "loss": 3.3895,
278
+ "mean_token_accuracy": 0.39159547463059424,
279
+ "num_tokens": 15238009.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 1.6085825747724316,
284
+ "grad_norm": 0.769548237323761,
285
+ "learning_rate": 1.678125e-05,
286
+ "loss": 3.3979,
287
+ "mean_token_accuracy": 0.39134200662374496,
288
+ "num_tokens": 15736398.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 1.660598179453836,
293
+ "grad_norm": 0.8247058391571045,
294
+ "learning_rate": 1.6677083333333333e-05,
295
+ "loss": 3.3884,
296
+ "mean_token_accuracy": 0.3924022614955902,
297
+ "num_tokens": 16235540.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 1.7126137841352405,
302
+ "grad_norm": 0.9121173024177551,
303
+ "learning_rate": 1.657291666666667e-05,
304
+ "loss": 3.3541,
305
+ "mean_token_accuracy": 0.3951391130685806,
306
+ "num_tokens": 16747733.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 1.7646293888166449,
311
+ "grad_norm": 1.3171864748001099,
312
+ "learning_rate": 1.6468750000000004e-05,
313
+ "loss": 3.3486,
314
+ "mean_token_accuracy": 0.39698211625218394,
315
+ "num_tokens": 17265239.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 1.8166449934980493,
320
+ "grad_norm": 4.172916412353516,
321
+ "learning_rate": 1.6364583333333336e-05,
322
+ "loss": 3.325,
323
+ "mean_token_accuracy": 0.39922175258398057,
324
+ "num_tokens": 17780508.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 1.8686605981794537,
329
+ "grad_norm": 0.4353245794773102,
330
+ "learning_rate": 1.6260416666666668e-05,
331
+ "loss": 3.2807,
332
+ "mean_token_accuracy": 0.4035400986671448,
333
+ "num_tokens": 18297602.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 1.9206762028608582,
338
+ "grad_norm": 3.6861395835876465,
339
+ "learning_rate": 1.6156250000000003e-05,
340
+ "loss": 3.2832,
341
+ "mean_token_accuracy": 0.4044442743062973,
342
+ "num_tokens": 18798625.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 1.9726918075422626,
347
+ "grad_norm": 1.1718441247940063,
348
+ "learning_rate": 1.6052083333333335e-05,
349
+ "loss": 3.2664,
350
+ "mean_token_accuracy": 0.40756076350808146,
351
+ "num_tokens": 19294344.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 2.020806241872562,
356
+ "grad_norm": 0.6570408940315247,
357
+ "learning_rate": 1.5947916666666667e-05,
358
+ "loss": 3.2056,
359
+ "mean_token_accuracy": 0.4098896440621969,
360
+ "num_tokens": 19761637.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 2.0728218465539663,
365
+ "grad_norm": 0.46285247802734375,
366
+ "learning_rate": 1.5843750000000002e-05,
367
+ "loss": 3.2396,
368
+ "mean_token_accuracy": 0.41089860200881956,
369
+ "num_tokens": 20265138.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 2.1248374512353707,
374
+ "grad_norm": 3.5917506217956543,
375
+ "learning_rate": 1.5739583333333334e-05,
376
+ "loss": 3.2205,
377
+ "mean_token_accuracy": 0.41081195920705793,
378
+ "num_tokens": 20777818.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 2.176853055916775,
383
+ "grad_norm": 0.6580970287322998,
384
+ "learning_rate": 1.5635416666666666e-05,
385
+ "loss": 3.225,
386
+ "mean_token_accuracy": 0.41244959980249407,
387
+ "num_tokens": 21282514.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 2.2288686605981796,
392
+ "grad_norm": 1.2673513889312744,
393
+ "learning_rate": 1.553125e-05,
394
+ "loss": 3.2143,
395
+ "mean_token_accuracy": 0.41195737645030023,
396
+ "num_tokens": 21796246.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 2.280884265279584,
401
+ "grad_norm": 1.9437355995178223,
402
+ "learning_rate": 1.5427083333333333e-05,
403
+ "loss": 3.2027,
404
+ "mean_token_accuracy": 0.41485874727368355,
405
+ "num_tokens": 22310915.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 2.3328998699609884,
410
+ "grad_norm": 0.4803381860256195,
411
+ "learning_rate": 1.532291666666667e-05,
412
+ "loss": 3.1743,
413
+ "mean_token_accuracy": 0.417393134534359,
414
+ "num_tokens": 22823271.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 2.384915474642393,
419
+ "grad_norm": 1.0227100849151611,
420
+ "learning_rate": 1.5218750000000002e-05,
421
+ "loss": 3.1751,
422
+ "mean_token_accuracy": 0.4173254922032356,
423
+ "num_tokens": 23326555.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 2.4369310793237973,
428
+ "grad_norm": 0.9217405915260315,
429
+ "learning_rate": 1.5114583333333334e-05,
430
+ "loss": 3.1336,
431
+ "mean_token_accuracy": 0.42112250030040743,
432
+ "num_tokens": 23832549.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 2.4889466840052017,
437
+ "grad_norm": 0.7527967095375061,
438
+ "learning_rate": 1.5010416666666668e-05,
439
+ "loss": 3.1469,
440
+ "mean_token_accuracy": 0.42071041762828826,
441
+ "num_tokens": 24338061.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 2.540962288686606,
446
+ "grad_norm": 0.370236873626709,
447
+ "learning_rate": 1.4906250000000001e-05,
448
+ "loss": 3.1461,
449
+ "mean_token_accuracy": 0.41948254629969595,
450
+ "num_tokens": 24854140.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 2.5929778933680105,
455
+ "grad_norm": 0.4003697335720062,
456
+ "learning_rate": 1.4802083333333335e-05,
457
+ "loss": 3.1162,
458
+ "mean_token_accuracy": 0.4244517408311367,
459
+ "num_tokens": 25367383.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 2.644993498049415,
464
+ "grad_norm": 2.082859516143799,
465
+ "learning_rate": 1.4697916666666667e-05,
466
+ "loss": 3.1256,
467
+ "mean_token_accuracy": 0.42284995764493943,
468
+ "num_tokens": 25872016.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 2.6970091027308194,
473
+ "grad_norm": 2.390314817428589,
474
+ "learning_rate": 1.4593750000000002e-05,
475
+ "loss": 3.0869,
476
+ "mean_token_accuracy": 0.42750845700502393,
477
+ "num_tokens": 26374448.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 2.749024707412224,
482
+ "grad_norm": 0.46641266345977783,
483
+ "learning_rate": 1.4489583333333334e-05,
484
+ "loss": 3.0982,
485
+ "mean_token_accuracy": 0.42524692341685294,
486
+ "num_tokens": 26892842.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 2.8010403120936282,
491
+ "grad_norm": 2.4539034366607666,
492
+ "learning_rate": 1.4385416666666668e-05,
493
+ "loss": 3.0921,
494
+ "mean_token_accuracy": 0.425856501609087,
495
+ "num_tokens": 27399620.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 2.8530559167750327,
500
+ "grad_norm": 0.5952461957931519,
501
+ "learning_rate": 1.4281250000000002e-05,
502
+ "loss": 3.0888,
503
+ "mean_token_accuracy": 0.4283280909061432,
504
+ "num_tokens": 27901792.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 2.9050715214564367,
509
+ "grad_norm": 0.8322259783744812,
510
+ "learning_rate": 1.4177083333333335e-05,
511
+ "loss": 3.0828,
512
+ "mean_token_accuracy": 0.4277593605220318,
513
+ "num_tokens": 28412070.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 2.957087126137841,
518
+ "grad_norm": 0.499178409576416,
519
+ "learning_rate": 1.4072916666666667e-05,
520
+ "loss": 3.0776,
521
+ "mean_token_accuracy": 0.42896960899233816,
522
+ "num_tokens": 28925255.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 3.0052015604681404,
527
+ "grad_norm": 0.5208362340927124,
528
+ "learning_rate": 1.3968750000000002e-05,
529
+ "loss": 3.0475,
530
+ "mean_token_accuracy": 0.4297164734956381,
531
+ "num_tokens": 29380955.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 3.057217165149545,
536
+ "grad_norm": 1.1485410928726196,
537
+ "learning_rate": 1.3864583333333334e-05,
538
+ "loss": 3.0694,
539
+ "mean_token_accuracy": 0.42906737625598906,
540
+ "num_tokens": 29891475.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 3.109232769830949,
545
+ "grad_norm": 0.44713732600212097,
546
+ "learning_rate": 1.3760416666666668e-05,
547
+ "loss": 3.0237,
548
+ "mean_token_accuracy": 0.43474498465657235,
549
+ "num_tokens": 30397348.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 3.1612483745123536,
554
+ "grad_norm": 0.6080917716026306,
555
+ "learning_rate": 1.3656250000000002e-05,
556
+ "loss": 3.0392,
557
+ "mean_token_accuracy": 0.4334587998688221,
558
+ "num_tokens": 30901701.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 3.213263979193758,
563
+ "grad_norm": 1.6787530183792114,
564
+ "learning_rate": 1.3552083333333335e-05,
565
+ "loss": 3.0626,
566
+ "mean_token_accuracy": 0.4313603535294533,
567
+ "num_tokens": 31412858.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 3.2652795838751625,
572
+ "grad_norm": 3.046581745147705,
573
+ "learning_rate": 1.3447916666666667e-05,
574
+ "loss": 3.0271,
575
+ "mean_token_accuracy": 0.43369641304016116,
576
+ "num_tokens": 31911006.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 3.317295188556567,
581
+ "grad_norm": 1.0174599885940552,
582
+ "learning_rate": 1.3343750000000002e-05,
583
+ "loss": 3.0229,
584
+ "mean_token_accuracy": 0.43341642022132876,
585
+ "num_tokens": 32422873.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 3.3693107932379713,
590
+ "grad_norm": 0.4790953993797302,
591
+ "learning_rate": 1.3239583333333334e-05,
592
+ "loss": 3.0186,
593
+ "mean_token_accuracy": 0.43539193645119667,
594
+ "num_tokens": 32937766.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 3.4213263979193758,
599
+ "grad_norm": 4.097532272338867,
600
+ "learning_rate": 1.3135416666666666e-05,
601
+ "loss": 2.998,
602
+ "mean_token_accuracy": 0.4385197140276432,
603
+ "num_tokens": 33440445.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 3.47334200260078,
608
+ "grad_norm": 0.36602917313575745,
609
+ "learning_rate": 1.3031250000000002e-05,
610
+ "loss": 3.0098,
611
+ "mean_token_accuracy": 0.4368562549352646,
612
+ "num_tokens": 33939105.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 3.5253576072821846,
617
+ "grad_norm": 1.6092300415039062,
618
+ "learning_rate": 1.2927083333333335e-05,
619
+ "loss": 3.0193,
620
+ "mean_token_accuracy": 0.43651641234755517,
621
+ "num_tokens": 34450716.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 3.577373211963589,
626
+ "grad_norm": 3.6081929206848145,
627
+ "learning_rate": 1.2822916666666667e-05,
628
+ "loss": 2.9884,
629
+ "mean_token_accuracy": 0.4398868255317211,
630
+ "num_tokens": 34958114.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 3.6293888166449935,
635
+ "grad_norm": 2.0380210876464844,
636
+ "learning_rate": 1.2718750000000003e-05,
637
+ "loss": 3.0131,
638
+ "mean_token_accuracy": 0.43571230471134187,
639
+ "num_tokens": 35468951.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 3.681404421326398,
644
+ "grad_norm": 0.5758602023124695,
645
+ "learning_rate": 1.2614583333333334e-05,
646
+ "loss": 2.9752,
647
+ "mean_token_accuracy": 0.4403461940586567,
648
+ "num_tokens": 35977001.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 3.7334200260078023,
653
+ "grad_norm": 0.914013147354126,
654
+ "learning_rate": 1.2510416666666666e-05,
655
+ "loss": 3.0002,
656
+ "mean_token_accuracy": 0.4371410049498081,
657
+ "num_tokens": 36491206.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 3.7854356306892067,
662
+ "grad_norm": 0.8788081407546997,
663
+ "learning_rate": 1.2406250000000002e-05,
664
+ "loss": 3.0013,
665
+ "mean_token_accuracy": 0.43753816187381744,
666
+ "num_tokens": 36996454.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 3.837451235370611,
671
+ "grad_norm": 0.35317686200141907,
672
+ "learning_rate": 1.2302083333333335e-05,
673
+ "loss": 2.9874,
674
+ "mean_token_accuracy": 0.44055561572313306,
675
+ "num_tokens": 37493079.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 3.8894668400520156,
680
+ "grad_norm": 0.40355926752090454,
681
+ "learning_rate": 1.2197916666666667e-05,
682
+ "loss": 2.9743,
683
+ "mean_token_accuracy": 0.4413572125136852,
684
+ "num_tokens": 38004972.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 3.94148244473342,
689
+ "grad_norm": 0.3518731892108917,
690
+ "learning_rate": 1.2093750000000003e-05,
691
+ "loss": 2.9602,
692
+ "mean_token_accuracy": 0.4420863904058933,
693
+ "num_tokens": 38534943.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 3.9934980494148244,
698
+ "grad_norm": 0.7066172361373901,
699
+ "learning_rate": 1.1989583333333335e-05,
700
+ "loss": 2.9588,
701
+ "mean_token_accuracy": 0.44266087040305135,
702
+ "num_tokens": 39052563.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 4.041612483745124,
707
+ "grad_norm": 2.8007071018218994,
708
+ "learning_rate": 1.1885416666666666e-05,
709
+ "loss": 2.9561,
710
+ "mean_token_accuracy": 0.44085964157774643,
711
+ "num_tokens": 39521519.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 4.093628088426528,
716
+ "grad_norm": 0.5129738450050354,
717
+ "learning_rate": 1.1781250000000002e-05,
718
+ "loss": 2.942,
719
+ "mean_token_accuracy": 0.4447008825838566,
720
+ "num_tokens": 40026574.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 4.145643693107933,
725
+ "grad_norm": 0.750156044960022,
726
+ "learning_rate": 1.1677083333333334e-05,
727
+ "loss": 2.9617,
728
+ "mean_token_accuracy": 0.4419982649385929,
729
+ "num_tokens": 40546106.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 4.197659297789337,
734
+ "grad_norm": 0.567778468132019,
735
+ "learning_rate": 1.1572916666666667e-05,
736
+ "loss": 2.9514,
737
+ "mean_token_accuracy": 0.44232149720191954,
738
+ "num_tokens": 41064511.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 4.249674902470741,
743
+ "grad_norm": 0.3298749029636383,
744
+ "learning_rate": 1.1468750000000003e-05,
745
+ "loss": 2.9526,
746
+ "mean_token_accuracy": 0.4424647718667984,
747
+ "num_tokens": 41568948.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 4.301690507152146,
752
+ "grad_norm": 0.36816585063934326,
753
+ "learning_rate": 1.1364583333333335e-05,
754
+ "loss": 2.9254,
755
+ "mean_token_accuracy": 0.44615685641765596,
756
+ "num_tokens": 42075864.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 4.35370611183355,
761
+ "grad_norm": 2.7220842838287354,
762
+ "learning_rate": 1.1260416666666666e-05,
763
+ "loss": 2.9571,
764
+ "mean_token_accuracy": 0.44221931844949725,
765
+ "num_tokens": 42594515.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 4.405721716514955,
770
+ "grad_norm": 0.37389975786209106,
771
+ "learning_rate": 1.1156250000000002e-05,
772
+ "loss": 2.9441,
773
+ "mean_token_accuracy": 0.442495983093977,
774
+ "num_tokens": 43111918.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 4.457737321196359,
779
+ "grad_norm": 0.40067073702812195,
780
+ "learning_rate": 1.1052083333333334e-05,
781
+ "loss": 2.9252,
782
+ "mean_token_accuracy": 0.4479030266404152,
783
+ "num_tokens": 43613796.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 4.5097529258777636,
788
+ "grad_norm": 0.9721180200576782,
789
+ "learning_rate": 1.0947916666666667e-05,
790
+ "loss": 2.9317,
791
+ "mean_token_accuracy": 0.4480686657130718,
792
+ "num_tokens": 44116168.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 4.561768530559168,
797
+ "grad_norm": 0.6541546583175659,
798
+ "learning_rate": 1.0843750000000001e-05,
799
+ "loss": 2.9164,
800
+ "mean_token_accuracy": 0.4477278359234333,
801
+ "num_tokens": 44619073.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 4.613784135240572,
806
+ "grad_norm": 0.9818729758262634,
807
+ "learning_rate": 1.0739583333333335e-05,
808
+ "loss": 2.9236,
809
+ "mean_token_accuracy": 0.44781008884310725,
810
+ "num_tokens": 45122295.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 4.665799739921977,
815
+ "grad_norm": 0.5139995813369751,
816
+ "learning_rate": 1.0635416666666667e-05,
817
+ "loss": 2.9245,
818
+ "mean_token_accuracy": 0.4462748847901821,
819
+ "num_tokens": 45640893.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 4.717815344603381,
824
+ "grad_norm": 0.5460729002952576,
825
+ "learning_rate": 1.0531250000000002e-05,
826
+ "loss": 2.8997,
827
+ "mean_token_accuracy": 0.45041476264595987,
828
+ "num_tokens": 46149628.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 4.769830949284786,
833
+ "grad_norm": 0.4323829412460327,
834
+ "learning_rate": 1.0427083333333334e-05,
835
+ "loss": 2.9241,
836
+ "mean_token_accuracy": 0.4480301558971405,
837
+ "num_tokens": 46658455.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 4.82184655396619,
842
+ "grad_norm": 0.368254154920578,
843
+ "learning_rate": 1.0322916666666667e-05,
844
+ "loss": 2.9198,
845
+ "mean_token_accuracy": 0.4473501190543175,
846
+ "num_tokens": 47152789.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 4.8738621586475945,
851
+ "grad_norm": 1.6135244369506836,
852
+ "learning_rate": 1.0218750000000001e-05,
853
+ "loss": 2.9327,
854
+ "mean_token_accuracy": 0.44697768464684484,
855
+ "num_tokens": 47664450.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 4.925877763328999,
860
+ "grad_norm": 0.5351427793502808,
861
+ "learning_rate": 1.0114583333333335e-05,
862
+ "loss": 2.9031,
863
+ "mean_token_accuracy": 0.4496559798717499,
864
+ "num_tokens": 48166797.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 4.977893368010403,
869
+ "grad_norm": 0.7081937193870544,
870
+ "learning_rate": 1.0010416666666667e-05,
871
+ "loss": 2.8935,
872
+ "mean_token_accuracy": 0.45092770233750346,
873
+ "num_tokens": 48674756.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 5.026007802340702,
878
+ "grad_norm": 35.31280517578125,
879
+ "learning_rate": 9.90625e-06,
880
+ "loss": 2.9106,
881
+ "mean_token_accuracy": 0.4502824319375528,
882
+ "num_tokens": 49132955.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 5.078023407022107,
887
+ "grad_norm": 0.49765506386756897,
888
+ "learning_rate": 9.802083333333334e-06,
889
+ "loss": 2.9015,
890
+ "mean_token_accuracy": 0.44894649609923365,
891
+ "num_tokens": 49642865.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 5.130039011703511,
896
+ "grad_norm": 0.4912125766277313,
897
+ "learning_rate": 9.697916666666667e-06,
898
+ "loss": 2.8704,
899
+ "mean_token_accuracy": 0.4528684295713902,
900
+ "num_tokens": 50145710.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 5.1820546163849155,
905
+ "grad_norm": 0.8951883912086487,
906
+ "learning_rate": 9.593750000000001e-06,
907
+ "loss": 2.8903,
908
+ "mean_token_accuracy": 0.45256839543581007,
909
+ "num_tokens": 50646447.0,
910
+ "step": 1000
911
+ },
912
+ {
913
+ "epoch": 5.23407022106632,
914
+ "grad_norm": 0.3062414824962616,
915
+ "learning_rate": 9.489583333333335e-06,
916
+ "loss": 2.9118,
917
+ "mean_token_accuracy": 0.44818254336714747,
918
+ "num_tokens": 51167514.0,
919
+ "step": 1010
920
+ },
921
+ {
922
+ "epoch": 5.286085825747724,
923
+ "grad_norm": 0.6343457102775574,
924
+ "learning_rate": 9.385416666666668e-06,
925
+ "loss": 2.8994,
926
+ "mean_token_accuracy": 0.44962315559387206,
927
+ "num_tokens": 51684111.0,
928
+ "step": 1020
929
+ },
930
+ {
931
+ "epoch": 5.338101430429129,
932
+ "grad_norm": 0.9417049288749695,
933
+ "learning_rate": 9.28125e-06,
934
+ "loss": 2.8761,
935
+ "mean_token_accuracy": 0.4527172662317753,
936
+ "num_tokens": 52189036.0,
937
+ "step": 1030
938
+ },
939
+ {
940
+ "epoch": 5.390117035110533,
941
+ "grad_norm": 0.6008946299552917,
942
+ "learning_rate": 9.177083333333334e-06,
943
+ "loss": 2.9054,
944
+ "mean_token_accuracy": 0.44979408457875253,
945
+ "num_tokens": 52700140.0,
946
+ "step": 1040
947
+ },
948
+ {
949
+ "epoch": 5.442132639791938,
950
+ "grad_norm": 1.5932005643844604,
951
+ "learning_rate": 9.072916666666668e-06,
952
+ "loss": 2.8849,
953
+ "mean_token_accuracy": 0.4512541800737381,
954
+ "num_tokens": 53209299.0,
955
+ "step": 1050
956
+ },
957
+ {
958
+ "epoch": 5.494148244473342,
959
+ "grad_norm": 0.4708751440048218,
960
+ "learning_rate": 8.968750000000001e-06,
961
+ "loss": 2.8747,
962
+ "mean_token_accuracy": 0.45460196733474734,
963
+ "num_tokens": 53708278.0,
964
+ "step": 1060
965
+ },
966
+ {
967
+ "epoch": 5.5461638491547465,
968
+ "grad_norm": 0.3820144832134247,
969
+ "learning_rate": 8.864583333333335e-06,
970
+ "loss": 2.8583,
971
+ "mean_token_accuracy": 0.4543687038123608,
972
+ "num_tokens": 54222052.0,
973
+ "step": 1070
974
+ },
975
+ {
976
+ "epoch": 5.598179453836151,
977
+ "grad_norm": 0.4965152442455292,
978
+ "learning_rate": 8.760416666666668e-06,
979
+ "loss": 2.9015,
980
+ "mean_token_accuracy": 0.45009259432554244,
981
+ "num_tokens": 54736716.0,
982
+ "step": 1080
983
+ },
984
+ {
985
+ "epoch": 5.650195058517555,
986
+ "grad_norm": 1.13825261592865,
987
+ "learning_rate": 8.65625e-06,
988
+ "loss": 2.8744,
989
+ "mean_token_accuracy": 0.4531993091106415,
990
+ "num_tokens": 55249854.0,
991
+ "step": 1090
992
+ },
993
+ {
994
+ "epoch": 5.70221066319896,
995
+ "grad_norm": 0.4815926253795624,
996
+ "learning_rate": 8.552083333333334e-06,
997
+ "loss": 2.85,
998
+ "mean_token_accuracy": 0.4550083190202713,
999
+ "num_tokens": 55756222.0,
1000
+ "step": 1100
1001
+ },
1002
+ {
1003
+ "epoch": 5.754226267880364,
1004
+ "grad_norm": 0.4123997986316681,
1005
+ "learning_rate": 8.447916666666668e-06,
1006
+ "loss": 2.8737,
1007
+ "mean_token_accuracy": 0.4540506564080715,
1008
+ "num_tokens": 56272172.0,
1009
+ "step": 1110
1010
+ },
1011
+ {
1012
+ "epoch": 5.806241872561769,
1013
+ "grad_norm": 0.3490988314151764,
1014
+ "learning_rate": 8.343750000000001e-06,
1015
+ "loss": 2.8835,
1016
+ "mean_token_accuracy": 0.45104856938123705,
1017
+ "num_tokens": 56780204.0,
1018
+ "step": 1120
1019
+ },
1020
+ {
1021
+ "epoch": 5.858257477243173,
1022
+ "grad_norm": 0.3921835124492645,
1023
+ "learning_rate": 8.239583333333335e-06,
1024
+ "loss": 2.8739,
1025
+ "mean_token_accuracy": 0.4529320389032364,
1026
+ "num_tokens": 57294606.0,
1027
+ "step": 1130
1028
+ },
1029
+ {
1030
+ "epoch": 5.9102730819245775,
1031
+ "grad_norm": 0.8325438499450684,
1032
+ "learning_rate": 8.135416666666668e-06,
1033
+ "loss": 2.8744,
1034
+ "mean_token_accuracy": 0.4542323037981987,
1035
+ "num_tokens": 57794088.0,
1036
+ "step": 1140
1037
+ },
1038
+ {
1039
+ "epoch": 5.962288686605982,
1040
+ "grad_norm": 0.3708273768424988,
1041
+ "learning_rate": 8.03125e-06,
1042
+ "loss": 2.8624,
1043
+ "mean_token_accuracy": 0.4541611663997173,
1044
+ "num_tokens": 58303085.0,
1045
+ "step": 1150
1046
+ },
1047
+ {
1048
+ "epoch": 6.010403120936281,
1049
+ "grad_norm": 0.45103487372398376,
1050
+ "learning_rate": 7.927083333333334e-06,
1051
+ "loss": 2.8571,
1052
+ "mean_token_accuracy": 0.4558314093061396,
1053
+ "num_tokens": 58763594.0,
1054
+ "step": 1160
1055
+ },
1056
+ {
1057
+ "epoch": 6.062418725617685,
1058
+ "grad_norm": 0.8588491082191467,
1059
+ "learning_rate": 7.822916666666668e-06,
1060
+ "loss": 2.8475,
1061
+ "mean_token_accuracy": 0.45537400618195534,
1062
+ "num_tokens": 59279111.0,
1063
+ "step": 1170
1064
+ },
1065
+ {
1066
+ "epoch": 6.11443433029909,
1067
+ "grad_norm": 0.32619649171829224,
1068
+ "learning_rate": 7.71875e-06,
1069
+ "loss": 2.8482,
1070
+ "mean_token_accuracy": 0.4566705487668514,
1071
+ "num_tokens": 59785889.0,
1072
+ "step": 1180
1073
+ },
1074
+ {
1075
+ "epoch": 6.166449934980494,
1076
+ "grad_norm": 1.3475053310394287,
1077
+ "learning_rate": 7.614583333333334e-06,
1078
+ "loss": 2.8583,
1079
+ "mean_token_accuracy": 0.4553842142224312,
1080
+ "num_tokens": 60293808.0,
1081
+ "step": 1190
1082
+ },
1083
+ {
1084
+ "epoch": 6.218465539661898,
1085
+ "grad_norm": 0.7148886322975159,
1086
+ "learning_rate": 7.510416666666668e-06,
1087
+ "loss": 2.8498,
1088
+ "mean_token_accuracy": 0.45562980622053145,
1089
+ "num_tokens": 60794974.0,
1090
+ "step": 1200
1091
+ },
1092
+ {
1093
+ "epoch": 6.270481144343303,
1094
+ "grad_norm": 1.6436187028884888,
1095
+ "learning_rate": 7.4062500000000005e-06,
1096
+ "loss": 2.8645,
1097
+ "mean_token_accuracy": 0.45401248037815095,
1098
+ "num_tokens": 61294331.0,
1099
+ "step": 1210
1100
+ },
1101
+ {
1102
+ "epoch": 6.322496749024707,
1103
+ "grad_norm": 0.30543598532676697,
1104
+ "learning_rate": 7.302083333333334e-06,
1105
+ "loss": 2.8751,
1106
+ "mean_token_accuracy": 0.4538280092179775,
1107
+ "num_tokens": 61804121.0,
1108
+ "step": 1220
1109
+ },
1110
+ {
1111
+ "epoch": 6.374512353706112,
1112
+ "grad_norm": 1.6045682430267334,
1113
+ "learning_rate": 7.197916666666668e-06,
1114
+ "loss": 2.8598,
1115
+ "mean_token_accuracy": 0.4542578525841236,
1116
+ "num_tokens": 62305866.0,
1117
+ "step": 1230
1118
+ },
1119
+ {
1120
+ "epoch": 6.426527958387516,
1121
+ "grad_norm": 0.2817009389400482,
1122
+ "learning_rate": 7.0937500000000005e-06,
1123
+ "loss": 2.8608,
1124
+ "mean_token_accuracy": 0.4549741417169571,
1125
+ "num_tokens": 62813145.0,
1126
+ "step": 1240
1127
+ },
1128
+ {
1129
+ "epoch": 6.4785435630689205,
1130
+ "grad_norm": 0.4654538631439209,
1131
+ "learning_rate": 6.989583333333334e-06,
1132
+ "loss": 2.8329,
1133
+ "mean_token_accuracy": 0.45891913771629333,
1134
+ "num_tokens": 63331319.0,
1135
+ "step": 1250
1136
+ },
1137
+ {
1138
+ "epoch": 6.530559167750325,
1139
+ "grad_norm": 1.1072373390197754,
1140
+ "learning_rate": 6.885416666666668e-06,
1141
+ "loss": 2.8505,
1142
+ "mean_token_accuracy": 0.45661833956837655,
1143
+ "num_tokens": 63846737.0,
1144
+ "step": 1260
1145
+ },
1146
+ {
1147
+ "epoch": 6.582574772431729,
1148
+ "grad_norm": 0.4172248840332031,
1149
+ "learning_rate": 6.7812500000000005e-06,
1150
+ "loss": 2.842,
1151
+ "mean_token_accuracy": 0.4565552145242691,
1152
+ "num_tokens": 64364625.0,
1153
+ "step": 1270
1154
+ },
1155
+ {
1156
+ "epoch": 6.634590377113134,
1157
+ "grad_norm": 0.4541364014148712,
1158
+ "learning_rate": 6.677083333333334e-06,
1159
+ "loss": 2.8368,
1160
+ "mean_token_accuracy": 0.45730073302984237,
1161
+ "num_tokens": 64878523.0,
1162
+ "step": 1280
1163
+ },
1164
+ {
1165
+ "epoch": 6.686605981794538,
1166
+ "grad_norm": 0.4843074679374695,
1167
+ "learning_rate": 6.572916666666668e-06,
1168
+ "loss": 2.8416,
1169
+ "mean_token_accuracy": 0.4561150461435318,
1170
+ "num_tokens": 65400334.0,
1171
+ "step": 1290
1172
+ },
1173
+ {
1174
+ "epoch": 6.738621586475943,
1175
+ "grad_norm": 0.3339775800704956,
1176
+ "learning_rate": 6.4687500000000005e-06,
1177
+ "loss": 2.8355,
1178
+ "mean_token_accuracy": 0.45750690847635267,
1179
+ "num_tokens": 65891422.0,
1180
+ "step": 1300
1181
+ },
1182
+ {
1183
+ "epoch": 6.790637191157347,
1184
+ "grad_norm": 0.3933147192001343,
1185
+ "learning_rate": 6.364583333333334e-06,
1186
+ "loss": 2.8366,
1187
+ "mean_token_accuracy": 0.4583583191037178,
1188
+ "num_tokens": 66398548.0,
1189
+ "step": 1310
1190
+ },
1191
+ {
1192
+ "epoch": 6.8426527958387515,
1193
+ "grad_norm": 0.5730970501899719,
1194
+ "learning_rate": 6.260416666666668e-06,
1195
+ "loss": 2.8281,
1196
+ "mean_token_accuracy": 0.459203827381134,
1197
+ "num_tokens": 66909808.0,
1198
+ "step": 1320
1199
+ },
1200
+ {
1201
+ "epoch": 6.894668400520156,
1202
+ "grad_norm": 0.34983688592910767,
1203
+ "learning_rate": 6.1562500000000006e-06,
1204
+ "loss": 2.8598,
1205
+ "mean_token_accuracy": 0.45596217811107637,
1206
+ "num_tokens": 67415434.0,
1207
+ "step": 1330
1208
+ },
1209
+ {
1210
+ "epoch": 6.94668400520156,
1211
+ "grad_norm": 0.5202476382255554,
1212
+ "learning_rate": 6.052083333333334e-06,
1213
+ "loss": 2.8361,
1214
+ "mean_token_accuracy": 0.4587762251496315,
1215
+ "num_tokens": 67926891.0,
1216
+ "step": 1340
1217
+ },
1218
+ {
1219
+ "epoch": 6.998699609882965,
1220
+ "grad_norm": 1.827202558517456,
1221
+ "learning_rate": 5.947916666666668e-06,
1222
+ "loss": 2.8464,
1223
+ "mean_token_accuracy": 0.4567068085074425,
1224
+ "num_tokens": 68435527.0,
1225
+ "step": 1350
1226
+ },
1227
+ {
1228
+ "epoch": 7.046814044213264,
1229
+ "grad_norm": 0.5215445160865784,
1230
+ "learning_rate": 5.843750000000001e-06,
1231
+ "loss": 2.8482,
1232
+ "mean_token_accuracy": 0.45589532884391576,
1233
+ "num_tokens": 68896515.0,
1234
+ "step": 1360
1235
+ },
1236
+ {
1237
+ "epoch": 7.098829648894668,
1238
+ "grad_norm": 0.8088448643684387,
1239
+ "learning_rate": 5.739583333333334e-06,
1240
+ "loss": 2.8337,
1241
+ "mean_token_accuracy": 0.4581175483763218,
1242
+ "num_tokens": 69402153.0,
1243
+ "step": 1370
1244
+ },
1245
+ {
1246
+ "epoch": 7.1508452535760725,
1247
+ "grad_norm": 0.8400851488113403,
1248
+ "learning_rate": 5.635416666666668e-06,
1249
+ "loss": 2.8347,
1250
+ "mean_token_accuracy": 0.45778577476739885,
1251
+ "num_tokens": 69918331.0,
1252
+ "step": 1380
1253
+ },
1254
+ {
1255
+ "epoch": 7.202860858257477,
1256
+ "grad_norm": 1.2368621826171875,
1257
+ "learning_rate": 5.531250000000001e-06,
1258
+ "loss": 2.851,
1259
+ "mean_token_accuracy": 0.457428352534771,
1260
+ "num_tokens": 70417431.0,
1261
+ "step": 1390
1262
+ },
1263
+ {
1264
+ "epoch": 7.254876462938881,
1265
+ "grad_norm": 0.3493196964263916,
1266
+ "learning_rate": 5.427083333333334e-06,
1267
+ "loss": 2.8379,
1268
+ "mean_token_accuracy": 0.4575755104422569,
1269
+ "num_tokens": 70927699.0,
1270
+ "step": 1400
1271
+ },
1272
+ {
1273
+ "epoch": 7.306892067620286,
1274
+ "grad_norm": 0.5054505467414856,
1275
+ "learning_rate": 5.322916666666668e-06,
1276
+ "loss": 2.8369,
1277
+ "mean_token_accuracy": 0.4592203006148338,
1278
+ "num_tokens": 71436412.0,
1279
+ "step": 1410
1280
+ },
1281
+ {
1282
+ "epoch": 7.35890767230169,
1283
+ "grad_norm": 0.8106542229652405,
1284
+ "learning_rate": 5.21875e-06,
1285
+ "loss": 2.8229,
1286
+ "mean_token_accuracy": 0.459158881008625,
1287
+ "num_tokens": 71952164.0,
1288
+ "step": 1420
1289
+ },
1290
+ {
1291
+ "epoch": 7.410923276983095,
1292
+ "grad_norm": 0.4714500606060028,
1293
+ "learning_rate": 5.114583333333334e-06,
1294
+ "loss": 2.8325,
1295
+ "mean_token_accuracy": 0.45578149855136874,
1296
+ "num_tokens": 72471078.0,
1297
+ "step": 1430
1298
+ },
1299
+ {
1300
+ "epoch": 7.462938881664499,
1301
+ "grad_norm": 0.4352336525917053,
1302
+ "learning_rate": 5.010416666666668e-06,
1303
+ "loss": 2.8232,
1304
+ "mean_token_accuracy": 0.4595529317855835,
1305
+ "num_tokens": 72980089.0,
1306
+ "step": 1440
1307
+ },
1308
+ {
1309
+ "epoch": 7.5149544863459035,
1310
+ "grad_norm": 0.38633960485458374,
1311
+ "learning_rate": 4.90625e-06,
1312
+ "loss": 2.7994,
1313
+ "mean_token_accuracy": 0.4613519378006458,
1314
+ "num_tokens": 73487248.0,
1315
+ "step": 1450
1316
+ },
1317
+ {
1318
+ "epoch": 7.566970091027308,
1319
+ "grad_norm": 0.4211936295032501,
1320
+ "learning_rate": 4.802083333333334e-06,
1321
+ "loss": 2.8164,
1322
+ "mean_token_accuracy": 0.4606427147984505,
1323
+ "num_tokens": 73991357.0,
1324
+ "step": 1460
1325
+ },
1326
+ {
1327
+ "epoch": 7.618985695708712,
1328
+ "grad_norm": 0.33615878224372864,
1329
+ "learning_rate": 4.697916666666667e-06,
1330
+ "loss": 2.8402,
1331
+ "mean_token_accuracy": 0.459219816327095,
1332
+ "num_tokens": 74502954.0,
1333
+ "step": 1470
1334
+ },
1335
+ {
1336
+ "epoch": 7.671001300390117,
1337
+ "grad_norm": 0.38883480429649353,
1338
+ "learning_rate": 4.59375e-06,
1339
+ "loss": 2.833,
1340
+ "mean_token_accuracy": 0.45845572650432587,
1341
+ "num_tokens": 75013948.0,
1342
+ "step": 1480
1343
+ },
1344
+ {
1345
+ "epoch": 7.723016905071521,
1346
+ "grad_norm": 0.546203076839447,
1347
+ "learning_rate": 4.4895833333333335e-06,
1348
+ "loss": 2.8084,
1349
+ "mean_token_accuracy": 0.46173520758748055,
1350
+ "num_tokens": 75525940.0,
1351
+ "step": 1490
1352
+ },
1353
+ {
1354
+ "epoch": 7.775032509752926,
1355
+ "grad_norm": 0.7886233329772949,
1356
+ "learning_rate": 4.385416666666667e-06,
1357
+ "loss": 2.8258,
1358
+ "mean_token_accuracy": 0.4596977703273296,
1359
+ "num_tokens": 76034878.0,
1360
+ "step": 1500
1361
+ }
1362
+ ],
1363
+ "logging_steps": 10,
1364
+ "max_steps": 1920,
1365
+ "num_input_tokens_seen": 0,
1366
+ "num_train_epochs": 10,
1367
+ "save_steps": 500,
1368
+ "stateful_callbacks": {
1369
+ "TrainerControl": {
1370
+ "args": {
1371
+ "should_epoch_stop": false,
1372
+ "should_evaluate": false,
1373
+ "should_log": false,
1374
+ "should_save": true,
1375
+ "should_training_stop": false
1376
+ },
1377
+ "attributes": {}
1378
+ }
1379
+ },
1380
+ "total_flos": 6.18349377958871e+16,
1381
+ "train_batch_size": 64,
1382
+ "trial_name": null,
1383
+ "trial_params": null
1384
+ }
checkpoint-1500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea990651e2aecabd57f309f323439c138096e78f77c92b4e54f46fe372d50d1d
3
+ size 5688
checkpoint-1920/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 512,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-1920/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-1920/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4086493d2e9f8bcf3f7568529059376e5b26a778b340b6136736c6e15206cde1
3
+ size 309900448
checkpoint-1920/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f2b253b8a375c117f692d402e3546be242b1130560a2968b6509bf749036212
3
+ size 619836730
checkpoint-1920/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b11a10749bfb1630d95eef94125f4590e8610c579d217d1f158e71ce518d72b
3
+ size 14244
checkpoint-1920/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b074489657b7fc573c9c21e352ead838a13e75dbe08202e4d2f38138b550463
3
+ size 1064
checkpoint-1920/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-1920/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-1920/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1920/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-1920/trainer_state.json ADDED
@@ -0,0 +1,1762 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 9.9518855656697,
6
+ "eval_steps": 500,
7
+ "global_step": 1920,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.05201560468140442,
14
+ "grad_norm": 3.158724784851074,
15
+ "learning_rate": 1.9906250000000003e-05,
16
+ "loss": 6.2302,
17
+ "mean_token_accuracy": 0.16946521885693072,
18
+ "num_tokens": 513852.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.10403120936280884,
23
+ "grad_norm": 10.176986694335938,
24
+ "learning_rate": 1.9802083333333335e-05,
25
+ "loss": 6.0745,
26
+ "mean_token_accuracy": 0.17949572652578355,
27
+ "num_tokens": 1032088.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.15604681404421328,
32
+ "grad_norm": 3.010871410369873,
33
+ "learning_rate": 1.9697916666666667e-05,
34
+ "loss": 5.8317,
35
+ "mean_token_accuracy": 0.1901865139603615,
36
+ "num_tokens": 1539389.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.20806241872561768,
41
+ "grad_norm": 3.298682451248169,
42
+ "learning_rate": 1.9593750000000002e-05,
43
+ "loss": 5.5987,
44
+ "mean_token_accuracy": 0.20738236755132675,
45
+ "num_tokens": 2057784.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.26007802340702213,
50
+ "grad_norm": 2.472743272781372,
51
+ "learning_rate": 1.9489583333333334e-05,
52
+ "loss": 5.4572,
53
+ "mean_token_accuracy": 0.2219402514398098,
54
+ "num_tokens": 2574094.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.31209362808842656,
59
+ "grad_norm": 2.3076207637786865,
60
+ "learning_rate": 1.938541666666667e-05,
61
+ "loss": 5.2817,
62
+ "mean_token_accuracy": 0.23026303350925445,
63
+ "num_tokens": 3076289.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.3641092327698309,
68
+ "grad_norm": 2.1481237411499023,
69
+ "learning_rate": 1.928125e-05,
70
+ "loss": 5.0596,
71
+ "mean_token_accuracy": 0.24168218560516835,
72
+ "num_tokens": 3586260.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.41612483745123535,
77
+ "grad_norm": 2.2509524822235107,
78
+ "learning_rate": 1.9177083333333336e-05,
79
+ "loss": 4.9117,
80
+ "mean_token_accuracy": 0.2510342214256525,
81
+ "num_tokens": 4094922.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.4681404421326398,
86
+ "grad_norm": 3.319535732269287,
87
+ "learning_rate": 1.907291666666667e-05,
88
+ "loss": 4.7624,
89
+ "mean_token_accuracy": 0.26219444051384927,
90
+ "num_tokens": 4606336.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.5201560468140443,
95
+ "grad_norm": 2.0256783962249756,
96
+ "learning_rate": 1.8968750000000004e-05,
97
+ "loss": 4.6138,
98
+ "mean_token_accuracy": 0.2705389905720949,
99
+ "num_tokens": 5110749.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.5721716514954487,
104
+ "grad_norm": 5.284979820251465,
105
+ "learning_rate": 1.8864583333333336e-05,
106
+ "loss": 4.5011,
107
+ "mean_token_accuracy": 0.27774968966841695,
108
+ "num_tokens": 5611872.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.6241872561768531,
113
+ "grad_norm": 15.897697448730469,
114
+ "learning_rate": 1.8760416666666668e-05,
115
+ "loss": 4.3486,
116
+ "mean_token_accuracy": 0.29455167055130005,
117
+ "num_tokens": 6123716.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.6762028608582574,
122
+ "grad_norm": 3.0891332626342773,
123
+ "learning_rate": 1.8656250000000003e-05,
124
+ "loss": 4.2372,
125
+ "mean_token_accuracy": 0.3053073361515999,
126
+ "num_tokens": 6630449.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.7282184655396619,
131
+ "grad_norm": 2.85463285446167,
132
+ "learning_rate": 1.8552083333333335e-05,
133
+ "loss": 4.161,
134
+ "mean_token_accuracy": 0.3164319314062595,
135
+ "num_tokens": 7145717.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.7802340702210663,
140
+ "grad_norm": 1.2611316442489624,
141
+ "learning_rate": 1.8447916666666667e-05,
142
+ "loss": 4.1002,
143
+ "mean_token_accuracy": 0.32258255779743195,
144
+ "num_tokens": 7655884.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.8322496749024707,
149
+ "grad_norm": 1.4083240032196045,
150
+ "learning_rate": 1.8343750000000002e-05,
151
+ "loss": 4.0138,
152
+ "mean_token_accuracy": 0.3296029016375542,
153
+ "num_tokens": 8159725.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.8842652795838751,
158
+ "grad_norm": 1.0340592861175537,
159
+ "learning_rate": 1.8239583333333334e-05,
160
+ "loss": 3.9163,
161
+ "mean_token_accuracy": 0.3385446682572365,
162
+ "num_tokens": 8668116.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.9362808842652796,
167
+ "grad_norm": 0.7984447479248047,
168
+ "learning_rate": 1.8135416666666666e-05,
169
+ "loss": 3.8628,
170
+ "mean_token_accuracy": 0.3455250822007656,
171
+ "num_tokens": 9182489.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.988296488946684,
176
+ "grad_norm": 1.244081735610962,
177
+ "learning_rate": 1.803125e-05,
178
+ "loss": 3.8025,
179
+ "mean_token_accuracy": 0.35223659574985505,
180
+ "num_tokens": 9676078.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 1.0364109232769831,
185
+ "grad_norm": 2.412661552429199,
186
+ "learning_rate": 1.7927083333333337e-05,
187
+ "loss": 3.7325,
188
+ "mean_token_accuracy": 0.35705321382831884,
189
+ "num_tokens": 10138114.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 1.0884265279583876,
194
+ "grad_norm": 1.7736923694610596,
195
+ "learning_rate": 1.782291666666667e-05,
196
+ "loss": 3.7052,
197
+ "mean_token_accuracy": 0.3590236373245716,
198
+ "num_tokens": 10656504.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 1.140442132639792,
203
+ "grad_norm": 0.7769147157669067,
204
+ "learning_rate": 1.7718750000000004e-05,
205
+ "loss": 3.6433,
206
+ "mean_token_accuracy": 0.36575346291065214,
207
+ "num_tokens": 11161374.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 1.1924577373211964,
212
+ "grad_norm": 0.9089329838752747,
213
+ "learning_rate": 1.7614583333333336e-05,
214
+ "loss": 3.653,
215
+ "mean_token_accuracy": 0.3651699565351009,
216
+ "num_tokens": 11670197.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 1.2444733420026008,
221
+ "grad_norm": 0.7330015897750854,
222
+ "learning_rate": 1.7510416666666668e-05,
223
+ "loss": 3.5843,
224
+ "mean_token_accuracy": 0.3718643538653851,
225
+ "num_tokens": 12174931.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 1.2964889466840053,
230
+ "grad_norm": 1.534440040588379,
231
+ "learning_rate": 1.7406250000000003e-05,
232
+ "loss": 3.5468,
233
+ "mean_token_accuracy": 0.3751932092010975,
234
+ "num_tokens": 12685023.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 1.3485045513654097,
239
+ "grad_norm": 1.9603444337844849,
240
+ "learning_rate": 1.7302083333333335e-05,
241
+ "loss": 3.4935,
242
+ "mean_token_accuracy": 0.3808542750775814,
243
+ "num_tokens": 13197338.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 1.4005201560468141,
248
+ "grad_norm": 0.5988335013389587,
249
+ "learning_rate": 1.7197916666666667e-05,
250
+ "loss": 3.4988,
251
+ "mean_token_accuracy": 0.3815221585333347,
252
+ "num_tokens": 13704900.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 1.4525357607282183,
257
+ "grad_norm": 2.7422847747802734,
258
+ "learning_rate": 1.7093750000000002e-05,
259
+ "loss": 3.4438,
260
+ "mean_token_accuracy": 0.3865826904773712,
261
+ "num_tokens": 14210835.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 1.5045513654096228,
266
+ "grad_norm": 1.9575722217559814,
267
+ "learning_rate": 1.6989583333333334e-05,
268
+ "loss": 3.417,
269
+ "mean_token_accuracy": 0.38725101202726364,
270
+ "num_tokens": 14726584.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 1.5565669700910272,
275
+ "grad_norm": 0.6955925226211548,
276
+ "learning_rate": 1.6885416666666666e-05,
277
+ "loss": 3.3895,
278
+ "mean_token_accuracy": 0.39159547463059424,
279
+ "num_tokens": 15238009.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 1.6085825747724316,
284
+ "grad_norm": 0.769548237323761,
285
+ "learning_rate": 1.678125e-05,
286
+ "loss": 3.3979,
287
+ "mean_token_accuracy": 0.39134200662374496,
288
+ "num_tokens": 15736398.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 1.660598179453836,
293
+ "grad_norm": 0.8247058391571045,
294
+ "learning_rate": 1.6677083333333333e-05,
295
+ "loss": 3.3884,
296
+ "mean_token_accuracy": 0.3924022614955902,
297
+ "num_tokens": 16235540.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 1.7126137841352405,
302
+ "grad_norm": 0.9121173024177551,
303
+ "learning_rate": 1.657291666666667e-05,
304
+ "loss": 3.3541,
305
+ "mean_token_accuracy": 0.3951391130685806,
306
+ "num_tokens": 16747733.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 1.7646293888166449,
311
+ "grad_norm": 1.3171864748001099,
312
+ "learning_rate": 1.6468750000000004e-05,
313
+ "loss": 3.3486,
314
+ "mean_token_accuracy": 0.39698211625218394,
315
+ "num_tokens": 17265239.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 1.8166449934980493,
320
+ "grad_norm": 4.172916412353516,
321
+ "learning_rate": 1.6364583333333336e-05,
322
+ "loss": 3.325,
323
+ "mean_token_accuracy": 0.39922175258398057,
324
+ "num_tokens": 17780508.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 1.8686605981794537,
329
+ "grad_norm": 0.4353245794773102,
330
+ "learning_rate": 1.6260416666666668e-05,
331
+ "loss": 3.2807,
332
+ "mean_token_accuracy": 0.4035400986671448,
333
+ "num_tokens": 18297602.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 1.9206762028608582,
338
+ "grad_norm": 3.6861395835876465,
339
+ "learning_rate": 1.6156250000000003e-05,
340
+ "loss": 3.2832,
341
+ "mean_token_accuracy": 0.4044442743062973,
342
+ "num_tokens": 18798625.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 1.9726918075422626,
347
+ "grad_norm": 1.1718441247940063,
348
+ "learning_rate": 1.6052083333333335e-05,
349
+ "loss": 3.2664,
350
+ "mean_token_accuracy": 0.40756076350808146,
351
+ "num_tokens": 19294344.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 2.020806241872562,
356
+ "grad_norm": 0.6570408940315247,
357
+ "learning_rate": 1.5947916666666667e-05,
358
+ "loss": 3.2056,
359
+ "mean_token_accuracy": 0.4098896440621969,
360
+ "num_tokens": 19761637.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 2.0728218465539663,
365
+ "grad_norm": 0.46285247802734375,
366
+ "learning_rate": 1.5843750000000002e-05,
367
+ "loss": 3.2396,
368
+ "mean_token_accuracy": 0.41089860200881956,
369
+ "num_tokens": 20265138.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 2.1248374512353707,
374
+ "grad_norm": 3.5917506217956543,
375
+ "learning_rate": 1.5739583333333334e-05,
376
+ "loss": 3.2205,
377
+ "mean_token_accuracy": 0.41081195920705793,
378
+ "num_tokens": 20777818.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 2.176853055916775,
383
+ "grad_norm": 0.6580970287322998,
384
+ "learning_rate": 1.5635416666666666e-05,
385
+ "loss": 3.225,
386
+ "mean_token_accuracy": 0.41244959980249407,
387
+ "num_tokens": 21282514.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 2.2288686605981796,
392
+ "grad_norm": 1.2673513889312744,
393
+ "learning_rate": 1.553125e-05,
394
+ "loss": 3.2143,
395
+ "mean_token_accuracy": 0.41195737645030023,
396
+ "num_tokens": 21796246.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 2.280884265279584,
401
+ "grad_norm": 1.9437355995178223,
402
+ "learning_rate": 1.5427083333333333e-05,
403
+ "loss": 3.2027,
404
+ "mean_token_accuracy": 0.41485874727368355,
405
+ "num_tokens": 22310915.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 2.3328998699609884,
410
+ "grad_norm": 0.4803381860256195,
411
+ "learning_rate": 1.532291666666667e-05,
412
+ "loss": 3.1743,
413
+ "mean_token_accuracy": 0.417393134534359,
414
+ "num_tokens": 22823271.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 2.384915474642393,
419
+ "grad_norm": 1.0227100849151611,
420
+ "learning_rate": 1.5218750000000002e-05,
421
+ "loss": 3.1751,
422
+ "mean_token_accuracy": 0.4173254922032356,
423
+ "num_tokens": 23326555.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 2.4369310793237973,
428
+ "grad_norm": 0.9217405915260315,
429
+ "learning_rate": 1.5114583333333334e-05,
430
+ "loss": 3.1336,
431
+ "mean_token_accuracy": 0.42112250030040743,
432
+ "num_tokens": 23832549.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 2.4889466840052017,
437
+ "grad_norm": 0.7527967095375061,
438
+ "learning_rate": 1.5010416666666668e-05,
439
+ "loss": 3.1469,
440
+ "mean_token_accuracy": 0.42071041762828826,
441
+ "num_tokens": 24338061.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 2.540962288686606,
446
+ "grad_norm": 0.370236873626709,
447
+ "learning_rate": 1.4906250000000001e-05,
448
+ "loss": 3.1461,
449
+ "mean_token_accuracy": 0.41948254629969595,
450
+ "num_tokens": 24854140.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 2.5929778933680105,
455
+ "grad_norm": 0.4003697335720062,
456
+ "learning_rate": 1.4802083333333335e-05,
457
+ "loss": 3.1162,
458
+ "mean_token_accuracy": 0.4244517408311367,
459
+ "num_tokens": 25367383.0,
460
+ "step": 500
461
+ },
462
+ {
463
+ "epoch": 2.644993498049415,
464
+ "grad_norm": 2.082859516143799,
465
+ "learning_rate": 1.4697916666666667e-05,
466
+ "loss": 3.1256,
467
+ "mean_token_accuracy": 0.42284995764493943,
468
+ "num_tokens": 25872016.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 2.6970091027308194,
473
+ "grad_norm": 2.390314817428589,
474
+ "learning_rate": 1.4593750000000002e-05,
475
+ "loss": 3.0869,
476
+ "mean_token_accuracy": 0.42750845700502393,
477
+ "num_tokens": 26374448.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 2.749024707412224,
482
+ "grad_norm": 0.46641266345977783,
483
+ "learning_rate": 1.4489583333333334e-05,
484
+ "loss": 3.0982,
485
+ "mean_token_accuracy": 0.42524692341685294,
486
+ "num_tokens": 26892842.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 2.8010403120936282,
491
+ "grad_norm": 2.4539034366607666,
492
+ "learning_rate": 1.4385416666666668e-05,
493
+ "loss": 3.0921,
494
+ "mean_token_accuracy": 0.425856501609087,
495
+ "num_tokens": 27399620.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 2.8530559167750327,
500
+ "grad_norm": 0.5952461957931519,
501
+ "learning_rate": 1.4281250000000002e-05,
502
+ "loss": 3.0888,
503
+ "mean_token_accuracy": 0.4283280909061432,
504
+ "num_tokens": 27901792.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 2.9050715214564367,
509
+ "grad_norm": 0.8322259783744812,
510
+ "learning_rate": 1.4177083333333335e-05,
511
+ "loss": 3.0828,
512
+ "mean_token_accuracy": 0.4277593605220318,
513
+ "num_tokens": 28412070.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 2.957087126137841,
518
+ "grad_norm": 0.499178409576416,
519
+ "learning_rate": 1.4072916666666667e-05,
520
+ "loss": 3.0776,
521
+ "mean_token_accuracy": 0.42896960899233816,
522
+ "num_tokens": 28925255.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 3.0052015604681404,
527
+ "grad_norm": 0.5208362340927124,
528
+ "learning_rate": 1.3968750000000002e-05,
529
+ "loss": 3.0475,
530
+ "mean_token_accuracy": 0.4297164734956381,
531
+ "num_tokens": 29380955.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 3.057217165149545,
536
+ "grad_norm": 1.1485410928726196,
537
+ "learning_rate": 1.3864583333333334e-05,
538
+ "loss": 3.0694,
539
+ "mean_token_accuracy": 0.42906737625598906,
540
+ "num_tokens": 29891475.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 3.109232769830949,
545
+ "grad_norm": 0.44713732600212097,
546
+ "learning_rate": 1.3760416666666668e-05,
547
+ "loss": 3.0237,
548
+ "mean_token_accuracy": 0.43474498465657235,
549
+ "num_tokens": 30397348.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 3.1612483745123536,
554
+ "grad_norm": 0.6080917716026306,
555
+ "learning_rate": 1.3656250000000002e-05,
556
+ "loss": 3.0392,
557
+ "mean_token_accuracy": 0.4334587998688221,
558
+ "num_tokens": 30901701.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 3.213263979193758,
563
+ "grad_norm": 1.6787530183792114,
564
+ "learning_rate": 1.3552083333333335e-05,
565
+ "loss": 3.0626,
566
+ "mean_token_accuracy": 0.4313603535294533,
567
+ "num_tokens": 31412858.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 3.2652795838751625,
572
+ "grad_norm": 3.046581745147705,
573
+ "learning_rate": 1.3447916666666667e-05,
574
+ "loss": 3.0271,
575
+ "mean_token_accuracy": 0.43369641304016116,
576
+ "num_tokens": 31911006.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 3.317295188556567,
581
+ "grad_norm": 1.0174599885940552,
582
+ "learning_rate": 1.3343750000000002e-05,
583
+ "loss": 3.0229,
584
+ "mean_token_accuracy": 0.43341642022132876,
585
+ "num_tokens": 32422873.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 3.3693107932379713,
590
+ "grad_norm": 0.4790953993797302,
591
+ "learning_rate": 1.3239583333333334e-05,
592
+ "loss": 3.0186,
593
+ "mean_token_accuracy": 0.43539193645119667,
594
+ "num_tokens": 32937766.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 3.4213263979193758,
599
+ "grad_norm": 4.097532272338867,
600
+ "learning_rate": 1.3135416666666666e-05,
601
+ "loss": 2.998,
602
+ "mean_token_accuracy": 0.4385197140276432,
603
+ "num_tokens": 33440445.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 3.47334200260078,
608
+ "grad_norm": 0.36602917313575745,
609
+ "learning_rate": 1.3031250000000002e-05,
610
+ "loss": 3.0098,
611
+ "mean_token_accuracy": 0.4368562549352646,
612
+ "num_tokens": 33939105.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 3.5253576072821846,
617
+ "grad_norm": 1.6092300415039062,
618
+ "learning_rate": 1.2927083333333335e-05,
619
+ "loss": 3.0193,
620
+ "mean_token_accuracy": 0.43651641234755517,
621
+ "num_tokens": 34450716.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 3.577373211963589,
626
+ "grad_norm": 3.6081929206848145,
627
+ "learning_rate": 1.2822916666666667e-05,
628
+ "loss": 2.9884,
629
+ "mean_token_accuracy": 0.4398868255317211,
630
+ "num_tokens": 34958114.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 3.6293888166449935,
635
+ "grad_norm": 2.0380210876464844,
636
+ "learning_rate": 1.2718750000000003e-05,
637
+ "loss": 3.0131,
638
+ "mean_token_accuracy": 0.43571230471134187,
639
+ "num_tokens": 35468951.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 3.681404421326398,
644
+ "grad_norm": 0.5758602023124695,
645
+ "learning_rate": 1.2614583333333334e-05,
646
+ "loss": 2.9752,
647
+ "mean_token_accuracy": 0.4403461940586567,
648
+ "num_tokens": 35977001.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 3.7334200260078023,
653
+ "grad_norm": 0.914013147354126,
654
+ "learning_rate": 1.2510416666666666e-05,
655
+ "loss": 3.0002,
656
+ "mean_token_accuracy": 0.4371410049498081,
657
+ "num_tokens": 36491206.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 3.7854356306892067,
662
+ "grad_norm": 0.8788081407546997,
663
+ "learning_rate": 1.2406250000000002e-05,
664
+ "loss": 3.0013,
665
+ "mean_token_accuracy": 0.43753816187381744,
666
+ "num_tokens": 36996454.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 3.837451235370611,
671
+ "grad_norm": 0.35317686200141907,
672
+ "learning_rate": 1.2302083333333335e-05,
673
+ "loss": 2.9874,
674
+ "mean_token_accuracy": 0.44055561572313306,
675
+ "num_tokens": 37493079.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 3.8894668400520156,
680
+ "grad_norm": 0.40355926752090454,
681
+ "learning_rate": 1.2197916666666667e-05,
682
+ "loss": 2.9743,
683
+ "mean_token_accuracy": 0.4413572125136852,
684
+ "num_tokens": 38004972.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 3.94148244473342,
689
+ "grad_norm": 0.3518731892108917,
690
+ "learning_rate": 1.2093750000000003e-05,
691
+ "loss": 2.9602,
692
+ "mean_token_accuracy": 0.4420863904058933,
693
+ "num_tokens": 38534943.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 3.9934980494148244,
698
+ "grad_norm": 0.7066172361373901,
699
+ "learning_rate": 1.1989583333333335e-05,
700
+ "loss": 2.9588,
701
+ "mean_token_accuracy": 0.44266087040305135,
702
+ "num_tokens": 39052563.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 4.041612483745124,
707
+ "grad_norm": 2.8007071018218994,
708
+ "learning_rate": 1.1885416666666666e-05,
709
+ "loss": 2.9561,
710
+ "mean_token_accuracy": 0.44085964157774643,
711
+ "num_tokens": 39521519.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 4.093628088426528,
716
+ "grad_norm": 0.5129738450050354,
717
+ "learning_rate": 1.1781250000000002e-05,
718
+ "loss": 2.942,
719
+ "mean_token_accuracy": 0.4447008825838566,
720
+ "num_tokens": 40026574.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 4.145643693107933,
725
+ "grad_norm": 0.750156044960022,
726
+ "learning_rate": 1.1677083333333334e-05,
727
+ "loss": 2.9617,
728
+ "mean_token_accuracy": 0.4419982649385929,
729
+ "num_tokens": 40546106.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 4.197659297789337,
734
+ "grad_norm": 0.567778468132019,
735
+ "learning_rate": 1.1572916666666667e-05,
736
+ "loss": 2.9514,
737
+ "mean_token_accuracy": 0.44232149720191954,
738
+ "num_tokens": 41064511.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 4.249674902470741,
743
+ "grad_norm": 0.3298749029636383,
744
+ "learning_rate": 1.1468750000000003e-05,
745
+ "loss": 2.9526,
746
+ "mean_token_accuracy": 0.4424647718667984,
747
+ "num_tokens": 41568948.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 4.301690507152146,
752
+ "grad_norm": 0.36816585063934326,
753
+ "learning_rate": 1.1364583333333335e-05,
754
+ "loss": 2.9254,
755
+ "mean_token_accuracy": 0.44615685641765596,
756
+ "num_tokens": 42075864.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 4.35370611183355,
761
+ "grad_norm": 2.7220842838287354,
762
+ "learning_rate": 1.1260416666666666e-05,
763
+ "loss": 2.9571,
764
+ "mean_token_accuracy": 0.44221931844949725,
765
+ "num_tokens": 42594515.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 4.405721716514955,
770
+ "grad_norm": 0.37389975786209106,
771
+ "learning_rate": 1.1156250000000002e-05,
772
+ "loss": 2.9441,
773
+ "mean_token_accuracy": 0.442495983093977,
774
+ "num_tokens": 43111918.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 4.457737321196359,
779
+ "grad_norm": 0.40067073702812195,
780
+ "learning_rate": 1.1052083333333334e-05,
781
+ "loss": 2.9252,
782
+ "mean_token_accuracy": 0.4479030266404152,
783
+ "num_tokens": 43613796.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 4.5097529258777636,
788
+ "grad_norm": 0.9721180200576782,
789
+ "learning_rate": 1.0947916666666667e-05,
790
+ "loss": 2.9317,
791
+ "mean_token_accuracy": 0.4480686657130718,
792
+ "num_tokens": 44116168.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 4.561768530559168,
797
+ "grad_norm": 0.6541546583175659,
798
+ "learning_rate": 1.0843750000000001e-05,
799
+ "loss": 2.9164,
800
+ "mean_token_accuracy": 0.4477278359234333,
801
+ "num_tokens": 44619073.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 4.613784135240572,
806
+ "grad_norm": 0.9818729758262634,
807
+ "learning_rate": 1.0739583333333335e-05,
808
+ "loss": 2.9236,
809
+ "mean_token_accuracy": 0.44781008884310725,
810
+ "num_tokens": 45122295.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 4.665799739921977,
815
+ "grad_norm": 0.5139995813369751,
816
+ "learning_rate": 1.0635416666666667e-05,
817
+ "loss": 2.9245,
818
+ "mean_token_accuracy": 0.4462748847901821,
819
+ "num_tokens": 45640893.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 4.717815344603381,
824
+ "grad_norm": 0.5460729002952576,
825
+ "learning_rate": 1.0531250000000002e-05,
826
+ "loss": 2.8997,
827
+ "mean_token_accuracy": 0.45041476264595987,
828
+ "num_tokens": 46149628.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 4.769830949284786,
833
+ "grad_norm": 0.4323829412460327,
834
+ "learning_rate": 1.0427083333333334e-05,
835
+ "loss": 2.9241,
836
+ "mean_token_accuracy": 0.4480301558971405,
837
+ "num_tokens": 46658455.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 4.82184655396619,
842
+ "grad_norm": 0.368254154920578,
843
+ "learning_rate": 1.0322916666666667e-05,
844
+ "loss": 2.9198,
845
+ "mean_token_accuracy": 0.4473501190543175,
846
+ "num_tokens": 47152789.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 4.8738621586475945,
851
+ "grad_norm": 1.6135244369506836,
852
+ "learning_rate": 1.0218750000000001e-05,
853
+ "loss": 2.9327,
854
+ "mean_token_accuracy": 0.44697768464684484,
855
+ "num_tokens": 47664450.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 4.925877763328999,
860
+ "grad_norm": 0.5351427793502808,
861
+ "learning_rate": 1.0114583333333335e-05,
862
+ "loss": 2.9031,
863
+ "mean_token_accuracy": 0.4496559798717499,
864
+ "num_tokens": 48166797.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 4.977893368010403,
869
+ "grad_norm": 0.7081937193870544,
870
+ "learning_rate": 1.0010416666666667e-05,
871
+ "loss": 2.8935,
872
+ "mean_token_accuracy": 0.45092770233750346,
873
+ "num_tokens": 48674756.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 5.026007802340702,
878
+ "grad_norm": 35.31280517578125,
879
+ "learning_rate": 9.90625e-06,
880
+ "loss": 2.9106,
881
+ "mean_token_accuracy": 0.4502824319375528,
882
+ "num_tokens": 49132955.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 5.078023407022107,
887
+ "grad_norm": 0.49765506386756897,
888
+ "learning_rate": 9.802083333333334e-06,
889
+ "loss": 2.9015,
890
+ "mean_token_accuracy": 0.44894649609923365,
891
+ "num_tokens": 49642865.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 5.130039011703511,
896
+ "grad_norm": 0.4912125766277313,
897
+ "learning_rate": 9.697916666666667e-06,
898
+ "loss": 2.8704,
899
+ "mean_token_accuracy": 0.4528684295713902,
900
+ "num_tokens": 50145710.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 5.1820546163849155,
905
+ "grad_norm": 0.8951883912086487,
906
+ "learning_rate": 9.593750000000001e-06,
907
+ "loss": 2.8903,
908
+ "mean_token_accuracy": 0.45256839543581007,
909
+ "num_tokens": 50646447.0,
910
+ "step": 1000
911
+ },
912
+ {
913
+ "epoch": 5.23407022106632,
914
+ "grad_norm": 0.3062414824962616,
915
+ "learning_rate": 9.489583333333335e-06,
916
+ "loss": 2.9118,
917
+ "mean_token_accuracy": 0.44818254336714747,
918
+ "num_tokens": 51167514.0,
919
+ "step": 1010
920
+ },
921
+ {
922
+ "epoch": 5.286085825747724,
923
+ "grad_norm": 0.6343457102775574,
924
+ "learning_rate": 9.385416666666668e-06,
925
+ "loss": 2.8994,
926
+ "mean_token_accuracy": 0.44962315559387206,
927
+ "num_tokens": 51684111.0,
928
+ "step": 1020
929
+ },
930
+ {
931
+ "epoch": 5.338101430429129,
932
+ "grad_norm": 0.9417049288749695,
933
+ "learning_rate": 9.28125e-06,
934
+ "loss": 2.8761,
935
+ "mean_token_accuracy": 0.4527172662317753,
936
+ "num_tokens": 52189036.0,
937
+ "step": 1030
938
+ },
939
+ {
940
+ "epoch": 5.390117035110533,
941
+ "grad_norm": 0.6008946299552917,
942
+ "learning_rate": 9.177083333333334e-06,
943
+ "loss": 2.9054,
944
+ "mean_token_accuracy": 0.44979408457875253,
945
+ "num_tokens": 52700140.0,
946
+ "step": 1040
947
+ },
948
+ {
949
+ "epoch": 5.442132639791938,
950
+ "grad_norm": 1.5932005643844604,
951
+ "learning_rate": 9.072916666666668e-06,
952
+ "loss": 2.8849,
953
+ "mean_token_accuracy": 0.4512541800737381,
954
+ "num_tokens": 53209299.0,
955
+ "step": 1050
956
+ },
957
+ {
958
+ "epoch": 5.494148244473342,
959
+ "grad_norm": 0.4708751440048218,
960
+ "learning_rate": 8.968750000000001e-06,
961
+ "loss": 2.8747,
962
+ "mean_token_accuracy": 0.45460196733474734,
963
+ "num_tokens": 53708278.0,
964
+ "step": 1060
965
+ },
966
+ {
967
+ "epoch": 5.5461638491547465,
968
+ "grad_norm": 0.3820144832134247,
969
+ "learning_rate": 8.864583333333335e-06,
970
+ "loss": 2.8583,
971
+ "mean_token_accuracy": 0.4543687038123608,
972
+ "num_tokens": 54222052.0,
973
+ "step": 1070
974
+ },
975
+ {
976
+ "epoch": 5.598179453836151,
977
+ "grad_norm": 0.4965152442455292,
978
+ "learning_rate": 8.760416666666668e-06,
979
+ "loss": 2.9015,
980
+ "mean_token_accuracy": 0.45009259432554244,
981
+ "num_tokens": 54736716.0,
982
+ "step": 1080
983
+ },
984
+ {
985
+ "epoch": 5.650195058517555,
986
+ "grad_norm": 1.13825261592865,
987
+ "learning_rate": 8.65625e-06,
988
+ "loss": 2.8744,
989
+ "mean_token_accuracy": 0.4531993091106415,
990
+ "num_tokens": 55249854.0,
991
+ "step": 1090
992
+ },
993
+ {
994
+ "epoch": 5.70221066319896,
995
+ "grad_norm": 0.4815926253795624,
996
+ "learning_rate": 8.552083333333334e-06,
997
+ "loss": 2.85,
998
+ "mean_token_accuracy": 0.4550083190202713,
999
+ "num_tokens": 55756222.0,
1000
+ "step": 1100
1001
+ },
1002
+ {
1003
+ "epoch": 5.754226267880364,
1004
+ "grad_norm": 0.4123997986316681,
1005
+ "learning_rate": 8.447916666666668e-06,
1006
+ "loss": 2.8737,
1007
+ "mean_token_accuracy": 0.4540506564080715,
1008
+ "num_tokens": 56272172.0,
1009
+ "step": 1110
1010
+ },
1011
+ {
1012
+ "epoch": 5.806241872561769,
1013
+ "grad_norm": 0.3490988314151764,
1014
+ "learning_rate": 8.343750000000001e-06,
1015
+ "loss": 2.8835,
1016
+ "mean_token_accuracy": 0.45104856938123705,
1017
+ "num_tokens": 56780204.0,
1018
+ "step": 1120
1019
+ },
1020
+ {
1021
+ "epoch": 5.858257477243173,
1022
+ "grad_norm": 0.3921835124492645,
1023
+ "learning_rate": 8.239583333333335e-06,
1024
+ "loss": 2.8739,
1025
+ "mean_token_accuracy": 0.4529320389032364,
1026
+ "num_tokens": 57294606.0,
1027
+ "step": 1130
1028
+ },
1029
+ {
1030
+ "epoch": 5.9102730819245775,
1031
+ "grad_norm": 0.8325438499450684,
1032
+ "learning_rate": 8.135416666666668e-06,
1033
+ "loss": 2.8744,
1034
+ "mean_token_accuracy": 0.4542323037981987,
1035
+ "num_tokens": 57794088.0,
1036
+ "step": 1140
1037
+ },
1038
+ {
1039
+ "epoch": 5.962288686605982,
1040
+ "grad_norm": 0.3708273768424988,
1041
+ "learning_rate": 8.03125e-06,
1042
+ "loss": 2.8624,
1043
+ "mean_token_accuracy": 0.4541611663997173,
1044
+ "num_tokens": 58303085.0,
1045
+ "step": 1150
1046
+ },
1047
+ {
1048
+ "epoch": 6.010403120936281,
1049
+ "grad_norm": 0.45103487372398376,
1050
+ "learning_rate": 7.927083333333334e-06,
1051
+ "loss": 2.8571,
1052
+ "mean_token_accuracy": 0.4558314093061396,
1053
+ "num_tokens": 58763594.0,
1054
+ "step": 1160
1055
+ },
1056
+ {
1057
+ "epoch": 6.062418725617685,
1058
+ "grad_norm": 0.8588491082191467,
1059
+ "learning_rate": 7.822916666666668e-06,
1060
+ "loss": 2.8475,
1061
+ "mean_token_accuracy": 0.45537400618195534,
1062
+ "num_tokens": 59279111.0,
1063
+ "step": 1170
1064
+ },
1065
+ {
1066
+ "epoch": 6.11443433029909,
1067
+ "grad_norm": 0.32619649171829224,
1068
+ "learning_rate": 7.71875e-06,
1069
+ "loss": 2.8482,
1070
+ "mean_token_accuracy": 0.4566705487668514,
1071
+ "num_tokens": 59785889.0,
1072
+ "step": 1180
1073
+ },
1074
+ {
1075
+ "epoch": 6.166449934980494,
1076
+ "grad_norm": 1.3475053310394287,
1077
+ "learning_rate": 7.614583333333334e-06,
1078
+ "loss": 2.8583,
1079
+ "mean_token_accuracy": 0.4553842142224312,
1080
+ "num_tokens": 60293808.0,
1081
+ "step": 1190
1082
+ },
1083
+ {
1084
+ "epoch": 6.218465539661898,
1085
+ "grad_norm": 0.7148886322975159,
1086
+ "learning_rate": 7.510416666666668e-06,
1087
+ "loss": 2.8498,
1088
+ "mean_token_accuracy": 0.45562980622053145,
1089
+ "num_tokens": 60794974.0,
1090
+ "step": 1200
1091
+ },
1092
+ {
1093
+ "epoch": 6.270481144343303,
1094
+ "grad_norm": 1.6436187028884888,
1095
+ "learning_rate": 7.4062500000000005e-06,
1096
+ "loss": 2.8645,
1097
+ "mean_token_accuracy": 0.45401248037815095,
1098
+ "num_tokens": 61294331.0,
1099
+ "step": 1210
1100
+ },
1101
+ {
1102
+ "epoch": 6.322496749024707,
1103
+ "grad_norm": 0.30543598532676697,
1104
+ "learning_rate": 7.302083333333334e-06,
1105
+ "loss": 2.8751,
1106
+ "mean_token_accuracy": 0.4538280092179775,
1107
+ "num_tokens": 61804121.0,
1108
+ "step": 1220
1109
+ },
1110
+ {
1111
+ "epoch": 6.374512353706112,
1112
+ "grad_norm": 1.6045682430267334,
1113
+ "learning_rate": 7.197916666666668e-06,
1114
+ "loss": 2.8598,
1115
+ "mean_token_accuracy": 0.4542578525841236,
1116
+ "num_tokens": 62305866.0,
1117
+ "step": 1230
1118
+ },
1119
+ {
1120
+ "epoch": 6.426527958387516,
1121
+ "grad_norm": 0.2817009389400482,
1122
+ "learning_rate": 7.0937500000000005e-06,
1123
+ "loss": 2.8608,
1124
+ "mean_token_accuracy": 0.4549741417169571,
1125
+ "num_tokens": 62813145.0,
1126
+ "step": 1240
1127
+ },
1128
+ {
1129
+ "epoch": 6.4785435630689205,
1130
+ "grad_norm": 0.4654538631439209,
1131
+ "learning_rate": 6.989583333333334e-06,
1132
+ "loss": 2.8329,
1133
+ "mean_token_accuracy": 0.45891913771629333,
1134
+ "num_tokens": 63331319.0,
1135
+ "step": 1250
1136
+ },
1137
+ {
1138
+ "epoch": 6.530559167750325,
1139
+ "grad_norm": 1.1072373390197754,
1140
+ "learning_rate": 6.885416666666668e-06,
1141
+ "loss": 2.8505,
1142
+ "mean_token_accuracy": 0.45661833956837655,
1143
+ "num_tokens": 63846737.0,
1144
+ "step": 1260
1145
+ },
1146
+ {
1147
+ "epoch": 6.582574772431729,
1148
+ "grad_norm": 0.4172248840332031,
1149
+ "learning_rate": 6.7812500000000005e-06,
1150
+ "loss": 2.842,
1151
+ "mean_token_accuracy": 0.4565552145242691,
1152
+ "num_tokens": 64364625.0,
1153
+ "step": 1270
1154
+ },
1155
+ {
1156
+ "epoch": 6.634590377113134,
1157
+ "grad_norm": 0.4541364014148712,
1158
+ "learning_rate": 6.677083333333334e-06,
1159
+ "loss": 2.8368,
1160
+ "mean_token_accuracy": 0.45730073302984237,
1161
+ "num_tokens": 64878523.0,
1162
+ "step": 1280
1163
+ },
1164
+ {
1165
+ "epoch": 6.686605981794538,
1166
+ "grad_norm": 0.4843074679374695,
1167
+ "learning_rate": 6.572916666666668e-06,
1168
+ "loss": 2.8416,
1169
+ "mean_token_accuracy": 0.4561150461435318,
1170
+ "num_tokens": 65400334.0,
1171
+ "step": 1290
1172
+ },
1173
+ {
1174
+ "epoch": 6.738621586475943,
1175
+ "grad_norm": 0.3339775800704956,
1176
+ "learning_rate": 6.4687500000000005e-06,
1177
+ "loss": 2.8355,
1178
+ "mean_token_accuracy": 0.45750690847635267,
1179
+ "num_tokens": 65891422.0,
1180
+ "step": 1300
1181
+ },
1182
+ {
1183
+ "epoch": 6.790637191157347,
1184
+ "grad_norm": 0.3933147192001343,
1185
+ "learning_rate": 6.364583333333334e-06,
1186
+ "loss": 2.8366,
1187
+ "mean_token_accuracy": 0.4583583191037178,
1188
+ "num_tokens": 66398548.0,
1189
+ "step": 1310
1190
+ },
1191
+ {
1192
+ "epoch": 6.8426527958387515,
1193
+ "grad_norm": 0.5730970501899719,
1194
+ "learning_rate": 6.260416666666668e-06,
1195
+ "loss": 2.8281,
1196
+ "mean_token_accuracy": 0.459203827381134,
1197
+ "num_tokens": 66909808.0,
1198
+ "step": 1320
1199
+ },
1200
+ {
1201
+ "epoch": 6.894668400520156,
1202
+ "grad_norm": 0.34983688592910767,
1203
+ "learning_rate": 6.1562500000000006e-06,
1204
+ "loss": 2.8598,
1205
+ "mean_token_accuracy": 0.45596217811107637,
1206
+ "num_tokens": 67415434.0,
1207
+ "step": 1330
1208
+ },
1209
+ {
1210
+ "epoch": 6.94668400520156,
1211
+ "grad_norm": 0.5202476382255554,
1212
+ "learning_rate": 6.052083333333334e-06,
1213
+ "loss": 2.8361,
1214
+ "mean_token_accuracy": 0.4587762251496315,
1215
+ "num_tokens": 67926891.0,
1216
+ "step": 1340
1217
+ },
1218
+ {
1219
+ "epoch": 6.998699609882965,
1220
+ "grad_norm": 1.827202558517456,
1221
+ "learning_rate": 5.947916666666668e-06,
1222
+ "loss": 2.8464,
1223
+ "mean_token_accuracy": 0.4567068085074425,
1224
+ "num_tokens": 68435527.0,
1225
+ "step": 1350
1226
+ },
1227
+ {
1228
+ "epoch": 7.046814044213264,
1229
+ "grad_norm": 0.5215445160865784,
1230
+ "learning_rate": 5.843750000000001e-06,
1231
+ "loss": 2.8482,
1232
+ "mean_token_accuracy": 0.45589532884391576,
1233
+ "num_tokens": 68896515.0,
1234
+ "step": 1360
1235
+ },
1236
+ {
1237
+ "epoch": 7.098829648894668,
1238
+ "grad_norm": 0.8088448643684387,
1239
+ "learning_rate": 5.739583333333334e-06,
1240
+ "loss": 2.8337,
1241
+ "mean_token_accuracy": 0.4581175483763218,
1242
+ "num_tokens": 69402153.0,
1243
+ "step": 1370
1244
+ },
1245
+ {
1246
+ "epoch": 7.1508452535760725,
1247
+ "grad_norm": 0.8400851488113403,
1248
+ "learning_rate": 5.635416666666668e-06,
1249
+ "loss": 2.8347,
1250
+ "mean_token_accuracy": 0.45778577476739885,
1251
+ "num_tokens": 69918331.0,
1252
+ "step": 1380
1253
+ },
1254
+ {
1255
+ "epoch": 7.202860858257477,
1256
+ "grad_norm": 1.2368621826171875,
1257
+ "learning_rate": 5.531250000000001e-06,
1258
+ "loss": 2.851,
1259
+ "mean_token_accuracy": 0.457428352534771,
1260
+ "num_tokens": 70417431.0,
1261
+ "step": 1390
1262
+ },
1263
+ {
1264
+ "epoch": 7.254876462938881,
1265
+ "grad_norm": 0.3493196964263916,
1266
+ "learning_rate": 5.427083333333334e-06,
1267
+ "loss": 2.8379,
1268
+ "mean_token_accuracy": 0.4575755104422569,
1269
+ "num_tokens": 70927699.0,
1270
+ "step": 1400
1271
+ },
1272
+ {
1273
+ "epoch": 7.306892067620286,
1274
+ "grad_norm": 0.5054505467414856,
1275
+ "learning_rate": 5.322916666666668e-06,
1276
+ "loss": 2.8369,
1277
+ "mean_token_accuracy": 0.4592203006148338,
1278
+ "num_tokens": 71436412.0,
1279
+ "step": 1410
1280
+ },
1281
+ {
1282
+ "epoch": 7.35890767230169,
1283
+ "grad_norm": 0.8106542229652405,
1284
+ "learning_rate": 5.21875e-06,
1285
+ "loss": 2.8229,
1286
+ "mean_token_accuracy": 0.459158881008625,
1287
+ "num_tokens": 71952164.0,
1288
+ "step": 1420
1289
+ },
1290
+ {
1291
+ "epoch": 7.410923276983095,
1292
+ "grad_norm": 0.4714500606060028,
1293
+ "learning_rate": 5.114583333333334e-06,
1294
+ "loss": 2.8325,
1295
+ "mean_token_accuracy": 0.45578149855136874,
1296
+ "num_tokens": 72471078.0,
1297
+ "step": 1430
1298
+ },
1299
+ {
1300
+ "epoch": 7.462938881664499,
1301
+ "grad_norm": 0.4352336525917053,
1302
+ "learning_rate": 5.010416666666668e-06,
1303
+ "loss": 2.8232,
1304
+ "mean_token_accuracy": 0.4595529317855835,
1305
+ "num_tokens": 72980089.0,
1306
+ "step": 1440
1307
+ },
1308
+ {
1309
+ "epoch": 7.5149544863459035,
1310
+ "grad_norm": 0.38633960485458374,
1311
+ "learning_rate": 4.90625e-06,
1312
+ "loss": 2.7994,
1313
+ "mean_token_accuracy": 0.4613519378006458,
1314
+ "num_tokens": 73487248.0,
1315
+ "step": 1450
1316
+ },
1317
+ {
1318
+ "epoch": 7.566970091027308,
1319
+ "grad_norm": 0.4211936295032501,
1320
+ "learning_rate": 4.802083333333334e-06,
1321
+ "loss": 2.8164,
1322
+ "mean_token_accuracy": 0.4606427147984505,
1323
+ "num_tokens": 73991357.0,
1324
+ "step": 1460
1325
+ },
1326
+ {
1327
+ "epoch": 7.618985695708712,
1328
+ "grad_norm": 0.33615878224372864,
1329
+ "learning_rate": 4.697916666666667e-06,
1330
+ "loss": 2.8402,
1331
+ "mean_token_accuracy": 0.459219816327095,
1332
+ "num_tokens": 74502954.0,
1333
+ "step": 1470
1334
+ },
1335
+ {
1336
+ "epoch": 7.671001300390117,
1337
+ "grad_norm": 0.38883480429649353,
1338
+ "learning_rate": 4.59375e-06,
1339
+ "loss": 2.833,
1340
+ "mean_token_accuracy": 0.45845572650432587,
1341
+ "num_tokens": 75013948.0,
1342
+ "step": 1480
1343
+ },
1344
+ {
1345
+ "epoch": 7.723016905071521,
1346
+ "grad_norm": 0.546203076839447,
1347
+ "learning_rate": 4.4895833333333335e-06,
1348
+ "loss": 2.8084,
1349
+ "mean_token_accuracy": 0.46173520758748055,
1350
+ "num_tokens": 75525940.0,
1351
+ "step": 1490
1352
+ },
1353
+ {
1354
+ "epoch": 7.775032509752926,
1355
+ "grad_norm": 0.7886233329772949,
1356
+ "learning_rate": 4.385416666666667e-06,
1357
+ "loss": 2.8258,
1358
+ "mean_token_accuracy": 0.4596977703273296,
1359
+ "num_tokens": 76034878.0,
1360
+ "step": 1500
1361
+ },
1362
+ {
1363
+ "epoch": 7.82704811443433,
1364
+ "grad_norm": 0.582435131072998,
1365
+ "learning_rate": 4.28125e-06,
1366
+ "loss": 2.849,
1367
+ "mean_token_accuracy": 0.45709013864398,
1368
+ "num_tokens": 76536548.0,
1369
+ "step": 1510
1370
+ },
1371
+ {
1372
+ "epoch": 7.879063719115734,
1373
+ "grad_norm": 0.703502357006073,
1374
+ "learning_rate": 4.1770833333333335e-06,
1375
+ "loss": 2.8172,
1376
+ "mean_token_accuracy": 0.4604157403111458,
1377
+ "num_tokens": 77050078.0,
1378
+ "step": 1520
1379
+ },
1380
+ {
1381
+ "epoch": 7.931079323797139,
1382
+ "grad_norm": 0.3695385456085205,
1383
+ "learning_rate": 4.072916666666667e-06,
1384
+ "loss": 2.8035,
1385
+ "mean_token_accuracy": 0.46285670101642606,
1386
+ "num_tokens": 77547926.0,
1387
+ "step": 1530
1388
+ },
1389
+ {
1390
+ "epoch": 7.983094928478543,
1391
+ "grad_norm": 0.6850471496582031,
1392
+ "learning_rate": 3.96875e-06,
1393
+ "loss": 2.7932,
1394
+ "mean_token_accuracy": 0.462927870452404,
1395
+ "num_tokens": 78055207.0,
1396
+ "step": 1540
1397
+ },
1398
+ {
1399
+ "epoch": 8.031209362808843,
1400
+ "grad_norm": 0.37858447432518005,
1401
+ "learning_rate": 3.8645833333333335e-06,
1402
+ "loss": 2.8145,
1403
+ "mean_token_accuracy": 0.4606696335045067,
1404
+ "num_tokens": 78525994.0,
1405
+ "step": 1550
1406
+ },
1407
+ {
1408
+ "epoch": 8.083224967490247,
1409
+ "grad_norm": 0.45084574818611145,
1410
+ "learning_rate": 3.7604166666666667e-06,
1411
+ "loss": 2.8039,
1412
+ "mean_token_accuracy": 0.46232729181647303,
1413
+ "num_tokens": 79038853.0,
1414
+ "step": 1560
1415
+ },
1416
+ {
1417
+ "epoch": 8.135240572171652,
1418
+ "grad_norm": 0.38596588373184204,
1419
+ "learning_rate": 3.65625e-06,
1420
+ "loss": 2.8225,
1421
+ "mean_token_accuracy": 0.4588643185794353,
1422
+ "num_tokens": 79554382.0,
1423
+ "step": 1570
1424
+ },
1425
+ {
1426
+ "epoch": 8.187256176853056,
1427
+ "grad_norm": 0.6875448226928711,
1428
+ "learning_rate": 3.552083333333334e-06,
1429
+ "loss": 2.8349,
1430
+ "mean_token_accuracy": 0.45943210944533347,
1431
+ "num_tokens": 80062934.0,
1432
+ "step": 1580
1433
+ },
1434
+ {
1435
+ "epoch": 8.23927178153446,
1436
+ "grad_norm": 0.44944027066230774,
1437
+ "learning_rate": 3.4479166666666668e-06,
1438
+ "loss": 2.8101,
1439
+ "mean_token_accuracy": 0.46102563589811324,
1440
+ "num_tokens": 80573264.0,
1441
+ "step": 1590
1442
+ },
1443
+ {
1444
+ "epoch": 8.291287386215865,
1445
+ "grad_norm": 0.9456131458282471,
1446
+ "learning_rate": 3.34375e-06,
1447
+ "loss": 2.7968,
1448
+ "mean_token_accuracy": 0.4618312932550907,
1449
+ "num_tokens": 81089403.0,
1450
+ "step": 1600
1451
+ },
1452
+ {
1453
+ "epoch": 8.34330299089727,
1454
+ "grad_norm": 0.35843464732170105,
1455
+ "learning_rate": 3.2395833333333336e-06,
1456
+ "loss": 2.8195,
1457
+ "mean_token_accuracy": 0.4609459452331066,
1458
+ "num_tokens": 81597420.0,
1459
+ "step": 1610
1460
+ },
1461
+ {
1462
+ "epoch": 8.395318595578674,
1463
+ "grad_norm": 0.44011804461479187,
1464
+ "learning_rate": 3.1354166666666668e-06,
1465
+ "loss": 2.8121,
1466
+ "mean_token_accuracy": 0.46082125306129457,
1467
+ "num_tokens": 82110599.0,
1468
+ "step": 1620
1469
+ },
1470
+ {
1471
+ "epoch": 8.447334200260078,
1472
+ "grad_norm": 0.9045904278755188,
1473
+ "learning_rate": 3.03125e-06,
1474
+ "loss": 2.8096,
1475
+ "mean_token_accuracy": 0.46105934455990794,
1476
+ "num_tokens": 82613564.0,
1477
+ "step": 1630
1478
+ },
1479
+ {
1480
+ "epoch": 8.499349804941483,
1481
+ "grad_norm": 0.4378410577774048,
1482
+ "learning_rate": 2.9270833333333336e-06,
1483
+ "loss": 2.8135,
1484
+ "mean_token_accuracy": 0.461653308570385,
1485
+ "num_tokens": 83120175.0,
1486
+ "step": 1640
1487
+ },
1488
+ {
1489
+ "epoch": 8.551365409622887,
1490
+ "grad_norm": 0.5772425532341003,
1491
+ "learning_rate": 2.822916666666667e-06,
1492
+ "loss": 2.8203,
1493
+ "mean_token_accuracy": 0.4603270493447781,
1494
+ "num_tokens": 83631755.0,
1495
+ "step": 1650
1496
+ },
1497
+ {
1498
+ "epoch": 8.603381014304292,
1499
+ "grad_norm": 0.4606432318687439,
1500
+ "learning_rate": 2.71875e-06,
1501
+ "loss": 2.8068,
1502
+ "mean_token_accuracy": 0.46261599510908125,
1503
+ "num_tokens": 84140572.0,
1504
+ "step": 1660
1505
+ },
1506
+ {
1507
+ "epoch": 8.655396618985696,
1508
+ "grad_norm": 0.605274498462677,
1509
+ "learning_rate": 2.6145833333333336e-06,
1510
+ "loss": 2.8211,
1511
+ "mean_token_accuracy": 0.4599962942302227,
1512
+ "num_tokens": 84652357.0,
1513
+ "step": 1670
1514
+ },
1515
+ {
1516
+ "epoch": 8.7074122236671,
1517
+ "grad_norm": 0.44417762756347656,
1518
+ "learning_rate": 2.510416666666667e-06,
1519
+ "loss": 2.7851,
1520
+ "mean_token_accuracy": 0.4644029013812542,
1521
+ "num_tokens": 85158917.0,
1522
+ "step": 1680
1523
+ },
1524
+ {
1525
+ "epoch": 8.759427828348505,
1526
+ "grad_norm": 0.37345197796821594,
1527
+ "learning_rate": 2.40625e-06,
1528
+ "loss": 2.8268,
1529
+ "mean_token_accuracy": 0.4594727732241154,
1530
+ "num_tokens": 85663276.0,
1531
+ "step": 1690
1532
+ },
1533
+ {
1534
+ "epoch": 8.81144343302991,
1535
+ "grad_norm": 0.6004917621612549,
1536
+ "learning_rate": 2.3020833333333332e-06,
1537
+ "loss": 2.8403,
1538
+ "mean_token_accuracy": 0.45820842683315277,
1539
+ "num_tokens": 86167477.0,
1540
+ "step": 1700
1541
+ },
1542
+ {
1543
+ "epoch": 8.863459037711314,
1544
+ "grad_norm": 0.38165897130966187,
1545
+ "learning_rate": 2.197916666666667e-06,
1546
+ "loss": 2.8076,
1547
+ "mean_token_accuracy": 0.4625224307179451,
1548
+ "num_tokens": 86678206.0,
1549
+ "step": 1710
1550
+ },
1551
+ {
1552
+ "epoch": 8.915474642392718,
1553
+ "grad_norm": 0.48859167098999023,
1554
+ "learning_rate": 2.09375e-06,
1555
+ "loss": 2.8216,
1556
+ "mean_token_accuracy": 0.4601791054010391,
1557
+ "num_tokens": 87178330.0,
1558
+ "step": 1720
1559
+ },
1560
+ {
1561
+ "epoch": 8.967490247074123,
1562
+ "grad_norm": 0.3046579957008362,
1563
+ "learning_rate": 1.9895833333333333e-06,
1564
+ "loss": 2.7935,
1565
+ "mean_token_accuracy": 0.46218492090702057,
1566
+ "num_tokens": 87683633.0,
1567
+ "step": 1730
1568
+ },
1569
+ {
1570
+ "epoch": 9.01560468140442,
1571
+ "grad_norm": 0.5711421966552734,
1572
+ "learning_rate": 1.8854166666666667e-06,
1573
+ "loss": 2.8012,
1574
+ "mean_token_accuracy": 0.46230730736577835,
1575
+ "num_tokens": 88138585.0,
1576
+ "step": 1740
1577
+ },
1578
+ {
1579
+ "epoch": 9.067620286085825,
1580
+ "grad_norm": 0.4970647394657135,
1581
+ "learning_rate": 1.78125e-06,
1582
+ "loss": 2.8216,
1583
+ "mean_token_accuracy": 0.4593446545302868,
1584
+ "num_tokens": 88661924.0,
1585
+ "step": 1750
1586
+ },
1587
+ {
1588
+ "epoch": 9.11963589076723,
1589
+ "grad_norm": 1.1503207683563232,
1590
+ "learning_rate": 1.6770833333333333e-06,
1591
+ "loss": 2.7998,
1592
+ "mean_token_accuracy": 0.4622166946530342,
1593
+ "num_tokens": 89176033.0,
1594
+ "step": 1760
1595
+ },
1596
+ {
1597
+ "epoch": 9.171651495448634,
1598
+ "grad_norm": 0.4958202838897705,
1599
+ "learning_rate": 1.5729166666666667e-06,
1600
+ "loss": 2.8121,
1601
+ "mean_token_accuracy": 0.4620626524090767,
1602
+ "num_tokens": 89679688.0,
1603
+ "step": 1770
1604
+ },
1605
+ {
1606
+ "epoch": 9.223667100130038,
1607
+ "grad_norm": 0.36465510725975037,
1608
+ "learning_rate": 1.4687500000000001e-06,
1609
+ "loss": 2.7833,
1610
+ "mean_token_accuracy": 0.4629977948963642,
1611
+ "num_tokens": 90191820.0,
1612
+ "step": 1780
1613
+ },
1614
+ {
1615
+ "epoch": 9.275682704811443,
1616
+ "grad_norm": 0.3522260785102844,
1617
+ "learning_rate": 1.3645833333333333e-06,
1618
+ "loss": 2.7935,
1619
+ "mean_token_accuracy": 0.4633957423269749,
1620
+ "num_tokens": 90696204.0,
1621
+ "step": 1790
1622
+ },
1623
+ {
1624
+ "epoch": 9.327698309492847,
1625
+ "grad_norm": 0.867915689945221,
1626
+ "learning_rate": 1.2604166666666667e-06,
1627
+ "loss": 2.7956,
1628
+ "mean_token_accuracy": 0.463359072804451,
1629
+ "num_tokens": 91196064.0,
1630
+ "step": 1800
1631
+ },
1632
+ {
1633
+ "epoch": 9.379713914174252,
1634
+ "grad_norm": 0.32881176471710205,
1635
+ "learning_rate": 1.1562500000000002e-06,
1636
+ "loss": 2.8095,
1637
+ "mean_token_accuracy": 0.4616400755941868,
1638
+ "num_tokens": 91707458.0,
1639
+ "step": 1810
1640
+ },
1641
+ {
1642
+ "epoch": 9.431729518855656,
1643
+ "grad_norm": 0.9605303406715393,
1644
+ "learning_rate": 1.0520833333333334e-06,
1645
+ "loss": 2.7994,
1646
+ "mean_token_accuracy": 0.463277330994606,
1647
+ "num_tokens": 92215685.0,
1648
+ "step": 1820
1649
+ },
1650
+ {
1651
+ "epoch": 9.48374512353706,
1652
+ "grad_norm": 0.401554137468338,
1653
+ "learning_rate": 9.479166666666667e-07,
1654
+ "loss": 2.8129,
1655
+ "mean_token_accuracy": 0.4616898998618126,
1656
+ "num_tokens": 92735287.0,
1657
+ "step": 1830
1658
+ },
1659
+ {
1660
+ "epoch": 9.535760728218465,
1661
+ "grad_norm": 0.5375267863273621,
1662
+ "learning_rate": 8.437500000000001e-07,
1663
+ "loss": 2.7958,
1664
+ "mean_token_accuracy": 0.46335128173232076,
1665
+ "num_tokens": 93238559.0,
1666
+ "step": 1840
1667
+ },
1668
+ {
1669
+ "epoch": 9.58777633289987,
1670
+ "grad_norm": 0.47346314787864685,
1671
+ "learning_rate": 7.395833333333334e-07,
1672
+ "loss": 2.8133,
1673
+ "mean_token_accuracy": 0.4614164501428604,
1674
+ "num_tokens": 93749855.0,
1675
+ "step": 1850
1676
+ },
1677
+ {
1678
+ "epoch": 9.639791937581274,
1679
+ "grad_norm": 0.7040103673934937,
1680
+ "learning_rate": 6.354166666666667e-07,
1681
+ "loss": 2.813,
1682
+ "mean_token_accuracy": 0.46037371680140493,
1683
+ "num_tokens": 94262461.0,
1684
+ "step": 1860
1685
+ },
1686
+ {
1687
+ "epoch": 9.691807542262678,
1688
+ "grad_norm": 0.34465914964675903,
1689
+ "learning_rate": 5.3125e-07,
1690
+ "loss": 2.7943,
1691
+ "mean_token_accuracy": 0.46172691956162454,
1692
+ "num_tokens": 94781134.0,
1693
+ "step": 1870
1694
+ },
1695
+ {
1696
+ "epoch": 9.743823146944083,
1697
+ "grad_norm": 0.6020010113716125,
1698
+ "learning_rate": 4.2708333333333336e-07,
1699
+ "loss": 2.8133,
1700
+ "mean_token_accuracy": 0.4611339412629604,
1701
+ "num_tokens": 95297236.0,
1702
+ "step": 1880
1703
+ },
1704
+ {
1705
+ "epoch": 9.795838751625487,
1706
+ "grad_norm": 0.4364703893661499,
1707
+ "learning_rate": 3.229166666666667e-07,
1708
+ "loss": 2.8168,
1709
+ "mean_token_accuracy": 0.4605915732681751,
1710
+ "num_tokens": 95814960.0,
1711
+ "step": 1890
1712
+ },
1713
+ {
1714
+ "epoch": 9.847854356306891,
1715
+ "grad_norm": 0.2786646783351898,
1716
+ "learning_rate": 2.1875e-07,
1717
+ "loss": 2.7972,
1718
+ "mean_token_accuracy": 0.4621445916593075,
1719
+ "num_tokens": 96318017.0,
1720
+ "step": 1900
1721
+ },
1722
+ {
1723
+ "epoch": 9.899869960988296,
1724
+ "grad_norm": 0.8361368179321289,
1725
+ "learning_rate": 1.1458333333333333e-07,
1726
+ "loss": 2.8178,
1727
+ "mean_token_accuracy": 0.46137167140841484,
1728
+ "num_tokens": 96821013.0,
1729
+ "step": 1910
1730
+ },
1731
+ {
1732
+ "epoch": 9.9518855656697,
1733
+ "grad_norm": 0.5781928300857544,
1734
+ "learning_rate": 1.0416666666666667e-08,
1735
+ "loss": 2.8006,
1736
+ "mean_token_accuracy": 0.4619719944894314,
1737
+ "num_tokens": 97314719.0,
1738
+ "step": 1920
1739
+ }
1740
+ ],
1741
+ "logging_steps": 10,
1742
+ "max_steps": 1920,
1743
+ "num_input_tokens_seen": 0,
1744
+ "num_train_epochs": 10,
1745
+ "save_steps": 500,
1746
+ "stateful_callbacks": {
1747
+ "TrainerControl": {
1748
+ "args": {
1749
+ "should_epoch_stop": false,
1750
+ "should_evaluate": false,
1751
+ "should_log": false,
1752
+ "should_save": true,
1753
+ "should_training_stop": true
1754
+ },
1755
+ "attributes": {}
1756
+ }
1757
+ },
1758
+ "total_flos": 7.91377641395159e+16,
1759
+ "train_batch_size": 64,
1760
+ "trial_name": null,
1761
+ "trial_params": null
1762
+ }
checkpoint-1920/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea990651e2aecabd57f309f323439c138096e78f77c92b4e54f46fe372d50d1d
3
+ size 5688
checkpoint-500/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 512,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
checkpoint-500/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }
checkpoint-500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1bc9ccc9f9684fc700511035d70f5284928551f9aed5fa58e2c5d56fddf0f73
3
+ size 309900448
checkpoint-500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:17ad5e6af42327b7174d16868511d6580268c9e480eaf2876d20aa2065340925
3
+ size 619836730
checkpoint-500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
3
+ size 14244
checkpoint-500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e887a572b04f7989a10ac8753c3380f39d20e1400e4ba26f960577fca8a5e6ed
3
+ size 1064
checkpoint-500/special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
checkpoint-500/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-500/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-500/tokenizer_config.json ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": true,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": true,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "clean_up_tokenization_spaces": false,
33
+ "eos_token": "</s>",
34
+ "extra_special_tokens": {},
35
+ "legacy": false,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "tokenizer_class": "LlamaTokenizer",
40
+ "unk_token": "<unk>",
41
+ "use_default_system_prompt": false
42
+ }
checkpoint-500/trainer_state.json ADDED
@@ -0,0 +1,484 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.5929778933680105,
6
+ "eval_steps": 500,
7
+ "global_step": 500,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.05201560468140442,
14
+ "grad_norm": 3.158724784851074,
15
+ "learning_rate": 1.9906250000000003e-05,
16
+ "loss": 6.2302,
17
+ "mean_token_accuracy": 0.16946521885693072,
18
+ "num_tokens": 513852.0,
19
+ "step": 10
20
+ },
21
+ {
22
+ "epoch": 0.10403120936280884,
23
+ "grad_norm": 10.176986694335938,
24
+ "learning_rate": 1.9802083333333335e-05,
25
+ "loss": 6.0745,
26
+ "mean_token_accuracy": 0.17949572652578355,
27
+ "num_tokens": 1032088.0,
28
+ "step": 20
29
+ },
30
+ {
31
+ "epoch": 0.15604681404421328,
32
+ "grad_norm": 3.010871410369873,
33
+ "learning_rate": 1.9697916666666667e-05,
34
+ "loss": 5.8317,
35
+ "mean_token_accuracy": 0.1901865139603615,
36
+ "num_tokens": 1539389.0,
37
+ "step": 30
38
+ },
39
+ {
40
+ "epoch": 0.20806241872561768,
41
+ "grad_norm": 3.298682451248169,
42
+ "learning_rate": 1.9593750000000002e-05,
43
+ "loss": 5.5987,
44
+ "mean_token_accuracy": 0.20738236755132675,
45
+ "num_tokens": 2057784.0,
46
+ "step": 40
47
+ },
48
+ {
49
+ "epoch": 0.26007802340702213,
50
+ "grad_norm": 2.472743272781372,
51
+ "learning_rate": 1.9489583333333334e-05,
52
+ "loss": 5.4572,
53
+ "mean_token_accuracy": 0.2219402514398098,
54
+ "num_tokens": 2574094.0,
55
+ "step": 50
56
+ },
57
+ {
58
+ "epoch": 0.31209362808842656,
59
+ "grad_norm": 2.3076207637786865,
60
+ "learning_rate": 1.938541666666667e-05,
61
+ "loss": 5.2817,
62
+ "mean_token_accuracy": 0.23026303350925445,
63
+ "num_tokens": 3076289.0,
64
+ "step": 60
65
+ },
66
+ {
67
+ "epoch": 0.3641092327698309,
68
+ "grad_norm": 2.1481237411499023,
69
+ "learning_rate": 1.928125e-05,
70
+ "loss": 5.0596,
71
+ "mean_token_accuracy": 0.24168218560516835,
72
+ "num_tokens": 3586260.0,
73
+ "step": 70
74
+ },
75
+ {
76
+ "epoch": 0.41612483745123535,
77
+ "grad_norm": 2.2509524822235107,
78
+ "learning_rate": 1.9177083333333336e-05,
79
+ "loss": 4.9117,
80
+ "mean_token_accuracy": 0.2510342214256525,
81
+ "num_tokens": 4094922.0,
82
+ "step": 80
83
+ },
84
+ {
85
+ "epoch": 0.4681404421326398,
86
+ "grad_norm": 3.319535732269287,
87
+ "learning_rate": 1.907291666666667e-05,
88
+ "loss": 4.7624,
89
+ "mean_token_accuracy": 0.26219444051384927,
90
+ "num_tokens": 4606336.0,
91
+ "step": 90
92
+ },
93
+ {
94
+ "epoch": 0.5201560468140443,
95
+ "grad_norm": 2.0256783962249756,
96
+ "learning_rate": 1.8968750000000004e-05,
97
+ "loss": 4.6138,
98
+ "mean_token_accuracy": 0.2705389905720949,
99
+ "num_tokens": 5110749.0,
100
+ "step": 100
101
+ },
102
+ {
103
+ "epoch": 0.5721716514954487,
104
+ "grad_norm": 5.284979820251465,
105
+ "learning_rate": 1.8864583333333336e-05,
106
+ "loss": 4.5011,
107
+ "mean_token_accuracy": 0.27774968966841695,
108
+ "num_tokens": 5611872.0,
109
+ "step": 110
110
+ },
111
+ {
112
+ "epoch": 0.6241872561768531,
113
+ "grad_norm": 15.897697448730469,
114
+ "learning_rate": 1.8760416666666668e-05,
115
+ "loss": 4.3486,
116
+ "mean_token_accuracy": 0.29455167055130005,
117
+ "num_tokens": 6123716.0,
118
+ "step": 120
119
+ },
120
+ {
121
+ "epoch": 0.6762028608582574,
122
+ "grad_norm": 3.0891332626342773,
123
+ "learning_rate": 1.8656250000000003e-05,
124
+ "loss": 4.2372,
125
+ "mean_token_accuracy": 0.3053073361515999,
126
+ "num_tokens": 6630449.0,
127
+ "step": 130
128
+ },
129
+ {
130
+ "epoch": 0.7282184655396619,
131
+ "grad_norm": 2.85463285446167,
132
+ "learning_rate": 1.8552083333333335e-05,
133
+ "loss": 4.161,
134
+ "mean_token_accuracy": 0.3164319314062595,
135
+ "num_tokens": 7145717.0,
136
+ "step": 140
137
+ },
138
+ {
139
+ "epoch": 0.7802340702210663,
140
+ "grad_norm": 1.2611316442489624,
141
+ "learning_rate": 1.8447916666666667e-05,
142
+ "loss": 4.1002,
143
+ "mean_token_accuracy": 0.32258255779743195,
144
+ "num_tokens": 7655884.0,
145
+ "step": 150
146
+ },
147
+ {
148
+ "epoch": 0.8322496749024707,
149
+ "grad_norm": 1.4083240032196045,
150
+ "learning_rate": 1.8343750000000002e-05,
151
+ "loss": 4.0138,
152
+ "mean_token_accuracy": 0.3296029016375542,
153
+ "num_tokens": 8159725.0,
154
+ "step": 160
155
+ },
156
+ {
157
+ "epoch": 0.8842652795838751,
158
+ "grad_norm": 1.0340592861175537,
159
+ "learning_rate": 1.8239583333333334e-05,
160
+ "loss": 3.9163,
161
+ "mean_token_accuracy": 0.3385446682572365,
162
+ "num_tokens": 8668116.0,
163
+ "step": 170
164
+ },
165
+ {
166
+ "epoch": 0.9362808842652796,
167
+ "grad_norm": 0.7984447479248047,
168
+ "learning_rate": 1.8135416666666666e-05,
169
+ "loss": 3.8628,
170
+ "mean_token_accuracy": 0.3455250822007656,
171
+ "num_tokens": 9182489.0,
172
+ "step": 180
173
+ },
174
+ {
175
+ "epoch": 0.988296488946684,
176
+ "grad_norm": 1.244081735610962,
177
+ "learning_rate": 1.803125e-05,
178
+ "loss": 3.8025,
179
+ "mean_token_accuracy": 0.35223659574985505,
180
+ "num_tokens": 9676078.0,
181
+ "step": 190
182
+ },
183
+ {
184
+ "epoch": 1.0364109232769831,
185
+ "grad_norm": 2.412661552429199,
186
+ "learning_rate": 1.7927083333333337e-05,
187
+ "loss": 3.7325,
188
+ "mean_token_accuracy": 0.35705321382831884,
189
+ "num_tokens": 10138114.0,
190
+ "step": 200
191
+ },
192
+ {
193
+ "epoch": 1.0884265279583876,
194
+ "grad_norm": 1.7736923694610596,
195
+ "learning_rate": 1.782291666666667e-05,
196
+ "loss": 3.7052,
197
+ "mean_token_accuracy": 0.3590236373245716,
198
+ "num_tokens": 10656504.0,
199
+ "step": 210
200
+ },
201
+ {
202
+ "epoch": 1.140442132639792,
203
+ "grad_norm": 0.7769147157669067,
204
+ "learning_rate": 1.7718750000000004e-05,
205
+ "loss": 3.6433,
206
+ "mean_token_accuracy": 0.36575346291065214,
207
+ "num_tokens": 11161374.0,
208
+ "step": 220
209
+ },
210
+ {
211
+ "epoch": 1.1924577373211964,
212
+ "grad_norm": 0.9089329838752747,
213
+ "learning_rate": 1.7614583333333336e-05,
214
+ "loss": 3.653,
215
+ "mean_token_accuracy": 0.3651699565351009,
216
+ "num_tokens": 11670197.0,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 1.2444733420026008,
221
+ "grad_norm": 0.7330015897750854,
222
+ "learning_rate": 1.7510416666666668e-05,
223
+ "loss": 3.5843,
224
+ "mean_token_accuracy": 0.3718643538653851,
225
+ "num_tokens": 12174931.0,
226
+ "step": 240
227
+ },
228
+ {
229
+ "epoch": 1.2964889466840053,
230
+ "grad_norm": 1.534440040588379,
231
+ "learning_rate": 1.7406250000000003e-05,
232
+ "loss": 3.5468,
233
+ "mean_token_accuracy": 0.3751932092010975,
234
+ "num_tokens": 12685023.0,
235
+ "step": 250
236
+ },
237
+ {
238
+ "epoch": 1.3485045513654097,
239
+ "grad_norm": 1.9603444337844849,
240
+ "learning_rate": 1.7302083333333335e-05,
241
+ "loss": 3.4935,
242
+ "mean_token_accuracy": 0.3808542750775814,
243
+ "num_tokens": 13197338.0,
244
+ "step": 260
245
+ },
246
+ {
247
+ "epoch": 1.4005201560468141,
248
+ "grad_norm": 0.5988335013389587,
249
+ "learning_rate": 1.7197916666666667e-05,
250
+ "loss": 3.4988,
251
+ "mean_token_accuracy": 0.3815221585333347,
252
+ "num_tokens": 13704900.0,
253
+ "step": 270
254
+ },
255
+ {
256
+ "epoch": 1.4525357607282183,
257
+ "grad_norm": 2.7422847747802734,
258
+ "learning_rate": 1.7093750000000002e-05,
259
+ "loss": 3.4438,
260
+ "mean_token_accuracy": 0.3865826904773712,
261
+ "num_tokens": 14210835.0,
262
+ "step": 280
263
+ },
264
+ {
265
+ "epoch": 1.5045513654096228,
266
+ "grad_norm": 1.9575722217559814,
267
+ "learning_rate": 1.6989583333333334e-05,
268
+ "loss": 3.417,
269
+ "mean_token_accuracy": 0.38725101202726364,
270
+ "num_tokens": 14726584.0,
271
+ "step": 290
272
+ },
273
+ {
274
+ "epoch": 1.5565669700910272,
275
+ "grad_norm": 0.6955925226211548,
276
+ "learning_rate": 1.6885416666666666e-05,
277
+ "loss": 3.3895,
278
+ "mean_token_accuracy": 0.39159547463059424,
279
+ "num_tokens": 15238009.0,
280
+ "step": 300
281
+ },
282
+ {
283
+ "epoch": 1.6085825747724316,
284
+ "grad_norm": 0.769548237323761,
285
+ "learning_rate": 1.678125e-05,
286
+ "loss": 3.3979,
287
+ "mean_token_accuracy": 0.39134200662374496,
288
+ "num_tokens": 15736398.0,
289
+ "step": 310
290
+ },
291
+ {
292
+ "epoch": 1.660598179453836,
293
+ "grad_norm": 0.8247058391571045,
294
+ "learning_rate": 1.6677083333333333e-05,
295
+ "loss": 3.3884,
296
+ "mean_token_accuracy": 0.3924022614955902,
297
+ "num_tokens": 16235540.0,
298
+ "step": 320
299
+ },
300
+ {
301
+ "epoch": 1.7126137841352405,
302
+ "grad_norm": 0.9121173024177551,
303
+ "learning_rate": 1.657291666666667e-05,
304
+ "loss": 3.3541,
305
+ "mean_token_accuracy": 0.3951391130685806,
306
+ "num_tokens": 16747733.0,
307
+ "step": 330
308
+ },
309
+ {
310
+ "epoch": 1.7646293888166449,
311
+ "grad_norm": 1.3171864748001099,
312
+ "learning_rate": 1.6468750000000004e-05,
313
+ "loss": 3.3486,
314
+ "mean_token_accuracy": 0.39698211625218394,
315
+ "num_tokens": 17265239.0,
316
+ "step": 340
317
+ },
318
+ {
319
+ "epoch": 1.8166449934980493,
320
+ "grad_norm": 4.172916412353516,
321
+ "learning_rate": 1.6364583333333336e-05,
322
+ "loss": 3.325,
323
+ "mean_token_accuracy": 0.39922175258398057,
324
+ "num_tokens": 17780508.0,
325
+ "step": 350
326
+ },
327
+ {
328
+ "epoch": 1.8686605981794537,
329
+ "grad_norm": 0.4353245794773102,
330
+ "learning_rate": 1.6260416666666668e-05,
331
+ "loss": 3.2807,
332
+ "mean_token_accuracy": 0.4035400986671448,
333
+ "num_tokens": 18297602.0,
334
+ "step": 360
335
+ },
336
+ {
337
+ "epoch": 1.9206762028608582,
338
+ "grad_norm": 3.6861395835876465,
339
+ "learning_rate": 1.6156250000000003e-05,
340
+ "loss": 3.2832,
341
+ "mean_token_accuracy": 0.4044442743062973,
342
+ "num_tokens": 18798625.0,
343
+ "step": 370
344
+ },
345
+ {
346
+ "epoch": 1.9726918075422626,
347
+ "grad_norm": 1.1718441247940063,
348
+ "learning_rate": 1.6052083333333335e-05,
349
+ "loss": 3.2664,
350
+ "mean_token_accuracy": 0.40756076350808146,
351
+ "num_tokens": 19294344.0,
352
+ "step": 380
353
+ },
354
+ {
355
+ "epoch": 2.020806241872562,
356
+ "grad_norm": 0.6570408940315247,
357
+ "learning_rate": 1.5947916666666667e-05,
358
+ "loss": 3.2056,
359
+ "mean_token_accuracy": 0.4098896440621969,
360
+ "num_tokens": 19761637.0,
361
+ "step": 390
362
+ },
363
+ {
364
+ "epoch": 2.0728218465539663,
365
+ "grad_norm": 0.46285247802734375,
366
+ "learning_rate": 1.5843750000000002e-05,
367
+ "loss": 3.2396,
368
+ "mean_token_accuracy": 0.41089860200881956,
369
+ "num_tokens": 20265138.0,
370
+ "step": 400
371
+ },
372
+ {
373
+ "epoch": 2.1248374512353707,
374
+ "grad_norm": 3.5917506217956543,
375
+ "learning_rate": 1.5739583333333334e-05,
376
+ "loss": 3.2205,
377
+ "mean_token_accuracy": 0.41081195920705793,
378
+ "num_tokens": 20777818.0,
379
+ "step": 410
380
+ },
381
+ {
382
+ "epoch": 2.176853055916775,
383
+ "grad_norm": 0.6580970287322998,
384
+ "learning_rate": 1.5635416666666666e-05,
385
+ "loss": 3.225,
386
+ "mean_token_accuracy": 0.41244959980249407,
387
+ "num_tokens": 21282514.0,
388
+ "step": 420
389
+ },
390
+ {
391
+ "epoch": 2.2288686605981796,
392
+ "grad_norm": 1.2673513889312744,
393
+ "learning_rate": 1.553125e-05,
394
+ "loss": 3.2143,
395
+ "mean_token_accuracy": 0.41195737645030023,
396
+ "num_tokens": 21796246.0,
397
+ "step": 430
398
+ },
399
+ {
400
+ "epoch": 2.280884265279584,
401
+ "grad_norm": 1.9437355995178223,
402
+ "learning_rate": 1.5427083333333333e-05,
403
+ "loss": 3.2027,
404
+ "mean_token_accuracy": 0.41485874727368355,
405
+ "num_tokens": 22310915.0,
406
+ "step": 440
407
+ },
408
+ {
409
+ "epoch": 2.3328998699609884,
410
+ "grad_norm": 0.4803381860256195,
411
+ "learning_rate": 1.532291666666667e-05,
412
+ "loss": 3.1743,
413
+ "mean_token_accuracy": 0.417393134534359,
414
+ "num_tokens": 22823271.0,
415
+ "step": 450
416
+ },
417
+ {
418
+ "epoch": 2.384915474642393,
419
+ "grad_norm": 1.0227100849151611,
420
+ "learning_rate": 1.5218750000000002e-05,
421
+ "loss": 3.1751,
422
+ "mean_token_accuracy": 0.4173254922032356,
423
+ "num_tokens": 23326555.0,
424
+ "step": 460
425
+ },
426
+ {
427
+ "epoch": 2.4369310793237973,
428
+ "grad_norm": 0.9217405915260315,
429
+ "learning_rate": 1.5114583333333334e-05,
430
+ "loss": 3.1336,
431
+ "mean_token_accuracy": 0.42112250030040743,
432
+ "num_tokens": 23832549.0,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 2.4889466840052017,
437
+ "grad_norm": 0.7527967095375061,
438
+ "learning_rate": 1.5010416666666668e-05,
439
+ "loss": 3.1469,
440
+ "mean_token_accuracy": 0.42071041762828826,
441
+ "num_tokens": 24338061.0,
442
+ "step": 480
443
+ },
444
+ {
445
+ "epoch": 2.540962288686606,
446
+ "grad_norm": 0.370236873626709,
447
+ "learning_rate": 1.4906250000000001e-05,
448
+ "loss": 3.1461,
449
+ "mean_token_accuracy": 0.41948254629969595,
450
+ "num_tokens": 24854140.0,
451
+ "step": 490
452
+ },
453
+ {
454
+ "epoch": 2.5929778933680105,
455
+ "grad_norm": 0.4003697335720062,
456
+ "learning_rate": 1.4802083333333335e-05,
457
+ "loss": 3.1162,
458
+ "mean_token_accuracy": 0.4244517408311367,
459
+ "num_tokens": 25367383.0,
460
+ "step": 500
461
+ }
462
+ ],
463
+ "logging_steps": 10,
464
+ "max_steps": 1920,
465
+ "num_input_tokens_seen": 0,
466
+ "num_train_epochs": 10,
467
+ "save_steps": 500,
468
+ "stateful_callbacks": {
469
+ "TrainerControl": {
470
+ "args": {
471
+ "should_epoch_stop": false,
472
+ "should_evaluate": false,
473
+ "should_log": false,
474
+ "should_save": true,
475
+ "should_training_stop": false
476
+ },
477
+ "attributes": {}
478
+ }
479
+ },
480
+ "total_flos": 2.062862731055923e+16,
481
+ "train_batch_size": 64,
482
+ "trial_name": null,
483
+ "trial_params": null
484
+ }
checkpoint-500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea990651e2aecabd57f309f323439c138096e78f77c92b4e54f46fe372d50d1d
3
+ size 5688
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 768,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1024,
14
+ "max_position_embeddings": 512,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 6,
18
+ "num_hidden_layers": 6,
19
+ "num_key_value_heads": 6,
20
+ "pad_token_id": 0,
21
+ "pretraining_tp": 1,
22
+ "rms_norm_eps": 1e-05,
23
+ "rope_scaling": null,
24
+ "rope_theta": 10000.0,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.51.3",
28
+ "use_cache": true,
29
+ "vocab_size": 32000
30
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.51.3"
7
+ }