pszemraj commited on
Commit
1058833
1 Parent(s): 55287d5

End of training

Browse files
Files changed (5) hide show
  1. README.md +2 -2
  2. all_results.json +17 -0
  3. eval_results.json +11 -0
  4. train_results.json +10 -0
  5. trainer_state.json +1575 -0
README.md CHANGED
@@ -15,11 +15,11 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  # jamba-H1024_L12-v0.12-fineweb-100k-xlong_16k-knowledge-inoc-concat-v1-vN
17
 
18
- This model is a fine-tuned version of [pszemraj/jamba-H1024_L12-v0.12-fineweb-100k-xlong_16k](https://huggingface.co/pszemraj/jamba-H1024_L12-v0.12-fineweb-100k-xlong_16k) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
  - Loss: 3.0366
21
  - Accuracy: 0.4514
22
- - Num Input Tokens Seen: 1677721600
23
 
24
  ## Model description
25
 
 
15
 
16
  # jamba-H1024_L12-v0.12-fineweb-100k-xlong_16k-knowledge-inoc-concat-v1-vN
17
 
18
+ This model is a fine-tuned version of [pszemraj/jamba-H1024_L12-v0.12-fineweb-100k-xlong_16k](https://huggingface.co/pszemraj/jamba-H1024_L12-v0.12-fineweb-100k-xlong_16k) on the BEE-spoke-data/knowledge-inoc-concat-v1 dataset.
19
  It achieves the following results on the evaluation set:
20
  - Loss: 3.0366
21
  - Accuracy: 0.4514
22
+ - Num Input Tokens Seen: 1975517184
23
 
24
  ## Model description
25
 
all_results.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.997349589186324,
3
+ "eval_accuracy": 0.45143400679566215,
4
+ "eval_loss": 3.0366320610046387,
5
+ "eval_runtime": 36.3403,
6
+ "eval_samples": 300,
7
+ "eval_samples_per_second": 8.255,
8
+ "eval_steps_per_second": 2.064,
9
+ "num_input_tokens_seen": 1975517184,
10
+ "perplexity": 20.83495408242377,
11
+ "total_flos": 9.732316690586272e+18,
12
+ "train_loss": 3.186332613039928,
13
+ "train_runtime": 34789.3198,
14
+ "train_samples": 60368,
15
+ "train_samples_per_second": 3.47,
16
+ "train_steps_per_second": 0.027
17
+ }
eval_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.997349589186324,
3
+ "eval_accuracy": 0.45143400679566215,
4
+ "eval_loss": 3.0366320610046387,
5
+ "eval_runtime": 36.3403,
6
+ "eval_samples": 300,
7
+ "eval_samples_per_second": 8.255,
8
+ "eval_steps_per_second": 2.064,
9
+ "num_input_tokens_seen": 1975517184,
10
+ "perplexity": 20.83495408242377
11
+ }
train_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.997349589186324,
3
+ "num_input_tokens_seen": 1975517184,
4
+ "total_flos": 9.732316690586272e+18,
5
+ "train_loss": 3.186332613039928,
6
+ "train_runtime": 34789.3198,
7
+ "train_samples": 60368,
8
+ "train_samples_per_second": 3.47,
9
+ "train_steps_per_second": 0.027
10
+ }
trainer_state.json ADDED
@@ -0,0 +1,1575 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.997349589186324,
5
+ "eval_steps": 200,
6
+ "global_step": 942,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.010601643254704479,
13
+ "grad_norm": 0.8797913789749146,
14
+ "learning_rate": 5.208333333333334e-06,
15
+ "loss": 3.3968,
16
+ "num_input_tokens_seen": 10485760,
17
+ "step": 5
18
+ },
19
+ {
20
+ "epoch": 0.021203286509408958,
21
+ "grad_norm": 0.5857719779014587,
22
+ "learning_rate": 1.0416666666666668e-05,
23
+ "loss": 3.3102,
24
+ "num_input_tokens_seen": 20971520,
25
+ "step": 10
26
+ },
27
+ {
28
+ "epoch": 0.031804929764113435,
29
+ "grad_norm": 0.38287150859832764,
30
+ "learning_rate": 1.5625e-05,
31
+ "loss": 3.3919,
32
+ "num_input_tokens_seen": 31457280,
33
+ "step": 15
34
+ },
35
+ {
36
+ "epoch": 0.042406573018817915,
37
+ "grad_norm": 0.2890233099460602,
38
+ "learning_rate": 2.0833333333333336e-05,
39
+ "loss": 3.3448,
40
+ "num_input_tokens_seen": 41943040,
41
+ "step": 20
42
+ },
43
+ {
44
+ "epoch": 0.053008216273522396,
45
+ "grad_norm": 0.28800511360168457,
46
+ "learning_rate": 2.604166666666667e-05,
47
+ "loss": 3.247,
48
+ "num_input_tokens_seen": 52428800,
49
+ "step": 25
50
+ },
51
+ {
52
+ "epoch": 0.06360985952822687,
53
+ "grad_norm": 0.5624825358390808,
54
+ "learning_rate": 3.125e-05,
55
+ "loss": 3.3736,
56
+ "num_input_tokens_seen": 62914560,
57
+ "step": 30
58
+ },
59
+ {
60
+ "epoch": 0.07421150278293136,
61
+ "grad_norm": 0.2708434760570526,
62
+ "learning_rate": 3.6458333333333336e-05,
63
+ "loss": 3.3105,
64
+ "num_input_tokens_seen": 73400320,
65
+ "step": 35
66
+ },
67
+ {
68
+ "epoch": 0.08481314603763583,
69
+ "grad_norm": 0.26101428270339966,
70
+ "learning_rate": 4.166666666666667e-05,
71
+ "loss": 3.3523,
72
+ "num_input_tokens_seen": 83886080,
73
+ "step": 40
74
+ },
75
+ {
76
+ "epoch": 0.09541478929234032,
77
+ "grad_norm": 0.22077707946300507,
78
+ "learning_rate": 4.6875e-05,
79
+ "loss": 3.3395,
80
+ "num_input_tokens_seen": 94371840,
81
+ "step": 45
82
+ },
83
+ {
84
+ "epoch": 0.10601643254704479,
85
+ "grad_norm": 0.2976844608783722,
86
+ "learning_rate": 4.9999382562611344e-05,
87
+ "loss": 3.2379,
88
+ "num_input_tokens_seen": 104857600,
89
+ "step": 50
90
+ },
91
+ {
92
+ "epoch": 0.11661807580174927,
93
+ "grad_norm": 0.32657739520072937,
94
+ "learning_rate": 4.999243674223826e-05,
95
+ "loss": 3.2281,
96
+ "num_input_tokens_seen": 115343360,
97
+ "step": 55
98
+ },
99
+ {
100
+ "epoch": 0.12721971905645374,
101
+ "grad_norm": 0.2382456362247467,
102
+ "learning_rate": 4.997777545616258e-05,
103
+ "loss": 3.2807,
104
+ "num_input_tokens_seen": 125829120,
105
+ "step": 60
106
+ },
107
+ {
108
+ "epoch": 0.13782136231115824,
109
+ "grad_norm": 0.4440530836582184,
110
+ "learning_rate": 4.99554032304996e-05,
111
+ "loss": 3.2192,
112
+ "num_input_tokens_seen": 136314880,
113
+ "step": 65
114
+ },
115
+ {
116
+ "epoch": 0.14842300556586271,
117
+ "grad_norm": 0.2103557139635086,
118
+ "learning_rate": 4.9925326971824345e-05,
119
+ "loss": 3.3152,
120
+ "num_input_tokens_seen": 146800640,
121
+ "step": 70
122
+ },
123
+ {
124
+ "epoch": 0.1590246488205672,
125
+ "grad_norm": 0.252693772315979,
126
+ "learning_rate": 4.988755596503948e-05,
127
+ "loss": 3.2393,
128
+ "num_input_tokens_seen": 157286400,
129
+ "step": 75
130
+ },
131
+ {
132
+ "epoch": 0.16962629207527166,
133
+ "grad_norm": 0.21133287250995636,
134
+ "learning_rate": 4.9842101870508904e-05,
135
+ "loss": 3.2727,
136
+ "num_input_tokens_seen": 167772160,
137
+ "step": 80
138
+ },
139
+ {
140
+ "epoch": 0.18022793532997614,
141
+ "grad_norm": 0.24876847863197327,
142
+ "learning_rate": 4.9788978720458104e-05,
143
+ "loss": 3.2566,
144
+ "num_input_tokens_seen": 178257920,
145
+ "step": 85
146
+ },
147
+ {
148
+ "epoch": 0.19082957858468064,
149
+ "grad_norm": 0.2968827486038208,
150
+ "learning_rate": 4.9728202914642183e-05,
151
+ "loss": 3.3258,
152
+ "num_input_tokens_seen": 188743680,
153
+ "step": 90
154
+ },
155
+ {
156
+ "epoch": 0.2014312218393851,
157
+ "grad_norm": 0.17700955271720886,
158
+ "learning_rate": 4.965979321528309e-05,
159
+ "loss": 3.2355,
160
+ "num_input_tokens_seen": 199229440,
161
+ "step": 95
162
+ },
163
+ {
164
+ "epoch": 0.21203286509408958,
165
+ "grad_norm": 0.20748387277126312,
166
+ "learning_rate": 4.9583770741277505e-05,
167
+ "loss": 3.195,
168
+ "num_input_tokens_seen": 209715200,
169
+ "step": 100
170
+ },
171
+ {
172
+ "epoch": 0.22263450834879406,
173
+ "grad_norm": 1.2878308296203613,
174
+ "learning_rate": 4.950015896167716e-05,
175
+ "loss": 3.3275,
176
+ "num_input_tokens_seen": 220200960,
177
+ "step": 105
178
+ },
179
+ {
180
+ "epoch": 0.23323615160349853,
181
+ "grad_norm": 0.26103636622428894,
182
+ "learning_rate": 4.9408983688443654e-05,
183
+ "loss": 3.3154,
184
+ "num_input_tokens_seen": 230686720,
185
+ "step": 110
186
+ },
187
+ {
188
+ "epoch": 0.24383779485820303,
189
+ "grad_norm": 0.2692287266254425,
190
+ "learning_rate": 4.931027306848004e-05,
191
+ "loss": 3.2573,
192
+ "num_input_tokens_seen": 241172480,
193
+ "step": 115
194
+ },
195
+ {
196
+ "epoch": 0.2544394381129075,
197
+ "grad_norm": 0.2487805187702179,
198
+ "learning_rate": 4.920405757494147e-05,
199
+ "loss": 3.2915,
200
+ "num_input_tokens_seen": 251658240,
201
+ "step": 120
202
+ },
203
+ {
204
+ "epoch": 0.26504108136761195,
205
+ "grad_norm": 0.17953425645828247,
206
+ "learning_rate": 4.9090369997827826e-05,
207
+ "loss": 3.1941,
208
+ "num_input_tokens_seen": 262144000,
209
+ "step": 125
210
+ },
211
+ {
212
+ "epoch": 0.2756427246223165,
213
+ "grad_norm": 0.19957853853702545,
214
+ "learning_rate": 4.896924543386099e-05,
215
+ "loss": 3.3033,
216
+ "num_input_tokens_seen": 272629760,
217
+ "step": 130
218
+ },
219
+ {
220
+ "epoch": 0.28624436787702096,
221
+ "grad_norm": 0.22015607357025146,
222
+ "learning_rate": 4.884072127565014e-05,
223
+ "loss": 3.2327,
224
+ "num_input_tokens_seen": 283115520,
225
+ "step": 135
226
+ },
227
+ {
228
+ "epoch": 0.29684601113172543,
229
+ "grad_norm": 0.241306334733963,
230
+ "learning_rate": 4.870483720014814e-05,
231
+ "loss": 3.2124,
232
+ "num_input_tokens_seen": 293601280,
233
+ "step": 140
234
+ },
235
+ {
236
+ "epoch": 0.3074476543864299,
237
+ "grad_norm": 0.49794742465019226,
238
+ "learning_rate": 4.85616351564028e-05,
239
+ "loss": 3.312,
240
+ "num_input_tokens_seen": 304087040,
241
+ "step": 145
242
+ },
243
+ {
244
+ "epoch": 0.3180492976411344,
245
+ "grad_norm": 0.18301662802696228,
246
+ "learning_rate": 4.8411159352606734e-05,
247
+ "loss": 3.295,
248
+ "num_input_tokens_seen": 314572800,
249
+ "step": 150
250
+ },
251
+ {
252
+ "epoch": 0.32865094089583885,
253
+ "grad_norm": 0.23481640219688416,
254
+ "learning_rate": 4.8253456242449704e-05,
255
+ "loss": 3.2555,
256
+ "num_input_tokens_seen": 325058560,
257
+ "step": 155
258
+ },
259
+ {
260
+ "epoch": 0.3392525841505433,
261
+ "grad_norm": 0.2195325493812561,
262
+ "learning_rate": 4.808857451077788e-05,
263
+ "loss": 3.1667,
264
+ "num_input_tokens_seen": 335544320,
265
+ "step": 160
266
+ },
267
+ {
268
+ "epoch": 0.3498542274052478,
269
+ "grad_norm": 0.23738664388656616,
270
+ "learning_rate": 4.7916565058564155e-05,
271
+ "loss": 3.2499,
272
+ "num_input_tokens_seen": 346030080,
273
+ "step": 165
274
+ },
275
+ {
276
+ "epoch": 0.36045587065995227,
277
+ "grad_norm": 0.3056640028953552,
278
+ "learning_rate": 4.7737480987194484e-05,
279
+ "loss": 3.2478,
280
+ "num_input_tokens_seen": 356515840,
281
+ "step": 170
282
+ },
283
+ {
284
+ "epoch": 0.37105751391465674,
285
+ "grad_norm": 0.18291524052619934,
286
+ "learning_rate": 4.755137758207479e-05,
287
+ "loss": 3.2648,
288
+ "num_input_tokens_seen": 367001600,
289
+ "step": 175
290
+ },
291
+ {
292
+ "epoch": 0.3816591571693613,
293
+ "grad_norm": 0.20183615386486053,
294
+ "learning_rate": 4.7358312295563734e-05,
295
+ "loss": 3.1717,
296
+ "num_input_tokens_seen": 377487360,
297
+ "step": 180
298
+ },
299
+ {
300
+ "epoch": 0.39226080042406575,
301
+ "grad_norm": 0.19222132861614227,
302
+ "learning_rate": 4.7158344729236454e-05,
303
+ "loss": 3.2288,
304
+ "num_input_tokens_seen": 387973120,
305
+ "step": 185
306
+ },
307
+ {
308
+ "epoch": 0.4028624436787702,
309
+ "grad_norm": 0.16234181821346283,
310
+ "learning_rate": 4.6951536615484854e-05,
311
+ "loss": 3.2316,
312
+ "num_input_tokens_seen": 398458880,
313
+ "step": 190
314
+ },
315
+ {
316
+ "epoch": 0.4134640869334747,
317
+ "grad_norm": 0.21459302306175232,
318
+ "learning_rate": 4.673795179846007e-05,
319
+ "loss": 3.1703,
320
+ "num_input_tokens_seen": 408944640,
321
+ "step": 195
322
+ },
323
+ {
324
+ "epoch": 0.42406573018817917,
325
+ "grad_norm": 0.1751609593629837,
326
+ "learning_rate": 4.651765621436303e-05,
327
+ "loss": 3.2013,
328
+ "num_input_tokens_seen": 419430400,
329
+ "step": 200
330
+ },
331
+ {
332
+ "epoch": 0.42406573018817917,
333
+ "eval_accuracy": 0.44791084254003133,
334
+ "eval_loss": 3.0653164386749268,
335
+ "eval_runtime": 36.089,
336
+ "eval_samples_per_second": 8.313,
337
+ "eval_steps_per_second": 2.078,
338
+ "num_input_tokens_seen": 419430400,
339
+ "step": 200
340
+ },
341
+ {
342
+ "epoch": 0.43466737344288364,
343
+ "grad_norm": 0.15440785884857178,
344
+ "learning_rate": 4.62907178710891e-05,
345
+ "loss": 3.2764,
346
+ "num_input_tokens_seen": 429916160,
347
+ "step": 205
348
+ },
349
+ {
350
+ "epoch": 0.4452690166975881,
351
+ "grad_norm": 0.19598723948001862,
352
+ "learning_rate": 4.60572068272333e-05,
353
+ "loss": 3.2128,
354
+ "num_input_tokens_seen": 440401920,
355
+ "step": 210
356
+ },
357
+ {
358
+ "epoch": 0.4558706599522926,
359
+ "grad_norm": 0.21214532852172852,
360
+ "learning_rate": 4.581719517046236e-05,
361
+ "loss": 3.1353,
362
+ "num_input_tokens_seen": 450887680,
363
+ "step": 215
364
+ },
365
+ {
366
+ "epoch": 0.46647230320699706,
367
+ "grad_norm": 0.181034654378891,
368
+ "learning_rate": 4.557075699526032e-05,
369
+ "loss": 3.2866,
370
+ "num_input_tokens_seen": 461373440,
371
+ "step": 220
372
+ },
373
+ {
374
+ "epoch": 0.47707394646170154,
375
+ "grad_norm": 0.16727791726589203,
376
+ "learning_rate": 4.531796838005477e-05,
377
+ "loss": 3.2111,
378
+ "num_input_tokens_seen": 471859200,
379
+ "step": 225
380
+ },
381
+ {
382
+ "epoch": 0.48767558971640607,
383
+ "grad_norm": 0.1804332137107849,
384
+ "learning_rate": 4.505890736373045e-05,
385
+ "loss": 3.297,
386
+ "num_input_tokens_seen": 482344960,
387
+ "step": 230
388
+ },
389
+ {
390
+ "epoch": 0.49827723297111054,
391
+ "grad_norm": 0.21492359042167664,
392
+ "learning_rate": 4.479365392153776e-05,
393
+ "loss": 3.1203,
394
+ "num_input_tokens_seen": 492830720,
395
+ "step": 235
396
+ },
397
+ {
398
+ "epoch": 0.508878876225815,
399
+ "grad_norm": 0.2664985656738281,
400
+ "learning_rate": 4.4522289940403404e-05,
401
+ "loss": 3.1352,
402
+ "num_input_tokens_seen": 503316480,
403
+ "step": 240
404
+ },
405
+ {
406
+ "epoch": 0.5194805194805194,
407
+ "grad_norm": 0.18706196546554565,
408
+ "learning_rate": 4.4244899193650933e-05,
409
+ "loss": 3.1747,
410
+ "num_input_tokens_seen": 513802240,
411
+ "step": 245
412
+ },
413
+ {
414
+ "epoch": 0.5300821627352239,
415
+ "grad_norm": 0.2427309900522232,
416
+ "learning_rate": 4.3961567315138885e-05,
417
+ "loss": 3.1677,
418
+ "num_input_tokens_seen": 524288000,
419
+ "step": 250
420
+ },
421
+ {
422
+ "epoch": 0.5406838059899285,
423
+ "grad_norm": 0.17540065944194794,
424
+ "learning_rate": 4.3672381772824615e-05,
425
+ "loss": 3.2012,
426
+ "num_input_tokens_seen": 534773760,
427
+ "step": 255
428
+ },
429
+ {
430
+ "epoch": 0.551285449244633,
431
+ "grad_norm": 0.31619423627853394,
432
+ "learning_rate": 4.3377431841761875e-05,
433
+ "loss": 3.1498,
434
+ "num_input_tokens_seen": 545259520,
435
+ "step": 260
436
+ },
437
+ {
438
+ "epoch": 0.5618870924993374,
439
+ "grad_norm": 0.24162785708904266,
440
+ "learning_rate": 4.307680857654052e-05,
441
+ "loss": 3.1703,
442
+ "num_input_tokens_seen": 555745280,
443
+ "step": 265
444
+ },
445
+ {
446
+ "epoch": 0.5724887357540419,
447
+ "grad_norm": 0.22926253080368042,
448
+ "learning_rate": 4.277060478317687e-05,
449
+ "loss": 3.2395,
450
+ "num_input_tokens_seen": 566231040,
451
+ "step": 270
452
+ },
453
+ {
454
+ "epoch": 0.5830903790087464,
455
+ "grad_norm": 0.23989850282669067,
456
+ "learning_rate": 4.245891499046338e-05,
457
+ "loss": 3.2383,
458
+ "num_input_tokens_seen": 576716800,
459
+ "step": 275
460
+ },
461
+ {
462
+ "epoch": 0.5936920222634509,
463
+ "grad_norm": 0.20367790758609772,
464
+ "learning_rate": 4.214183542078646e-05,
465
+ "loss": 3.1489,
466
+ "num_input_tokens_seen": 587202560,
467
+ "step": 280
468
+ },
469
+ {
470
+ "epoch": 0.6042936655181553,
471
+ "grad_norm": 0.22007331252098083,
472
+ "learning_rate": 4.1819463960421454e-05,
473
+ "loss": 3.1707,
474
+ "num_input_tokens_seen": 597688320,
475
+ "step": 285
476
+ },
477
+ {
478
+ "epoch": 0.6148953087728598,
479
+ "grad_norm": 0.19839680194854736,
480
+ "learning_rate": 4.149190012931402e-05,
481
+ "loss": 3.1679,
482
+ "num_input_tokens_seen": 608174080,
483
+ "step": 290
484
+ },
485
+ {
486
+ "epoch": 0.6254969520275643,
487
+ "grad_norm": 0.23358359932899475,
488
+ "learning_rate": 4.1159245050357065e-05,
489
+ "loss": 3.1629,
490
+ "num_input_tokens_seen": 618659840,
491
+ "step": 295
492
+ },
493
+ {
494
+ "epoch": 0.6360985952822688,
495
+ "grad_norm": 0.16982465982437134,
496
+ "learning_rate": 4.082160141817293e-05,
497
+ "loss": 3.16,
498
+ "num_input_tokens_seen": 629145600,
499
+ "step": 300
500
+ },
501
+ {
502
+ "epoch": 0.6467002385369732,
503
+ "grad_norm": 0.3546576201915741,
504
+ "learning_rate": 4.0479073467410286e-05,
505
+ "loss": 3.2448,
506
+ "num_input_tokens_seen": 639631360,
507
+ "step": 305
508
+ },
509
+ {
510
+ "epoch": 0.6573018817916777,
511
+ "grad_norm": 0.16487599909305573,
512
+ "learning_rate": 4.0131766940565715e-05,
513
+ "loss": 3.1789,
514
+ "num_input_tokens_seen": 650117120,
515
+ "step": 310
516
+ },
517
+ {
518
+ "epoch": 0.6679035250463822,
519
+ "grad_norm": 0.26232755184173584,
520
+ "learning_rate": 3.9779789055339656e-05,
521
+ "loss": 3.1994,
522
+ "num_input_tokens_seen": 660602880,
523
+ "step": 315
524
+ },
525
+ {
526
+ "epoch": 0.6785051683010866,
527
+ "grad_norm": 0.17949418723583221,
528
+ "learning_rate": 3.9423248471537065e-05,
529
+ "loss": 3.1464,
530
+ "num_input_tokens_seen": 671088640,
531
+ "step": 320
532
+ },
533
+ {
534
+ "epoch": 0.6891068115557911,
535
+ "grad_norm": 0.22808890044689178,
536
+ "learning_rate": 3.9062255257522794e-05,
537
+ "loss": 3.2207,
538
+ "num_input_tokens_seen": 681574400,
539
+ "step": 325
540
+ },
541
+ {
542
+ "epoch": 0.6997084548104956,
543
+ "grad_norm": 0.20621077716350555,
544
+ "learning_rate": 3.8696920856242174e-05,
545
+ "loss": 3.1688,
546
+ "num_input_tokens_seen": 692060160,
547
+ "step": 330
548
+ },
549
+ {
550
+ "epoch": 0.7103100980652001,
551
+ "grad_norm": 0.20659266412258148,
552
+ "learning_rate": 3.8327358050817234e-05,
553
+ "loss": 3.1726,
554
+ "num_input_tokens_seen": 702545920,
555
+ "step": 335
556
+ },
557
+ {
558
+ "epoch": 0.7209117413199045,
559
+ "grad_norm": 0.23239997029304504,
560
+ "learning_rate": 3.7953680929729215e-05,
561
+ "loss": 3.2254,
562
+ "num_input_tokens_seen": 713031680,
563
+ "step": 340
564
+ },
565
+ {
566
+ "epoch": 0.731513384574609,
567
+ "grad_norm": 0.15853458642959595,
568
+ "learning_rate": 3.757600485159805e-05,
569
+ "loss": 3.1453,
570
+ "num_input_tokens_seen": 723517440,
571
+ "step": 345
572
+ },
573
+ {
574
+ "epoch": 0.7421150278293135,
575
+ "grad_norm": 0.28123387694358826,
576
+ "learning_rate": 3.719444640956981e-05,
577
+ "loss": 3.1849,
578
+ "num_input_tokens_seen": 734003200,
579
+ "step": 350
580
+ },
581
+ {
582
+ "epoch": 0.7527166710840181,
583
+ "grad_norm": 0.16030676662921906,
584
+ "learning_rate": 3.680912339532296e-05,
585
+ "loss": 3.187,
586
+ "num_input_tokens_seen": 744488960,
587
+ "step": 355
588
+ },
589
+ {
590
+ "epoch": 0.7633183143387225,
591
+ "grad_norm": 0.1698087602853775,
592
+ "learning_rate": 3.6420154762704686e-05,
593
+ "loss": 3.1603,
594
+ "num_input_tokens_seen": 754974720,
595
+ "step": 360
596
+ },
597
+ {
598
+ "epoch": 0.773919957593427,
599
+ "grad_norm": 0.1804792284965515,
600
+ "learning_rate": 3.602766059100838e-05,
601
+ "loss": 3.1501,
602
+ "num_input_tokens_seen": 765460480,
603
+ "step": 365
604
+ },
605
+ {
606
+ "epoch": 0.7845216008481315,
607
+ "grad_norm": 0.22060543298721313,
608
+ "learning_rate": 3.563176204790374e-05,
609
+ "loss": 3.2135,
610
+ "num_input_tokens_seen": 775946240,
611
+ "step": 370
612
+ },
613
+ {
614
+ "epoch": 0.795123244102836,
615
+ "grad_norm": 0.2411007434129715,
616
+ "learning_rate": 3.523258135203087e-05,
617
+ "loss": 3.1263,
618
+ "num_input_tokens_seen": 786432000,
619
+ "step": 375
620
+ },
621
+ {
622
+ "epoch": 0.8057248873575404,
623
+ "grad_norm": 0.17039397358894348,
624
+ "learning_rate": 3.483024173526985e-05,
625
+ "loss": 3.1278,
626
+ "num_input_tokens_seen": 796917760,
627
+ "step": 380
628
+ },
629
+ {
630
+ "epoch": 0.8163265306122449,
631
+ "grad_norm": 0.17096886038780212,
632
+ "learning_rate": 3.442486740469766e-05,
633
+ "loss": 3.1805,
634
+ "num_input_tokens_seen": 807403520,
635
+ "step": 385
636
+ },
637
+ {
638
+ "epoch": 0.8269281738669494,
639
+ "grad_norm": 0.20285804569721222,
640
+ "learning_rate": 3.401658350424389e-05,
641
+ "loss": 3.1093,
642
+ "num_input_tokens_seen": 817889280,
643
+ "step": 390
644
+ },
645
+ {
646
+ "epoch": 0.8375298171216539,
647
+ "grad_norm": 0.17684973776340485,
648
+ "learning_rate": 3.360551607605735e-05,
649
+ "loss": 3.1336,
650
+ "num_input_tokens_seen": 828375040,
651
+ "step": 395
652
+ },
653
+ {
654
+ "epoch": 0.8481314603763583,
655
+ "grad_norm": 0.16076605021953583,
656
+ "learning_rate": 3.3191792021595316e-05,
657
+ "loss": 3.1976,
658
+ "num_input_tokens_seen": 838860800,
659
+ "step": 400
660
+ },
661
+ {
662
+ "epoch": 0.8481314603763583,
663
+ "eval_accuracy": 0.4506036745406824,
664
+ "eval_loss": 3.043421983718872,
665
+ "eval_runtime": 36.2661,
666
+ "eval_samples_per_second": 8.272,
667
+ "eval_steps_per_second": 2.068,
668
+ "num_input_tokens_seen": 838860800,
669
+ "step": 400
670
+ },
671
+ {
672
+ "epoch": 0.8587331036310628,
673
+ "grad_norm": 0.1959790736436844,
674
+ "learning_rate": 3.277553906244756e-05,
675
+ "loss": 3.2013,
676
+ "num_input_tokens_seen": 849346560,
677
+ "step": 405
678
+ },
679
+ {
680
+ "epoch": 0.8693347468857673,
681
+ "grad_norm": 0.15397347509860992,
682
+ "learning_rate": 3.23568857009071e-05,
683
+ "loss": 3.124,
684
+ "num_input_tokens_seen": 859832320,
685
+ "step": 410
686
+ },
687
+ {
688
+ "epoch": 0.8799363901404718,
689
+ "grad_norm": 0.22286681830883026,
690
+ "learning_rate": 3.193596118030005e-05,
691
+ "loss": 3.1723,
692
+ "num_input_tokens_seen": 870318080,
693
+ "step": 415
694
+ },
695
+ {
696
+ "epoch": 0.8905380333951762,
697
+ "grad_norm": 0.22909095883369446,
698
+ "learning_rate": 3.1512895445086636e-05,
699
+ "loss": 3.1068,
700
+ "num_input_tokens_seen": 880803840,
701
+ "step": 420
702
+ },
703
+ {
704
+ "epoch": 0.9011396766498807,
705
+ "grad_norm": 0.24190495908260345,
706
+ "learning_rate": 3.108781910074578e-05,
707
+ "loss": 3.143,
708
+ "num_input_tokens_seen": 891289600,
709
+ "step": 425
710
+ },
711
+ {
712
+ "epoch": 0.9117413199045852,
713
+ "grad_norm": 0.24453699588775635,
714
+ "learning_rate": 3.0660863373455595e-05,
715
+ "loss": 3.1667,
716
+ "num_input_tokens_seen": 901775360,
717
+ "step": 430
718
+ },
719
+ {
720
+ "epoch": 0.9223429631592897,
721
+ "grad_norm": 0.1636463701725006,
722
+ "learning_rate": 3.0232160069582332e-05,
723
+ "loss": 3.1123,
724
+ "num_input_tokens_seen": 912261120,
725
+ "step": 435
726
+ },
727
+ {
728
+ "epoch": 0.9329446064139941,
729
+ "grad_norm": 0.14330972731113434,
730
+ "learning_rate": 2.9801841534990115e-05,
731
+ "loss": 3.1616,
732
+ "num_input_tokens_seen": 922746880,
733
+ "step": 440
734
+ },
735
+ {
736
+ "epoch": 0.9435462496686986,
737
+ "grad_norm": 0.14314919710159302,
738
+ "learning_rate": 2.9370040614184245e-05,
739
+ "loss": 3.1241,
740
+ "num_input_tokens_seen": 933232640,
741
+ "step": 445
742
+ },
743
+ {
744
+ "epoch": 0.9541478929234031,
745
+ "grad_norm": 0.15261518955230713,
746
+ "learning_rate": 2.893689060930045e-05,
747
+ "loss": 3.1573,
748
+ "num_input_tokens_seen": 943718400,
749
+ "step": 450
750
+ },
751
+ {
752
+ "epoch": 0.9647495361781077,
753
+ "grad_norm": 0.14767758548259735,
754
+ "learning_rate": 2.8502525238952916e-05,
755
+ "loss": 3.2165,
756
+ "num_input_tokens_seen": 954204160,
757
+ "step": 455
758
+ },
759
+ {
760
+ "epoch": 0.9753511794328121,
761
+ "grad_norm": 0.161110982298851,
762
+ "learning_rate": 2.8067078596953796e-05,
763
+ "loss": 3.1457,
764
+ "num_input_tokens_seen": 964689920,
765
+ "step": 460
766
+ },
767
+ {
768
+ "epoch": 0.9859528226875166,
769
+ "grad_norm": 0.15843427181243896,
770
+ "learning_rate": 2.7630685110916778e-05,
771
+ "loss": 3.143,
772
+ "num_input_tokens_seen": 975175680,
773
+ "step": 465
774
+ },
775
+ {
776
+ "epoch": 0.9965544659422211,
777
+ "grad_norm": 0.24187405407428741,
778
+ "learning_rate": 2.7193479500757685e-05,
779
+ "loss": 3.2156,
780
+ "num_input_tokens_seen": 985661440,
781
+ "step": 470
782
+ },
783
+ {
784
+ "epoch": 1.0071561091969254,
785
+ "grad_norm": 0.15035971999168396,
786
+ "learning_rate": 2.675559673710485e-05,
787
+ "loss": 3.0852,
788
+ "num_input_tokens_seen": 996147200,
789
+ "step": 475
790
+ },
791
+ {
792
+ "epoch": 1.01775775245163,
793
+ "grad_norm": 0.17717382311820984,
794
+ "learning_rate": 2.631717199963199e-05,
795
+ "loss": 3.1221,
796
+ "num_input_tokens_seen": 1006632960,
797
+ "step": 480
798
+ },
799
+ {
800
+ "epoch": 1.0283593957063344,
801
+ "grad_norm": 0.1390335112810135,
802
+ "learning_rate": 2.5878340635326686e-05,
803
+ "loss": 3.1358,
804
+ "num_input_tokens_seen": 1017118720,
805
+ "step": 485
806
+ },
807
+ {
808
+ "epoch": 1.0389610389610389,
809
+ "grad_norm": 0.14454010128974915,
810
+ "learning_rate": 2.5439238116707102e-05,
811
+ "loss": 3.1307,
812
+ "num_input_tokens_seen": 1027604480,
813
+ "step": 490
814
+ },
815
+ {
816
+ "epoch": 1.0495626822157433,
817
+ "grad_norm": 0.19984786212444305,
818
+ "learning_rate": 2.5e-05,
819
+ "loss": 3.1269,
820
+ "num_input_tokens_seen": 1038090240,
821
+ "step": 495
822
+ },
823
+ {
824
+ "epoch": 1.0601643254704478,
825
+ "grad_norm": 0.263656347990036,
826
+ "learning_rate": 2.45607618832929e-05,
827
+ "loss": 3.1076,
828
+ "num_input_tokens_seen": 1048576000,
829
+ "step": 500
830
+ },
831
+ {
832
+ "epoch": 1.0707659687251523,
833
+ "grad_norm": 0.15432648360729218,
834
+ "learning_rate": 2.412165936467332e-05,
835
+ "loss": 3.1218,
836
+ "num_input_tokens_seen": 1059061760,
837
+ "step": 505
838
+ },
839
+ {
840
+ "epoch": 1.081367611979857,
841
+ "grad_norm": 0.24721471965312958,
842
+ "learning_rate": 2.368282800036801e-05,
843
+ "loss": 3.1035,
844
+ "num_input_tokens_seen": 1069547520,
845
+ "step": 510
846
+ },
847
+ {
848
+ "epoch": 1.0919692552345615,
849
+ "grad_norm": 0.1441834270954132,
850
+ "learning_rate": 2.3244403262895153e-05,
851
+ "loss": 3.1801,
852
+ "num_input_tokens_seen": 1080033280,
853
+ "step": 515
854
+ },
855
+ {
856
+ "epoch": 1.102570898489266,
857
+ "grad_norm": 0.1535598337650299,
858
+ "learning_rate": 2.280652049924232e-05,
859
+ "loss": 3.2002,
860
+ "num_input_tokens_seen": 1090519040,
861
+ "step": 520
862
+ },
863
+ {
864
+ "epoch": 1.1131725417439704,
865
+ "grad_norm": 0.15322482585906982,
866
+ "learning_rate": 2.2369314889083235e-05,
867
+ "loss": 3.1208,
868
+ "num_input_tokens_seen": 1101004800,
869
+ "step": 525
870
+ },
871
+ {
872
+ "epoch": 1.1237741849986749,
873
+ "grad_norm": 0.1468251496553421,
874
+ "learning_rate": 2.1932921403046207e-05,
875
+ "loss": 3.1917,
876
+ "num_input_tokens_seen": 1111490560,
877
+ "step": 530
878
+ },
879
+ {
880
+ "epoch": 1.1343758282533793,
881
+ "grad_norm": 0.1530437469482422,
882
+ "learning_rate": 2.1497474761047086e-05,
883
+ "loss": 3.0797,
884
+ "num_input_tokens_seen": 1121976320,
885
+ "step": 535
886
+ },
887
+ {
888
+ "epoch": 1.1449774715080838,
889
+ "grad_norm": 0.1411171704530716,
890
+ "learning_rate": 2.106310939069956e-05,
891
+ "loss": 3.1136,
892
+ "num_input_tokens_seen": 1132462080,
893
+ "step": 540
894
+ },
895
+ {
896
+ "epoch": 1.1555791147627883,
897
+ "grad_norm": 0.14499180018901825,
898
+ "learning_rate": 2.0629959385815757e-05,
899
+ "loss": 3.1427,
900
+ "num_input_tokens_seen": 1142947840,
901
+ "step": 545
902
+ },
903
+ {
904
+ "epoch": 1.1661807580174928,
905
+ "grad_norm": 0.14953380823135376,
906
+ "learning_rate": 2.019815846500988e-05,
907
+ "loss": 3.1948,
908
+ "num_input_tokens_seen": 1153433600,
909
+ "step": 550
910
+ },
911
+ {
912
+ "epoch": 1.1767824012721972,
913
+ "grad_norm": 0.1349666863679886,
914
+ "learning_rate": 1.9767839930417673e-05,
915
+ "loss": 3.1605,
916
+ "num_input_tokens_seen": 1163919360,
917
+ "step": 555
918
+ },
919
+ {
920
+ "epoch": 1.1873840445269017,
921
+ "grad_norm": 0.144802987575531,
922
+ "learning_rate": 1.9339136626544407e-05,
923
+ "loss": 3.1857,
924
+ "num_input_tokens_seen": 1174405120,
925
+ "step": 560
926
+ },
927
+ {
928
+ "epoch": 1.1979856877816062,
929
+ "grad_norm": 0.16848242282867432,
930
+ "learning_rate": 1.891218089925423e-05,
931
+ "loss": 3.18,
932
+ "num_input_tokens_seen": 1184890880,
933
+ "step": 565
934
+ },
935
+ {
936
+ "epoch": 1.2085873310363107,
937
+ "grad_norm": 0.15472151339054108,
938
+ "learning_rate": 1.8487104554913363e-05,
939
+ "loss": 3.1939,
940
+ "num_input_tokens_seen": 1195376640,
941
+ "step": 570
942
+ },
943
+ {
944
+ "epoch": 1.2191889742910151,
945
+ "grad_norm": 0.1529063582420349,
946
+ "learning_rate": 1.806403881969996e-05,
947
+ "loss": 3.142,
948
+ "num_input_tokens_seen": 1205862400,
949
+ "step": 575
950
+ },
951
+ {
952
+ "epoch": 1.2297906175457196,
953
+ "grad_norm": 0.1687614619731903,
954
+ "learning_rate": 1.764311429909291e-05,
955
+ "loss": 3.1344,
956
+ "num_input_tokens_seen": 1216348160,
957
+ "step": 580
958
+ },
959
+ {
960
+ "epoch": 1.240392260800424,
961
+ "grad_norm": 0.13812561333179474,
962
+ "learning_rate": 1.7224460937552446e-05,
963
+ "loss": 3.1626,
964
+ "num_input_tokens_seen": 1226833920,
965
+ "step": 585
966
+ },
967
+ {
968
+ "epoch": 1.2509939040551286,
969
+ "grad_norm": 0.16453585028648376,
970
+ "learning_rate": 1.6808207978404683e-05,
971
+ "loss": 3.1616,
972
+ "num_input_tokens_seen": 1237319680,
973
+ "step": 590
974
+ },
975
+ {
976
+ "epoch": 1.261595547309833,
977
+ "grad_norm": 0.13337315618991852,
978
+ "learning_rate": 1.6394483923942655e-05,
979
+ "loss": 3.1413,
980
+ "num_input_tokens_seen": 1247805440,
981
+ "step": 595
982
+ },
983
+ {
984
+ "epoch": 1.2721971905645375,
985
+ "grad_norm": 0.2029571831226349,
986
+ "learning_rate": 1.5983416495756116e-05,
987
+ "loss": 3.1485,
988
+ "num_input_tokens_seen": 1258291200,
989
+ "step": 600
990
+ },
991
+ {
992
+ "epoch": 1.2721971905645375,
993
+ "eval_accuracy": 0.45126899835195017,
994
+ "eval_loss": 3.0374794006347656,
995
+ "eval_runtime": 36.1798,
996
+ "eval_samples_per_second": 8.292,
997
+ "eval_steps_per_second": 2.073,
998
+ "num_input_tokens_seen": 1258291200,
999
+ "step": 600
1000
+ },
1001
+ {
1002
+ "epoch": 1.282798833819242,
1003
+ "grad_norm": 0.21333394944667816,
1004
+ "learning_rate": 1.5575132595302352e-05,
1005
+ "loss": 3.1131,
1006
+ "num_input_tokens_seen": 1268776960,
1007
+ "step": 605
1008
+ },
1009
+ {
1010
+ "epoch": 1.2934004770739465,
1011
+ "grad_norm": 0.1743020862340927,
1012
+ "learning_rate": 1.516975826473015e-05,
1013
+ "loss": 3.1177,
1014
+ "num_input_tokens_seen": 1279262720,
1015
+ "step": 610
1016
+ },
1017
+ {
1018
+ "epoch": 1.304002120328651,
1019
+ "grad_norm": 0.15571139752864838,
1020
+ "learning_rate": 1.4767418647969133e-05,
1021
+ "loss": 3.1626,
1022
+ "num_input_tokens_seen": 1289748480,
1023
+ "step": 615
1024
+ },
1025
+ {
1026
+ "epoch": 1.3146037635833554,
1027
+ "grad_norm": 0.16648022830486298,
1028
+ "learning_rate": 1.4368237952096258e-05,
1029
+ "loss": 3.1987,
1030
+ "num_input_tokens_seen": 1300234240,
1031
+ "step": 620
1032
+ },
1033
+ {
1034
+ "epoch": 1.3252054068380599,
1035
+ "grad_norm": 0.2865128219127655,
1036
+ "learning_rate": 1.3972339408991626e-05,
1037
+ "loss": 3.176,
1038
+ "num_input_tokens_seen": 1310720000,
1039
+ "step": 625
1040
+ },
1041
+ {
1042
+ "epoch": 1.3358070500927643,
1043
+ "grad_norm": 0.14998364448547363,
1044
+ "learning_rate": 1.357984523729533e-05,
1045
+ "loss": 3.1568,
1046
+ "num_input_tokens_seen": 1321205760,
1047
+ "step": 630
1048
+ },
1049
+ {
1050
+ "epoch": 1.3464086933474688,
1051
+ "grad_norm": 0.14780345559120178,
1052
+ "learning_rate": 1.3190876604677043e-05,
1053
+ "loss": 3.1895,
1054
+ "num_input_tokens_seen": 1331691520,
1055
+ "step": 635
1056
+ },
1057
+ {
1058
+ "epoch": 1.3570103366021733,
1059
+ "grad_norm": 0.15796662867069244,
1060
+ "learning_rate": 1.2805553590430197e-05,
1061
+ "loss": 3.1708,
1062
+ "num_input_tokens_seen": 1342177280,
1063
+ "step": 640
1064
+ },
1065
+ {
1066
+ "epoch": 1.3676119798568778,
1067
+ "grad_norm": 0.14260126650333405,
1068
+ "learning_rate": 1.2423995148401954e-05,
1069
+ "loss": 3.1649,
1070
+ "num_input_tokens_seen": 1352663040,
1071
+ "step": 645
1072
+ },
1073
+ {
1074
+ "epoch": 1.3782136231115822,
1075
+ "grad_norm": 0.16667188704013824,
1076
+ "learning_rate": 1.2046319070270792e-05,
1077
+ "loss": 3.1345,
1078
+ "num_input_tokens_seen": 1363148800,
1079
+ "step": 650
1080
+ },
1081
+ {
1082
+ "epoch": 1.3888152663662867,
1083
+ "grad_norm": 0.13283655047416687,
1084
+ "learning_rate": 1.1672641949182769e-05,
1085
+ "loss": 3.1653,
1086
+ "num_input_tokens_seen": 1373634560,
1087
+ "step": 655
1088
+ },
1089
+ {
1090
+ "epoch": 1.3994169096209912,
1091
+ "grad_norm": 0.1479809582233429,
1092
+ "learning_rate": 1.130307914375783e-05,
1093
+ "loss": 3.2014,
1094
+ "num_input_tokens_seen": 1384120320,
1095
+ "step": 660
1096
+ },
1097
+ {
1098
+ "epoch": 1.4100185528756957,
1099
+ "grad_norm": 0.1443646103143692,
1100
+ "learning_rate": 1.093774474247721e-05,
1101
+ "loss": 3.1561,
1102
+ "num_input_tokens_seen": 1394606080,
1103
+ "step": 665
1104
+ },
1105
+ {
1106
+ "epoch": 1.4206201961304001,
1107
+ "grad_norm": 0.15595485270023346,
1108
+ "learning_rate": 1.0576751528462935e-05,
1109
+ "loss": 3.1774,
1110
+ "num_input_tokens_seen": 1405091840,
1111
+ "step": 670
1112
+ },
1113
+ {
1114
+ "epoch": 1.4312218393851046,
1115
+ "grad_norm": 0.1376049816608429,
1116
+ "learning_rate": 1.0220210944660338e-05,
1117
+ "loss": 3.1102,
1118
+ "num_input_tokens_seen": 1415577600,
1119
+ "step": 675
1120
+ },
1121
+ {
1122
+ "epoch": 1.4418234826398093,
1123
+ "grad_norm": 0.14685603976249695,
1124
+ "learning_rate": 9.868233059434288e-06,
1125
+ "loss": 3.2077,
1126
+ "num_input_tokens_seen": 1426063360,
1127
+ "step": 680
1128
+ },
1129
+ {
1130
+ "epoch": 1.4524251258945138,
1131
+ "grad_norm": 0.13503292202949524,
1132
+ "learning_rate": 9.520926532589725e-06,
1133
+ "loss": 3.1899,
1134
+ "num_input_tokens_seen": 1436549120,
1135
+ "step": 685
1136
+ },
1137
+ {
1138
+ "epoch": 1.4630267691492183,
1139
+ "grad_norm": 0.18555864691734314,
1140
+ "learning_rate": 9.178398581827085e-06,
1141
+ "loss": 3.2424,
1142
+ "num_input_tokens_seen": 1447034880,
1143
+ "step": 690
1144
+ },
1145
+ {
1146
+ "epoch": 1.4736284124039227,
1147
+ "grad_norm": 0.26128634810447693,
1148
+ "learning_rate": 8.840754949642935e-06,
1149
+ "loss": 3.1609,
1150
+ "num_input_tokens_seen": 1457520640,
1151
+ "step": 695
1152
+ },
1153
+ {
1154
+ "epoch": 1.4842300556586272,
1155
+ "grad_norm": 0.13384391367435455,
1156
+ "learning_rate": 8.50809987068598e-06,
1157
+ "loss": 3.1455,
1158
+ "num_input_tokens_seen": 1468006400,
1159
+ "step": 700
1160
+ },
1161
+ {
1162
+ "epoch": 1.4948316989133317,
1163
+ "grad_norm": 0.20566639304161072,
1164
+ "learning_rate": 8.180536039578545e-06,
1165
+ "loss": 3.1995,
1166
+ "num_input_tokens_seen": 1478492160,
1167
+ "step": 705
1168
+ },
1169
+ {
1170
+ "epoch": 1.5054333421680361,
1171
+ "grad_norm": 0.13730423152446747,
1172
+ "learning_rate": 7.858164579213547e-06,
1173
+ "loss": 3.134,
1174
+ "num_input_tokens_seen": 1488977920,
1175
+ "step": 710
1176
+ },
1177
+ {
1178
+ "epoch": 1.5160349854227406,
1179
+ "grad_norm": 0.14655794203281403,
1180
+ "learning_rate": 7.541085009536625e-06,
1181
+ "loss": 3.0949,
1182
+ "num_input_tokens_seen": 1499463680,
1183
+ "step": 715
1184
+ },
1185
+ {
1186
+ "epoch": 1.526636628677445,
1187
+ "grad_norm": 0.12526313960552216,
1188
+ "learning_rate": 7.2293952168231316e-06,
1189
+ "loss": 3.1615,
1190
+ "num_input_tokens_seen": 1509949440,
1191
+ "step": 720
1192
+ },
1193
+ {
1194
+ "epoch": 1.5372382719321496,
1195
+ "grad_norm": 0.1351662427186966,
1196
+ "learning_rate": 6.923191423459482e-06,
1197
+ "loss": 3.1805,
1198
+ "num_input_tokens_seen": 1520435200,
1199
+ "step": 725
1200
+ },
1201
+ {
1202
+ "epoch": 1.547839915186854,
1203
+ "grad_norm": 0.1686793714761734,
1204
+ "learning_rate": 6.622568158238126e-06,
1205
+ "loss": 3.1779,
1206
+ "num_input_tokens_seen": 1530920960,
1207
+ "step": 730
1208
+ },
1209
+ {
1210
+ "epoch": 1.5584415584415585,
1211
+ "grad_norm": 0.14492954313755035,
1212
+ "learning_rate": 6.327618227175389e-06,
1213
+ "loss": 3.1046,
1214
+ "num_input_tokens_seen": 1541406720,
1215
+ "step": 735
1216
+ },
1217
+ {
1218
+ "epoch": 1.569043201696263,
1219
+ "grad_norm": 0.14910686016082764,
1220
+ "learning_rate": 6.0384326848611225e-06,
1221
+ "loss": 3.0803,
1222
+ "num_input_tokens_seen": 1551892480,
1223
+ "step": 740
1224
+ },
1225
+ {
1226
+ "epoch": 1.5796448449509675,
1227
+ "grad_norm": 0.1342857927083969,
1228
+ "learning_rate": 5.755100806349076e-06,
1229
+ "loss": 3.062,
1230
+ "num_input_tokens_seen": 1562378240,
1231
+ "step": 745
1232
+ },
1233
+ {
1234
+ "epoch": 1.590246488205672,
1235
+ "grad_norm": 0.1598527878522873,
1236
+ "learning_rate": 5.4777100595965994e-06,
1237
+ "loss": 3.2036,
1238
+ "num_input_tokens_seen": 1572864000,
1239
+ "step": 750
1240
+ },
1241
+ {
1242
+ "epoch": 1.6008481314603764,
1243
+ "grad_norm": 0.1773335188627243,
1244
+ "learning_rate": 5.206346078462249e-06,
1245
+ "loss": 3.1892,
1246
+ "num_input_tokens_seen": 1583349760,
1247
+ "step": 755
1248
+ },
1249
+ {
1250
+ "epoch": 1.6114497747150809,
1251
+ "grad_norm": 0.17586533725261688,
1252
+ "learning_rate": 4.941092636269554e-06,
1253
+ "loss": 3.1995,
1254
+ "num_input_tokens_seen": 1593835520,
1255
+ "step": 760
1256
+ },
1257
+ {
1258
+ "epoch": 1.6220514179697854,
1259
+ "grad_norm": 0.16583110392093658,
1260
+ "learning_rate": 4.682031619945238e-06,
1261
+ "loss": 3.1309,
1262
+ "num_input_tokens_seen": 1604321280,
1263
+ "step": 765
1264
+ },
1265
+ {
1266
+ "epoch": 1.6326530612244898,
1267
+ "grad_norm": 0.1462288796901703,
1268
+ "learning_rate": 4.4292430047396914e-06,
1269
+ "loss": 3.1376,
1270
+ "num_input_tokens_seen": 1614807040,
1271
+ "step": 770
1272
+ },
1273
+ {
1274
+ "epoch": 1.6432547044791943,
1275
+ "grad_norm": 0.13279996812343597,
1276
+ "learning_rate": 4.182804829537654e-06,
1277
+ "loss": 3.1541,
1278
+ "num_input_tokens_seen": 1625292800,
1279
+ "step": 775
1280
+ },
1281
+ {
1282
+ "epoch": 1.6538563477338988,
1283
+ "grad_norm": 0.18102097511291504,
1284
+ "learning_rate": 3.942793172766699e-06,
1285
+ "loss": 3.1894,
1286
+ "num_input_tokens_seen": 1635778560,
1287
+ "step": 780
1288
+ },
1289
+ {
1290
+ "epoch": 1.6644579909886033,
1291
+ "grad_norm": 0.16354134678840637,
1292
+ "learning_rate": 3.709282128910907e-06,
1293
+ "loss": 3.1664,
1294
+ "num_input_tokens_seen": 1646264320,
1295
+ "step": 785
1296
+ },
1297
+ {
1298
+ "epoch": 1.6750596342433077,
1299
+ "grad_norm": 0.32594841718673706,
1300
+ "learning_rate": 3.4823437856369794e-06,
1301
+ "loss": 3.1721,
1302
+ "num_input_tokens_seen": 1656750080,
1303
+ "step": 790
1304
+ },
1305
+ {
1306
+ "epoch": 1.6856612774980122,
1307
+ "grad_norm": 0.13957597315311432,
1308
+ "learning_rate": 3.2620482015399302e-06,
1309
+ "loss": 3.2471,
1310
+ "num_input_tokens_seen": 1667235840,
1311
+ "step": 795
1312
+ },
1313
+ {
1314
+ "epoch": 1.6962629207527167,
1315
+ "grad_norm": 0.13924147188663483,
1316
+ "learning_rate": 3.0484633845151488e-06,
1317
+ "loss": 3.1871,
1318
+ "num_input_tokens_seen": 1677721600,
1319
+ "step": 800
1320
+ },
1321
+ {
1322
+ "epoch": 1.6962629207527167,
1323
+ "eval_accuracy": 0.4514018596512645,
1324
+ "eval_loss": 3.0366272926330566,
1325
+ "eval_runtime": 36.0138,
1326
+ "eval_samples_per_second": 8.33,
1327
+ "eval_steps_per_second": 2.083,
1328
+ "num_input_tokens_seen": 1677721600,
1329
+ "step": 800
1330
+ },
1331
+ {
1332
+ "epoch": 1.7068645640074211,
1333
+ "grad_norm": 0.14513356983661652,
1334
+ "learning_rate": 2.841655270763549e-06,
1335
+ "loss": 3.066,
1336
+ "num_input_tokens_seen": 1688207360,
1337
+ "step": 805
1338
+ },
1339
+ {
1340
+ "epoch": 1.7174662072621256,
1341
+ "grad_norm": 0.14862333238124847,
1342
+ "learning_rate": 2.6416877044362685e-06,
1343
+ "loss": 3.1206,
1344
+ "num_input_tokens_seen": 1698693120,
1345
+ "step": 810
1346
+ },
1347
+ {
1348
+ "epoch": 1.72806785051683,
1349
+ "grad_norm": 0.13608209788799286,
1350
+ "learning_rate": 2.448622417925214e-06,
1351
+ "loss": 3.2281,
1352
+ "num_input_tokens_seen": 1709178880,
1353
+ "step": 815
1354
+ },
1355
+ {
1356
+ "epoch": 1.7386694937715346,
1357
+ "grad_norm": 0.1472754031419754,
1358
+ "learning_rate": 2.2625190128055168e-06,
1359
+ "loss": 3.2023,
1360
+ "num_input_tokens_seen": 1719664640,
1361
+ "step": 820
1362
+ },
1363
+ {
1364
+ "epoch": 1.749271137026239,
1365
+ "grad_norm": 0.17672620713710785,
1366
+ "learning_rate": 2.0834349414358495e-06,
1367
+ "loss": 3.2217,
1368
+ "num_input_tokens_seen": 1730150400,
1369
+ "step": 825
1370
+ },
1371
+ {
1372
+ "epoch": 1.7598727802809435,
1373
+ "grad_norm": 0.14799125492572784,
1374
+ "learning_rate": 1.911425489222127e-06,
1375
+ "loss": 3.2231,
1376
+ "num_input_tokens_seen": 1740636160,
1377
+ "step": 830
1378
+ },
1379
+ {
1380
+ "epoch": 1.770474423535648,
1381
+ "grad_norm": 0.14448776841163635,
1382
+ "learning_rate": 1.7465437575502952e-06,
1383
+ "loss": 3.1126,
1384
+ "num_input_tokens_seen": 1751121920,
1385
+ "step": 835
1386
+ },
1387
+ {
1388
+ "epoch": 1.7810760667903525,
1389
+ "grad_norm": 0.14082539081573486,
1390
+ "learning_rate": 1.5888406473932692e-06,
1391
+ "loss": 3.0933,
1392
+ "num_input_tokens_seen": 1761607680,
1393
+ "step": 840
1394
+ },
1395
+ {
1396
+ "epoch": 1.791677710045057,
1397
+ "grad_norm": 0.18886087834835052,
1398
+ "learning_rate": 1.4383648435972007e-06,
1399
+ "loss": 3.2596,
1400
+ "num_input_tokens_seen": 1772093440,
1401
+ "step": 845
1402
+ },
1403
+ {
1404
+ "epoch": 1.8022793532997614,
1405
+ "grad_norm": 0.14818327128887177,
1406
+ "learning_rate": 1.2951627998518623e-06,
1407
+ "loss": 3.1292,
1408
+ "num_input_tokens_seen": 1782579200,
1409
+ "step": 850
1410
+ },
1411
+ {
1412
+ "epoch": 1.8128809965544659,
1413
+ "grad_norm": 0.14874930679798126,
1414
+ "learning_rate": 1.1592787243498631e-06,
1415
+ "loss": 3.1284,
1416
+ "num_input_tokens_seen": 1793064960,
1417
+ "step": 855
1418
+ },
1419
+ {
1420
+ "epoch": 1.8234826398091704,
1421
+ "grad_norm": 0.16616405546665192,
1422
+ "learning_rate": 1.0307545661390139e-06,
1423
+ "loss": 3.1422,
1424
+ "num_input_tokens_seen": 1803550720,
1425
+ "step": 860
1426
+ },
1427
+ {
1428
+ "epoch": 1.8340842830638748,
1429
+ "grad_norm": 0.15211746096611023,
1430
+ "learning_rate": 9.09630002172182e-07,
1431
+ "loss": 3.1793,
1432
+ "num_input_tokens_seen": 1814036480,
1433
+ "step": 865
1434
+ },
1435
+ {
1436
+ "epoch": 1.8446859263185793,
1437
+ "grad_norm": 0.14233224093914032,
1438
+ "learning_rate": 7.959424250585323e-07,
1439
+ "loss": 3.1624,
1440
+ "num_input_tokens_seen": 1824522240,
1441
+ "step": 870
1442
+ },
1443
+ {
1444
+ "epoch": 1.8552875695732838,
1445
+ "grad_norm": 0.13079190254211426,
1446
+ "learning_rate": 6.897269315199628e-07,
1447
+ "loss": 3.2029,
1448
+ "num_input_tokens_seen": 1835008000,
1449
+ "step": 875
1450
+ },
1451
+ {
1452
+ "epoch": 1.8658892128279883,
1453
+ "grad_norm": 0.1364782601594925,
1454
+ "learning_rate": 5.910163115563471e-07,
1455
+ "loss": 3.1872,
1456
+ "num_input_tokens_seen": 1845493760,
1457
+ "step": 880
1458
+ },
1459
+ {
1460
+ "epoch": 1.8764908560826927,
1461
+ "grad_norm": 0.22711250185966492,
1462
+ "learning_rate": 4.998410383228458e-07,
1463
+ "loss": 3.1676,
1464
+ "num_input_tokens_seen": 1855979520,
1465
+ "step": 885
1466
+ },
1467
+ {
1468
+ "epoch": 1.8870924993373972,
1469
+ "grad_norm": 0.21140585839748383,
1470
+ "learning_rate": 4.162292587224947e-07,
1471
+ "loss": 3.1227,
1472
+ "num_input_tokens_seen": 1866465280,
1473
+ "step": 890
1474
+ },
1475
+ {
1476
+ "epoch": 1.8976941425921017,
1477
+ "grad_norm": 0.2973221242427826,
1478
+ "learning_rate": 3.4020678471690934e-07,
1479
+ "loss": 3.1285,
1480
+ "num_input_tokens_seen": 1876951040,
1481
+ "step": 895
1482
+ },
1483
+ {
1484
+ "epoch": 1.9082957858468061,
1485
+ "grad_norm": 0.15094168484210968,
1486
+ "learning_rate": 2.7179708535781943e-07,
1487
+ "loss": 3.152,
1488
+ "num_input_tokens_seen": 1887436800,
1489
+ "step": 900
1490
+ },
1491
+ {
1492
+ "epoch": 1.9188974291015106,
1493
+ "grad_norm": 0.14399217069149017,
1494
+ "learning_rate": 2.110212795418953e-07,
1495
+ "loss": 3.1971,
1496
+ "num_input_tokens_seen": 1897922560,
1497
+ "step": 905
1498
+ },
1499
+ {
1500
+ "epoch": 1.929499072356215,
1501
+ "grad_norm": 0.15141591429710388,
1502
+ "learning_rate": 1.578981294910936e-07,
1503
+ "loss": 3.1604,
1504
+ "num_input_tokens_seen": 1908408320,
1505
+ "step": 910
1506
+ },
1507
+ {
1508
+ "epoch": 1.9401007156109196,
1509
+ "grad_norm": 0.15116803348064423,
1510
+ "learning_rate": 1.1244403496052658e-07,
1511
+ "loss": 3.0712,
1512
+ "num_input_tokens_seen": 1918894080,
1513
+ "step": 915
1514
+ },
1515
+ {
1516
+ "epoch": 1.950702358865624,
1517
+ "grad_norm": 0.1515650451183319,
1518
+ "learning_rate": 7.46730281756619e-08,
1519
+ "loss": 3.0719,
1520
+ "num_input_tokens_seen": 1929379840,
1521
+ "step": 920
1522
+ },
1523
+ {
1524
+ "epoch": 1.9613040021203285,
1525
+ "grad_norm": 0.1448003053665161,
1526
+ "learning_rate": 4.4596769500407366e-08,
1527
+ "loss": 3.2699,
1528
+ "num_input_tokens_seen": 1939865600,
1529
+ "step": 925
1530
+ },
1531
+ {
1532
+ "epoch": 1.971905645375033,
1533
+ "grad_norm": 0.18116047978401184,
1534
+ "learning_rate": 2.2224543837423562e-08,
1535
+ "loss": 3.1134,
1536
+ "num_input_tokens_seen": 1950351360,
1537
+ "step": 930
1538
+ },
1539
+ {
1540
+ "epoch": 1.9825072886297375,
1541
+ "grad_norm": 0.3333103358745575,
1542
+ "learning_rate": 7.563257761744601e-09,
1543
+ "loss": 3.1083,
1544
+ "num_input_tokens_seen": 1960837120,
1545
+ "step": 935
1546
+ },
1547
+ {
1548
+ "epoch": 1.993108931884442,
1549
+ "grad_norm": 0.15106463432312012,
1550
+ "learning_rate": 6.174373886586037e-10,
1551
+ "loss": 3.2087,
1552
+ "num_input_tokens_seen": 1971322880,
1553
+ "step": 940
1554
+ },
1555
+ {
1556
+ "epoch": 1.997349589186324,
1557
+ "num_input_tokens_seen": 1975517184,
1558
+ "step": 942,
1559
+ "total_flos": 9.732316690586272e+18,
1560
+ "train_loss": 3.186332613039928,
1561
+ "train_runtime": 34789.3198,
1562
+ "train_samples_per_second": 3.47,
1563
+ "train_steps_per_second": 0.027
1564
+ }
1565
+ ],
1566
+ "logging_steps": 5,
1567
+ "max_steps": 942,
1568
+ "num_input_tokens_seen": 1975517184,
1569
+ "num_train_epochs": 2,
1570
+ "save_steps": 100,
1571
+ "total_flos": 9.732316690586272e+18,
1572
+ "train_batch_size": 4,
1573
+ "trial_name": null,
1574
+ "trial_params": null
1575
+ }