antonpolishko commited on
Commit
7537637
·
verified ·
1 Parent(s): 21cd8f4

Model save

Browse files
README.md ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: google/gemma-2-9b
3
+ library_name: transformers
4
+ model_name: gemma-2-9b-overfit-ua
5
+ tags:
6
+ - generated_from_trainer
7
+ - trl
8
+ - sft
9
+ licence: license
10
+ ---
11
+
12
+ # Model Card for gemma-2-9b-overfit-ua
13
+
14
+ This model is a fine-tuned version of [google/gemma-2-9b](https://huggingface.co/google/gemma-2-9b).
15
+ It has been trained using [TRL](https://github.com/huggingface/trl).
16
+
17
+ ## Quick start
18
+
19
+ ```python
20
+ from transformers import pipeline
21
+
22
+ question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
23
+ generator = pipeline("text-generation", model="antonpolishko/gemma-2-9b-overfit-ua", device="cuda")
24
+ output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
25
+ print(output["generated_text"])
26
+ ```
27
+
28
+ ## Training procedure
29
+
30
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/polyagent/huggingface/runs/bm5sjqkj)
31
+
32
+ This model was trained with SFT.
33
+
34
+ ### Framework versions
35
+
36
+ - TRL: 0.12.1
37
+ - Transformers: 4.46.3
38
+ - Pytorch: 2.6.0.dev20241113+rocm6.2
39
+ - Datasets: 3.1.0
40
+ - Tokenizers: 0.20.3
41
+
42
+ ## Citations
43
+
44
+
45
+
46
+ Cite TRL as:
47
+
48
+ ```bibtex
49
+ @misc{vonwerra2022trl,
50
+ title = {{TRL: Transformer Reinforcement Learning}},
51
+ author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
52
+ year = 2020,
53
+ journal = {GitHub repository},
54
+ publisher = {GitHub},
55
+ howpublished = {\url{https://github.com/huggingface/trl}}
56
+ }
57
+ ```
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 7.187967826039144e+18,
4
+ "train_loss": 1.5352961912832626,
5
+ "train_runtime": 6906.6629,
6
+ "train_samples": 95663,
7
+ "train_samples_per_second": 10.159,
8
+ "train_steps_per_second": 0.159
9
+ }
generation_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 2,
4
+ "cache_implementation": "hybrid",
5
+ "eos_token_id": 1,
6
+ "pad_token_id": 0,
7
+ "transformers_version": "4.46.3"
8
+ }
runs/Dec26_12-49-24_mia1-gpu-109/events.out.tfevents.1735217422.mia1-gpu-109.3741323.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ca898dee7bc5598eb359a8b99bbd7eb416e4a8c73be50f880c7376fde881e4c8
3
- size 53072
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ea7495af360f8178bec6659203e26a255f5e8f05d0b602e18a38f45972aac57
3
+ size 53426
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 7.187967826039144e+18,
4
+ "train_loss": 1.5352961912832626,
5
+ "train_runtime": 6906.6629,
6
+ "train_samples": 95663,
7
+ "train_samples_per_second": 10.159,
8
+ "train_steps_per_second": 0.159
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1606 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0,
5
+ "eval_steps": 300,
6
+ "global_step": 1098,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.00273224043715847,
13
+ "grad_norm": 84.5,
14
+ "learning_rate": 1e-06,
15
+ "loss": 2.8857,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.01366120218579235,
20
+ "grad_norm": 40.75,
21
+ "learning_rate": 1.9999630266778663e-06,
22
+ "loss": 2.6209,
23
+ "step": 5
24
+ },
25
+ {
26
+ "epoch": 0.0273224043715847,
27
+ "grad_norm": 16.875,
28
+ "learning_rate": 1.999737088499184e-06,
29
+ "loss": 2.1772,
30
+ "step": 10
31
+ },
32
+ {
33
+ "epoch": 0.040983606557377046,
34
+ "grad_norm": 9.75,
35
+ "learning_rate": 1.999305799228933e-06,
36
+ "loss": 1.96,
37
+ "step": 15
38
+ },
39
+ {
40
+ "epoch": 0.0546448087431694,
41
+ "grad_norm": 6.03125,
42
+ "learning_rate": 1.998669247456129e-06,
43
+ "loss": 1.8595,
44
+ "step": 20
45
+ },
46
+ {
47
+ "epoch": 0.06830601092896176,
48
+ "grad_norm": 4.8125,
49
+ "learning_rate": 1.997827563931747e-06,
50
+ "loss": 1.7983,
51
+ "step": 25
52
+ },
53
+ {
54
+ "epoch": 0.08196721311475409,
55
+ "grad_norm": 3.90625,
56
+ "learning_rate": 1.9967809215418726e-06,
57
+ "loss": 1.7237,
58
+ "step": 30
59
+ },
60
+ {
61
+ "epoch": 0.09562841530054644,
62
+ "grad_norm": 2.765625,
63
+ "learning_rate": 1.9955295352721854e-06,
64
+ "loss": 1.6815,
65
+ "step": 35
66
+ },
67
+ {
68
+ "epoch": 0.1092896174863388,
69
+ "grad_norm": 2.625,
70
+ "learning_rate": 1.9940736621637997e-06,
71
+ "loss": 1.6637,
72
+ "step": 40
73
+ },
74
+ {
75
+ "epoch": 0.12295081967213115,
76
+ "grad_norm": 2.484375,
77
+ "learning_rate": 1.992413601260471e-06,
78
+ "loss": 1.6501,
79
+ "step": 45
80
+ },
81
+ {
82
+ "epoch": 0.1366120218579235,
83
+ "grad_norm": 2.171875,
84
+ "learning_rate": 1.990549693547166e-06,
85
+ "loss": 1.6316,
86
+ "step": 50
87
+ },
88
+ {
89
+ "epoch": 0.15027322404371585,
90
+ "grad_norm": 1.984375,
91
+ "learning_rate": 1.988482321880025e-06,
92
+ "loss": 1.6138,
93
+ "step": 55
94
+ },
95
+ {
96
+ "epoch": 0.16393442622950818,
97
+ "grad_norm": 2.28125,
98
+ "learning_rate": 1.9862119109077224e-06,
99
+ "loss": 1.6096,
100
+ "step": 60
101
+ },
102
+ {
103
+ "epoch": 0.17759562841530055,
104
+ "grad_norm": 1.921875,
105
+ "learning_rate": 1.983738926984239e-06,
106
+ "loss": 1.5883,
107
+ "step": 65
108
+ },
109
+ {
110
+ "epoch": 0.1912568306010929,
111
+ "grad_norm": 1.8828125,
112
+ "learning_rate": 1.9810638780730727e-06,
113
+ "loss": 1.5932,
114
+ "step": 70
115
+ },
116
+ {
117
+ "epoch": 0.20491803278688525,
118
+ "grad_norm": 2.078125,
119
+ "learning_rate": 1.9781873136428984e-06,
120
+ "loss": 1.5829,
121
+ "step": 75
122
+ },
123
+ {
124
+ "epoch": 0.2185792349726776,
125
+ "grad_norm": 1.9140625,
126
+ "learning_rate": 1.975109824554707e-06,
127
+ "loss": 1.5835,
128
+ "step": 80
129
+ },
130
+ {
131
+ "epoch": 0.23224043715846995,
132
+ "grad_norm": 1.828125,
133
+ "learning_rate": 1.971832042940436e-06,
134
+ "loss": 1.5746,
135
+ "step": 85
136
+ },
137
+ {
138
+ "epoch": 0.2459016393442623,
139
+ "grad_norm": 1.8046875,
140
+ "learning_rate": 1.968354642073129e-06,
141
+ "loss": 1.5765,
142
+ "step": 90
143
+ },
144
+ {
145
+ "epoch": 0.25956284153005466,
146
+ "grad_norm": 1.75,
147
+ "learning_rate": 1.964678336228642e-06,
148
+ "loss": 1.5685,
149
+ "step": 95
150
+ },
151
+ {
152
+ "epoch": 0.273224043715847,
153
+ "grad_norm": 1.8671875,
154
+ "learning_rate": 1.960803880538925e-06,
155
+ "loss": 1.564,
156
+ "step": 100
157
+ },
158
+ {
159
+ "epoch": 0.28688524590163933,
160
+ "grad_norm": 1.8046875,
161
+ "learning_rate": 1.9567320708369176e-06,
162
+ "loss": 1.5642,
163
+ "step": 105
164
+ },
165
+ {
166
+ "epoch": 0.3005464480874317,
167
+ "grad_norm": 1.8984375,
168
+ "learning_rate": 1.9524637434930776e-06,
169
+ "loss": 1.5723,
170
+ "step": 110
171
+ },
172
+ {
173
+ "epoch": 0.31420765027322406,
174
+ "grad_norm": 1.828125,
175
+ "learning_rate": 1.9479997752435884e-06,
176
+ "loss": 1.5712,
177
+ "step": 115
178
+ },
179
+ {
180
+ "epoch": 0.32786885245901637,
181
+ "grad_norm": 1.921875,
182
+ "learning_rate": 1.943341083010272e-06,
183
+ "loss": 1.5609,
184
+ "step": 120
185
+ },
186
+ {
187
+ "epoch": 0.34153005464480873,
188
+ "grad_norm": 1.6796875,
189
+ "learning_rate": 1.9384886237122493e-06,
190
+ "loss": 1.5596,
191
+ "step": 125
192
+ },
193
+ {
194
+ "epoch": 0.3551912568306011,
195
+ "grad_norm": 1.90625,
196
+ "learning_rate": 1.9334433940693826e-06,
197
+ "loss": 1.5582,
198
+ "step": 130
199
+ },
200
+ {
201
+ "epoch": 0.36885245901639346,
202
+ "grad_norm": 1.6875,
203
+ "learning_rate": 1.928206430397546e-06,
204
+ "loss": 1.5554,
205
+ "step": 135
206
+ },
207
+ {
208
+ "epoch": 0.3825136612021858,
209
+ "grad_norm": 1.71875,
210
+ "learning_rate": 1.9227788083957586e-06,
211
+ "loss": 1.5653,
212
+ "step": 140
213
+ },
214
+ {
215
+ "epoch": 0.39617486338797814,
216
+ "grad_norm": 1.6875,
217
+ "learning_rate": 1.9171616429252344e-06,
218
+ "loss": 1.5515,
219
+ "step": 145
220
+ },
221
+ {
222
+ "epoch": 0.4098360655737705,
223
+ "grad_norm": 1.6796875,
224
+ "learning_rate": 1.9113560877803796e-06,
225
+ "loss": 1.5509,
226
+ "step": 150
227
+ },
228
+ {
229
+ "epoch": 0.42349726775956287,
230
+ "grad_norm": 1.65625,
231
+ "learning_rate": 1.9053633354517988e-06,
232
+ "loss": 1.5523,
233
+ "step": 155
234
+ },
235
+ {
236
+ "epoch": 0.4371584699453552,
237
+ "grad_norm": 1.6875,
238
+ "learning_rate": 1.8991846168813544e-06,
239
+ "loss": 1.5588,
240
+ "step": 160
241
+ },
242
+ {
243
+ "epoch": 0.45081967213114754,
244
+ "grad_norm": 1.7578125,
245
+ "learning_rate": 1.89282120120932e-06,
246
+ "loss": 1.5507,
247
+ "step": 165
248
+ },
249
+ {
250
+ "epoch": 0.4644808743169399,
251
+ "grad_norm": 1.7109375,
252
+ "learning_rate": 1.8862743955136963e-06,
253
+ "loss": 1.5467,
254
+ "step": 170
255
+ },
256
+ {
257
+ "epoch": 0.4781420765027322,
258
+ "grad_norm": 1.6875,
259
+ "learning_rate": 1.8795455445417286e-06,
260
+ "loss": 1.5491,
261
+ "step": 175
262
+ },
263
+ {
264
+ "epoch": 0.4918032786885246,
265
+ "grad_norm": 1.65625,
266
+ "learning_rate": 1.8726360304336893e-06,
267
+ "loss": 1.5416,
268
+ "step": 180
269
+ },
270
+ {
271
+ "epoch": 0.505464480874317,
272
+ "grad_norm": 1.6796875,
273
+ "learning_rate": 1.8655472724389796e-06,
274
+ "loss": 1.5483,
275
+ "step": 185
276
+ },
277
+ {
278
+ "epoch": 0.5191256830601093,
279
+ "grad_norm": 1.65625,
280
+ "learning_rate": 1.858280726624609e-06,
281
+ "loss": 1.5543,
282
+ "step": 190
283
+ },
284
+ {
285
+ "epoch": 0.5327868852459017,
286
+ "grad_norm": 1.71875,
287
+ "learning_rate": 1.8508378855761095e-06,
288
+ "loss": 1.5393,
289
+ "step": 195
290
+ },
291
+ {
292
+ "epoch": 0.546448087431694,
293
+ "grad_norm": 1.7265625,
294
+ "learning_rate": 1.8432202780909538e-06,
295
+ "loss": 1.5435,
296
+ "step": 200
297
+ },
298
+ {
299
+ "epoch": 0.5601092896174863,
300
+ "grad_norm": 1.7421875,
301
+ "learning_rate": 1.83542946886453e-06,
302
+ "loss": 1.549,
303
+ "step": 205
304
+ },
305
+ {
306
+ "epoch": 0.5737704918032787,
307
+ "grad_norm": 1.75,
308
+ "learning_rate": 1.8274670581687478e-06,
309
+ "loss": 1.5422,
310
+ "step": 210
311
+ },
312
+ {
313
+ "epoch": 0.587431693989071,
314
+ "grad_norm": 1.671875,
315
+ "learning_rate": 1.819334681523331e-06,
316
+ "loss": 1.5413,
317
+ "step": 215
318
+ },
319
+ {
320
+ "epoch": 0.6010928961748634,
321
+ "grad_norm": 1.671875,
322
+ "learning_rate": 1.811034009359877e-06,
323
+ "loss": 1.5403,
324
+ "step": 220
325
+ },
326
+ {
327
+ "epoch": 0.6147540983606558,
328
+ "grad_norm": 1.703125,
329
+ "learning_rate": 1.8025667466787391e-06,
330
+ "loss": 1.5293,
331
+ "step": 225
332
+ },
333
+ {
334
+ "epoch": 0.6284153005464481,
335
+ "grad_norm": 1.6953125,
336
+ "learning_rate": 1.7939346326988125e-06,
337
+ "loss": 1.5472,
338
+ "step": 230
339
+ },
340
+ {
341
+ "epoch": 0.6420765027322405,
342
+ "grad_norm": 1.6953125,
343
+ "learning_rate": 1.7851394405002884e-06,
344
+ "loss": 1.5436,
345
+ "step": 235
346
+ },
347
+ {
348
+ "epoch": 0.6557377049180327,
349
+ "grad_norm": 1.6875,
350
+ "learning_rate": 1.7761829766604554e-06,
351
+ "loss": 1.5341,
352
+ "step": 240
353
+ },
354
+ {
355
+ "epoch": 0.6693989071038251,
356
+ "grad_norm": 1.7421875,
357
+ "learning_rate": 1.7670670808826191e-06,
358
+ "loss": 1.5287,
359
+ "step": 245
360
+ },
361
+ {
362
+ "epoch": 0.6830601092896175,
363
+ "grad_norm": 1.703125,
364
+ "learning_rate": 1.7577936256182167e-06,
365
+ "loss": 1.5389,
366
+ "step": 250
367
+ },
368
+ {
369
+ "epoch": 0.6967213114754098,
370
+ "grad_norm": 1.6875,
371
+ "learning_rate": 1.7483645156822069e-06,
372
+ "loss": 1.5431,
373
+ "step": 255
374
+ },
375
+ {
376
+ "epoch": 0.7103825136612022,
377
+ "grad_norm": 1.7265625,
378
+ "learning_rate": 1.7387816878618117e-06,
379
+ "loss": 1.5424,
380
+ "step": 260
381
+ },
382
+ {
383
+ "epoch": 0.7240437158469946,
384
+ "grad_norm": 1.703125,
385
+ "learning_rate": 1.729047110518689e-06,
386
+ "loss": 1.5294,
387
+ "step": 265
388
+ },
389
+ {
390
+ "epoch": 0.7377049180327869,
391
+ "grad_norm": 1.7109375,
392
+ "learning_rate": 1.7191627831846222e-06,
393
+ "loss": 1.5256,
394
+ "step": 270
395
+ },
396
+ {
397
+ "epoch": 0.7513661202185792,
398
+ "grad_norm": 1.8515625,
399
+ "learning_rate": 1.7091307361508055e-06,
400
+ "loss": 1.5359,
401
+ "step": 275
402
+ },
403
+ {
404
+ "epoch": 0.7650273224043715,
405
+ "grad_norm": 1.640625,
406
+ "learning_rate": 1.6989530300508123e-06,
407
+ "loss": 1.5318,
408
+ "step": 280
409
+ },
410
+ {
411
+ "epoch": 0.7786885245901639,
412
+ "grad_norm": 1.78125,
413
+ "learning_rate": 1.6886317554373302e-06,
414
+ "loss": 1.5344,
415
+ "step": 285
416
+ },
417
+ {
418
+ "epoch": 0.7923497267759563,
419
+ "grad_norm": 1.65625,
420
+ "learning_rate": 1.6781690323527509e-06,
421
+ "loss": 1.5272,
422
+ "step": 290
423
+ },
424
+ {
425
+ "epoch": 0.8060109289617486,
426
+ "grad_norm": 1.6875,
427
+ "learning_rate": 1.6675670098937032e-06,
428
+ "loss": 1.5293,
429
+ "step": 295
430
+ },
431
+ {
432
+ "epoch": 0.819672131147541,
433
+ "grad_norm": 1.6484375,
434
+ "learning_rate": 1.6568278657696162e-06,
435
+ "loss": 1.5401,
436
+ "step": 300
437
+ },
438
+ {
439
+ "epoch": 0.819672131147541,
440
+ "eval_loss": 1.5333800315856934,
441
+ "eval_runtime": 19.9646,
442
+ "eval_samples_per_second": 35.463,
443
+ "eval_steps_per_second": 1.152,
444
+ "step": 300
445
+ },
446
+ {
447
+ "epoch": 0.8333333333333334,
448
+ "grad_norm": 1.6640625,
449
+ "learning_rate": 1.6459538058554087e-06,
450
+ "loss": 1.5176,
451
+ "step": 305
452
+ },
453
+ {
454
+ "epoch": 0.8469945355191257,
455
+ "grad_norm": 1.671875,
456
+ "learning_rate": 1.6349470637383888e-06,
457
+ "loss": 1.5312,
458
+ "step": 310
459
+ },
460
+ {
461
+ "epoch": 0.860655737704918,
462
+ "grad_norm": 1.65625,
463
+ "learning_rate": 1.6238099002594669e-06,
464
+ "loss": 1.5391,
465
+ "step": 315
466
+ },
467
+ {
468
+ "epoch": 0.8743169398907104,
469
+ "grad_norm": 1.7421875,
470
+ "learning_rate": 1.6125446030487642e-06,
471
+ "loss": 1.5248,
472
+ "step": 320
473
+ },
474
+ {
475
+ "epoch": 0.8879781420765027,
476
+ "grad_norm": 1.6875,
477
+ "learning_rate": 1.6011534860557236e-06,
478
+ "loss": 1.5381,
479
+ "step": 325
480
+ },
481
+ {
482
+ "epoch": 0.9016393442622951,
483
+ "grad_norm": 1.6484375,
484
+ "learning_rate": 1.5896388890738127e-06,
485
+ "loss": 1.5216,
486
+ "step": 330
487
+ },
488
+ {
489
+ "epoch": 0.9153005464480874,
490
+ "grad_norm": 1.6640625,
491
+ "learning_rate": 1.578003177259917e-06,
492
+ "loss": 1.524,
493
+ "step": 335
494
+ },
495
+ {
496
+ "epoch": 0.9289617486338798,
497
+ "grad_norm": 1.65625,
498
+ "learning_rate": 1.566248740648527e-06,
499
+ "loss": 1.5255,
500
+ "step": 340
501
+ },
502
+ {
503
+ "epoch": 0.9426229508196722,
504
+ "grad_norm": 1.671875,
505
+ "learning_rate": 1.5543779936608106e-06,
506
+ "loss": 1.5213,
507
+ "step": 345
508
+ },
509
+ {
510
+ "epoch": 0.9562841530054644,
511
+ "grad_norm": 1.6875,
512
+ "learning_rate": 1.5423933746086793e-06,
513
+ "loss": 1.517,
514
+ "step": 350
515
+ },
516
+ {
517
+ "epoch": 0.9699453551912568,
518
+ "grad_norm": 1.6796875,
519
+ "learning_rate": 1.5302973451939472e-06,
520
+ "loss": 1.524,
521
+ "step": 355
522
+ },
523
+ {
524
+ "epoch": 0.9836065573770492,
525
+ "grad_norm": 1.8203125,
526
+ "learning_rate": 1.5180923900026845e-06,
527
+ "loss": 1.5224,
528
+ "step": 360
529
+ },
530
+ {
531
+ "epoch": 0.9972677595628415,
532
+ "grad_norm": 1.6875,
533
+ "learning_rate": 1.5057810159948714e-06,
534
+ "loss": 1.5148,
535
+ "step": 365
536
+ },
537
+ {
538
+ "epoch": 1.010928961748634,
539
+ "grad_norm": 1.6328125,
540
+ "learning_rate": 1.493365751989454e-06,
541
+ "loss": 1.5175,
542
+ "step": 370
543
+ },
544
+ {
545
+ "epoch": 1.0245901639344261,
546
+ "grad_norm": 1.609375,
547
+ "learning_rate": 1.4808491481449143e-06,
548
+ "loss": 1.5084,
549
+ "step": 375
550
+ },
551
+ {
552
+ "epoch": 1.0382513661202186,
553
+ "grad_norm": 1.625,
554
+ "learning_rate": 1.4682337754354533e-06,
555
+ "loss": 1.5141,
556
+ "step": 380
557
+ },
558
+ {
559
+ "epoch": 1.0519125683060109,
560
+ "grad_norm": 1.6796875,
561
+ "learning_rate": 1.4555222251228995e-06,
562
+ "loss": 1.52,
563
+ "step": 385
564
+ },
565
+ {
566
+ "epoch": 1.0655737704918034,
567
+ "grad_norm": 1.671875,
568
+ "learning_rate": 1.442717108224452e-06,
569
+ "loss": 1.5167,
570
+ "step": 390
571
+ },
572
+ {
573
+ "epoch": 1.0792349726775956,
574
+ "grad_norm": 1.6640625,
575
+ "learning_rate": 1.4298210549763628e-06,
576
+ "loss": 1.5087,
577
+ "step": 395
578
+ },
579
+ {
580
+ "epoch": 1.092896174863388,
581
+ "grad_norm": 1.6875,
582
+ "learning_rate": 1.4168367142936734e-06,
583
+ "loss": 1.5049,
584
+ "step": 400
585
+ },
586
+ {
587
+ "epoch": 1.1065573770491803,
588
+ "grad_norm": 1.6796875,
589
+ "learning_rate": 1.4037667532261142e-06,
590
+ "loss": 1.5102,
591
+ "step": 405
592
+ },
593
+ {
594
+ "epoch": 1.1202185792349726,
595
+ "grad_norm": 1.671875,
596
+ "learning_rate": 1.3906138564102792e-06,
597
+ "loss": 1.5156,
598
+ "step": 410
599
+ },
600
+ {
601
+ "epoch": 1.133879781420765,
602
+ "grad_norm": 1.609375,
603
+ "learning_rate": 1.3773807255181874e-06,
604
+ "loss": 1.5162,
605
+ "step": 415
606
+ },
607
+ {
608
+ "epoch": 1.1475409836065573,
609
+ "grad_norm": 1.6171875,
610
+ "learning_rate": 1.3640700787023462e-06,
611
+ "loss": 1.5099,
612
+ "step": 420
613
+ },
614
+ {
615
+ "epoch": 1.1612021857923498,
616
+ "grad_norm": 1.6640625,
617
+ "learning_rate": 1.3506846500374283e-06,
618
+ "loss": 1.5159,
619
+ "step": 425
620
+ },
621
+ {
622
+ "epoch": 1.174863387978142,
623
+ "grad_norm": 1.671875,
624
+ "learning_rate": 1.337227188958679e-06,
625
+ "loss": 1.5145,
626
+ "step": 430
627
+ },
628
+ {
629
+ "epoch": 1.1885245901639343,
630
+ "grad_norm": 3.078125,
631
+ "learning_rate": 1.3237004596971687e-06,
632
+ "loss": 1.5066,
633
+ "step": 435
634
+ },
635
+ {
636
+ "epoch": 1.2021857923497268,
637
+ "grad_norm": 1.6484375,
638
+ "learning_rate": 1.3101072407120055e-06,
639
+ "loss": 1.5144,
640
+ "step": 440
641
+ },
642
+ {
643
+ "epoch": 1.215846994535519,
644
+ "grad_norm": 1.6953125,
645
+ "learning_rate": 1.2964503241196256e-06,
646
+ "loss": 1.5124,
647
+ "step": 445
648
+ },
649
+ {
650
+ "epoch": 1.2295081967213115,
651
+ "grad_norm": 1.65625,
652
+ "learning_rate": 1.2827325151202782e-06,
653
+ "loss": 1.5052,
654
+ "step": 450
655
+ },
656
+ {
657
+ "epoch": 1.2431693989071038,
658
+ "grad_norm": 1.7734375,
659
+ "learning_rate": 1.2689566314218228e-06,
660
+ "loss": 1.5103,
661
+ "step": 455
662
+ },
663
+ {
664
+ "epoch": 1.2568306010928962,
665
+ "grad_norm": 1.6953125,
666
+ "learning_rate": 1.255125502660958e-06,
667
+ "loss": 1.5078,
668
+ "step": 460
669
+ },
670
+ {
671
+ "epoch": 1.2704918032786885,
672
+ "grad_norm": 1.625,
673
+ "learning_rate": 1.2412419698220001e-06,
674
+ "loss": 1.5134,
675
+ "step": 465
676
+ },
677
+ {
678
+ "epoch": 1.2841530054644807,
679
+ "grad_norm": 1.640625,
680
+ "learning_rate": 1.2273088846533302e-06,
681
+ "loss": 1.5064,
682
+ "step": 470
683
+ },
684
+ {
685
+ "epoch": 1.2978142076502732,
686
+ "grad_norm": 1.6640625,
687
+ "learning_rate": 1.2133291090816295e-06,
688
+ "loss": 1.5124,
689
+ "step": 475
690
+ },
691
+ {
692
+ "epoch": 1.3114754098360657,
693
+ "grad_norm": 1.609375,
694
+ "learning_rate": 1.1993055146240272e-06,
695
+ "loss": 1.5144,
696
+ "step": 480
697
+ },
698
+ {
699
+ "epoch": 1.325136612021858,
700
+ "grad_norm": 1.65625,
701
+ "learning_rate": 1.185240981798273e-06,
702
+ "loss": 1.5221,
703
+ "step": 485
704
+ },
705
+ {
706
+ "epoch": 1.3387978142076502,
707
+ "grad_norm": 1.671875,
708
+ "learning_rate": 1.171138399531068e-06,
709
+ "loss": 1.5079,
710
+ "step": 490
711
+ },
712
+ {
713
+ "epoch": 1.3524590163934427,
714
+ "grad_norm": 1.6640625,
715
+ "learning_rate": 1.1570006645646614e-06,
716
+ "loss": 1.5145,
717
+ "step": 495
718
+ },
719
+ {
720
+ "epoch": 1.366120218579235,
721
+ "grad_norm": 1.640625,
722
+ "learning_rate": 1.1428306808618454e-06,
723
+ "loss": 1.5043,
724
+ "step": 500
725
+ },
726
+ {
727
+ "epoch": 1.3797814207650272,
728
+ "grad_norm": 1.671875,
729
+ "learning_rate": 1.1286313590094686e-06,
730
+ "loss": 1.5171,
731
+ "step": 505
732
+ },
733
+ {
734
+ "epoch": 1.3934426229508197,
735
+ "grad_norm": 1.640625,
736
+ "learning_rate": 1.1144056156205831e-06,
737
+ "loss": 1.5114,
738
+ "step": 510
739
+ },
740
+ {
741
+ "epoch": 1.4071038251366121,
742
+ "grad_norm": 1.6484375,
743
+ "learning_rate": 1.100156372735361e-06,
744
+ "loss": 1.5032,
745
+ "step": 515
746
+ },
747
+ {
748
+ "epoch": 1.4207650273224044,
749
+ "grad_norm": 1.6328125,
750
+ "learning_rate": 1.0858865572208891e-06,
751
+ "loss": 1.5151,
752
+ "step": 520
753
+ },
754
+ {
755
+ "epoch": 1.4344262295081966,
756
+ "grad_norm": 1.625,
757
+ "learning_rate": 1.071599100169978e-06,
758
+ "loss": 1.5015,
759
+ "step": 525
760
+ },
761
+ {
762
+ "epoch": 1.4480874316939891,
763
+ "grad_norm": 1.65625,
764
+ "learning_rate": 1.0572969362990997e-06,
765
+ "loss": 1.5075,
766
+ "step": 530
767
+ },
768
+ {
769
+ "epoch": 1.4617486338797814,
770
+ "grad_norm": 2.65625,
771
+ "learning_rate": 1.042983003345582e-06,
772
+ "loss": 1.5117,
773
+ "step": 535
774
+ },
775
+ {
776
+ "epoch": 1.4754098360655736,
777
+ "grad_norm": 1.6328125,
778
+ "learning_rate": 1.0286602414641815e-06,
779
+ "loss": 1.5036,
780
+ "step": 540
781
+ },
782
+ {
783
+ "epoch": 1.489071038251366,
784
+ "grad_norm": 1.703125,
785
+ "learning_rate": 1.0143315926231624e-06,
786
+ "loss": 1.527,
787
+ "step": 545
788
+ },
789
+ {
790
+ "epoch": 1.5027322404371586,
791
+ "grad_norm": 1.6484375,
792
+ "learning_rate": 1e-06,
793
+ "loss": 1.5065,
794
+ "step": 550
795
+ },
796
+ {
797
+ "epoch": 1.5163934426229508,
798
+ "grad_norm": 1.65625,
799
+ "learning_rate": 9.856684073768378e-07,
800
+ "loss": 1.4989,
801
+ "step": 555
802
+ },
803
+ {
804
+ "epoch": 1.530054644808743,
805
+ "grad_norm": 1.6640625,
806
+ "learning_rate": 9.713397585358188e-07,
807
+ "loss": 1.5066,
808
+ "step": 560
809
+ },
810
+ {
811
+ "epoch": 1.5437158469945356,
812
+ "grad_norm": 1.6484375,
813
+ "learning_rate": 9.57016996654418e-07,
814
+ "loss": 1.5024,
815
+ "step": 565
816
+ },
817
+ {
818
+ "epoch": 1.5573770491803278,
819
+ "grad_norm": 1.6484375,
820
+ "learning_rate": 9.427030637009002e-07,
821
+ "loss": 1.5095,
822
+ "step": 570
823
+ },
824
+ {
825
+ "epoch": 1.57103825136612,
826
+ "grad_norm": 1.671875,
827
+ "learning_rate": 9.28400899830022e-07,
828
+ "loss": 1.5144,
829
+ "step": 575
830
+ },
831
+ {
832
+ "epoch": 1.5846994535519126,
833
+ "grad_norm": 1.640625,
834
+ "learning_rate": 9.141134427791109e-07,
835
+ "loss": 1.5191,
836
+ "step": 580
837
+ },
838
+ {
839
+ "epoch": 1.598360655737705,
840
+ "grad_norm": 1.6640625,
841
+ "learning_rate": 8.998436272646393e-07,
842
+ "loss": 1.5177,
843
+ "step": 585
844
+ },
845
+ {
846
+ "epoch": 1.6120218579234973,
847
+ "grad_norm": 1.6796875,
848
+ "learning_rate": 8.85594384379417e-07,
849
+ "loss": 1.5098,
850
+ "step": 590
851
+ },
852
+ {
853
+ "epoch": 1.6256830601092895,
854
+ "grad_norm": 1.6875,
855
+ "learning_rate": 8.713686409905313e-07,
856
+ "loss": 1.503,
857
+ "step": 595
858
+ },
859
+ {
860
+ "epoch": 1.639344262295082,
861
+ "grad_norm": 1.6640625,
862
+ "learning_rate": 8.571693191381544e-07,
863
+ "loss": 1.5091,
864
+ "step": 600
865
+ },
866
+ {
867
+ "epoch": 1.639344262295082,
868
+ "eval_loss": 1.5181070566177368,
869
+ "eval_runtime": 19.9478,
870
+ "eval_samples_per_second": 35.493,
871
+ "eval_steps_per_second": 1.153,
872
+ "step": 600
873
+ },
874
+ {
875
+ "epoch": 1.6530054644808743,
876
+ "grad_norm": 1.6640625,
877
+ "learning_rate": 8.429993354353388e-07,
878
+ "loss": 1.5003,
879
+ "step": 605
880
+ },
881
+ {
882
+ "epoch": 1.6666666666666665,
883
+ "grad_norm": 1.625,
884
+ "learning_rate": 8.288616004689319e-07,
885
+ "loss": 1.5068,
886
+ "step": 610
887
+ },
888
+ {
889
+ "epoch": 1.680327868852459,
890
+ "grad_norm": 1.6953125,
891
+ "learning_rate": 8.147590182017269e-07,
892
+ "loss": 1.5067,
893
+ "step": 615
894
+ },
895
+ {
896
+ "epoch": 1.6939890710382515,
897
+ "grad_norm": 1.640625,
898
+ "learning_rate": 8.006944853759732e-07,
899
+ "loss": 1.5069,
900
+ "step": 620
901
+ },
902
+ {
903
+ "epoch": 1.7076502732240437,
904
+ "grad_norm": 1.7734375,
905
+ "learning_rate": 7.866708909183702e-07,
906
+ "loss": 1.4937,
907
+ "step": 625
908
+ },
909
+ {
910
+ "epoch": 1.721311475409836,
911
+ "grad_norm": 1.640625,
912
+ "learning_rate": 7.726911153466697e-07,
913
+ "loss": 1.5089,
914
+ "step": 630
915
+ },
916
+ {
917
+ "epoch": 1.7349726775956285,
918
+ "grad_norm": 1.6484375,
919
+ "learning_rate": 7.587580301779999e-07,
920
+ "loss": 1.511,
921
+ "step": 635
922
+ },
923
+ {
924
+ "epoch": 1.748633879781421,
925
+ "grad_norm": 1.6875,
926
+ "learning_rate": 7.448744973390422e-07,
927
+ "loss": 1.5025,
928
+ "step": 640
929
+ },
930
+ {
931
+ "epoch": 1.762295081967213,
932
+ "grad_norm": 1.640625,
933
+ "learning_rate": 7.310433685781777e-07,
934
+ "loss": 1.5007,
935
+ "step": 645
936
+ },
937
+ {
938
+ "epoch": 1.7759562841530054,
939
+ "grad_norm": 1.671875,
940
+ "learning_rate": 7.172674848797217e-07,
941
+ "loss": 1.4922,
942
+ "step": 650
943
+ },
944
+ {
945
+ "epoch": 1.789617486338798,
946
+ "grad_norm": 1.640625,
947
+ "learning_rate": 7.035496758803743e-07,
948
+ "loss": 1.505,
949
+ "step": 655
950
+ },
951
+ {
952
+ "epoch": 1.8032786885245902,
953
+ "grad_norm": 1.625,
954
+ "learning_rate": 6.898927592879944e-07,
955
+ "loss": 1.5111,
956
+ "step": 660
957
+ },
958
+ {
959
+ "epoch": 1.8169398907103824,
960
+ "grad_norm": 1.640625,
961
+ "learning_rate": 6.762995403028314e-07,
962
+ "loss": 1.5145,
963
+ "step": 665
964
+ },
965
+ {
966
+ "epoch": 1.830601092896175,
967
+ "grad_norm": 1.8359375,
968
+ "learning_rate": 6.627728110413213e-07,
969
+ "loss": 1.5097,
970
+ "step": 670
971
+ },
972
+ {
973
+ "epoch": 1.8442622950819674,
974
+ "grad_norm": 1.609375,
975
+ "learning_rate": 6.493153499625719e-07,
976
+ "loss": 1.5087,
977
+ "step": 675
978
+ },
979
+ {
980
+ "epoch": 1.8579234972677594,
981
+ "grad_norm": 2.0625,
982
+ "learning_rate": 6.359299212976534e-07,
983
+ "loss": 1.5044,
984
+ "step": 680
985
+ },
986
+ {
987
+ "epoch": 1.8715846994535519,
988
+ "grad_norm": 1.640625,
989
+ "learning_rate": 6.226192744818124e-07,
990
+ "loss": 1.5105,
991
+ "step": 685
992
+ },
993
+ {
994
+ "epoch": 1.8852459016393444,
995
+ "grad_norm": 1.6484375,
996
+ "learning_rate": 6.093861435897207e-07,
997
+ "loss": 1.5109,
998
+ "step": 690
999
+ },
1000
+ {
1001
+ "epoch": 1.8989071038251366,
1002
+ "grad_norm": 1.6640625,
1003
+ "learning_rate": 5.962332467738857e-07,
1004
+ "loss": 1.5103,
1005
+ "step": 695
1006
+ },
1007
+ {
1008
+ "epoch": 1.9125683060109289,
1009
+ "grad_norm": 1.6796875,
1010
+ "learning_rate": 5.83163285706327e-07,
1011
+ "loss": 1.5023,
1012
+ "step": 700
1013
+ },
1014
+ {
1015
+ "epoch": 1.9262295081967213,
1016
+ "grad_norm": 1.6640625,
1017
+ "learning_rate": 5.701789450236376e-07,
1018
+ "loss": 1.5084,
1019
+ "step": 705
1020
+ },
1021
+ {
1022
+ "epoch": 1.9398907103825138,
1023
+ "grad_norm": 1.6484375,
1024
+ "learning_rate": 5.57282891775548e-07,
1025
+ "loss": 1.5063,
1026
+ "step": 710
1027
+ },
1028
+ {
1029
+ "epoch": 1.9535519125683058,
1030
+ "grad_norm": 1.6328125,
1031
+ "learning_rate": 5.444777748771006e-07,
1032
+ "loss": 1.5074,
1033
+ "step": 715
1034
+ },
1035
+ {
1036
+ "epoch": 1.9672131147540983,
1037
+ "grad_norm": 1.6640625,
1038
+ "learning_rate": 5.317662245645469e-07,
1039
+ "loss": 1.5037,
1040
+ "step": 720
1041
+ },
1042
+ {
1043
+ "epoch": 1.9808743169398908,
1044
+ "grad_norm": 1.6796875,
1045
+ "learning_rate": 5.191508518550855e-07,
1046
+ "loss": 1.5085,
1047
+ "step": 725
1048
+ },
1049
+ {
1050
+ "epoch": 1.994535519125683,
1051
+ "grad_norm": 1.6015625,
1052
+ "learning_rate": 5.066342480105459e-07,
1053
+ "loss": 1.5082,
1054
+ "step": 730
1055
+ },
1056
+ {
1057
+ "epoch": 2.0081967213114753,
1058
+ "grad_norm": 1.6484375,
1059
+ "learning_rate": 4.942189840051287e-07,
1060
+ "loss": 1.5054,
1061
+ "step": 735
1062
+ },
1063
+ {
1064
+ "epoch": 2.021857923497268,
1065
+ "grad_norm": 1.84375,
1066
+ "learning_rate": 4.819076099973152e-07,
1067
+ "loss": 1.4999,
1068
+ "step": 740
1069
+ },
1070
+ {
1071
+ "epoch": 2.0355191256830603,
1072
+ "grad_norm": 1.6171875,
1073
+ "learning_rate": 4.697026548060528e-07,
1074
+ "loss": 1.5094,
1075
+ "step": 745
1076
+ },
1077
+ {
1078
+ "epoch": 2.0491803278688523,
1079
+ "grad_norm": 1.703125,
1080
+ "learning_rate": 4.5760662539132077e-07,
1081
+ "loss": 1.5065,
1082
+ "step": 750
1083
+ },
1084
+ {
1085
+ "epoch": 2.0628415300546448,
1086
+ "grad_norm": 1.921875,
1087
+ "learning_rate": 4.4562200633918943e-07,
1088
+ "loss": 1.5013,
1089
+ "step": 755
1090
+ },
1091
+ {
1092
+ "epoch": 2.0765027322404372,
1093
+ "grad_norm": 1.625,
1094
+ "learning_rate": 4.337512593514728e-07,
1095
+ "loss": 1.5041,
1096
+ "step": 760
1097
+ },
1098
+ {
1099
+ "epoch": 2.0901639344262297,
1100
+ "grad_norm": 1.6640625,
1101
+ "learning_rate": 4.2199682274008255e-07,
1102
+ "loss": 1.5083,
1103
+ "step": 765
1104
+ },
1105
+ {
1106
+ "epoch": 2.1038251366120218,
1107
+ "grad_norm": 1.6171875,
1108
+ "learning_rate": 4.103611109261872e-07,
1109
+ "loss": 1.4982,
1110
+ "step": 770
1111
+ },
1112
+ {
1113
+ "epoch": 2.1174863387978142,
1114
+ "grad_norm": 1.6328125,
1115
+ "learning_rate": 3.9884651394427625e-07,
1116
+ "loss": 1.5051,
1117
+ "step": 775
1118
+ },
1119
+ {
1120
+ "epoch": 2.1311475409836067,
1121
+ "grad_norm": 1.671875,
1122
+ "learning_rate": 3.8745539695123577e-07,
1123
+ "loss": 1.4924,
1124
+ "step": 780
1125
+ },
1126
+ {
1127
+ "epoch": 2.1448087431693987,
1128
+ "grad_norm": 1.65625,
1129
+ "learning_rate": 3.761900997405332e-07,
1130
+ "loss": 1.5113,
1131
+ "step": 785
1132
+ },
1133
+ {
1134
+ "epoch": 2.158469945355191,
1135
+ "grad_norm": 1.6484375,
1136
+ "learning_rate": 3.6505293626161127e-07,
1137
+ "loss": 1.5058,
1138
+ "step": 790
1139
+ },
1140
+ {
1141
+ "epoch": 2.1721311475409837,
1142
+ "grad_norm": 1.625,
1143
+ "learning_rate": 3.5404619414459147e-07,
1144
+ "loss": 1.5019,
1145
+ "step": 795
1146
+ },
1147
+ {
1148
+ "epoch": 2.185792349726776,
1149
+ "grad_norm": 1.640625,
1150
+ "learning_rate": 3.4317213423038384e-07,
1151
+ "loss": 1.4948,
1152
+ "step": 800
1153
+ },
1154
+ {
1155
+ "epoch": 2.199453551912568,
1156
+ "grad_norm": 1.625,
1157
+ "learning_rate": 3.32432990106297e-07,
1158
+ "loss": 1.5092,
1159
+ "step": 805
1160
+ },
1161
+ {
1162
+ "epoch": 2.2131147540983607,
1163
+ "grad_norm": 1.734375,
1164
+ "learning_rate": 3.2183096764724914e-07,
1165
+ "loss": 1.5048,
1166
+ "step": 810
1167
+ },
1168
+ {
1169
+ "epoch": 2.226775956284153,
1170
+ "grad_norm": 1.8984375,
1171
+ "learning_rate": 3.1136824456267006e-07,
1172
+ "loss": 1.5005,
1173
+ "step": 815
1174
+ },
1175
+ {
1176
+ "epoch": 2.240437158469945,
1177
+ "grad_norm": 1.7421875,
1178
+ "learning_rate": 3.01046969949188e-07,
1179
+ "loss": 1.5097,
1180
+ "step": 820
1181
+ },
1182
+ {
1183
+ "epoch": 2.2540983606557377,
1184
+ "grad_norm": 1.6640625,
1185
+ "learning_rate": 2.908692638491945e-07,
1186
+ "loss": 1.5054,
1187
+ "step": 825
1188
+ },
1189
+ {
1190
+ "epoch": 2.26775956284153,
1191
+ "grad_norm": 1.65625,
1192
+ "learning_rate": 2.80837216815378e-07,
1193
+ "loss": 1.5131,
1194
+ "step": 830
1195
+ },
1196
+ {
1197
+ "epoch": 2.281420765027322,
1198
+ "grad_norm": 1.671875,
1199
+ "learning_rate": 2.7095288948131114e-07,
1200
+ "loss": 1.4999,
1201
+ "step": 835
1202
+ },
1203
+ {
1204
+ "epoch": 2.2950819672131146,
1205
+ "grad_norm": 1.6328125,
1206
+ "learning_rate": 2.6121831213818826e-07,
1207
+ "loss": 1.4989,
1208
+ "step": 840
1209
+ },
1210
+ {
1211
+ "epoch": 2.308743169398907,
1212
+ "grad_norm": 1.6953125,
1213
+ "learning_rate": 2.51635484317793e-07,
1214
+ "loss": 1.495,
1215
+ "step": 845
1216
+ },
1217
+ {
1218
+ "epoch": 2.3224043715846996,
1219
+ "grad_norm": 1.703125,
1220
+ "learning_rate": 2.4220637438178313e-07,
1221
+ "loss": 1.5125,
1222
+ "step": 850
1223
+ },
1224
+ {
1225
+ "epoch": 2.3360655737704916,
1226
+ "grad_norm": 1.703125,
1227
+ "learning_rate": 2.3293291911738078e-07,
1228
+ "loss": 1.5092,
1229
+ "step": 855
1230
+ },
1231
+ {
1232
+ "epoch": 2.349726775956284,
1233
+ "grad_norm": 1.6640625,
1234
+ "learning_rate": 2.2381702333954433e-07,
1235
+ "loss": 1.4965,
1236
+ "step": 860
1237
+ },
1238
+ {
1239
+ "epoch": 2.3633879781420766,
1240
+ "grad_norm": 1.6328125,
1241
+ "learning_rate": 2.148605594997115e-07,
1242
+ "loss": 1.497,
1243
+ "step": 865
1244
+ },
1245
+ {
1246
+ "epoch": 2.3770491803278686,
1247
+ "grad_norm": 1.765625,
1248
+ "learning_rate": 2.0606536730118763e-07,
1249
+ "loss": 1.5081,
1250
+ "step": 870
1251
+ },
1252
+ {
1253
+ "epoch": 2.390710382513661,
1254
+ "grad_norm": 1.625,
1255
+ "learning_rate": 1.9743325332126105e-07,
1256
+ "loss": 1.5091,
1257
+ "step": 875
1258
+ },
1259
+ {
1260
+ "epoch": 2.4043715846994536,
1261
+ "grad_norm": 1.6171875,
1262
+ "learning_rate": 1.8896599064012298e-07,
1263
+ "loss": 1.5045,
1264
+ "step": 880
1265
+ },
1266
+ {
1267
+ "epoch": 2.418032786885246,
1268
+ "grad_norm": 1.640625,
1269
+ "learning_rate": 1.8066531847666888e-07,
1270
+ "loss": 1.5008,
1271
+ "step": 885
1272
+ },
1273
+ {
1274
+ "epoch": 2.431693989071038,
1275
+ "grad_norm": 1.65625,
1276
+ "learning_rate": 1.7253294183125222e-07,
1277
+ "loss": 1.511,
1278
+ "step": 890
1279
+ },
1280
+ {
1281
+ "epoch": 2.4453551912568305,
1282
+ "grad_norm": 1.6328125,
1283
+ "learning_rate": 1.645705311354697e-07,
1284
+ "loss": 1.4998,
1285
+ "step": 895
1286
+ },
1287
+ {
1288
+ "epoch": 2.459016393442623,
1289
+ "grad_norm": 1.65625,
1290
+ "learning_rate": 1.5677972190904621e-07,
1291
+ "loss": 1.5021,
1292
+ "step": 900
1293
+ },
1294
+ {
1295
+ "epoch": 2.459016393442623,
1296
+ "eval_loss": 1.5161317586898804,
1297
+ "eval_runtime": 19.9284,
1298
+ "eval_samples_per_second": 35.527,
1299
+ "eval_steps_per_second": 1.154,
1300
+ "step": 900
1301
+ },
1302
+ {
1303
+ "epoch": 2.4726775956284155,
1304
+ "grad_norm": 1.6328125,
1305
+ "learning_rate": 1.4916211442389048e-07,
1306
+ "loss": 1.502,
1307
+ "step": 905
1308
+ },
1309
+ {
1310
+ "epoch": 2.4863387978142075,
1311
+ "grad_norm": 1.625,
1312
+ "learning_rate": 1.4171927337539104e-07,
1313
+ "loss": 1.5012,
1314
+ "step": 910
1315
+ },
1316
+ {
1317
+ "epoch": 2.5,
1318
+ "grad_norm": 1.640625,
1319
+ "learning_rate": 1.344527275610202e-07,
1320
+ "loss": 1.5019,
1321
+ "step": 915
1322
+ },
1323
+ {
1324
+ "epoch": 2.5136612021857925,
1325
+ "grad_norm": 1.640625,
1326
+ "learning_rate": 1.273639695663108e-07,
1327
+ "loss": 1.5085,
1328
+ "step": 920
1329
+ },
1330
+ {
1331
+ "epoch": 2.527322404371585,
1332
+ "grad_norm": 1.75,
1333
+ "learning_rate": 1.204544554582716e-07,
1334
+ "loss": 1.4973,
1335
+ "step": 925
1336
+ },
1337
+ {
1338
+ "epoch": 2.540983606557377,
1339
+ "grad_norm": 1.671875,
1340
+ "learning_rate": 1.1372560448630375e-07,
1341
+ "loss": 1.5037,
1342
+ "step": 930
1343
+ },
1344
+ {
1345
+ "epoch": 2.5546448087431695,
1346
+ "grad_norm": 1.65625,
1347
+ "learning_rate": 1.0717879879068004e-07,
1348
+ "loss": 1.5005,
1349
+ "step": 935
1350
+ },
1351
+ {
1352
+ "epoch": 2.5683060109289615,
1353
+ "grad_norm": 1.6796875,
1354
+ "learning_rate": 1.0081538311864568e-07,
1355
+ "loss": 1.5024,
1356
+ "step": 940
1357
+ },
1358
+ {
1359
+ "epoch": 2.581967213114754,
1360
+ "grad_norm": 1.625,
1361
+ "learning_rate": 9.463666454820118e-08,
1362
+ "loss": 1.4988,
1363
+ "step": 945
1364
+ },
1365
+ {
1366
+ "epoch": 2.5956284153005464,
1367
+ "grad_norm": 1.625,
1368
+ "learning_rate": 8.864391221962064e-08,
1369
+ "loss": 1.5053,
1370
+ "step": 950
1371
+ },
1372
+ {
1373
+ "epoch": 2.609289617486339,
1374
+ "grad_norm": 1.6328125,
1375
+ "learning_rate": 8.28383570747655e-08,
1376
+ "loss": 1.5044,
1377
+ "step": 955
1378
+ },
1379
+ {
1380
+ "epoch": 2.6229508196721314,
1381
+ "grad_norm": 1.609375,
1382
+ "learning_rate": 7.722119160424112e-08,
1383
+ "loss": 1.4995,
1384
+ "step": 960
1385
+ },
1386
+ {
1387
+ "epoch": 2.6366120218579234,
1388
+ "grad_norm": 1.640625,
1389
+ "learning_rate": 7.179356960245409e-08,
1390
+ "loss": 1.5122,
1391
+ "step": 965
1392
+ },
1393
+ {
1394
+ "epoch": 2.650273224043716,
1395
+ "grad_norm": 1.7890625,
1396
+ "learning_rate": 6.655660593061718e-08,
1397
+ "loss": 1.5054,
1398
+ "step": 970
1399
+ },
1400
+ {
1401
+ "epoch": 2.663934426229508,
1402
+ "grad_norm": 1.6484375,
1403
+ "learning_rate": 6.151137628775049e-08,
1404
+ "loss": 1.5108,
1405
+ "step": 975
1406
+ },
1407
+ {
1408
+ "epoch": 2.6775956284153004,
1409
+ "grad_norm": 1.609375,
1410
+ "learning_rate": 5.665891698972769e-08,
1411
+ "loss": 1.5003,
1412
+ "step": 980
1413
+ },
1414
+ {
1415
+ "epoch": 2.691256830601093,
1416
+ "grad_norm": 1.640625,
1417
+ "learning_rate": 5.200022475641153e-08,
1418
+ "loss": 1.5015,
1419
+ "step": 985
1420
+ },
1421
+ {
1422
+ "epoch": 2.7049180327868854,
1423
+ "grad_norm": 1.609375,
1424
+ "learning_rate": 4.75362565069225e-08,
1425
+ "loss": 1.5002,
1426
+ "step": 990
1427
+ },
1428
+ {
1429
+ "epoch": 2.718579234972678,
1430
+ "grad_norm": 1.640625,
1431
+ "learning_rate": 4.326792916308242e-08,
1432
+ "loss": 1.5029,
1433
+ "step": 995
1434
+ },
1435
+ {
1436
+ "epoch": 2.73224043715847,
1437
+ "grad_norm": 1.7421875,
1438
+ "learning_rate": 3.919611946107493e-08,
1439
+ "loss": 1.5068,
1440
+ "step": 1000
1441
+ },
1442
+ {
1443
+ "epoch": 2.7459016393442623,
1444
+ "grad_norm": 1.921875,
1445
+ "learning_rate": 3.532166377135814e-08,
1446
+ "loss": 1.4961,
1447
+ "step": 1005
1448
+ },
1449
+ {
1450
+ "epoch": 2.7595628415300544,
1451
+ "grad_norm": 1.6953125,
1452
+ "learning_rate": 3.164535792687095e-08,
1453
+ "loss": 1.5,
1454
+ "step": 1010
1455
+ },
1456
+ {
1457
+ "epoch": 2.773224043715847,
1458
+ "grad_norm": 1.6171875,
1459
+ "learning_rate": 2.8167957059564095e-08,
1460
+ "loss": 1.5035,
1461
+ "step": 1015
1462
+ },
1463
+ {
1464
+ "epoch": 2.7868852459016393,
1465
+ "grad_norm": 1.6484375,
1466
+ "learning_rate": 2.4890175445293147e-08,
1467
+ "loss": 1.5007,
1468
+ "step": 1020
1469
+ },
1470
+ {
1471
+ "epoch": 2.800546448087432,
1472
+ "grad_norm": 1.6328125,
1473
+ "learning_rate": 2.1812686357101428e-08,
1474
+ "loss": 1.505,
1475
+ "step": 1025
1476
+ },
1477
+ {
1478
+ "epoch": 2.8142076502732243,
1479
+ "grad_norm": 1.6484375,
1480
+ "learning_rate": 1.8936121926927507e-08,
1481
+ "loss": 1.5066,
1482
+ "step": 1030
1483
+ },
1484
+ {
1485
+ "epoch": 2.8278688524590163,
1486
+ "grad_norm": 1.640625,
1487
+ "learning_rate": 1.6261073015761072e-08,
1488
+ "loss": 1.502,
1489
+ "step": 1035
1490
+ },
1491
+ {
1492
+ "epoch": 2.841530054644809,
1493
+ "grad_norm": 1.75,
1494
+ "learning_rate": 1.3788089092277688e-08,
1495
+ "loss": 1.5016,
1496
+ "step": 1040
1497
+ },
1498
+ {
1499
+ "epoch": 2.855191256830601,
1500
+ "grad_norm": 1.65625,
1501
+ "learning_rate": 1.1517678119975061e-08,
1502
+ "loss": 1.5024,
1503
+ "step": 1045
1504
+ },
1505
+ {
1506
+ "epoch": 2.8688524590163933,
1507
+ "grad_norm": 1.609375,
1508
+ "learning_rate": 9.450306452834178e-09,
1509
+ "loss": 1.5,
1510
+ "step": 1050
1511
+ },
1512
+ {
1513
+ "epoch": 2.8825136612021858,
1514
+ "grad_norm": 1.734375,
1515
+ "learning_rate": 7.586398739528932e-09,
1516
+ "loss": 1.4989,
1517
+ "step": 1055
1518
+ },
1519
+ {
1520
+ "epoch": 2.8961748633879782,
1521
+ "grad_norm": 1.7890625,
1522
+ "learning_rate": 5.926337836199891e-09,
1523
+ "loss": 1.507,
1524
+ "step": 1060
1525
+ },
1526
+ {
1527
+ "epoch": 2.9098360655737707,
1528
+ "grad_norm": 1.7734375,
1529
+ "learning_rate": 4.470464727814538e-09,
1530
+ "loss": 1.5119,
1531
+ "step": 1065
1532
+ },
1533
+ {
1534
+ "epoch": 2.9234972677595628,
1535
+ "grad_norm": 1.7734375,
1536
+ "learning_rate": 3.219078458127078e-09,
1537
+ "loss": 1.5072,
1538
+ "step": 1070
1539
+ },
1540
+ {
1541
+ "epoch": 2.9371584699453552,
1542
+ "grad_norm": 1.6328125,
1543
+ "learning_rate": 2.172436068252792e-09,
1544
+ "loss": 1.5074,
1545
+ "step": 1075
1546
+ },
1547
+ {
1548
+ "epoch": 2.9508196721311473,
1549
+ "grad_norm": 1.6171875,
1550
+ "learning_rate": 1.330752543871161e-09,
1551
+ "loss": 1.4947,
1552
+ "step": 1080
1553
+ },
1554
+ {
1555
+ "epoch": 2.9644808743169397,
1556
+ "grad_norm": 1.671875,
1557
+ "learning_rate": 6.942007710665221e-10,
1558
+ "loss": 1.5025,
1559
+ "step": 1085
1560
+ },
1561
+ {
1562
+ "epoch": 2.978142076502732,
1563
+ "grad_norm": 1.6640625,
1564
+ "learning_rate": 2.6291150081603207e-10,
1565
+ "loss": 1.5014,
1566
+ "step": 1090
1567
+ },
1568
+ {
1569
+ "epoch": 2.9918032786885247,
1570
+ "grad_norm": 1.609375,
1571
+ "learning_rate": 3.697332213348225e-11,
1572
+ "loss": 1.4964,
1573
+ "step": 1095
1574
+ },
1575
+ {
1576
+ "epoch": 3.0,
1577
+ "step": 1098,
1578
+ "total_flos": 7.187967826039144e+18,
1579
+ "train_loss": 1.5352961912832626,
1580
+ "train_runtime": 6906.6629,
1581
+ "train_samples_per_second": 10.159,
1582
+ "train_steps_per_second": 0.159
1583
+ }
1584
+ ],
1585
+ "logging_steps": 5,
1586
+ "max_steps": 1098,
1587
+ "num_input_tokens_seen": 0,
1588
+ "num_train_epochs": 3,
1589
+ "save_steps": 300,
1590
+ "stateful_callbacks": {
1591
+ "TrainerControl": {
1592
+ "args": {
1593
+ "should_epoch_stop": false,
1594
+ "should_evaluate": false,
1595
+ "should_log": false,
1596
+ "should_save": true,
1597
+ "should_training_stop": true
1598
+ },
1599
+ "attributes": {}
1600
+ }
1601
+ },
1602
+ "total_flos": 7.187967826039144e+18,
1603
+ "train_batch_size": 8,
1604
+ "trial_name": null,
1605
+ "trial_params": null
1606
+ }