antonpolishko commited on
Commit
974e73b
·
verified ·
1 Parent(s): c301e21

Model save

Browse files
README.md ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: meta-llama/Llama-3.1-8B
3
+ library_name: transformers
4
+ model_name: llama-3_1-8b-overfit-ua
5
+ tags:
6
+ - generated_from_trainer
7
+ - trl
8
+ - sft
9
+ licence: license
10
+ ---
11
+
12
+ # Model Card for llama-3_1-8b-overfit-ua
13
+
14
+ This model is a fine-tuned version of [meta-llama/Llama-3.1-8B](https://huggingface.co/meta-llama/Llama-3.1-8B).
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/llama-3_1-8b-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/e2est3jv)
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": 6.427401199279931e+18,
4
+ "train_loss": 1.711617823146263,
5
+ "train_runtime": 5145.0339,
6
+ "train_samples": 95663,
7
+ "train_samples_per_second": 13.545,
8
+ "train_steps_per_second": 0.212
9
+ }
generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 128000,
4
+ "do_sample": true,
5
+ "eos_token_id": 128001,
6
+ "temperature": 0.6,
7
+ "top_p": 0.9,
8
+ "transformers_version": "4.46.3"
9
+ }
runs/Dec26_12-08-20_mia1-gpu-110/events.out.tfevents.1735214952.mia1-gpu-110.3773944.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7a9d2c1f73e98bf878dd4f8cd6855cb864dea5e95da7596a422dad90da6a2554
3
- size 52652
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f07e9c40b0852d6075483db246b6a6194366aa39fd072c30691d0ce0de6cdd45
3
+ size 53006
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 6.427401199279931e+18,
4
+ "train_loss": 1.711617823146263,
5
+ "train_runtime": 5145.0339,
6
+ "train_samples": 95663,
7
+ "train_samples_per_second": 13.545,
8
+ "train_steps_per_second": 0.212
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1592 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0,
5
+ "eval_steps": 300,
6
+ "global_step": 1089,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0027548209366391185,
13
+ "grad_norm": 10.8125,
14
+ "learning_rate": 1e-06,
15
+ "loss": 1.9168,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.013774104683195593,
20
+ "grad_norm": 8.4375,
21
+ "learning_rate": 1.999962411893365e-06,
22
+ "loss": 1.9099,
23
+ "step": 5
24
+ },
25
+ {
26
+ "epoch": 0.027548209366391185,
27
+ "grad_norm": 4.90625,
28
+ "learning_rate": 1.9997327170302815e-06,
29
+ "loss": 1.8629,
30
+ "step": 10
31
+ },
32
+ {
33
+ "epoch": 0.04132231404958678,
34
+ "grad_norm": 3.671875,
35
+ "learning_rate": 1.99929425749243e-06,
36
+ "loss": 1.8376,
37
+ "step": 15
38
+ },
39
+ {
40
+ "epoch": 0.05509641873278237,
41
+ "grad_norm": 2.828125,
42
+ "learning_rate": 1.998647124839145e-06,
43
+ "loss": 1.8137,
44
+ "step": 20
45
+ },
46
+ {
47
+ "epoch": 0.06887052341597796,
48
+ "grad_norm": 2.640625,
49
+ "learning_rate": 1.997791454204984e-06,
50
+ "loss": 1.8055,
51
+ "step": 25
52
+ },
53
+ {
54
+ "epoch": 0.08264462809917356,
55
+ "grad_norm": 2.234375,
56
+ "learning_rate": 1.9967274242715065e-06,
57
+ "loss": 1.7913,
58
+ "step": 30
59
+ },
60
+ {
61
+ "epoch": 0.09641873278236915,
62
+ "grad_norm": 2.21875,
63
+ "learning_rate": 1.995455257229964e-06,
64
+ "loss": 1.7897,
65
+ "step": 35
66
+ },
67
+ {
68
+ "epoch": 0.11019283746556474,
69
+ "grad_norm": 1.8828125,
70
+ "learning_rate": 1.9939752187349e-06,
71
+ "loss": 1.7752,
72
+ "step": 40
73
+ },
74
+ {
75
+ "epoch": 0.12396694214876033,
76
+ "grad_norm": 1.8671875,
77
+ "learning_rate": 1.9922876178486764e-06,
78
+ "loss": 1.7802,
79
+ "step": 45
80
+ },
81
+ {
82
+ "epoch": 0.13774104683195593,
83
+ "grad_norm": 1.65625,
84
+ "learning_rate": 1.9903928069769356e-06,
85
+ "loss": 1.7686,
86
+ "step": 50
87
+ },
88
+ {
89
+ "epoch": 0.15151515151515152,
90
+ "grad_norm": 1.765625,
91
+ "learning_rate": 1.9882911817950105e-06,
92
+ "loss": 1.7702,
93
+ "step": 55
94
+ },
95
+ {
96
+ "epoch": 0.1652892561983471,
97
+ "grad_norm": 1.7265625,
98
+ "learning_rate": 1.985983181165299e-06,
99
+ "loss": 1.7618,
100
+ "step": 60
101
+ },
102
+ {
103
+ "epoch": 0.1790633608815427,
104
+ "grad_norm": 1.609375,
105
+ "learning_rate": 1.98346928704562e-06,
106
+ "loss": 1.7627,
107
+ "step": 65
108
+ },
109
+ {
110
+ "epoch": 0.1928374655647383,
111
+ "grad_norm": 1.5625,
112
+ "learning_rate": 1.9807500243885744e-06,
113
+ "loss": 1.7665,
114
+ "step": 70
115
+ },
116
+ {
117
+ "epoch": 0.2066115702479339,
118
+ "grad_norm": 1.515625,
119
+ "learning_rate": 1.9778259610319187e-06,
120
+ "loss": 1.755,
121
+ "step": 75
122
+ },
123
+ {
124
+ "epoch": 0.22038567493112948,
125
+ "grad_norm": 1.5234375,
126
+ "learning_rate": 1.9746977075799933e-06,
127
+ "loss": 1.7574,
128
+ "step": 80
129
+ },
130
+ {
131
+ "epoch": 0.23415977961432508,
132
+ "grad_norm": 1.4609375,
133
+ "learning_rate": 1.9713659172762126e-06,
134
+ "loss": 1.7529,
135
+ "step": 85
136
+ },
137
+ {
138
+ "epoch": 0.24793388429752067,
139
+ "grad_norm": 1.453125,
140
+ "learning_rate": 1.9678312858666578e-06,
141
+ "loss": 1.7417,
142
+ "step": 90
143
+ },
144
+ {
145
+ "epoch": 0.26170798898071623,
146
+ "grad_norm": 1.421875,
147
+ "learning_rate": 1.964094551454788e-06,
148
+ "loss": 1.7509,
149
+ "step": 95
150
+ },
151
+ {
152
+ "epoch": 0.27548209366391185,
153
+ "grad_norm": 1.5,
154
+ "learning_rate": 1.960156494347309e-06,
155
+ "loss": 1.7486,
156
+ "step": 100
157
+ },
158
+ {
159
+ "epoch": 0.2892561983471074,
160
+ "grad_norm": 1.3984375,
161
+ "learning_rate": 1.9560179368912327e-06,
162
+ "loss": 1.7531,
163
+ "step": 105
164
+ },
165
+ {
166
+ "epoch": 0.30303030303030304,
167
+ "grad_norm": 1.40625,
168
+ "learning_rate": 1.95167974330215e-06,
169
+ "loss": 1.7435,
170
+ "step": 110
171
+ },
172
+ {
173
+ "epoch": 0.3168044077134986,
174
+ "grad_norm": 1.4140625,
175
+ "learning_rate": 1.9471428194837667e-06,
176
+ "loss": 1.7403,
177
+ "step": 115
178
+ },
179
+ {
180
+ "epoch": 0.3305785123966942,
181
+ "grad_norm": 1.3984375,
182
+ "learning_rate": 1.9424081128387337e-06,
183
+ "loss": 1.7435,
184
+ "step": 120
185
+ },
186
+ {
187
+ "epoch": 0.3443526170798898,
188
+ "grad_norm": 1.40625,
189
+ "learning_rate": 1.9374766120708077e-06,
190
+ "loss": 1.75,
191
+ "step": 125
192
+ },
193
+ {
194
+ "epoch": 0.3581267217630854,
195
+ "grad_norm": 1.3828125,
196
+ "learning_rate": 1.932349346978389e-06,
197
+ "loss": 1.7469,
198
+ "step": 130
199
+ },
200
+ {
201
+ "epoch": 0.371900826446281,
202
+ "grad_norm": 1.359375,
203
+ "learning_rate": 1.927027388239482e-06,
204
+ "loss": 1.734,
205
+ "step": 135
206
+ },
207
+ {
208
+ "epoch": 0.3856749311294766,
209
+ "grad_norm": 1.40625,
210
+ "learning_rate": 1.921511847188112e-06,
211
+ "loss": 1.7385,
212
+ "step": 140
213
+ },
214
+ {
215
+ "epoch": 0.39944903581267216,
216
+ "grad_norm": 1.390625,
217
+ "learning_rate": 1.915803875582259e-06,
218
+ "loss": 1.736,
219
+ "step": 145
220
+ },
221
+ {
222
+ "epoch": 0.4132231404958678,
223
+ "grad_norm": 1.34375,
224
+ "learning_rate": 1.9099046653633437e-06,
225
+ "loss": 1.7336,
226
+ "step": 150
227
+ },
228
+ {
229
+ "epoch": 0.42699724517906334,
230
+ "grad_norm": 1.3515625,
231
+ "learning_rate": 1.9038154484073284e-06,
232
+ "loss": 1.7324,
233
+ "step": 155
234
+ },
235
+ {
236
+ "epoch": 0.44077134986225897,
237
+ "grad_norm": 1.3828125,
238
+ "learning_rate": 1.8975374962674753e-06,
239
+ "loss": 1.7406,
240
+ "step": 160
241
+ },
242
+ {
243
+ "epoch": 0.45454545454545453,
244
+ "grad_norm": 1.3359375,
245
+ "learning_rate": 1.8910721199088195e-06,
246
+ "loss": 1.7461,
247
+ "step": 165
248
+ },
249
+ {
250
+ "epoch": 0.46831955922865015,
251
+ "grad_norm": 1.3515625,
252
+ "learning_rate": 1.8844206694344138e-06,
253
+ "loss": 1.7296,
254
+ "step": 170
255
+ },
256
+ {
257
+ "epoch": 0.4820936639118457,
258
+ "grad_norm": 1.390625,
259
+ "learning_rate": 1.877584533803398e-06,
260
+ "loss": 1.7286,
261
+ "step": 175
262
+ },
263
+ {
264
+ "epoch": 0.49586776859504134,
265
+ "grad_norm": 1.3203125,
266
+ "learning_rate": 1.8705651405409566e-06,
267
+ "loss": 1.7225,
268
+ "step": 180
269
+ },
270
+ {
271
+ "epoch": 0.509641873278237,
272
+ "grad_norm": 1.328125,
273
+ "learning_rate": 1.8633639554402234e-06,
274
+ "loss": 1.7366,
275
+ "step": 185
276
+ },
277
+ {
278
+ "epoch": 0.5234159779614325,
279
+ "grad_norm": 1.3828125,
280
+ "learning_rate": 1.8559824822561913e-06,
281
+ "loss": 1.7353,
282
+ "step": 190
283
+ },
284
+ {
285
+ "epoch": 0.5371900826446281,
286
+ "grad_norm": 1.3515625,
287
+ "learning_rate": 1.8484222623917e-06,
288
+ "loss": 1.7223,
289
+ "step": 195
290
+ },
291
+ {
292
+ "epoch": 0.5509641873278237,
293
+ "grad_norm": 1.328125,
294
+ "learning_rate": 1.8406848745755578e-06,
295
+ "loss": 1.7256,
296
+ "step": 200
297
+ },
298
+ {
299
+ "epoch": 0.5647382920110193,
300
+ "grad_norm": 1.3671875,
301
+ "learning_rate": 1.832771934532872e-06,
302
+ "loss": 1.7288,
303
+ "step": 205
304
+ },
305
+ {
306
+ "epoch": 0.5785123966942148,
307
+ "grad_norm": 1.34375,
308
+ "learning_rate": 1.8246850946476505e-06,
309
+ "loss": 1.7247,
310
+ "step": 210
311
+ },
312
+ {
313
+ "epoch": 0.5922865013774105,
314
+ "grad_norm": 1.3359375,
315
+ "learning_rate": 1.8164260436177524e-06,
316
+ "loss": 1.7221,
317
+ "step": 215
318
+ },
319
+ {
320
+ "epoch": 0.6060606060606061,
321
+ "grad_norm": 1.3515625,
322
+ "learning_rate": 1.8079965061022518e-06,
323
+ "loss": 1.7224,
324
+ "step": 220
325
+ },
326
+ {
327
+ "epoch": 0.6198347107438017,
328
+ "grad_norm": 1.328125,
329
+ "learning_rate": 1.7993982423612941e-06,
330
+ "loss": 1.7206,
331
+ "step": 225
332
+ },
333
+ {
334
+ "epoch": 0.6336088154269972,
335
+ "grad_norm": 1.3359375,
336
+ "learning_rate": 1.7906330478885174e-06,
337
+ "loss": 1.7252,
338
+ "step": 230
339
+ },
340
+ {
341
+ "epoch": 0.6473829201101928,
342
+ "grad_norm": 1.359375,
343
+ "learning_rate": 1.7817027530361174e-06,
344
+ "loss": 1.725,
345
+ "step": 235
346
+ },
347
+ {
348
+ "epoch": 0.6611570247933884,
349
+ "grad_norm": 1.3359375,
350
+ "learning_rate": 1.7726092226326315e-06,
351
+ "loss": 1.7189,
352
+ "step": 240
353
+ },
354
+ {
355
+ "epoch": 0.6749311294765841,
356
+ "grad_norm": 1.3359375,
357
+ "learning_rate": 1.7633543555935245e-06,
358
+ "loss": 1.719,
359
+ "step": 245
360
+ },
361
+ {
362
+ "epoch": 0.6887052341597796,
363
+ "grad_norm": 1.3359375,
364
+ "learning_rate": 1.7539400845246564e-06,
365
+ "loss": 1.7121,
366
+ "step": 250
367
+ },
368
+ {
369
+ "epoch": 0.7024793388429752,
370
+ "grad_norm": 1.3046875,
371
+ "learning_rate": 1.744368375318715e-06,
372
+ "loss": 1.7198,
373
+ "step": 255
374
+ },
375
+ {
376
+ "epoch": 0.7162534435261708,
377
+ "grad_norm": 1.3359375,
378
+ "learning_rate": 1.7346412267446958e-06,
379
+ "loss": 1.7149,
380
+ "step": 260
381
+ },
382
+ {
383
+ "epoch": 0.7300275482093664,
384
+ "grad_norm": 1.3515625,
385
+ "learning_rate": 1.724760670030521e-06,
386
+ "loss": 1.713,
387
+ "step": 265
388
+ },
389
+ {
390
+ "epoch": 0.743801652892562,
391
+ "grad_norm": 1.3046875,
392
+ "learning_rate": 1.7147287684388738e-06,
393
+ "loss": 1.7192,
394
+ "step": 270
395
+ },
396
+ {
397
+ "epoch": 0.7575757575757576,
398
+ "grad_norm": 1.359375,
399
+ "learning_rate": 1.7045476168363498e-06,
400
+ "loss": 1.721,
401
+ "step": 275
402
+ },
403
+ {
404
+ "epoch": 0.7713498622589532,
405
+ "grad_norm": 1.3359375,
406
+ "learning_rate": 1.6942193412560043e-06,
407
+ "loss": 1.7102,
408
+ "step": 280
409
+ },
410
+ {
411
+ "epoch": 0.7851239669421488,
412
+ "grad_norm": 1.34375,
413
+ "learning_rate": 1.6837460984533934e-06,
414
+ "loss": 1.7218,
415
+ "step": 285
416
+ },
417
+ {
418
+ "epoch": 0.7988980716253443,
419
+ "grad_norm": 1.2890625,
420
+ "learning_rate": 1.6731300754562008e-06,
421
+ "loss": 1.7107,
422
+ "step": 290
423
+ },
424
+ {
425
+ "epoch": 0.8126721763085399,
426
+ "grad_norm": 1.34375,
427
+ "learning_rate": 1.6623734891075385e-06,
428
+ "loss": 1.7138,
429
+ "step": 295
430
+ },
431
+ {
432
+ "epoch": 0.8264462809917356,
433
+ "grad_norm": 1.3203125,
434
+ "learning_rate": 1.6514785856030272e-06,
435
+ "loss": 1.7112,
436
+ "step": 300
437
+ },
438
+ {
439
+ "epoch": 0.8264462809917356,
440
+ "eval_loss": 1.7193405628204346,
441
+ "eval_runtime": 8.403,
442
+ "eval_samples_per_second": 83.779,
443
+ "eval_steps_per_second": 2.618,
444
+ "step": 300
445
+ },
446
+ {
447
+ "epoch": 0.8402203856749312,
448
+ "grad_norm": 1.359375,
449
+ "learning_rate": 1.640447640021744e-06,
450
+ "loss": 1.7211,
451
+ "step": 305
452
+ },
453
+ {
454
+ "epoch": 0.8539944903581267,
455
+ "grad_norm": 1.2890625,
456
+ "learning_rate": 1.6292829558511376e-06,
457
+ "loss": 1.7142,
458
+ "step": 310
459
+ },
460
+ {
461
+ "epoch": 0.8677685950413223,
462
+ "grad_norm": 1.328125,
463
+ "learning_rate": 1.6179868645060162e-06,
464
+ "loss": 1.7028,
465
+ "step": 315
466
+ },
467
+ {
468
+ "epoch": 0.8815426997245179,
469
+ "grad_norm": 1.34375,
470
+ "learning_rate": 1.6065617248416967e-06,
471
+ "loss": 1.7091,
472
+ "step": 320
473
+ },
474
+ {
475
+ "epoch": 0.8953168044077136,
476
+ "grad_norm": 1.3125,
477
+ "learning_rate": 1.59500992266143e-06,
478
+ "loss": 1.7143,
479
+ "step": 325
480
+ },
481
+ {
482
+ "epoch": 0.9090909090909091,
483
+ "grad_norm": 1.3515625,
484
+ "learning_rate": 1.5833338702181959e-06,
485
+ "loss": 1.7105,
486
+ "step": 330
487
+ },
488
+ {
489
+ "epoch": 0.9228650137741047,
490
+ "grad_norm": 1.40625,
491
+ "learning_rate": 1.5715360057109744e-06,
492
+ "loss": 1.7146,
493
+ "step": 335
494
+ },
495
+ {
496
+ "epoch": 0.9366391184573003,
497
+ "grad_norm": 1.4140625,
498
+ "learning_rate": 1.5596187927755993e-06,
499
+ "loss": 1.7145,
500
+ "step": 340
501
+ },
502
+ {
503
+ "epoch": 0.9504132231404959,
504
+ "grad_norm": 1.3828125,
505
+ "learning_rate": 1.5475847199703033e-06,
506
+ "loss": 1.7099,
507
+ "step": 345
508
+ },
509
+ {
510
+ "epoch": 0.9641873278236914,
511
+ "grad_norm": 1.3359375,
512
+ "learning_rate": 1.535436300256053e-06,
513
+ "loss": 1.7143,
514
+ "step": 350
515
+ },
516
+ {
517
+ "epoch": 0.977961432506887,
518
+ "grad_norm": 1.3828125,
519
+ "learning_rate": 1.523176070471793e-06,
520
+ "loss": 1.7131,
521
+ "step": 355
522
+ },
523
+ {
524
+ "epoch": 0.9917355371900827,
525
+ "grad_norm": 1.34375,
526
+ "learning_rate": 1.5108065908047014e-06,
527
+ "loss": 1.7248,
528
+ "step": 360
529
+ },
530
+ {
531
+ "epoch": 1.0055096418732783,
532
+ "grad_norm": 1.3203125,
533
+ "learning_rate": 1.4983304442555698e-06,
534
+ "loss": 1.7079,
535
+ "step": 365
536
+ },
537
+ {
538
+ "epoch": 1.019283746556474,
539
+ "grad_norm": 1.3671875,
540
+ "learning_rate": 1.4857502360994204e-06,
541
+ "loss": 1.712,
542
+ "step": 370
543
+ },
544
+ {
545
+ "epoch": 1.0330578512396693,
546
+ "grad_norm": 1.34375,
547
+ "learning_rate": 1.4730685933414714e-06,
548
+ "loss": 1.703,
549
+ "step": 375
550
+ },
551
+ {
552
+ "epoch": 1.046831955922865,
553
+ "grad_norm": 1.34375,
554
+ "learning_rate": 1.4602881641685643e-06,
555
+ "loss": 1.6944,
556
+ "step": 380
557
+ },
558
+ {
559
+ "epoch": 1.0606060606060606,
560
+ "grad_norm": 1.375,
561
+ "learning_rate": 1.4474116173961668e-06,
562
+ "loss": 1.7032,
563
+ "step": 385
564
+ },
565
+ {
566
+ "epoch": 1.0743801652892562,
567
+ "grad_norm": 1.375,
568
+ "learning_rate": 1.4344416419110728e-06,
569
+ "loss": 1.7122,
570
+ "step": 390
571
+ },
572
+ {
573
+ "epoch": 1.0881542699724518,
574
+ "grad_norm": 1.359375,
575
+ "learning_rate": 1.4213809461099033e-06,
576
+ "loss": 1.703,
577
+ "step": 395
578
+ },
579
+ {
580
+ "epoch": 1.1019283746556474,
581
+ "grad_norm": 1.3515625,
582
+ "learning_rate": 1.4082322573335422e-06,
583
+ "loss": 1.703,
584
+ "step": 400
585
+ },
586
+ {
587
+ "epoch": 1.115702479338843,
588
+ "grad_norm": 1.328125,
589
+ "learning_rate": 1.394998321297608e-06,
590
+ "loss": 1.7024,
591
+ "step": 405
592
+ },
593
+ {
594
+ "epoch": 1.1294765840220387,
595
+ "grad_norm": 1.359375,
596
+ "learning_rate": 1.3816819015190943e-06,
597
+ "loss": 1.7058,
598
+ "step": 410
599
+ },
600
+ {
601
+ "epoch": 1.1432506887052343,
602
+ "grad_norm": 1.34375,
603
+ "learning_rate": 1.3682857787392905e-06,
604
+ "loss": 1.6987,
605
+ "step": 415
606
+ },
607
+ {
608
+ "epoch": 1.1570247933884297,
609
+ "grad_norm": 1.375,
610
+ "learning_rate": 1.3548127503431038e-06,
611
+ "loss": 1.7029,
612
+ "step": 420
613
+ },
614
+ {
615
+ "epoch": 1.1707988980716253,
616
+ "grad_norm": 1.375,
617
+ "learning_rate": 1.3412656297749135e-06,
618
+ "loss": 1.6998,
619
+ "step": 425
620
+ },
621
+ {
622
+ "epoch": 1.184573002754821,
623
+ "grad_norm": 1.3828125,
624
+ "learning_rate": 1.327647245951058e-06,
625
+ "loss": 1.7051,
626
+ "step": 430
627
+ },
628
+ {
629
+ "epoch": 1.1983471074380165,
630
+ "grad_norm": 1.3359375,
631
+ "learning_rate": 1.3139604426691072e-06,
632
+ "loss": 1.7065,
633
+ "step": 435
634
+ },
635
+ {
636
+ "epoch": 1.2121212121212122,
637
+ "grad_norm": 1.3359375,
638
+ "learning_rate": 1.300208078014014e-06,
639
+ "loss": 1.7019,
640
+ "step": 440
641
+ },
642
+ {
643
+ "epoch": 1.2258953168044078,
644
+ "grad_norm": 1.359375,
645
+ "learning_rate": 1.2863930237612896e-06,
646
+ "loss": 1.6999,
647
+ "step": 445
648
+ },
649
+ {
650
+ "epoch": 1.2396694214876034,
651
+ "grad_norm": 1.3203125,
652
+ "learning_rate": 1.2725181647773174e-06,
653
+ "loss": 1.7046,
654
+ "step": 450
655
+ },
656
+ {
657
+ "epoch": 1.2534435261707988,
658
+ "grad_norm": 1.390625,
659
+ "learning_rate": 1.2585863984169343e-06,
660
+ "loss": 1.7069,
661
+ "step": 455
662
+ },
663
+ {
664
+ "epoch": 1.2672176308539944,
665
+ "grad_norm": 1.359375,
666
+ "learning_rate": 1.2446006339184035e-06,
667
+ "loss": 1.7025,
668
+ "step": 460
669
+ },
670
+ {
671
+ "epoch": 1.28099173553719,
672
+ "grad_norm": 1.40625,
673
+ "learning_rate": 1.2305637917959058e-06,
674
+ "loss": 1.7016,
675
+ "step": 465
676
+ },
677
+ {
678
+ "epoch": 1.2947658402203857,
679
+ "grad_norm": 1.3203125,
680
+ "learning_rate": 1.2164788032296755e-06,
681
+ "loss": 1.6997,
682
+ "step": 470
683
+ },
684
+ {
685
+ "epoch": 1.3085399449035813,
686
+ "grad_norm": 1.328125,
687
+ "learning_rate": 1.2023486094539124e-06,
688
+ "loss": 1.7003,
689
+ "step": 475
690
+ },
691
+ {
692
+ "epoch": 1.322314049586777,
693
+ "grad_norm": 1.3671875,
694
+ "learning_rate": 1.1881761611425888e-06,
695
+ "loss": 1.6885,
696
+ "step": 480
697
+ },
698
+ {
699
+ "epoch": 1.3360881542699725,
700
+ "grad_norm": 1.375,
701
+ "learning_rate": 1.1739644177932907e-06,
702
+ "loss": 1.7,
703
+ "step": 485
704
+ },
705
+ {
706
+ "epoch": 1.3498622589531681,
707
+ "grad_norm": 1.34375,
708
+ "learning_rate": 1.159716347109213e-06,
709
+ "loss": 1.6989,
710
+ "step": 490
711
+ },
712
+ {
713
+ "epoch": 1.3636363636363638,
714
+ "grad_norm": 1.3671875,
715
+ "learning_rate": 1.1454349243794419e-06,
716
+ "loss": 1.7009,
717
+ "step": 495
718
+ },
719
+ {
720
+ "epoch": 1.3774104683195592,
721
+ "grad_norm": 1.359375,
722
+ "learning_rate": 1.1311231318576545e-06,
723
+ "loss": 1.7007,
724
+ "step": 500
725
+ },
726
+ {
727
+ "epoch": 1.3911845730027548,
728
+ "grad_norm": 1.3828125,
729
+ "learning_rate": 1.1167839581393628e-06,
730
+ "loss": 1.6918,
731
+ "step": 505
732
+ },
733
+ {
734
+ "epoch": 1.4049586776859504,
735
+ "grad_norm": 1.296875,
736
+ "learning_rate": 1.1024203975378335e-06,
737
+ "loss": 1.6975,
738
+ "step": 510
739
+ },
740
+ {
741
+ "epoch": 1.418732782369146,
742
+ "grad_norm": 1.3203125,
743
+ "learning_rate": 1.0880354494588138e-06,
744
+ "loss": 1.6998,
745
+ "step": 515
746
+ },
747
+ {
748
+ "epoch": 1.4325068870523416,
749
+ "grad_norm": 1.375,
750
+ "learning_rate": 1.073632117774195e-06,
751
+ "loss": 1.6973,
752
+ "step": 520
753
+ },
754
+ {
755
+ "epoch": 1.4462809917355373,
756
+ "grad_norm": 1.296875,
757
+ "learning_rate": 1.0592134101947417e-06,
758
+ "loss": 1.7023,
759
+ "step": 525
760
+ },
761
+ {
762
+ "epoch": 1.4600550964187327,
763
+ "grad_norm": 1.3125,
764
+ "learning_rate": 1.0447823376420206e-06,
765
+ "loss": 1.707,
766
+ "step": 530
767
+ },
768
+ {
769
+ "epoch": 1.4738292011019283,
770
+ "grad_norm": 1.34375,
771
+ "learning_rate": 1.0303419136196575e-06,
772
+ "loss": 1.6916,
773
+ "step": 535
774
+ },
775
+ {
776
+ "epoch": 1.487603305785124,
777
+ "grad_norm": 1.2890625,
778
+ "learning_rate": 1.0158951535840576e-06,
779
+ "loss": 1.6977,
780
+ "step": 540
781
+ },
782
+ {
783
+ "epoch": 1.5013774104683195,
784
+ "grad_norm": 1.265625,
785
+ "learning_rate": 1.0014450743147145e-06,
786
+ "loss": 1.699,
787
+ "step": 545
788
+ },
789
+ {
790
+ "epoch": 1.5151515151515151,
791
+ "grad_norm": 1.3046875,
792
+ "learning_rate": 9.869946932842466e-07,
793
+ "loss": 1.6907,
794
+ "step": 550
795
+ },
796
+ {
797
+ "epoch": 1.5289256198347108,
798
+ "grad_norm": 1.3046875,
799
+ "learning_rate": 9.725470280282855e-07,
800
+ "loss": 1.7001,
801
+ "step": 555
802
+ },
803
+ {
804
+ "epoch": 1.5426997245179064,
805
+ "grad_norm": 1.3125,
806
+ "learning_rate": 9.581050955153545e-07,
807
+ "loss": 1.7054,
808
+ "step": 560
809
+ },
810
+ {
811
+ "epoch": 1.556473829201102,
812
+ "grad_norm": 1.296875,
813
+ "learning_rate": 9.43671911516861e-07,
814
+ "loss": 1.687,
815
+ "step": 565
816
+ },
817
+ {
818
+ "epoch": 1.5702479338842976,
819
+ "grad_norm": 1.3203125,
820
+ "learning_rate": 9.292504899773453e-07,
821
+ "loss": 1.7055,
822
+ "step": 570
823
+ },
824
+ {
825
+ "epoch": 1.5840220385674932,
826
+ "grad_norm": 1.28125,
827
+ "learning_rate": 9.148438423851041e-07,
828
+ "loss": 1.6975,
829
+ "step": 575
830
+ },
831
+ {
832
+ "epoch": 1.5977961432506889,
833
+ "grad_norm": 1.2890625,
834
+ "learning_rate": 9.00454977143331e-07,
835
+ "loss": 1.6997,
836
+ "step": 580
837
+ },
838
+ {
839
+ "epoch": 1.6115702479338843,
840
+ "grad_norm": 1.296875,
841
+ "learning_rate": 8.860868989419017e-07,
842
+ "loss": 1.6983,
843
+ "step": 585
844
+ },
845
+ {
846
+ "epoch": 1.6253443526170799,
847
+ "grad_norm": 1.3515625,
848
+ "learning_rate": 8.717426081299308e-07,
849
+ "loss": 1.6995,
850
+ "step": 590
851
+ },
852
+ {
853
+ "epoch": 1.6391184573002755,
854
+ "grad_norm": 1.296875,
855
+ "learning_rate": 8.574251000892386e-07,
856
+ "loss": 1.6948,
857
+ "step": 595
858
+ },
859
+ {
860
+ "epoch": 1.6528925619834711,
861
+ "grad_norm": 1.3515625,
862
+ "learning_rate": 8.431373646088549e-07,
863
+ "loss": 1.6961,
864
+ "step": 600
865
+ },
866
+ {
867
+ "epoch": 1.6528925619834711,
868
+ "eval_loss": 1.7039618492126465,
869
+ "eval_runtime": 8.3736,
870
+ "eval_samples_per_second": 84.073,
871
+ "eval_steps_per_second": 2.627,
872
+ "step": 600
873
+ },
874
+ {
875
+ "epoch": 1.6666666666666665,
876
+ "grad_norm": 1.2890625,
877
+ "learning_rate": 8.288823852606882e-07,
878
+ "loss": 1.6915,
879
+ "step": 605
880
+ },
881
+ {
882
+ "epoch": 1.6804407713498621,
883
+ "grad_norm": 1.296875,
884
+ "learning_rate": 8.14663138776496e-07,
885
+ "loss": 1.7006,
886
+ "step": 610
887
+ },
888
+ {
889
+ "epoch": 1.6942148760330578,
890
+ "grad_norm": 1.2890625,
891
+ "learning_rate": 8.004825944262805e-07,
892
+ "loss": 1.7029,
893
+ "step": 615
894
+ },
895
+ {
896
+ "epoch": 1.7079889807162534,
897
+ "grad_norm": 1.375,
898
+ "learning_rate": 7.863437133982471e-07,
899
+ "loss": 1.6942,
900
+ "step": 620
901
+ },
902
+ {
903
+ "epoch": 1.721763085399449,
904
+ "grad_norm": 1.3125,
905
+ "learning_rate": 7.722494481804445e-07,
906
+ "loss": 1.6927,
907
+ "step": 625
908
+ },
909
+ {
910
+ "epoch": 1.7355371900826446,
911
+ "grad_norm": 1.3046875,
912
+ "learning_rate": 7.582027419442268e-07,
913
+ "loss": 1.6929,
914
+ "step": 630
915
+ },
916
+ {
917
+ "epoch": 1.7493112947658402,
918
+ "grad_norm": 1.28125,
919
+ "learning_rate": 7.442065279296578e-07,
920
+ "loss": 1.702,
921
+ "step": 635
922
+ },
923
+ {
924
+ "epoch": 1.7630853994490359,
925
+ "grad_norm": 1.28125,
926
+ "learning_rate": 7.302637288329915e-07,
927
+ "loss": 1.6971,
928
+ "step": 640
929
+ },
930
+ {
931
+ "epoch": 1.7768595041322315,
932
+ "grad_norm": 1.2734375,
933
+ "learning_rate": 7.163772561963519e-07,
934
+ "loss": 1.6923,
935
+ "step": 645
936
+ },
937
+ {
938
+ "epoch": 1.790633608815427,
939
+ "grad_norm": 1.28125,
940
+ "learning_rate": 7.02550009799745e-07,
941
+ "loss": 1.6883,
942
+ "step": 650
943
+ },
944
+ {
945
+ "epoch": 1.8044077134986227,
946
+ "grad_norm": 1.3046875,
947
+ "learning_rate": 6.887848770555234e-07,
948
+ "loss": 1.6991,
949
+ "step": 655
950
+ },
951
+ {
952
+ "epoch": 1.8181818181818183,
953
+ "grad_norm": 1.296875,
954
+ "learning_rate": 6.750847324054374e-07,
955
+ "loss": 1.7069,
956
+ "step": 660
957
+ },
958
+ {
959
+ "epoch": 1.8319559228650137,
960
+ "grad_norm": 1.2890625,
961
+ "learning_rate": 6.614524367203906e-07,
962
+ "loss": 1.6979,
963
+ "step": 665
964
+ },
965
+ {
966
+ "epoch": 1.8457300275482094,
967
+ "grad_norm": 1.3046875,
968
+ "learning_rate": 6.478908367030338e-07,
969
+ "loss": 1.6829,
970
+ "step": 670
971
+ },
972
+ {
973
+ "epoch": 1.859504132231405,
974
+ "grad_norm": 1.3125,
975
+ "learning_rate": 6.344027642933128e-07,
976
+ "loss": 1.692,
977
+ "step": 675
978
+ },
979
+ {
980
+ "epoch": 1.8732782369146006,
981
+ "grad_norm": 1.28125,
982
+ "learning_rate": 6.209910360771033e-07,
983
+ "loss": 1.703,
984
+ "step": 680
985
+ },
986
+ {
987
+ "epoch": 1.887052341597796,
988
+ "grad_norm": 1.328125,
989
+ "learning_rate": 6.076584526980484e-07,
990
+ "loss": 1.706,
991
+ "step": 685
992
+ },
993
+ {
994
+ "epoch": 1.9008264462809916,
995
+ "grad_norm": 1.2890625,
996
+ "learning_rate": 5.944077982727285e-07,
997
+ "loss": 1.6906,
998
+ "step": 690
999
+ },
1000
+ {
1001
+ "epoch": 1.9146005509641872,
1002
+ "grad_norm": 1.2734375,
1003
+ "learning_rate": 5.812418398092787e-07,
1004
+ "loss": 1.6982,
1005
+ "step": 695
1006
+ },
1007
+ {
1008
+ "epoch": 1.9283746556473829,
1009
+ "grad_norm": 1.3359375,
1010
+ "learning_rate": 5.681633266295834e-07,
1011
+ "loss": 1.6917,
1012
+ "step": 700
1013
+ },
1014
+ {
1015
+ "epoch": 1.9421487603305785,
1016
+ "grad_norm": 1.28125,
1017
+ "learning_rate": 5.551749897951582e-07,
1018
+ "loss": 1.7014,
1019
+ "step": 705
1020
+ },
1021
+ {
1022
+ "epoch": 1.955922865013774,
1023
+ "grad_norm": 1.2734375,
1024
+ "learning_rate": 5.422795415368518e-07,
1025
+ "loss": 1.6861,
1026
+ "step": 710
1027
+ },
1028
+ {
1029
+ "epoch": 1.9696969696969697,
1030
+ "grad_norm": 1.3046875,
1031
+ "learning_rate": 5.294796746884745e-07,
1032
+ "loss": 1.6953,
1033
+ "step": 715
1034
+ },
1035
+ {
1036
+ "epoch": 1.9834710743801653,
1037
+ "grad_norm": 1.3359375,
1038
+ "learning_rate": 5.167780621244801e-07,
1039
+ "loss": 1.6973,
1040
+ "step": 720
1041
+ },
1042
+ {
1043
+ "epoch": 1.997245179063361,
1044
+ "grad_norm": 1.2890625,
1045
+ "learning_rate": 5.041773562018135e-07,
1046
+ "loss": 1.7019,
1047
+ "step": 725
1048
+ },
1049
+ {
1050
+ "epoch": 2.0110192837465566,
1051
+ "grad_norm": 1.265625,
1052
+ "learning_rate": 4.91680188206047e-07,
1053
+ "loss": 1.7011,
1054
+ "step": 730
1055
+ },
1056
+ {
1057
+ "epoch": 2.024793388429752,
1058
+ "grad_norm": 1.265625,
1059
+ "learning_rate": 4.792891678019115e-07,
1060
+ "loss": 1.7013,
1061
+ "step": 735
1062
+ },
1063
+ {
1064
+ "epoch": 2.038567493112948,
1065
+ "grad_norm": 1.2734375,
1066
+ "learning_rate": 4.6700688248834664e-07,
1067
+ "loss": 1.6895,
1068
+ "step": 740
1069
+ },
1070
+ {
1071
+ "epoch": 2.0523415977961434,
1072
+ "grad_norm": 1.3125,
1073
+ "learning_rate": 4.548358970581757e-07,
1074
+ "loss": 1.7029,
1075
+ "step": 745
1076
+ },
1077
+ {
1078
+ "epoch": 2.0661157024793386,
1079
+ "grad_norm": 1.2734375,
1080
+ "learning_rate": 4.427787530625278e-07,
1081
+ "loss": 1.6931,
1082
+ "step": 750
1083
+ },
1084
+ {
1085
+ "epoch": 2.0798898071625342,
1086
+ "grad_norm": 1.2734375,
1087
+ "learning_rate": 4.3083796828010675e-07,
1088
+ "loss": 1.6886,
1089
+ "step": 755
1090
+ },
1091
+ {
1092
+ "epoch": 2.09366391184573,
1093
+ "grad_norm": 1.3046875,
1094
+ "learning_rate": 4.190160361914292e-07,
1095
+ "loss": 1.6907,
1096
+ "step": 760
1097
+ },
1098
+ {
1099
+ "epoch": 2.1074380165289255,
1100
+ "grad_norm": 1.296875,
1101
+ "learning_rate": 4.07315425458134e-07,
1102
+ "loss": 1.6924,
1103
+ "step": 765
1104
+ },
1105
+ {
1106
+ "epoch": 2.121212121212121,
1107
+ "grad_norm": 1.28125,
1108
+ "learning_rate": 3.9573857940747537e-07,
1109
+ "loss": 1.7019,
1110
+ "step": 770
1111
+ },
1112
+ {
1113
+ "epoch": 2.1349862258953167,
1114
+ "grad_norm": 1.28125,
1115
+ "learning_rate": 3.8428791552210594e-07,
1116
+ "loss": 1.6975,
1117
+ "step": 775
1118
+ },
1119
+ {
1120
+ "epoch": 2.1487603305785123,
1121
+ "grad_norm": 1.2890625,
1122
+ "learning_rate": 3.729658249352563e-07,
1123
+ "loss": 1.6986,
1124
+ "step": 780
1125
+ },
1126
+ {
1127
+ "epoch": 2.162534435261708,
1128
+ "grad_norm": 1.3125,
1129
+ "learning_rate": 3.6177467193141886e-07,
1130
+ "loss": 1.6893,
1131
+ "step": 785
1132
+ },
1133
+ {
1134
+ "epoch": 2.1763085399449036,
1135
+ "grad_norm": 1.2734375,
1136
+ "learning_rate": 3.5071679345263537e-07,
1137
+ "loss": 1.6833,
1138
+ "step": 790
1139
+ },
1140
+ {
1141
+ "epoch": 2.190082644628099,
1142
+ "grad_norm": 1.2734375,
1143
+ "learning_rate": 3.397944986104968e-07,
1144
+ "loss": 1.693,
1145
+ "step": 795
1146
+ },
1147
+ {
1148
+ "epoch": 2.203856749311295,
1149
+ "grad_norm": 1.3046875,
1150
+ "learning_rate": 3.290100682039516e-07,
1151
+ "loss": 1.6978,
1152
+ "step": 800
1153
+ },
1154
+ {
1155
+ "epoch": 2.2176308539944904,
1156
+ "grad_norm": 1.28125,
1157
+ "learning_rate": 3.1836575424303034e-07,
1158
+ "loss": 1.7019,
1159
+ "step": 805
1160
+ },
1161
+ {
1162
+ "epoch": 2.231404958677686,
1163
+ "grad_norm": 1.2890625,
1164
+ "learning_rate": 3.078637794785791e-07,
1165
+ "loss": 1.6977,
1166
+ "step": 810
1167
+ },
1168
+ {
1169
+ "epoch": 2.2451790633608817,
1170
+ "grad_norm": 1.28125,
1171
+ "learning_rate": 2.9750633693810224e-07,
1172
+ "loss": 1.6898,
1173
+ "step": 815
1174
+ },
1175
+ {
1176
+ "epoch": 2.2589531680440773,
1177
+ "grad_norm": 1.28125,
1178
+ "learning_rate": 2.872955894678153e-07,
1179
+ "loss": 1.6915,
1180
+ "step": 820
1181
+ },
1182
+ {
1183
+ "epoch": 2.2727272727272725,
1184
+ "grad_norm": 1.265625,
1185
+ "learning_rate": 2.7723366928099754e-07,
1186
+ "loss": 1.6922,
1187
+ "step": 825
1188
+ },
1189
+ {
1190
+ "epoch": 2.2865013774104685,
1191
+ "grad_norm": 1.3046875,
1192
+ "learning_rate": 2.673226775127422e-07,
1193
+ "loss": 1.6922,
1194
+ "step": 830
1195
+ },
1196
+ {
1197
+ "epoch": 2.3002754820936637,
1198
+ "grad_norm": 1.3046875,
1199
+ "learning_rate": 2.5756468378119533e-07,
1200
+ "loss": 1.6873,
1201
+ "step": 835
1202
+ },
1203
+ {
1204
+ "epoch": 2.3140495867768593,
1205
+ "grad_norm": 1.265625,
1206
+ "learning_rate": 2.4796172575537934e-07,
1207
+ "loss": 1.7068,
1208
+ "step": 840
1209
+ },
1210
+ {
1211
+ "epoch": 2.327823691460055,
1212
+ "grad_norm": 1.3203125,
1213
+ "learning_rate": 2.3851580872968435e-07,
1214
+ "loss": 1.6993,
1215
+ "step": 845
1216
+ },
1217
+ {
1218
+ "epoch": 2.3415977961432506,
1219
+ "grad_norm": 1.3046875,
1220
+ "learning_rate": 2.292289052051224e-07,
1221
+ "loss": 1.6992,
1222
+ "step": 850
1223
+ },
1224
+ {
1225
+ "epoch": 2.355371900826446,
1226
+ "grad_norm": 1.2578125,
1227
+ "learning_rate": 2.2010295447742743e-07,
1228
+ "loss": 1.6891,
1229
+ "step": 855
1230
+ },
1231
+ {
1232
+ "epoch": 2.369146005509642,
1233
+ "grad_norm": 1.296875,
1234
+ "learning_rate": 2.111398622320927e-07,
1235
+ "loss": 1.6968,
1236
+ "step": 860
1237
+ },
1238
+ {
1239
+ "epoch": 2.3829201101928374,
1240
+ "grad_norm": 1.296875,
1241
+ "learning_rate": 2.0234150014642305e-07,
1242
+ "loss": 1.6946,
1243
+ "step": 865
1244
+ },
1245
+ {
1246
+ "epoch": 2.396694214876033,
1247
+ "grad_norm": 1.265625,
1248
+ "learning_rate": 1.937097054986915e-07,
1249
+ "loss": 1.6892,
1250
+ "step": 870
1251
+ },
1252
+ {
1253
+ "epoch": 2.4104683195592287,
1254
+ "grad_norm": 1.2578125,
1255
+ "learning_rate": 1.8524628078447602e-07,
1256
+ "loss": 1.6915,
1257
+ "step": 875
1258
+ },
1259
+ {
1260
+ "epoch": 2.4242424242424243,
1261
+ "grad_norm": 1.2890625,
1262
+ "learning_rate": 1.769529933402637e-07,
1263
+ "loss": 1.6946,
1264
+ "step": 880
1265
+ },
1266
+ {
1267
+ "epoch": 2.43801652892562,
1268
+ "grad_norm": 1.2578125,
1269
+ "learning_rate": 1.6883157497439349e-07,
1270
+ "loss": 1.6975,
1271
+ "step": 885
1272
+ },
1273
+ {
1274
+ "epoch": 2.4517906336088156,
1275
+ "grad_norm": 1.2734375,
1276
+ "learning_rate": 1.6088372160541962e-07,
1277
+ "loss": 1.6871,
1278
+ "step": 890
1279
+ },
1280
+ {
1281
+ "epoch": 2.465564738292011,
1282
+ "grad_norm": 1.2734375,
1283
+ "learning_rate": 1.531110929079681e-07,
1284
+ "loss": 1.6909,
1285
+ "step": 895
1286
+ },
1287
+ {
1288
+ "epoch": 2.479338842975207,
1289
+ "grad_norm": 1.28125,
1290
+ "learning_rate": 1.4551531196616396e-07,
1291
+ "loss": 1.6908,
1292
+ "step": 900
1293
+ },
1294
+ {
1295
+ "epoch": 2.479338842975207,
1296
+ "eval_loss": 1.7025996446609497,
1297
+ "eval_runtime": 8.3873,
1298
+ "eval_samples_per_second": 83.936,
1299
+ "eval_steps_per_second": 2.623,
1300
+ "step": 900
1301
+ },
1302
+ {
1303
+ "epoch": 2.4931129476584024,
1304
+ "grad_norm": 1.265625,
1305
+ "learning_rate": 1.3809796493469728e-07,
1306
+ "loss": 1.6981,
1307
+ "step": 905
1308
+ },
1309
+ {
1310
+ "epoch": 2.5068870523415976,
1311
+ "grad_norm": 1.3046875,
1312
+ "learning_rate": 1.3086060070760196e-07,
1313
+ "loss": 1.6902,
1314
+ "step": 910
1315
+ },
1316
+ {
1317
+ "epoch": 2.5206611570247937,
1318
+ "grad_norm": 1.3046875,
1319
+ "learning_rate": 1.23804730594814e-07,
1320
+ "loss": 1.6964,
1321
+ "step": 915
1322
+ },
1323
+ {
1324
+ "epoch": 2.534435261707989,
1325
+ "grad_norm": 1.2890625,
1326
+ "learning_rate": 1.1693182800658042e-07,
1327
+ "loss": 1.6884,
1328
+ "step": 920
1329
+ },
1330
+ {
1331
+ "epoch": 2.5482093663911844,
1332
+ "grad_norm": 1.2890625,
1333
+ "learning_rate": 1.102433281457802e-07,
1334
+ "loss": 1.6969,
1335
+ "step": 925
1336
+ },
1337
+ {
1338
+ "epoch": 2.56198347107438,
1339
+ "grad_norm": 1.265625,
1340
+ "learning_rate": 1.0374062770822411e-07,
1341
+ "loss": 1.7003,
1342
+ "step": 930
1343
+ },
1344
+ {
1345
+ "epoch": 2.5757575757575757,
1346
+ "grad_norm": 1.2578125,
1347
+ "learning_rate": 9.742508459099707e-08,
1348
+ "loss": 1.7095,
1349
+ "step": 935
1350
+ },
1351
+ {
1352
+ "epoch": 2.5895316804407713,
1353
+ "grad_norm": 1.2890625,
1354
+ "learning_rate": 9.129801760890076e-08,
1355
+ "loss": 1.7026,
1356
+ "step": 940
1357
+ },
1358
+ {
1359
+ "epoch": 2.603305785123967,
1360
+ "grad_norm": 1.2578125,
1361
+ "learning_rate": 8.536070621905811e-08,
1362
+ "loss": 1.6964,
1363
+ "step": 945
1364
+ },
1365
+ {
1366
+ "epoch": 2.6170798898071626,
1367
+ "grad_norm": 1.3046875,
1368
+ "learning_rate": 7.961439025373617e-08,
1369
+ "loss": 1.6984,
1370
+ "step": 950
1371
+ },
1372
+ {
1373
+ "epoch": 2.630853994490358,
1374
+ "grad_norm": 1.28125,
1375
+ "learning_rate": 7.40602696614444e-08,
1376
+ "loss": 1.7022,
1377
+ "step": 955
1378
+ },
1379
+ {
1380
+ "epoch": 2.644628099173554,
1381
+ "grad_norm": 1.2734375,
1382
+ "learning_rate": 6.869950425636095e-08,
1383
+ "loss": 1.6955,
1384
+ "step": 960
1385
+ },
1386
+ {
1387
+ "epoch": 2.6584022038567494,
1388
+ "grad_norm": 1.265625,
1389
+ "learning_rate": 6.353321347613815e-08,
1390
+ "loss": 1.6962,
1391
+ "step": 965
1392
+ },
1393
+ {
1394
+ "epoch": 2.672176308539945,
1395
+ "grad_norm": 1.3828125,
1396
+ "learning_rate": 5.856247614814292e-08,
1397
+ "loss": 1.6914,
1398
+ "step": 970
1399
+ },
1400
+ {
1401
+ "epoch": 2.6859504132231407,
1402
+ "grad_norm": 1.296875,
1403
+ "learning_rate": 5.3788330264174506e-08,
1404
+ "loss": 1.6934,
1405
+ "step": 975
1406
+ },
1407
+ {
1408
+ "epoch": 2.6997245179063363,
1409
+ "grad_norm": 1.28125,
1410
+ "learning_rate": 4.921177276371069e-08,
1411
+ "loss": 1.6947,
1412
+ "step": 980
1413
+ },
1414
+ {
1415
+ "epoch": 2.7134986225895315,
1416
+ "grad_norm": 1.2890625,
1417
+ "learning_rate": 4.483375932572597e-08,
1418
+ "loss": 1.6929,
1419
+ "step": 985
1420
+ },
1421
+ {
1422
+ "epoch": 2.7272727272727275,
1423
+ "grad_norm": 1.2890625,
1424
+ "learning_rate": 4.0655204169127156e-08,
1425
+ "loss": 1.6944,
1426
+ "step": 990
1427
+ },
1428
+ {
1429
+ "epoch": 2.7410468319559227,
1430
+ "grad_norm": 1.2734375,
1431
+ "learning_rate": 3.667697986184526e-08,
1432
+ "loss": 1.6898,
1433
+ "step": 995
1434
+ },
1435
+ {
1436
+ "epoch": 2.7548209366391183,
1437
+ "grad_norm": 1.2578125,
1438
+ "learning_rate": 3.2899917138625055e-08,
1439
+ "loss": 1.7061,
1440
+ "step": 1000
1441
+ },
1442
+ {
1443
+ "epoch": 2.768595041322314,
1444
+ "grad_norm": 1.3125,
1445
+ "learning_rate": 2.9324804727551055e-08,
1446
+ "loss": 1.6974,
1447
+ "step": 1005
1448
+ },
1449
+ {
1450
+ "epoch": 2.7823691460055096,
1451
+ "grad_norm": 1.2734375,
1452
+ "learning_rate": 2.5952389185344925e-08,
1453
+ "loss": 1.6892,
1454
+ "step": 1010
1455
+ },
1456
+ {
1457
+ "epoch": 2.796143250688705,
1458
+ "grad_norm": 1.2421875,
1459
+ "learning_rate": 2.2783374741469186e-08,
1460
+ "loss": 1.696,
1461
+ "step": 1015
1462
+ },
1463
+ {
1464
+ "epoch": 2.809917355371901,
1465
+ "grad_norm": 1.28125,
1466
+ "learning_rate": 1.9818423151069406e-08,
1467
+ "loss": 1.6879,
1468
+ "step": 1020
1469
+ },
1470
+ {
1471
+ "epoch": 2.8236914600550964,
1472
+ "grad_norm": 1.3125,
1473
+ "learning_rate": 1.705815355678619e-08,
1474
+ "loss": 1.6943,
1475
+ "step": 1025
1476
+ },
1477
+ {
1478
+ "epoch": 2.837465564738292,
1479
+ "grad_norm": 1.3046875,
1480
+ "learning_rate": 1.4503142359465925e-08,
1481
+ "loss": 1.6919,
1482
+ "step": 1030
1483
+ },
1484
+ {
1485
+ "epoch": 2.8512396694214877,
1486
+ "grad_norm": 1.2421875,
1487
+ "learning_rate": 1.215392309779617e-08,
1488
+ "loss": 1.6907,
1489
+ "step": 1035
1490
+ },
1491
+ {
1492
+ "epoch": 2.8650137741046833,
1493
+ "grad_norm": 1.2734375,
1494
+ "learning_rate": 1.0010986336891458e-08,
1495
+ "loss": 1.704,
1496
+ "step": 1040
1497
+ },
1498
+ {
1499
+ "epoch": 2.878787878787879,
1500
+ "grad_norm": 1.2578125,
1501
+ "learning_rate": 8.074779565854117e-09,
1502
+ "loss": 1.691,
1503
+ "step": 1045
1504
+ },
1505
+ {
1506
+ "epoch": 2.8925619834710745,
1507
+ "grad_norm": 1.3203125,
1508
+ "learning_rate": 6.34570710432869e-09,
1509
+ "loss": 1.6975,
1510
+ "step": 1050
1511
+ },
1512
+ {
1513
+ "epoch": 2.90633608815427,
1514
+ "grad_norm": 1.2734375,
1515
+ "learning_rate": 4.824130018072026e-09,
1516
+ "loss": 1.6918,
1517
+ "step": 1055
1518
+ },
1519
+ {
1520
+ "epoch": 2.9201101928374653,
1521
+ "grad_norm": 1.2890625,
1522
+ "learning_rate": 3.5103660435551465e-09,
1523
+ "loss": 1.6933,
1524
+ "step": 1060
1525
+ },
1526
+ {
1527
+ "epoch": 2.9338842975206614,
1528
+ "grad_norm": 1.3046875,
1529
+ "learning_rate": 2.4046895216136563e-09,
1530
+ "loss": 1.6872,
1531
+ "step": 1065
1532
+ },
1533
+ {
1534
+ "epoch": 2.9476584022038566,
1535
+ "grad_norm": 1.28125,
1536
+ "learning_rate": 1.5073313401594568e-09,
1537
+ "loss": 1.696,
1538
+ "step": 1070
1539
+ },
1540
+ {
1541
+ "epoch": 2.9614325068870526,
1542
+ "grad_norm": 1.234375,
1543
+ "learning_rate": 8.184788859667557e-10,
1544
+ "loss": 1.6964,
1545
+ "step": 1075
1546
+ },
1547
+ {
1548
+ "epoch": 2.975206611570248,
1549
+ "grad_norm": 1.3203125,
1550
+ "learning_rate": 3.3827600554170444e-10,
1551
+ "loss": 1.6941,
1552
+ "step": 1080
1553
+ },
1554
+ {
1555
+ "epoch": 2.9889807162534434,
1556
+ "grad_norm": 1.3125,
1557
+ "learning_rate": 6.682297508464608e-11,
1558
+ "loss": 1.6993,
1559
+ "step": 1085
1560
+ },
1561
+ {
1562
+ "epoch": 3.0,
1563
+ "step": 1089,
1564
+ "total_flos": 6.427401199279931e+18,
1565
+ "train_loss": 1.711617823146263,
1566
+ "train_runtime": 5145.0339,
1567
+ "train_samples_per_second": 13.545,
1568
+ "train_steps_per_second": 0.212
1569
+ }
1570
+ ],
1571
+ "logging_steps": 5,
1572
+ "max_steps": 1089,
1573
+ "num_input_tokens_seen": 0,
1574
+ "num_train_epochs": 3,
1575
+ "save_steps": 300,
1576
+ "stateful_callbacks": {
1577
+ "TrainerControl": {
1578
+ "args": {
1579
+ "should_epoch_stop": false,
1580
+ "should_evaluate": false,
1581
+ "should_log": false,
1582
+ "should_save": true,
1583
+ "should_training_stop": true
1584
+ },
1585
+ "attributes": {}
1586
+ }
1587
+ },
1588
+ "total_flos": 6.427401199279931e+18,
1589
+ "train_batch_size": 8,
1590
+ "trial_name": null,
1591
+ "trial_params": null
1592
+ }