polejowska
commited on
Commit
•
681580b
1
Parent(s):
dbbd150
End of training
Browse files- trainer_state.json +154 -14
trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"best_metric": 1.
|
3 |
-
"best_model_checkpoint": "cdetr-r50-cd45rb-all-4ah\\checkpoint-
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -148,18 +148,158 @@
|
|
148 |
"step": 23030
|
149 |
},
|
150 |
{
|
151 |
-
"epoch":
|
152 |
-
"
|
153 |
-
"
|
154 |
-
"
|
155 |
-
|
156 |
-
|
157 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
158 |
}
|
159 |
],
|
160 |
-
"max_steps":
|
161 |
-
"num_train_epochs":
|
162 |
-
"total_flos":
|
163 |
"trial_name": null,
|
164 |
"trial_params": null
|
165 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 1.7968701124191284,
|
3 |
+
"best_model_checkpoint": "cdetr-r50-cd45rb-all-4ah\\checkpoint-46060",
|
4 |
+
"epoch": 20.0,
|
5 |
+
"global_step": 46060,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
148 |
"step": 23030
|
149 |
},
|
150 |
{
|
151 |
+
"epoch": 11.0,
|
152 |
+
"learning_rate": 4.503473729917499e-06,
|
153 |
+
"loss": 1.8177,
|
154 |
+
"step": 25333
|
155 |
+
},
|
156 |
+
{
|
157 |
+
"epoch": 11.0,
|
158 |
+
"eval_loss": 1.8496007919311523,
|
159 |
+
"eval_runtime": 388.6824,
|
160 |
+
"eval_samples_per_second": 4.582,
|
161 |
+
"eval_steps_per_second": 0.574,
|
162 |
+
"step": 25333
|
163 |
+
},
|
164 |
+
{
|
165 |
+
"epoch": 12.0,
|
166 |
+
"learning_rate": 4.003690838037343e-06,
|
167 |
+
"loss": 1.8092,
|
168 |
+
"step": 27636
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"epoch": 12.0,
|
172 |
+
"eval_loss": 1.8583861589431763,
|
173 |
+
"eval_runtime": 382.9251,
|
174 |
+
"eval_samples_per_second": 4.651,
|
175 |
+
"eval_steps_per_second": 0.582,
|
176 |
+
"step": 27636
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 13.0,
|
180 |
+
"learning_rate": 3.5039079461571867e-06,
|
181 |
+
"loss": 1.8016,
|
182 |
+
"step": 29939
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 13.0,
|
186 |
+
"eval_loss": 1.8277424573898315,
|
187 |
+
"eval_runtime": 375.0626,
|
188 |
+
"eval_samples_per_second": 4.749,
|
189 |
+
"eval_steps_per_second": 0.595,
|
190 |
+
"step": 29939
|
191 |
+
},
|
192 |
+
{
|
193 |
+
"epoch": 14.0,
|
194 |
+
"learning_rate": 3.0039079461571864e-06,
|
195 |
+
"loss": 1.7961,
|
196 |
+
"step": 32242
|
197 |
+
},
|
198 |
+
{
|
199 |
+
"epoch": 14.0,
|
200 |
+
"eval_loss": 1.8229725360870361,
|
201 |
+
"eval_runtime": 215.0084,
|
202 |
+
"eval_samples_per_second": 8.283,
|
203 |
+
"eval_steps_per_second": 1.037,
|
204 |
+
"step": 32242
|
205 |
+
},
|
206 |
+
{
|
207 |
+
"epoch": 15.0,
|
208 |
+
"learning_rate": 2.50412505427703e-06,
|
209 |
+
"loss": 1.7853,
|
210 |
+
"step": 34545
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"epoch": 15.0,
|
214 |
+
"eval_loss": 1.8130303621292114,
|
215 |
+
"eval_runtime": 207.7071,
|
216 |
+
"eval_samples_per_second": 8.575,
|
217 |
+
"eval_steps_per_second": 1.074,
|
218 |
+
"step": 34545
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 16.0,
|
222 |
+
"learning_rate": 2.0043421623968738e-06,
|
223 |
+
"loss": 1.7769,
|
224 |
+
"step": 36848
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 16.0,
|
228 |
+
"eval_loss": 1.8098373413085938,
|
229 |
+
"eval_runtime": 212.184,
|
230 |
+
"eval_samples_per_second": 8.394,
|
231 |
+
"eval_steps_per_second": 1.051,
|
232 |
+
"step": 36848
|
233 |
+
},
|
234 |
+
{
|
235 |
+
"epoch": 17.0,
|
236 |
+
"learning_rate": 1.5045592705167174e-06,
|
237 |
+
"loss": 1.7708,
|
238 |
+
"step": 39151
|
239 |
+
},
|
240 |
+
{
|
241 |
+
"epoch": 17.0,
|
242 |
+
"eval_loss": 1.8029736280441284,
|
243 |
+
"eval_runtime": 208.316,
|
244 |
+
"eval_samples_per_second": 8.55,
|
245 |
+
"eval_steps_per_second": 1.07,
|
246 |
+
"step": 39151
|
247 |
+
},
|
248 |
+
{
|
249 |
+
"epoch": 18.0,
|
250 |
+
"learning_rate": 1.0049934867564047e-06,
|
251 |
+
"loss": 1.7626,
|
252 |
+
"step": 41454
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 18.0,
|
256 |
+
"eval_loss": 1.7993708848953247,
|
257 |
+
"eval_runtime": 207.7204,
|
258 |
+
"eval_samples_per_second": 8.574,
|
259 |
+
"eval_steps_per_second": 1.074,
|
260 |
+
"step": 41454
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 19.0,
|
264 |
+
"learning_rate": 5.049934867564047e-07,
|
265 |
+
"loss": 1.7596,
|
266 |
+
"step": 43757
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 19.0,
|
270 |
+
"eval_loss": 1.8018128871917725,
|
271 |
+
"eval_runtime": 211.9911,
|
272 |
+
"eval_samples_per_second": 8.401,
|
273 |
+
"eval_steps_per_second": 1.052,
|
274 |
+
"step": 43757
|
275 |
+
},
|
276 |
+
{
|
277 |
+
"epoch": 20.0,
|
278 |
+
"learning_rate": 5.210594876248372e-09,
|
279 |
+
"loss": 1.7547,
|
280 |
+
"step": 46060
|
281 |
+
},
|
282 |
+
{
|
283 |
+
"epoch": 20.0,
|
284 |
+
"eval_loss": 1.7968701124191284,
|
285 |
+
"eval_runtime": 439.0653,
|
286 |
+
"eval_samples_per_second": 4.056,
|
287 |
+
"eval_steps_per_second": 0.508,
|
288 |
+
"step": 46060
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 20.0,
|
292 |
+
"step": 46060,
|
293 |
+
"total_flos": 1.838847807523584e+20,
|
294 |
+
"train_loss": 0.8917218147440159,
|
295 |
+
"train_runtime": 42292.62,
|
296 |
+
"train_samples_per_second": 8.711,
|
297 |
+
"train_steps_per_second": 1.089
|
298 |
}
|
299 |
],
|
300 |
+
"max_steps": 46060,
|
301 |
+
"num_train_epochs": 20,
|
302 |
+
"total_flos": 1.838847807523584e+20,
|
303 |
"trial_name": null,
|
304 |
"trial_params": null
|
305 |
}
|