End of training
Browse files
all_results.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_accuracy": 0.6854754440961337,
|
4 |
+
"eval_loss": 0.8600627183914185,
|
5 |
+
"eval_runtime": 25.7064,
|
6 |
+
"eval_samples_per_second": 111.684,
|
7 |
+
"eval_steps_per_second": 3.501,
|
8 |
+
"total_flos": 6.004415924974301e+18,
|
9 |
+
"train_loss": 1.1543370756772484,
|
10 |
+
"train_runtime": 1817.3296,
|
11 |
+
"train_samples_per_second": 42.653,
|
12 |
+
"train_steps_per_second": 0.333
|
13 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_accuracy": 0.6854754440961337,
|
4 |
+
"eval_loss": 0.8600627183914185,
|
5 |
+
"eval_runtime": 25.7064,
|
6 |
+
"eval_samples_per_second": 111.684,
|
7 |
+
"eval_steps_per_second": 3.501
|
8 |
+
}
|
runs/Jan08_17-02-48_cd58e1e980d8/events.out.tfevents.1673199309.cd58e1e980d8.23.5
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:acc2d15897565dac51505f8b014d73a035f651fbde741bf1f2a7726ccc2b731b
|
3 |
+
size 363
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"total_flos": 6.004415924974301e+18,
|
4 |
+
"train_loss": 1.1543370756772484,
|
5 |
+
"train_runtime": 1817.3296,
|
6 |
+
"train_samples_per_second": 42.653,
|
7 |
+
"train_steps_per_second": 0.333
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,412 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.6854754440961337,
|
3 |
+
"best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuned-FER2013-0.0001/checkpoint-606",
|
4 |
+
"epoch": 3.0,
|
5 |
+
"global_step": 606,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.05,
|
12 |
+
"learning_rate": 1.6393442622950818e-05,
|
13 |
+
"loss": 1.9514,
|
14 |
+
"step": 10
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.1,
|
18 |
+
"learning_rate": 3.2786885245901635e-05,
|
19 |
+
"loss": 1.7098,
|
20 |
+
"step": 20
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.15,
|
24 |
+
"learning_rate": 4.918032786885246e-05,
|
25 |
+
"loss": 1.5711,
|
26 |
+
"step": 30
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.2,
|
30 |
+
"learning_rate": 6.557377049180327e-05,
|
31 |
+
"loss": 1.4081,
|
32 |
+
"step": 40
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.25,
|
36 |
+
"learning_rate": 8.19672131147541e-05,
|
37 |
+
"loss": 1.4072,
|
38 |
+
"step": 50
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.3,
|
42 |
+
"learning_rate": 9.836065573770493e-05,
|
43 |
+
"loss": 1.382,
|
44 |
+
"step": 60
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.35,
|
48 |
+
"learning_rate": 9.834862385321102e-05,
|
49 |
+
"loss": 1.3336,
|
50 |
+
"step": 70
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.4,
|
54 |
+
"learning_rate": 9.651376146788991e-05,
|
55 |
+
"loss": 1.3143,
|
56 |
+
"step": 80
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.45,
|
60 |
+
"learning_rate": 9.467889908256882e-05,
|
61 |
+
"loss": 1.3,
|
62 |
+
"step": 90
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.5,
|
66 |
+
"learning_rate": 9.284403669724772e-05,
|
67 |
+
"loss": 1.3335,
|
68 |
+
"step": 100
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.54,
|
72 |
+
"learning_rate": 9.100917431192661e-05,
|
73 |
+
"loss": 1.291,
|
74 |
+
"step": 110
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.59,
|
78 |
+
"learning_rate": 8.917431192660551e-05,
|
79 |
+
"loss": 1.2489,
|
80 |
+
"step": 120
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.64,
|
84 |
+
"learning_rate": 8.733944954128442e-05,
|
85 |
+
"loss": 1.2776,
|
86 |
+
"step": 130
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.69,
|
90 |
+
"learning_rate": 8.550458715596331e-05,
|
91 |
+
"loss": 1.2443,
|
92 |
+
"step": 140
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.74,
|
96 |
+
"learning_rate": 8.366972477064221e-05,
|
97 |
+
"loss": 1.214,
|
98 |
+
"step": 150
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.79,
|
102 |
+
"learning_rate": 8.183486238532112e-05,
|
103 |
+
"loss": 1.2379,
|
104 |
+
"step": 160
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.84,
|
108 |
+
"learning_rate": 8e-05,
|
109 |
+
"loss": 1.1672,
|
110 |
+
"step": 170
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.89,
|
114 |
+
"learning_rate": 7.816513761467891e-05,
|
115 |
+
"loss": 1.1764,
|
116 |
+
"step": 180
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.94,
|
120 |
+
"learning_rate": 7.63302752293578e-05,
|
121 |
+
"loss": 1.2006,
|
122 |
+
"step": 190
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.99,
|
126 |
+
"learning_rate": 7.44954128440367e-05,
|
127 |
+
"loss": 1.1632,
|
128 |
+
"step": 200
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 1.0,
|
132 |
+
"eval_accuracy": 0.6290491118077325,
|
133 |
+
"eval_loss": 0.9974912405014038,
|
134 |
+
"eval_runtime": 25.6751,
|
135 |
+
"eval_samples_per_second": 111.821,
|
136 |
+
"eval_steps_per_second": 3.505,
|
137 |
+
"step": 202
|
138 |
+
},
|
139 |
+
{
|
140 |
+
"epoch": 1.04,
|
141 |
+
"learning_rate": 7.266055045871561e-05,
|
142 |
+
"loss": 1.1413,
|
143 |
+
"step": 210
|
144 |
+
},
|
145 |
+
{
|
146 |
+
"epoch": 1.09,
|
147 |
+
"learning_rate": 7.08256880733945e-05,
|
148 |
+
"loss": 1.1803,
|
149 |
+
"step": 220
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 1.14,
|
153 |
+
"learning_rate": 6.89908256880734e-05,
|
154 |
+
"loss": 1.1526,
|
155 |
+
"step": 230
|
156 |
+
},
|
157 |
+
{
|
158 |
+
"epoch": 1.19,
|
159 |
+
"learning_rate": 6.715596330275231e-05,
|
160 |
+
"loss": 1.0817,
|
161 |
+
"step": 240
|
162 |
+
},
|
163 |
+
{
|
164 |
+
"epoch": 1.24,
|
165 |
+
"learning_rate": 6.53211009174312e-05,
|
166 |
+
"loss": 1.1358,
|
167 |
+
"step": 250
|
168 |
+
},
|
169 |
+
{
|
170 |
+
"epoch": 1.29,
|
171 |
+
"learning_rate": 6.34862385321101e-05,
|
172 |
+
"loss": 1.1496,
|
173 |
+
"step": 260
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 1.34,
|
177 |
+
"learning_rate": 6.165137614678899e-05,
|
178 |
+
"loss": 1.1992,
|
179 |
+
"step": 270
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 1.39,
|
183 |
+
"learning_rate": 5.9816513761467896e-05,
|
184 |
+
"loss": 1.1061,
|
185 |
+
"step": 280
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 1.44,
|
189 |
+
"learning_rate": 5.798165137614679e-05,
|
190 |
+
"loss": 1.1059,
|
191 |
+
"step": 290
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 1.49,
|
195 |
+
"learning_rate": 5.61467889908257e-05,
|
196 |
+
"loss": 1.1208,
|
197 |
+
"step": 300
|
198 |
+
},
|
199 |
+
{
|
200 |
+
"epoch": 1.53,
|
201 |
+
"learning_rate": 5.4311926605504594e-05,
|
202 |
+
"loss": 1.1173,
|
203 |
+
"step": 310
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"epoch": 1.58,
|
207 |
+
"learning_rate": 5.247706422018349e-05,
|
208 |
+
"loss": 1.1132,
|
209 |
+
"step": 320
|
210 |
+
},
|
211 |
+
{
|
212 |
+
"epoch": 1.63,
|
213 |
+
"learning_rate": 5.064220183486239e-05,
|
214 |
+
"loss": 1.0441,
|
215 |
+
"step": 330
|
216 |
+
},
|
217 |
+
{
|
218 |
+
"epoch": 1.68,
|
219 |
+
"learning_rate": 4.880733944954129e-05,
|
220 |
+
"loss": 1.0603,
|
221 |
+
"step": 340
|
222 |
+
},
|
223 |
+
{
|
224 |
+
"epoch": 1.73,
|
225 |
+
"learning_rate": 4.697247706422019e-05,
|
226 |
+
"loss": 1.1201,
|
227 |
+
"step": 350
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"epoch": 1.78,
|
231 |
+
"learning_rate": 4.513761467889909e-05,
|
232 |
+
"loss": 1.0933,
|
233 |
+
"step": 360
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 1.83,
|
237 |
+
"learning_rate": 4.3302752293577984e-05,
|
238 |
+
"loss": 1.0863,
|
239 |
+
"step": 370
|
240 |
+
},
|
241 |
+
{
|
242 |
+
"epoch": 1.88,
|
243 |
+
"learning_rate": 4.146788990825689e-05,
|
244 |
+
"loss": 1.0781,
|
245 |
+
"step": 380
|
246 |
+
},
|
247 |
+
{
|
248 |
+
"epoch": 1.93,
|
249 |
+
"learning_rate": 3.9633027522935785e-05,
|
250 |
+
"loss": 1.1139,
|
251 |
+
"step": 390
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"epoch": 1.98,
|
255 |
+
"learning_rate": 3.779816513761468e-05,
|
256 |
+
"loss": 1.0563,
|
257 |
+
"step": 400
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 2.0,
|
261 |
+
"eval_accuracy": 0.6614420062695925,
|
262 |
+
"eval_loss": 0.9350310564041138,
|
263 |
+
"eval_runtime": 27.29,
|
264 |
+
"eval_samples_per_second": 105.203,
|
265 |
+
"eval_steps_per_second": 3.298,
|
266 |
+
"step": 404
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 2.03,
|
270 |
+
"learning_rate": 3.596330275229358e-05,
|
271 |
+
"loss": 1.0391,
|
272 |
+
"step": 410
|
273 |
+
},
|
274 |
+
{
|
275 |
+
"epoch": 2.08,
|
276 |
+
"learning_rate": 3.4128440366972484e-05,
|
277 |
+
"loss": 0.974,
|
278 |
+
"step": 420
|
279 |
+
},
|
280 |
+
{
|
281 |
+
"epoch": 2.13,
|
282 |
+
"learning_rate": 3.229357798165138e-05,
|
283 |
+
"loss": 1.0585,
|
284 |
+
"step": 430
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 2.18,
|
288 |
+
"learning_rate": 3.0458715596330278e-05,
|
289 |
+
"loss": 1.0451,
|
290 |
+
"step": 440
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 2.23,
|
294 |
+
"learning_rate": 2.862385321100918e-05,
|
295 |
+
"loss": 0.9934,
|
296 |
+
"step": 450
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 2.28,
|
300 |
+
"learning_rate": 2.6788990825688076e-05,
|
301 |
+
"loss": 1.015,
|
302 |
+
"step": 460
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 2.33,
|
306 |
+
"learning_rate": 2.4954128440366977e-05,
|
307 |
+
"loss": 1.0505,
|
308 |
+
"step": 470
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 2.38,
|
312 |
+
"learning_rate": 2.3119266055045874e-05,
|
313 |
+
"loss": 1.0035,
|
314 |
+
"step": 480
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 2.43,
|
318 |
+
"learning_rate": 2.1284403669724774e-05,
|
319 |
+
"loss": 0.9752,
|
320 |
+
"step": 490
|
321 |
+
},
|
322 |
+
{
|
323 |
+
"epoch": 2.48,
|
324 |
+
"learning_rate": 1.944954128440367e-05,
|
325 |
+
"loss": 1.0124,
|
326 |
+
"step": 500
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 2.52,
|
330 |
+
"learning_rate": 1.7614678899082572e-05,
|
331 |
+
"loss": 1.0122,
|
332 |
+
"step": 510
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 2.57,
|
336 |
+
"learning_rate": 1.577981651376147e-05,
|
337 |
+
"loss": 1.0435,
|
338 |
+
"step": 520
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 2.62,
|
342 |
+
"learning_rate": 1.3944954128440368e-05,
|
343 |
+
"loss": 1.0102,
|
344 |
+
"step": 530
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 2.67,
|
348 |
+
"learning_rate": 1.2110091743119267e-05,
|
349 |
+
"loss": 1.0476,
|
350 |
+
"step": 540
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 2.72,
|
354 |
+
"learning_rate": 1.0275229357798166e-05,
|
355 |
+
"loss": 0.9989,
|
356 |
+
"step": 550
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 2.77,
|
360 |
+
"learning_rate": 8.440366972477065e-06,
|
361 |
+
"loss": 0.9549,
|
362 |
+
"step": 560
|
363 |
+
},
|
364 |
+
{
|
365 |
+
"epoch": 2.82,
|
366 |
+
"learning_rate": 6.605504587155964e-06,
|
367 |
+
"loss": 0.9696,
|
368 |
+
"step": 570
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 2.87,
|
372 |
+
"learning_rate": 4.7706422018348626e-06,
|
373 |
+
"loss": 1.0032,
|
374 |
+
"step": 580
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 2.92,
|
378 |
+
"learning_rate": 2.935779816513762e-06,
|
379 |
+
"loss": 0.9941,
|
380 |
+
"step": 590
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 2.97,
|
384 |
+
"learning_rate": 1.1009174311926608e-06,
|
385 |
+
"loss": 0.9564,
|
386 |
+
"step": 600
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 3.0,
|
390 |
+
"eval_accuracy": 0.6854754440961337,
|
391 |
+
"eval_loss": 0.8600627183914185,
|
392 |
+
"eval_runtime": 25.8777,
|
393 |
+
"eval_samples_per_second": 110.945,
|
394 |
+
"eval_steps_per_second": 3.478,
|
395 |
+
"step": 606
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 3.0,
|
399 |
+
"step": 606,
|
400 |
+
"total_flos": 6.004415924974301e+18,
|
401 |
+
"train_loss": 1.1543370756772484,
|
402 |
+
"train_runtime": 1817.3296,
|
403 |
+
"train_samples_per_second": 42.653,
|
404 |
+
"train_steps_per_second": 0.333
|
405 |
+
}
|
406 |
+
],
|
407 |
+
"max_steps": 606,
|
408 |
+
"num_train_epochs": 3,
|
409 |
+
"total_flos": 6.004415924974301e+18,
|
410 |
+
"trial_name": null,
|
411 |
+
"trial_params": null
|
412 |
+
}
|