End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +469 -0
all_results.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_accuracy": 0.9665621116174469,
|
4 |
+
"eval_loss": 0.09518735855817795,
|
5 |
+
"eval_runtime": 166.01,
|
6 |
+
"eval_samples_per_second": 101.783,
|
7 |
+
"eval_steps_per_second": 0.801,
|
8 |
+
"total_flos": 1.1536340155092173e+19,
|
9 |
+
"train_loss": 0.19495156454661536,
|
10 |
+
"train_runtime": 10434.9174,
|
11 |
+
"train_samples_per_second": 34.004,
|
12 |
+
"train_steps_per_second": 0.066
|
13 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"eval_accuracy": 0.9665621116174469,
|
4 |
+
"eval_loss": 0.09518735855817795,
|
5 |
+
"eval_runtime": 166.01,
|
6 |
+
"eval_samples_per_second": 101.783,
|
7 |
+
"eval_steps_per_second": 0.801
|
8 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 3.0,
|
3 |
+
"total_flos": 1.1536340155092173e+19,
|
4 |
+
"train_loss": 0.19495156454661536,
|
5 |
+
"train_runtime": 10434.9174,
|
6 |
+
"train_samples_per_second": 34.004,
|
7 |
+
"train_steps_per_second": 0.066
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,469 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.9665621116174469,
|
3 |
+
"best_model_checkpoint": "SwinMark2/checkpoint-693",
|
4 |
+
"epoch": 2.996756756756757,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 693,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.04,
|
13 |
+
"learning_rate": 7.142857142857143e-06,
|
14 |
+
"loss": 1.3098,
|
15 |
+
"step": 10
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 0.09,
|
19 |
+
"learning_rate": 1.4285714285714285e-05,
|
20 |
+
"loss": 1.1138,
|
21 |
+
"step": 20
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"epoch": 0.13,
|
25 |
+
"learning_rate": 2.1428571428571428e-05,
|
26 |
+
"loss": 0.8719,
|
27 |
+
"step": 30
|
28 |
+
},
|
29 |
+
{
|
30 |
+
"epoch": 0.17,
|
31 |
+
"learning_rate": 2.857142857142857e-05,
|
32 |
+
"loss": 0.6339,
|
33 |
+
"step": 40
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 0.22,
|
37 |
+
"learning_rate": 3.571428571428572e-05,
|
38 |
+
"loss": 0.4509,
|
39 |
+
"step": 50
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.26,
|
43 |
+
"learning_rate": 4.2857142857142856e-05,
|
44 |
+
"loss": 0.3485,
|
45 |
+
"step": 60
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.3,
|
49 |
+
"learning_rate": 5e-05,
|
50 |
+
"loss": 0.3103,
|
51 |
+
"step": 70
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.35,
|
55 |
+
"learning_rate": 4.919743178170145e-05,
|
56 |
+
"loss": 0.2529,
|
57 |
+
"step": 80
|
58 |
+
},
|
59 |
+
{
|
60 |
+
"epoch": 0.39,
|
61 |
+
"learning_rate": 4.839486356340289e-05,
|
62 |
+
"loss": 0.2251,
|
63 |
+
"step": 90
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"epoch": 0.43,
|
67 |
+
"learning_rate": 4.759229534510434e-05,
|
68 |
+
"loss": 0.2148,
|
69 |
+
"step": 100
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.48,
|
73 |
+
"learning_rate": 4.678972712680578e-05,
|
74 |
+
"loss": 0.2153,
|
75 |
+
"step": 110
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 0.52,
|
79 |
+
"learning_rate": 4.598715890850723e-05,
|
80 |
+
"loss": 0.1991,
|
81 |
+
"step": 120
|
82 |
+
},
|
83 |
+
{
|
84 |
+
"epoch": 0.56,
|
85 |
+
"learning_rate": 4.5184590690208673e-05,
|
86 |
+
"loss": 0.1931,
|
87 |
+
"step": 130
|
88 |
+
},
|
89 |
+
{
|
90 |
+
"epoch": 0.61,
|
91 |
+
"learning_rate": 4.438202247191011e-05,
|
92 |
+
"loss": 0.1893,
|
93 |
+
"step": 140
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.65,
|
97 |
+
"learning_rate": 4.357945425361156e-05,
|
98 |
+
"loss": 0.1806,
|
99 |
+
"step": 150
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.69,
|
103 |
+
"learning_rate": 4.2776886035313e-05,
|
104 |
+
"loss": 0.1581,
|
105 |
+
"step": 160
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"epoch": 0.74,
|
109 |
+
"learning_rate": 4.197431781701445e-05,
|
110 |
+
"loss": 0.1654,
|
111 |
+
"step": 170
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"epoch": 0.78,
|
115 |
+
"learning_rate": 4.117174959871589e-05,
|
116 |
+
"loss": 0.1606,
|
117 |
+
"step": 180
|
118 |
+
},
|
119 |
+
{
|
120 |
+
"epoch": 0.82,
|
121 |
+
"learning_rate": 4.036918138041734e-05,
|
122 |
+
"loss": 0.1579,
|
123 |
+
"step": 190
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 0.86,
|
127 |
+
"learning_rate": 3.956661316211878e-05,
|
128 |
+
"loss": 0.1584,
|
129 |
+
"step": 200
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.91,
|
133 |
+
"learning_rate": 3.876404494382023e-05,
|
134 |
+
"loss": 0.1931,
|
135 |
+
"step": 210
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.95,
|
139 |
+
"learning_rate": 3.796147672552167e-05,
|
140 |
+
"loss": 0.1474,
|
141 |
+
"step": 220
|
142 |
+
},
|
143 |
+
{
|
144 |
+
"epoch": 0.99,
|
145 |
+
"learning_rate": 3.715890850722312e-05,
|
146 |
+
"loss": 0.1407,
|
147 |
+
"step": 230
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"epoch": 1.0,
|
151 |
+
"eval_accuracy": 0.9586317097709652,
|
152 |
+
"eval_loss": 0.12301469594240189,
|
153 |
+
"eval_runtime": 259.331,
|
154 |
+
"eval_samples_per_second": 65.156,
|
155 |
+
"eval_steps_per_second": 0.513,
|
156 |
+
"step": 231
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 1.04,
|
160 |
+
"learning_rate": 3.6356340288924564e-05,
|
161 |
+
"loss": 0.1519,
|
162 |
+
"step": 240
|
163 |
+
},
|
164 |
+
{
|
165 |
+
"epoch": 1.08,
|
166 |
+
"learning_rate": 3.555377207062601e-05,
|
167 |
+
"loss": 0.1442,
|
168 |
+
"step": 250
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"epoch": 1.12,
|
172 |
+
"learning_rate": 3.4751203852327454e-05,
|
173 |
+
"loss": 0.1409,
|
174 |
+
"step": 260
|
175 |
+
},
|
176 |
+
{
|
177 |
+
"epoch": 1.17,
|
178 |
+
"learning_rate": 3.394863563402889e-05,
|
179 |
+
"loss": 0.1468,
|
180 |
+
"step": 270
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"epoch": 1.21,
|
184 |
+
"learning_rate": 3.314606741573034e-05,
|
185 |
+
"loss": 0.1337,
|
186 |
+
"step": 280
|
187 |
+
},
|
188 |
+
{
|
189 |
+
"epoch": 1.25,
|
190 |
+
"learning_rate": 3.234349919743178e-05,
|
191 |
+
"loss": 0.146,
|
192 |
+
"step": 290
|
193 |
+
},
|
194 |
+
{
|
195 |
+
"epoch": 1.3,
|
196 |
+
"learning_rate": 3.154093097913323e-05,
|
197 |
+
"loss": 0.1486,
|
198 |
+
"step": 300
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 1.34,
|
202 |
+
"learning_rate": 3.073836276083467e-05,
|
203 |
+
"loss": 0.1386,
|
204 |
+
"step": 310
|
205 |
+
},
|
206 |
+
{
|
207 |
+
"epoch": 1.38,
|
208 |
+
"learning_rate": 2.9935794542536115e-05,
|
209 |
+
"loss": 0.13,
|
210 |
+
"step": 320
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"epoch": 1.43,
|
214 |
+
"learning_rate": 2.913322632423756e-05,
|
215 |
+
"loss": 0.1192,
|
216 |
+
"step": 330
|
217 |
+
},
|
218 |
+
{
|
219 |
+
"epoch": 1.47,
|
220 |
+
"learning_rate": 2.8330658105939005e-05,
|
221 |
+
"loss": 0.1418,
|
222 |
+
"step": 340
|
223 |
+
},
|
224 |
+
{
|
225 |
+
"epoch": 1.51,
|
226 |
+
"learning_rate": 2.752808988764045e-05,
|
227 |
+
"loss": 0.1171,
|
228 |
+
"step": 350
|
229 |
+
},
|
230 |
+
{
|
231 |
+
"epoch": 1.56,
|
232 |
+
"learning_rate": 2.6725521669341895e-05,
|
233 |
+
"loss": 0.1251,
|
234 |
+
"step": 360
|
235 |
+
},
|
236 |
+
{
|
237 |
+
"epoch": 1.6,
|
238 |
+
"learning_rate": 2.592295345104334e-05,
|
239 |
+
"loss": 0.1261,
|
240 |
+
"step": 370
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 1.64,
|
244 |
+
"learning_rate": 2.5120385232744786e-05,
|
245 |
+
"loss": 0.1282,
|
246 |
+
"step": 380
|
247 |
+
},
|
248 |
+
{
|
249 |
+
"epoch": 1.69,
|
250 |
+
"learning_rate": 2.4317817014446228e-05,
|
251 |
+
"loss": 0.1235,
|
252 |
+
"step": 390
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 1.73,
|
256 |
+
"learning_rate": 2.3515248796147673e-05,
|
257 |
+
"loss": 0.1269,
|
258 |
+
"step": 400
|
259 |
+
},
|
260 |
+
{
|
261 |
+
"epoch": 1.77,
|
262 |
+
"learning_rate": 2.2712680577849118e-05,
|
263 |
+
"loss": 0.119,
|
264 |
+
"step": 410
|
265 |
+
},
|
266 |
+
{
|
267 |
+
"epoch": 1.82,
|
268 |
+
"learning_rate": 2.1910112359550563e-05,
|
269 |
+
"loss": 0.1296,
|
270 |
+
"step": 420
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 1.86,
|
274 |
+
"learning_rate": 2.1107544141252008e-05,
|
275 |
+
"loss": 0.1114,
|
276 |
+
"step": 430
|
277 |
+
},
|
278 |
+
{
|
279 |
+
"epoch": 1.9,
|
280 |
+
"learning_rate": 2.030497592295345e-05,
|
281 |
+
"loss": 0.1334,
|
282 |
+
"step": 440
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 1.95,
|
286 |
+
"learning_rate": 1.9502407704654895e-05,
|
287 |
+
"loss": 0.118,
|
288 |
+
"step": 450
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 1.99,
|
292 |
+
"learning_rate": 1.869983948635634e-05,
|
293 |
+
"loss": 0.1209,
|
294 |
+
"step": 460
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 2.0,
|
298 |
+
"eval_accuracy": 0.9630111854175297,
|
299 |
+
"eval_loss": 0.10658775269985199,
|
300 |
+
"eval_runtime": 231.585,
|
301 |
+
"eval_samples_per_second": 72.962,
|
302 |
+
"eval_steps_per_second": 0.574,
|
303 |
+
"step": 462
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 2.03,
|
307 |
+
"learning_rate": 1.7897271268057786e-05,
|
308 |
+
"loss": 0.1038,
|
309 |
+
"step": 470
|
310 |
+
},
|
311 |
+
{
|
312 |
+
"epoch": 2.08,
|
313 |
+
"learning_rate": 1.709470304975923e-05,
|
314 |
+
"loss": 0.1159,
|
315 |
+
"step": 480
|
316 |
+
},
|
317 |
+
{
|
318 |
+
"epoch": 2.12,
|
319 |
+
"learning_rate": 1.6292134831460676e-05,
|
320 |
+
"loss": 0.0992,
|
321 |
+
"step": 490
|
322 |
+
},
|
323 |
+
{
|
324 |
+
"epoch": 2.16,
|
325 |
+
"learning_rate": 1.548956661316212e-05,
|
326 |
+
"loss": 0.1166,
|
327 |
+
"step": 500
|
328 |
+
},
|
329 |
+
{
|
330 |
+
"epoch": 2.21,
|
331 |
+
"learning_rate": 1.4686998394863563e-05,
|
332 |
+
"loss": 0.1046,
|
333 |
+
"step": 510
|
334 |
+
},
|
335 |
+
{
|
336 |
+
"epoch": 2.25,
|
337 |
+
"learning_rate": 1.3884430176565008e-05,
|
338 |
+
"loss": 0.0994,
|
339 |
+
"step": 520
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 2.29,
|
343 |
+
"learning_rate": 1.3081861958266453e-05,
|
344 |
+
"loss": 0.124,
|
345 |
+
"step": 530
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 2.34,
|
349 |
+
"learning_rate": 1.2279293739967898e-05,
|
350 |
+
"loss": 0.1113,
|
351 |
+
"step": 540
|
352 |
+
},
|
353 |
+
{
|
354 |
+
"epoch": 2.38,
|
355 |
+
"learning_rate": 1.1476725521669344e-05,
|
356 |
+
"loss": 0.1106,
|
357 |
+
"step": 550
|
358 |
+
},
|
359 |
+
{
|
360 |
+
"epoch": 2.42,
|
361 |
+
"learning_rate": 1.0674157303370787e-05,
|
362 |
+
"loss": 0.1083,
|
363 |
+
"step": 560
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 2.46,
|
367 |
+
"learning_rate": 9.871589085072232e-06,
|
368 |
+
"loss": 0.1108,
|
369 |
+
"step": 570
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 2.51,
|
373 |
+
"learning_rate": 9.069020866773677e-06,
|
374 |
+
"loss": 0.1093,
|
375 |
+
"step": 580
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"epoch": 2.55,
|
379 |
+
"learning_rate": 8.26645264847512e-06,
|
380 |
+
"loss": 0.0975,
|
381 |
+
"step": 590
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 2.59,
|
385 |
+
"learning_rate": 7.463884430176565e-06,
|
386 |
+
"loss": 0.1041,
|
387 |
+
"step": 600
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 2.64,
|
391 |
+
"learning_rate": 6.66131621187801e-06,
|
392 |
+
"loss": 0.1157,
|
393 |
+
"step": 610
|
394 |
+
},
|
395 |
+
{
|
396 |
+
"epoch": 2.68,
|
397 |
+
"learning_rate": 5.858747993579455e-06,
|
398 |
+
"loss": 0.1005,
|
399 |
+
"step": 620
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 2.72,
|
403 |
+
"learning_rate": 5.056179775280899e-06,
|
404 |
+
"loss": 0.1097,
|
405 |
+
"step": 630
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"epoch": 2.77,
|
409 |
+
"learning_rate": 4.253611556982344e-06,
|
410 |
+
"loss": 0.1102,
|
411 |
+
"step": 640
|
412 |
+
},
|
413 |
+
{
|
414 |
+
"epoch": 2.81,
|
415 |
+
"learning_rate": 3.4510433386837885e-06,
|
416 |
+
"loss": 0.111,
|
417 |
+
"step": 650
|
418 |
+
},
|
419 |
+
{
|
420 |
+
"epoch": 2.85,
|
421 |
+
"learning_rate": 2.648475120385233e-06,
|
422 |
+
"loss": 0.0956,
|
423 |
+
"step": 660
|
424 |
+
},
|
425 |
+
{
|
426 |
+
"epoch": 2.9,
|
427 |
+
"learning_rate": 1.8459069020866775e-06,
|
428 |
+
"loss": 0.1079,
|
429 |
+
"step": 670
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 2.94,
|
433 |
+
"learning_rate": 1.043338683788122e-06,
|
434 |
+
"loss": 0.1047,
|
435 |
+
"step": 680
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"epoch": 2.98,
|
439 |
+
"learning_rate": 2.407704654895666e-07,
|
440 |
+
"loss": 0.0987,
|
441 |
+
"step": 690
|
442 |
+
},
|
443 |
+
{
|
444 |
+
"epoch": 3.0,
|
445 |
+
"eval_accuracy": 0.9665621116174469,
|
446 |
+
"eval_loss": 0.09518735855817795,
|
447 |
+
"eval_runtime": 181.4053,
|
448 |
+
"eval_samples_per_second": 93.145,
|
449 |
+
"eval_steps_per_second": 0.733,
|
450 |
+
"step": 693
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 3.0,
|
454 |
+
"step": 693,
|
455 |
+
"total_flos": 1.1536340155092173e+19,
|
456 |
+
"train_loss": 0.19495156454661536,
|
457 |
+
"train_runtime": 10434.9174,
|
458 |
+
"train_samples_per_second": 34.004,
|
459 |
+
"train_steps_per_second": 0.066
|
460 |
+
}
|
461 |
+
],
|
462 |
+
"logging_steps": 10,
|
463 |
+
"max_steps": 693,
|
464 |
+
"num_train_epochs": 3,
|
465 |
+
"save_steps": 500,
|
466 |
+
"total_flos": 1.1536340155092173e+19,
|
467 |
+
"trial_name": null,
|
468 |
+
"trial_params": null
|
469 |
+
}
|