Commit
Β·
7836e48
1
Parent(s):
759a93d
Update to latest model
Browse files- README.md +6 -6
- checkpoint-11092/trainer_state.json +0 -3156
- {checkpoint-11092 β checkpoint-20331}/config.json +0 -0
- {checkpoint-11092 β checkpoint-20331}/model.safetensors +1 -1
- {checkpoint-11092 β checkpoint-20331}/optimizer.pt +1 -1
- {checkpoint-11092 β checkpoint-20331}/rng_state.pth +1 -1
- {checkpoint-11092 β checkpoint-20331}/scheduler.pt +1 -1
- checkpoint-20331/trainer_state.json +0 -0
- {checkpoint-11092 β checkpoint-20331}/training_args.bin +1 -1
- model.safetensors +1 -1
- runs/{Dec09_07-00-37_e11b5028ca4b/events.out.tfevents.1733727645.e11b5028ca4b.82.0 β Dec17_07-54-46_31406a8e7e2f/events.out.tfevents.1734422095.31406a8e7e2f.61.0} +2 -2
- runs/{Dec09_07-00-37_e11b5028ca4b/events.out.tfevents.1733763045.e11b5028ca4b.82.1 β Dec17_07-54-46_31406a8e7e2f/events.out.tfevents.1734490506.31406a8e7e2f.61.1} +2 -2
- training_args.bin +1 -1
- training_params.json +2 -2
README.md
CHANGED
@@ -18,14 +18,14 @@ widget:
|
|
18 |
- Problem type: Image Classification
|
19 |
|
20 |
## Validation Metrics
|
21 |
-
loss: 0.
|
22 |
|
23 |
-
f1: 0.
|
24 |
|
25 |
-
precision: 0.
|
26 |
|
27 |
-
recall: 0.
|
28 |
|
29 |
-
auc: 0.
|
30 |
|
31 |
-
accuracy: 0.
|
|
|
18 |
- Problem type: Image Classification
|
19 |
|
20 |
## Validation Metrics
|
21 |
+
loss: 0.10143255442380905
|
22 |
|
23 |
+
f1: 0.9866905430258446
|
24 |
|
25 |
+
precision: 0.9828849676983897
|
26 |
|
27 |
+
recall: 0.9905257020697698
|
28 |
|
29 |
+
auc: 0.9958292510810062
|
30 |
|
31 |
+
accuracy: 0.9801937412222262
|
checkpoint-11092/trainer_state.json
DELETED
@@ -1,3156 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": 0.16565780341625214,
|
3 |
-
"best_model_checkpoint": "autotrain-ai-image-detect-20241209-0148/checkpoint-11092",
|
4 |
-
"epoch": 1.0,
|
5 |
-
"eval_steps": 500,
|
6 |
-
"global_step": 11092,
|
7 |
-
"is_hyper_param_search": false,
|
8 |
-
"is_local_process_zero": true,
|
9 |
-
"is_world_process_zero": true,
|
10 |
-
"log_history": [
|
11 |
-
{
|
12 |
-
"epoch": 0.0022538766678687344,
|
13 |
-
"grad_norm": 41.54821014404297,
|
14 |
-
"learning_rate": 3.999949863013569e-05,
|
15 |
-
"loss": 0.0706,
|
16 |
-
"step": 25
|
17 |
-
},
|
18 |
-
{
|
19 |
-
"epoch": 0.004507753335737469,
|
20 |
-
"grad_norm": 0.026578206568956375,
|
21 |
-
"learning_rate": 3.999799454567995e-05,
|
22 |
-
"loss": 0.0649,
|
23 |
-
"step": 50
|
24 |
-
},
|
25 |
-
{
|
26 |
-
"epoch": 0.006761630003606202,
|
27 |
-
"grad_norm": 0.012844717130064964,
|
28 |
-
"learning_rate": 3.999548782204303e-05,
|
29 |
-
"loss": 0.139,
|
30 |
-
"step": 75
|
31 |
-
},
|
32 |
-
{
|
33 |
-
"epoch": 0.009015506671474938,
|
34 |
-
"grad_norm": 0.06785742193460464,
|
35 |
-
"learning_rate": 3.999197858490449e-05,
|
36 |
-
"loss": 0.0872,
|
37 |
-
"step": 100
|
38 |
-
},
|
39 |
-
{
|
40 |
-
"epoch": 0.01126938333934367,
|
41 |
-
"grad_norm": 0.28117385506629944,
|
42 |
-
"learning_rate": 3.998746701020692e-05,
|
43 |
-
"loss": 0.1982,
|
44 |
-
"step": 125
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.013523260007212405,
|
48 |
-
"grad_norm": 15.71075439453125,
|
49 |
-
"learning_rate": 3.998195332414708e-05,
|
50 |
-
"loss": 0.1397,
|
51 |
-
"step": 150
|
52 |
-
},
|
53 |
-
{
|
54 |
-
"epoch": 0.01577713667508114,
|
55 |
-
"grad_norm": 28.885644912719727,
|
56 |
-
"learning_rate": 3.997543780316456e-05,
|
57 |
-
"loss": 0.1616,
|
58 |
-
"step": 175
|
59 |
-
},
|
60 |
-
{
|
61 |
-
"epoch": 0.018031013342949875,
|
62 |
-
"grad_norm": 0.12888827919960022,
|
63 |
-
"learning_rate": 3.9967920773927955e-05,
|
64 |
-
"loss": 0.0577,
|
65 |
-
"step": 200
|
66 |
-
},
|
67 |
-
{
|
68 |
-
"epoch": 0.020284890010818606,
|
69 |
-
"grad_norm": 1.3943430185317993,
|
70 |
-
"learning_rate": 3.9959402613318466e-05,
|
71 |
-
"loss": 0.1144,
|
72 |
-
"step": 225
|
73 |
-
},
|
74 |
-
{
|
75 |
-
"epoch": 0.02253876667868734,
|
76 |
-
"grad_norm": 0.10762884467840195,
|
77 |
-
"learning_rate": 3.9949883748410986e-05,
|
78 |
-
"loss": 0.0898,
|
79 |
-
"step": 250
|
80 |
-
},
|
81 |
-
{
|
82 |
-
"epoch": 0.024792643346556075,
|
83 |
-
"grad_norm": 0.22058086097240448,
|
84 |
-
"learning_rate": 3.993936465645272e-05,
|
85 |
-
"loss": 0.086,
|
86 |
-
"step": 275
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.02704652001442481,
|
90 |
-
"grad_norm": 0.7955367565155029,
|
91 |
-
"learning_rate": 3.992784586483923e-05,
|
92 |
-
"loss": 0.0848,
|
93 |
-
"step": 300
|
94 |
-
},
|
95 |
-
{
|
96 |
-
"epoch": 0.029300396682293544,
|
97 |
-
"grad_norm": 0.08179175853729248,
|
98 |
-
"learning_rate": 3.9915327951088025e-05,
|
99 |
-
"loss": 0.1402,
|
100 |
-
"step": 325
|
101 |
-
},
|
102 |
-
{
|
103 |
-
"epoch": 0.03155427335016228,
|
104 |
-
"grad_norm": 71.69986724853516,
|
105 |
-
"learning_rate": 3.990181154280957e-05,
|
106 |
-
"loss": 0.0636,
|
107 |
-
"step": 350
|
108 |
-
},
|
109 |
-
{
|
110 |
-
"epoch": 0.033808150018031016,
|
111 |
-
"grad_norm": 14.672368049621582,
|
112 |
-
"learning_rate": 3.988729731767585e-05,
|
113 |
-
"loss": 0.0808,
|
114 |
-
"step": 375
|
115 |
-
},
|
116 |
-
{
|
117 |
-
"epoch": 0.03606202668589975,
|
118 |
-
"grad_norm": 55.48202133178711,
|
119 |
-
"learning_rate": 3.987178600338637e-05,
|
120 |
-
"loss": 0.118,
|
121 |
-
"step": 400
|
122 |
-
},
|
123 |
-
{
|
124 |
-
"epoch": 0.038315903353768485,
|
125 |
-
"grad_norm": 0.3038710057735443,
|
126 |
-
"learning_rate": 3.985527837763168e-05,
|
127 |
-
"loss": 0.07,
|
128 |
-
"step": 425
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 0.04056978002163721,
|
132 |
-
"grad_norm": 0.5395505428314209,
|
133 |
-
"learning_rate": 3.9837775268054396e-05,
|
134 |
-
"loss": 0.1537,
|
135 |
-
"step": 450
|
136 |
-
},
|
137 |
-
{
|
138 |
-
"epoch": 0.04282365668950595,
|
139 |
-
"grad_norm": 41.55247497558594,
|
140 |
-
"learning_rate": 3.9819277552207684e-05,
|
141 |
-
"loss": 0.1318,
|
142 |
-
"step": 475
|
143 |
-
},
|
144 |
-
{
|
145 |
-
"epoch": 0.04507753335737468,
|
146 |
-
"grad_norm": 0.6121176481246948,
|
147 |
-
"learning_rate": 3.979978615751127e-05,
|
148 |
-
"loss": 0.1667,
|
149 |
-
"step": 500
|
150 |
-
},
|
151 |
-
{
|
152 |
-
"epoch": 0.047331410025243416,
|
153 |
-
"grad_norm": 36.30903625488281,
|
154 |
-
"learning_rate": 3.9779302061204947e-05,
|
155 |
-
"loss": 0.1401,
|
156 |
-
"step": 525
|
157 |
-
},
|
158 |
-
{
|
159 |
-
"epoch": 0.04958528669311215,
|
160 |
-
"grad_norm": 0.037001315504312515,
|
161 |
-
"learning_rate": 3.975782629029957e-05,
|
162 |
-
"loss": 0.1211,
|
163 |
-
"step": 550
|
164 |
-
},
|
165 |
-
{
|
166 |
-
"epoch": 0.051839163360980885,
|
167 |
-
"grad_norm": 18.61513900756836,
|
168 |
-
"learning_rate": 3.973535992152557e-05,
|
169 |
-
"loss": 0.1649,
|
170 |
-
"step": 575
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.05409304002884962,
|
174 |
-
"grad_norm": 0.05658584460616112,
|
175 |
-
"learning_rate": 3.971190408127899e-05,
|
176 |
-
"loss": 0.1064,
|
177 |
-
"step": 600
|
178 |
-
},
|
179 |
-
{
|
180 |
-
"epoch": 0.056346916696718354,
|
181 |
-
"grad_norm": 20.563526153564453,
|
182 |
-
"learning_rate": 3.968745994556496e-05,
|
183 |
-
"loss": 0.118,
|
184 |
-
"step": 625
|
185 |
-
},
|
186 |
-
{
|
187 |
-
"epoch": 0.05860079336458709,
|
188 |
-
"grad_norm": 16.269672393798828,
|
189 |
-
"learning_rate": 3.966202873993878e-05,
|
190 |
-
"loss": 0.1738,
|
191 |
-
"step": 650
|
192 |
-
},
|
193 |
-
{
|
194 |
-
"epoch": 0.06085467003245582,
|
195 |
-
"grad_norm": 0.014949072152376175,
|
196 |
-
"learning_rate": 3.963561173944447e-05,
|
197 |
-
"loss": 0.1419,
|
198 |
-
"step": 675
|
199 |
-
},
|
200 |
-
{
|
201 |
-
"epoch": 0.06310854670032456,
|
202 |
-
"grad_norm": 16.951595306396484,
|
203 |
-
"learning_rate": 3.960821026855082e-05,
|
204 |
-
"loss": 0.1341,
|
205 |
-
"step": 700
|
206 |
-
},
|
207 |
-
{
|
208 |
-
"epoch": 0.0653624233681933,
|
209 |
-
"grad_norm": 0.2742099463939667,
|
210 |
-
"learning_rate": 3.957982570108501e-05,
|
211 |
-
"loss": 0.1038,
|
212 |
-
"step": 725
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 0.06761630003606203,
|
216 |
-
"grad_norm": 0.0967467799782753,
|
217 |
-
"learning_rate": 3.9550459460163694e-05,
|
218 |
-
"loss": 0.1592,
|
219 |
-
"step": 750
|
220 |
-
},
|
221 |
-
{
|
222 |
-
"epoch": 0.06987017670393077,
|
223 |
-
"grad_norm": 12.86486530303955,
|
224 |
-
"learning_rate": 3.9520113018121725e-05,
|
225 |
-
"loss": 0.1989,
|
226 |
-
"step": 775
|
227 |
-
},
|
228 |
-
{
|
229 |
-
"epoch": 0.0721240533717995,
|
230 |
-
"grad_norm": 0.03231130912899971,
|
231 |
-
"learning_rate": 3.948878789643823e-05,
|
232 |
-
"loss": 0.0891,
|
233 |
-
"step": 800
|
234 |
-
},
|
235 |
-
{
|
236 |
-
"epoch": 0.07437793003966824,
|
237 |
-
"grad_norm": 0.26412704586982727,
|
238 |
-
"learning_rate": 3.945648566566043e-05,
|
239 |
-
"loss": 0.1243,
|
240 |
-
"step": 825
|
241 |
-
},
|
242 |
-
{
|
243 |
-
"epoch": 0.07663180670753697,
|
244 |
-
"grad_norm": 0.028534913435578346,
|
245 |
-
"learning_rate": 3.942320794532482e-05,
|
246 |
-
"loss": 0.113,
|
247 |
-
"step": 850
|
248 |
-
},
|
249 |
-
{
|
250 |
-
"epoch": 0.0788856833754057,
|
251 |
-
"grad_norm": 4.5355353355407715,
|
252 |
-
"learning_rate": 3.938895640387601e-05,
|
253 |
-
"loss": 0.113,
|
254 |
-
"step": 875
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 0.08113956004327443,
|
258 |
-
"grad_norm": 19.036169052124023,
|
259 |
-
"learning_rate": 3.9353732758583066e-05,
|
260 |
-
"loss": 0.2539,
|
261 |
-
"step": 900
|
262 |
-
},
|
263 |
-
{
|
264 |
-
"epoch": 0.08339343671114316,
|
265 |
-
"grad_norm": 0.2818792462348938,
|
266 |
-
"learning_rate": 3.931753877545343e-05,
|
267 |
-
"loss": 0.0756,
|
268 |
-
"step": 925
|
269 |
-
},
|
270 |
-
{
|
271 |
-
"epoch": 0.0856473133790119,
|
272 |
-
"grad_norm": 15.56667709350586,
|
273 |
-
"learning_rate": 3.928037626914433e-05,
|
274 |
-
"loss": 0.1667,
|
275 |
-
"step": 950
|
276 |
-
},
|
277 |
-
{
|
278 |
-
"epoch": 0.08790119004688063,
|
279 |
-
"grad_norm": 0.27570584416389465,
|
280 |
-
"learning_rate": 3.924224710287186e-05,
|
281 |
-
"loss": 0.1486,
|
282 |
-
"step": 975
|
283 |
-
},
|
284 |
-
{
|
285 |
-
"epoch": 0.09015506671474936,
|
286 |
-
"grad_norm": 0.0981006845831871,
|
287 |
-
"learning_rate": 3.920315318831748e-05,
|
288 |
-
"loss": 0.1393,
|
289 |
-
"step": 1000
|
290 |
-
},
|
291 |
-
{
|
292 |
-
"epoch": 0.0924089433826181,
|
293 |
-
"grad_norm": 1.0749387741088867,
|
294 |
-
"learning_rate": 3.9163096485532276e-05,
|
295 |
-
"loss": 0.1619,
|
296 |
-
"step": 1025
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 0.09466282005048683,
|
300 |
-
"grad_norm": 0.04413340985774994,
|
301 |
-
"learning_rate": 3.912207900283861e-05,
|
302 |
-
"loss": 0.0503,
|
303 |
-
"step": 1050
|
304 |
-
},
|
305 |
-
{
|
306 |
-
"epoch": 0.09691669671835557,
|
307 |
-
"grad_norm": 0.8058507442474365,
|
308 |
-
"learning_rate": 3.908010279672945e-05,
|
309 |
-
"loss": 0.0981,
|
310 |
-
"step": 1075
|
311 |
-
},
|
312 |
-
{
|
313 |
-
"epoch": 0.0991705733862243,
|
314 |
-
"grad_norm": 0.022279851138591766,
|
315 |
-
"learning_rate": 3.9037169971765265e-05,
|
316 |
-
"loss": 0.1993,
|
317 |
-
"step": 1100
|
318 |
-
},
|
319 |
-
{
|
320 |
-
"epoch": 0.10142445005409303,
|
321 |
-
"grad_norm": 14.819859504699707,
|
322 |
-
"learning_rate": 3.899328268046854e-05,
|
323 |
-
"loss": 0.0935,
|
324 |
-
"step": 1125
|
325 |
-
},
|
326 |
-
{
|
327 |
-
"epoch": 0.10367832672196177,
|
328 |
-
"grad_norm": 7.201342582702637,
|
329 |
-
"learning_rate": 3.894844312321578e-05,
|
330 |
-
"loss": 0.0976,
|
331 |
-
"step": 1150
|
332 |
-
},
|
333 |
-
{
|
334 |
-
"epoch": 0.1059322033898305,
|
335 |
-
"grad_norm": 10.411704063415527,
|
336 |
-
"learning_rate": 3.890265354812727e-05,
|
337 |
-
"loss": 0.1177,
|
338 |
-
"step": 1175
|
339 |
-
},
|
340 |
-
{
|
341 |
-
"epoch": 0.10818608005769924,
|
342 |
-
"grad_norm": 0.008358177728950977,
|
343 |
-
"learning_rate": 3.885591625095432e-05,
|
344 |
-
"loss": 0.1715,
|
345 |
-
"step": 1200
|
346 |
-
},
|
347 |
-
{
|
348 |
-
"epoch": 0.11043995672556797,
|
349 |
-
"grad_norm": 1.033854365348816,
|
350 |
-
"learning_rate": 3.880823357496415e-05,
|
351 |
-
"loss": 0.1786,
|
352 |
-
"step": 1225
|
353 |
-
},
|
354 |
-
{
|
355 |
-
"epoch": 0.11269383339343671,
|
356 |
-
"grad_norm": 0.022488640621304512,
|
357 |
-
"learning_rate": 3.875960791082245e-05,
|
358 |
-
"loss": 0.0552,
|
359 |
-
"step": 1250
|
360 |
-
},
|
361 |
-
{
|
362 |
-
"epoch": 0.11494771006130544,
|
363 |
-
"grad_norm": 25.363340377807617,
|
364 |
-
"learning_rate": 3.8710041696473484e-05,
|
365 |
-
"loss": 0.1337,
|
366 |
-
"step": 1275
|
367 |
-
},
|
368 |
-
{
|
369 |
-
"epoch": 0.11720158672917418,
|
370 |
-
"grad_norm": 15.832186698913574,
|
371 |
-
"learning_rate": 3.865953741701785e-05,
|
372 |
-
"loss": 0.111,
|
373 |
-
"step": 1300
|
374 |
-
},
|
375 |
-
{
|
376 |
-
"epoch": 0.11945546339704291,
|
377 |
-
"grad_norm": 0.22804483771324158,
|
378 |
-
"learning_rate": 3.8608097604587946e-05,
|
379 |
-
"loss": 0.1831,
|
380 |
-
"step": 1325
|
381 |
-
},
|
382 |
-
{
|
383 |
-
"epoch": 0.12170934006491164,
|
384 |
-
"grad_norm": 14.083056449890137,
|
385 |
-
"learning_rate": 3.855572483822094e-05,
|
386 |
-
"loss": 0.1434,
|
387 |
-
"step": 1350
|
388 |
-
},
|
389 |
-
{
|
390 |
-
"epoch": 0.12396321673278038,
|
391 |
-
"grad_norm": 0.06291620433330536,
|
392 |
-
"learning_rate": 3.850242174372951e-05,
|
393 |
-
"loss": 0.1374,
|
394 |
-
"step": 1375
|
395 |
-
},
|
396 |
-
{
|
397 |
-
"epoch": 0.12621709340064913,
|
398 |
-
"grad_norm": 15.938265800476074,
|
399 |
-
"learning_rate": 3.844819099357017e-05,
|
400 |
-
"loss": 0.1469,
|
401 |
-
"step": 1400
|
402 |
-
},
|
403 |
-
{
|
404 |
-
"epoch": 0.12847097006851785,
|
405 |
-
"grad_norm": 28.19304656982422,
|
406 |
-
"learning_rate": 3.8393035306709315e-05,
|
407 |
-
"loss": 0.1276,
|
408 |
-
"step": 1425
|
409 |
-
},
|
410 |
-
{
|
411 |
-
"epoch": 0.1307248467363866,
|
412 |
-
"grad_norm": 1.189284086227417,
|
413 |
-
"learning_rate": 3.8336957448486874e-05,
|
414 |
-
"loss": 0.0754,
|
415 |
-
"step": 1450
|
416 |
-
},
|
417 |
-
{
|
418 |
-
"epoch": 0.13297872340425532,
|
419 |
-
"grad_norm": 9.746376037597656,
|
420 |
-
"learning_rate": 3.8279960230477655e-05,
|
421 |
-
"loss": 0.1164,
|
422 |
-
"step": 1475
|
423 |
-
},
|
424 |
-
{
|
425 |
-
"epoch": 0.13523260007212407,
|
426 |
-
"grad_norm": 0.04917048290371895,
|
427 |
-
"learning_rate": 3.822204651035041e-05,
|
428 |
-
"loss": 0.1273,
|
429 |
-
"step": 1500
|
430 |
-
},
|
431 |
-
{
|
432 |
-
"epoch": 0.13748647673999279,
|
433 |
-
"grad_norm": 1.262009620666504,
|
434 |
-
"learning_rate": 3.816321919172453e-05,
|
435 |
-
"loss": 0.2363,
|
436 |
-
"step": 1525
|
437 |
-
},
|
438 |
-
{
|
439 |
-
"epoch": 0.13974035340786153,
|
440 |
-
"grad_norm": 0.042943213135004044,
|
441 |
-
"learning_rate": 3.810348122402451e-05,
|
442 |
-
"loss": 0.1137,
|
443 |
-
"step": 1550
|
444 |
-
},
|
445 |
-
{
|
446 |
-
"epoch": 0.14199423007573025,
|
447 |
-
"grad_norm": 0.04536544904112816,
|
448 |
-
"learning_rate": 3.8042835602332e-05,
|
449 |
-
"loss": 0.105,
|
450 |
-
"step": 1575
|
451 |
-
},
|
452 |
-
{
|
453 |
-
"epoch": 0.144248106743599,
|
454 |
-
"grad_norm": 0.18310292065143585,
|
455 |
-
"learning_rate": 3.798128536723572e-05,
|
456 |
-
"loss": 0.1096,
|
457 |
-
"step": 1600
|
458 |
-
},
|
459 |
-
{
|
460 |
-
"epoch": 0.14650198341146772,
|
461 |
-
"grad_norm": 0.1673455536365509,
|
462 |
-
"learning_rate": 3.791883360467899e-05,
|
463 |
-
"loss": 0.2513,
|
464 |
-
"step": 1625
|
465 |
-
},
|
466 |
-
{
|
467 |
-
"epoch": 0.14875586007933647,
|
468 |
-
"grad_norm": 2.193483591079712,
|
469 |
-
"learning_rate": 3.785548344580496e-05,
|
470 |
-
"loss": 0.1124,
|
471 |
-
"step": 1650
|
472 |
-
},
|
473 |
-
{
|
474 |
-
"epoch": 0.1510097367472052,
|
475 |
-
"grad_norm": 0.516349196434021,
|
476 |
-
"learning_rate": 3.77912380667997e-05,
|
477 |
-
"loss": 0.115,
|
478 |
-
"step": 1675
|
479 |
-
},
|
480 |
-
{
|
481 |
-
"epoch": 0.15326361341507394,
|
482 |
-
"grad_norm": 0.1462392657995224,
|
483 |
-
"learning_rate": 3.772610068873289e-05,
|
484 |
-
"loss": 0.1605,
|
485 |
-
"step": 1700
|
486 |
-
},
|
487 |
-
{
|
488 |
-
"epoch": 0.15551749008294266,
|
489 |
-
"grad_norm": 0.3099835515022278,
|
490 |
-
"learning_rate": 3.7660074577396384e-05,
|
491 |
-
"loss": 0.2527,
|
492 |
-
"step": 1725
|
493 |
-
},
|
494 |
-
{
|
495 |
-
"epoch": 0.1577713667508114,
|
496 |
-
"grad_norm": 89.47386169433594,
|
497 |
-
"learning_rate": 3.759316304314042e-05,
|
498 |
-
"loss": 0.1028,
|
499 |
-
"step": 1750
|
500 |
-
},
|
501 |
-
{
|
502 |
-
"epoch": 0.16002524341868013,
|
503 |
-
"grad_norm": 26.115930557250977,
|
504 |
-
"learning_rate": 3.752536944070769e-05,
|
505 |
-
"loss": 0.1671,
|
506 |
-
"step": 1775
|
507 |
-
},
|
508 |
-
{
|
509 |
-
"epoch": 0.16227912008654885,
|
510 |
-
"grad_norm": 0.8346800208091736,
|
511 |
-
"learning_rate": 3.7456697169065135e-05,
|
512 |
-
"loss": 0.0907,
|
513 |
-
"step": 1800
|
514 |
-
},
|
515 |
-
{
|
516 |
-
"epoch": 0.1645329967544176,
|
517 |
-
"grad_norm": 1.5394320487976074,
|
518 |
-
"learning_rate": 3.7387149671233475e-05,
|
519 |
-
"loss": 0.1599,
|
520 |
-
"step": 1825
|
521 |
-
},
|
522 |
-
{
|
523 |
-
"epoch": 0.16678687342228632,
|
524 |
-
"grad_norm": 71.76346588134766,
|
525 |
-
"learning_rate": 3.731673043411468e-05,
|
526 |
-
"loss": 0.1996,
|
527 |
-
"step": 1850
|
528 |
-
},
|
529 |
-
{
|
530 |
-
"epoch": 0.16904075009015507,
|
531 |
-
"grad_norm": 0.13989679515361786,
|
532 |
-
"learning_rate": 3.724544298831709e-05,
|
533 |
-
"loss": 0.1484,
|
534 |
-
"step": 1875
|
535 |
-
},
|
536 |
-
{
|
537 |
-
"epoch": 0.1712946267580238,
|
538 |
-
"grad_norm": 0.12155357003211975,
|
539 |
-
"learning_rate": 3.71732909079784e-05,
|
540 |
-
"loss": 0.0917,
|
541 |
-
"step": 1900
|
542 |
-
},
|
543 |
-
{
|
544 |
-
"epoch": 0.17354850342589254,
|
545 |
-
"grad_norm": 0.2155969738960266,
|
546 |
-
"learning_rate": 3.7100277810586487e-05,
|
547 |
-
"loss": 0.1469,
|
548 |
-
"step": 1925
|
549 |
-
},
|
550 |
-
{
|
551 |
-
"epoch": 0.17580238009376126,
|
552 |
-
"grad_norm": 0.21151651442050934,
|
553 |
-
"learning_rate": 3.702640735679803e-05,
|
554 |
-
"loss": 0.1285,
|
555 |
-
"step": 1950
|
556 |
-
},
|
557 |
-
{
|
558 |
-
"epoch": 0.17805625676163,
|
559 |
-
"grad_norm": 1.40923273563385,
|
560 |
-
"learning_rate": 3.695168325025494e-05,
|
561 |
-
"loss": 0.0989,
|
562 |
-
"step": 1975
|
563 |
-
},
|
564 |
-
{
|
565 |
-
"epoch": 0.18031013342949873,
|
566 |
-
"grad_norm": 0.06402026861906052,
|
567 |
-
"learning_rate": 3.687610923739877e-05,
|
568 |
-
"loss": 0.15,
|
569 |
-
"step": 2000
|
570 |
-
},
|
571 |
-
{
|
572 |
-
"epoch": 0.18256401009736747,
|
573 |
-
"grad_norm": 1.6097108125686646,
|
574 |
-
"learning_rate": 3.679968910728277e-05,
|
575 |
-
"loss": 0.1419,
|
576 |
-
"step": 2025
|
577 |
-
},
|
578 |
-
{
|
579 |
-
"epoch": 0.1848178867652362,
|
580 |
-
"grad_norm": 13.362088203430176,
|
581 |
-
"learning_rate": 3.672242669138194e-05,
|
582 |
-
"loss": 0.2005,
|
583 |
-
"step": 2050
|
584 |
-
},
|
585 |
-
{
|
586 |
-
"epoch": 0.18707176343310494,
|
587 |
-
"grad_norm": 0.5157915353775024,
|
588 |
-
"learning_rate": 3.664432586340101e-05,
|
589 |
-
"loss": 0.1103,
|
590 |
-
"step": 2075
|
591 |
-
},
|
592 |
-
{
|
593 |
-
"epoch": 0.18932564010097366,
|
594 |
-
"grad_norm": 11.879634857177734,
|
595 |
-
"learning_rate": 3.656539053908011e-05,
|
596 |
-
"loss": 0.1569,
|
597 |
-
"step": 2100
|
598 |
-
},
|
599 |
-
{
|
600 |
-
"epoch": 0.1915795167688424,
|
601 |
-
"grad_norm": 7.176632404327393,
|
602 |
-
"learning_rate": 3.648562467599854e-05,
|
603 |
-
"loss": 0.1201,
|
604 |
-
"step": 2125
|
605 |
-
},
|
606 |
-
{
|
607 |
-
"epoch": 0.19383339343671113,
|
608 |
-
"grad_norm": 0.20655395090579987,
|
609 |
-
"learning_rate": 3.640503227337628e-05,
|
610 |
-
"loss": 0.0605,
|
611 |
-
"step": 2150
|
612 |
-
},
|
613 |
-
{
|
614 |
-
"epoch": 0.19608727010457988,
|
615 |
-
"grad_norm": 24.909950256347656,
|
616 |
-
"learning_rate": 3.632361737187354e-05,
|
617 |
-
"loss": 0.2595,
|
618 |
-
"step": 2175
|
619 |
-
},
|
620 |
-
{
|
621 |
-
"epoch": 0.1983411467724486,
|
622 |
-
"grad_norm": 11.23094654083252,
|
623 |
-
"learning_rate": 3.624138405338812e-05,
|
624 |
-
"loss": 0.0956,
|
625 |
-
"step": 2200
|
626 |
-
},
|
627 |
-
{
|
628 |
-
"epoch": 0.20059502344031735,
|
629 |
-
"grad_norm": 0.4831882119178772,
|
630 |
-
"learning_rate": 3.61583364408508e-05,
|
631 |
-
"loss": 0.101,
|
632 |
-
"step": 2225
|
633 |
-
},
|
634 |
-
{
|
635 |
-
"epoch": 0.20284890010818607,
|
636 |
-
"grad_norm": 0.07537912577390671,
|
637 |
-
"learning_rate": 3.607447869801861e-05,
|
638 |
-
"loss": 0.1745,
|
639 |
-
"step": 2250
|
640 |
-
},
|
641 |
-
{
|
642 |
-
"epoch": 0.20510277677605482,
|
643 |
-
"grad_norm": 73.84308624267578,
|
644 |
-
"learning_rate": 3.5989815029266045e-05,
|
645 |
-
"loss": 0.0903,
|
646 |
-
"step": 2275
|
647 |
-
},
|
648 |
-
{
|
649 |
-
"epoch": 0.20735665344392354,
|
650 |
-
"grad_norm": 23.107955932617188,
|
651 |
-
"learning_rate": 3.590434967937434e-05,
|
652 |
-
"loss": 0.1492,
|
653 |
-
"step": 2300
|
654 |
-
},
|
655 |
-
{
|
656 |
-
"epoch": 0.2096105301117923,
|
657 |
-
"grad_norm": 0.5104714632034302,
|
658 |
-
"learning_rate": 3.5818086933318564e-05,
|
659 |
-
"loss": 0.1237,
|
660 |
-
"step": 2325
|
661 |
-
},
|
662 |
-
{
|
663 |
-
"epoch": 0.211864406779661,
|
664 |
-
"grad_norm": 0.11333534121513367,
|
665 |
-
"learning_rate": 3.573103111605286e-05,
|
666 |
-
"loss": 0.1466,
|
667 |
-
"step": 2350
|
668 |
-
},
|
669 |
-
{
|
670 |
-
"epoch": 0.21411828344752976,
|
671 |
-
"grad_norm": 3.127657651901245,
|
672 |
-
"learning_rate": 3.5643186592293534e-05,
|
673 |
-
"loss": 0.1375,
|
674 |
-
"step": 2375
|
675 |
-
},
|
676 |
-
{
|
677 |
-
"epoch": 0.21637216011539848,
|
678 |
-
"grad_norm": 15.03669548034668,
|
679 |
-
"learning_rate": 3.555455776630031e-05,
|
680 |
-
"loss": 0.1278,
|
681 |
-
"step": 2400
|
682 |
-
},
|
683 |
-
{
|
684 |
-
"epoch": 0.21862603678326722,
|
685 |
-
"grad_norm": 0.13146854937076569,
|
686 |
-
"learning_rate": 3.546514908165542e-05,
|
687 |
-
"loss": 0.0952,
|
688 |
-
"step": 2425
|
689 |
-
},
|
690 |
-
{
|
691 |
-
"epoch": 0.22087991345113595,
|
692 |
-
"grad_norm": 0.2824071943759918,
|
693 |
-
"learning_rate": 3.537496502104087e-05,
|
694 |
-
"loss": 0.154,
|
695 |
-
"step": 2450
|
696 |
-
},
|
697 |
-
{
|
698 |
-
"epoch": 0.2231337901190047,
|
699 |
-
"grad_norm": 0.16681243479251862,
|
700 |
-
"learning_rate": 3.5284010106013685e-05,
|
701 |
-
"loss": 0.2106,
|
702 |
-
"step": 2475
|
703 |
-
},
|
704 |
-
{
|
705 |
-
"epoch": 0.22538766678687341,
|
706 |
-
"grad_norm": 28.65338897705078,
|
707 |
-
"learning_rate": 3.519228889677922e-05,
|
708 |
-
"loss": 0.0779,
|
709 |
-
"step": 2500
|
710 |
-
},
|
711 |
-
{
|
712 |
-
"epoch": 0.22764154345474216,
|
713 |
-
"grad_norm": 0.7323271632194519,
|
714 |
-
"learning_rate": 3.509980599196247e-05,
|
715 |
-
"loss": 0.1528,
|
716 |
-
"step": 2525
|
717 |
-
},
|
718 |
-
{
|
719 |
-
"epoch": 0.22989542012261088,
|
720 |
-
"grad_norm": 0.08046358078718185,
|
721 |
-
"learning_rate": 3.500656602837761e-05,
|
722 |
-
"loss": 0.1075,
|
723 |
-
"step": 2550
|
724 |
-
},
|
725 |
-
{
|
726 |
-
"epoch": 0.23214929679047963,
|
727 |
-
"grad_norm": 17.188133239746094,
|
728 |
-
"learning_rate": 3.491257368079541e-05,
|
729 |
-
"loss": 0.1009,
|
730 |
-
"step": 2575
|
731 |
-
},
|
732 |
-
{
|
733 |
-
"epoch": 0.23440317345834835,
|
734 |
-
"grad_norm": 19.240558624267578,
|
735 |
-
"learning_rate": 3.481783366170892e-05,
|
736 |
-
"loss": 0.1451,
|
737 |
-
"step": 2600
|
738 |
-
},
|
739 |
-
{
|
740 |
-
"epoch": 0.2366570501262171,
|
741 |
-
"grad_norm": 0.0762796625494957,
|
742 |
-
"learning_rate": 3.472235072109721e-05,
|
743 |
-
"loss": 0.1475,
|
744 |
-
"step": 2625
|
745 |
-
},
|
746 |
-
{
|
747 |
-
"epoch": 0.23891092679408582,
|
748 |
-
"grad_norm": 0.09147282689809799,
|
749 |
-
"learning_rate": 3.462612964618716e-05,
|
750 |
-
"loss": 0.1088,
|
751 |
-
"step": 2650
|
752 |
-
},
|
753 |
-
{
|
754 |
-
"epoch": 0.24116480346195457,
|
755 |
-
"grad_norm": 129.14137268066406,
|
756 |
-
"learning_rate": 3.4529175261213515e-05,
|
757 |
-
"loss": 0.122,
|
758 |
-
"step": 2675
|
759 |
-
},
|
760 |
-
{
|
761 |
-
"epoch": 0.2434186801298233,
|
762 |
-
"grad_norm": 61.75165557861328,
|
763 |
-
"learning_rate": 3.4431492427176946e-05,
|
764 |
-
"loss": 0.134,
|
765 |
-
"step": 2700
|
766 |
-
},
|
767 |
-
{
|
768 |
-
"epoch": 0.24567255679769204,
|
769 |
-
"grad_norm": 9.533876419067383,
|
770 |
-
"learning_rate": 3.433308604160037e-05,
|
771 |
-
"loss": 0.2032,
|
772 |
-
"step": 2725
|
773 |
-
},
|
774 |
-
{
|
775 |
-
"epoch": 0.24792643346556076,
|
776 |
-
"grad_norm": 16.249788284301758,
|
777 |
-
"learning_rate": 3.423396103828343e-05,
|
778 |
-
"loss": 0.0909,
|
779 |
-
"step": 2750
|
780 |
-
},
|
781 |
-
{
|
782 |
-
"epoch": 0.2501803101334295,
|
783 |
-
"grad_norm": 0.6528722047805786,
|
784 |
-
"learning_rate": 3.413412238705505e-05,
|
785 |
-
"loss": 0.1521,
|
786 |
-
"step": 2775
|
787 |
-
},
|
788 |
-
{
|
789 |
-
"epoch": 0.25243418680129825,
|
790 |
-
"grad_norm": 0.07963704317808151,
|
791 |
-
"learning_rate": 3.403357509352434e-05,
|
792 |
-
"loss": 0.0892,
|
793 |
-
"step": 2800
|
794 |
-
},
|
795 |
-
{
|
796 |
-
"epoch": 0.25468806346916695,
|
797 |
-
"grad_norm": 0.8912085294723511,
|
798 |
-
"learning_rate": 3.39323241988296e-05,
|
799 |
-
"loss": 0.2469,
|
800 |
-
"step": 2825
|
801 |
-
},
|
802 |
-
{
|
803 |
-
"epoch": 0.2569419401370357,
|
804 |
-
"grad_norm": 0.3295014202594757,
|
805 |
-
"learning_rate": 3.3830374779385545e-05,
|
806 |
-
"loss": 0.1307,
|
807 |
-
"step": 2850
|
808 |
-
},
|
809 |
-
{
|
810 |
-
"epoch": 0.25919581680490444,
|
811 |
-
"grad_norm": 0.8810415267944336,
|
812 |
-
"learning_rate": 3.3727731946628854e-05,
|
813 |
-
"loss": 0.0884,
|
814 |
-
"step": 2875
|
815 |
-
},
|
816 |
-
{
|
817 |
-
"epoch": 0.2614496934727732,
|
818 |
-
"grad_norm": 9.193936347961426,
|
819 |
-
"learning_rate": 3.3624400846761825e-05,
|
820 |
-
"loss": 0.1281,
|
821 |
-
"step": 2900
|
822 |
-
},
|
823 |
-
{
|
824 |
-
"epoch": 0.2637035701406419,
|
825 |
-
"grad_norm": 0.2354651391506195,
|
826 |
-
"learning_rate": 3.352038666049442e-05,
|
827 |
-
"loss": 0.0692,
|
828 |
-
"step": 2925
|
829 |
-
},
|
830 |
-
{
|
831 |
-
"epoch": 0.26595744680851063,
|
832 |
-
"grad_norm": 0.6887637376785278,
|
833 |
-
"learning_rate": 3.3415694602784474e-05,
|
834 |
-
"loss": 0.1294,
|
835 |
-
"step": 2950
|
836 |
-
},
|
837 |
-
{
|
838 |
-
"epoch": 0.2682113234763794,
|
839 |
-
"grad_norm": 0.6869497895240784,
|
840 |
-
"learning_rate": 3.331032992257627e-05,
|
841 |
-
"loss": 0.1143,
|
842 |
-
"step": 2975
|
843 |
-
},
|
844 |
-
{
|
845 |
-
"epoch": 0.27046520014424813,
|
846 |
-
"grad_norm": 0.2631952166557312,
|
847 |
-
"learning_rate": 3.320429790253735e-05,
|
848 |
-
"loss": 0.0998,
|
849 |
-
"step": 3000
|
850 |
-
},
|
851 |
-
{
|
852 |
-
"epoch": 0.2727190768121168,
|
853 |
-
"grad_norm": 54.48784637451172,
|
854 |
-
"learning_rate": 3.309760385879367e-05,
|
855 |
-
"loss": 0.0434,
|
856 |
-
"step": 3025
|
857 |
-
},
|
858 |
-
{
|
859 |
-
"epoch": 0.27497295347998557,
|
860 |
-
"grad_norm": 0.14939600229263306,
|
861 |
-
"learning_rate": 3.299025314066304e-05,
|
862 |
-
"loss": 0.0712,
|
863 |
-
"step": 3050
|
864 |
-
},
|
865 |
-
{
|
866 |
-
"epoch": 0.2772268301478543,
|
867 |
-
"grad_norm": 13.989568710327148,
|
868 |
-
"learning_rate": 3.288225113038697e-05,
|
869 |
-
"loss": 0.0866,
|
870 |
-
"step": 3075
|
871 |
-
},
|
872 |
-
{
|
873 |
-
"epoch": 0.27948070681572307,
|
874 |
-
"grad_norm": 0.02706342749297619,
|
875 |
-
"learning_rate": 3.277360324286077e-05,
|
876 |
-
"loss": 0.1525,
|
877 |
-
"step": 3100
|
878 |
-
},
|
879 |
-
{
|
880 |
-
"epoch": 0.28173458348359176,
|
881 |
-
"grad_norm": 0.04102005437016487,
|
882 |
-
"learning_rate": 3.266431492536212e-05,
|
883 |
-
"loss": 0.1479,
|
884 |
-
"step": 3125
|
885 |
-
},
|
886 |
-
{
|
887 |
-
"epoch": 0.2839884601514605,
|
888 |
-
"grad_norm": 0.026412546634674072,
|
889 |
-
"learning_rate": 3.255439165727789e-05,
|
890 |
-
"loss": 0.1455,
|
891 |
-
"step": 3150
|
892 |
-
},
|
893 |
-
{
|
894 |
-
"epoch": 0.28624233681932926,
|
895 |
-
"grad_norm": 0.07349425554275513,
|
896 |
-
"learning_rate": 3.244383894982951e-05,
|
897 |
-
"loss": 0.1582,
|
898 |
-
"step": 3175
|
899 |
-
},
|
900 |
-
{
|
901 |
-
"epoch": 0.288496213487198,
|
902 |
-
"grad_norm": 21.036027908325195,
|
903 |
-
"learning_rate": 3.233266234579655e-05,
|
904 |
-
"loss": 0.1244,
|
905 |
-
"step": 3200
|
906 |
-
},
|
907 |
-
{
|
908 |
-
"epoch": 0.2907500901550667,
|
909 |
-
"grad_norm": 0.17163363099098206,
|
910 |
-
"learning_rate": 3.222086741923891e-05,
|
911 |
-
"loss": 0.0904,
|
912 |
-
"step": 3225
|
913 |
-
},
|
914 |
-
{
|
915 |
-
"epoch": 0.29300396682293545,
|
916 |
-
"grad_norm": 30.365720748901367,
|
917 |
-
"learning_rate": 3.2108459775217286e-05,
|
918 |
-
"loss": 0.19,
|
919 |
-
"step": 3250
|
920 |
-
},
|
921 |
-
{
|
922 |
-
"epoch": 0.2952578434908042,
|
923 |
-
"grad_norm": 3.574190378189087,
|
924 |
-
"learning_rate": 3.199544504951222e-05,
|
925 |
-
"loss": 0.1075,
|
926 |
-
"step": 3275
|
927 |
-
},
|
928 |
-
{
|
929 |
-
"epoch": 0.29751172015867294,
|
930 |
-
"grad_norm": 0.05780789256095886,
|
931 |
-
"learning_rate": 3.1881828908341484e-05,
|
932 |
-
"loss": 0.1706,
|
933 |
-
"step": 3300
|
934 |
-
},
|
935 |
-
{
|
936 |
-
"epoch": 0.29976559682654164,
|
937 |
-
"grad_norm": 0.05321686342358589,
|
938 |
-
"learning_rate": 3.1767617048075995e-05,
|
939 |
-
"loss": 0.1113,
|
940 |
-
"step": 3325
|
941 |
-
},
|
942 |
-
{
|
943 |
-
"epoch": 0.3020194734944104,
|
944 |
-
"grad_norm": 0.05100090056657791,
|
945 |
-
"learning_rate": 3.165281519495424e-05,
|
946 |
-
"loss": 0.0566,
|
947 |
-
"step": 3350
|
948 |
-
},
|
949 |
-
{
|
950 |
-
"epoch": 0.30427335016227913,
|
951 |
-
"grad_norm": 3.367283344268799,
|
952 |
-
"learning_rate": 3.153742910479518e-05,
|
953 |
-
"loss": 0.0947,
|
954 |
-
"step": 3375
|
955 |
-
},
|
956 |
-
{
|
957 |
-
"epoch": 0.3065272268301479,
|
958 |
-
"grad_norm": 0.9839688539505005,
|
959 |
-
"learning_rate": 3.142146456270965e-05,
|
960 |
-
"loss": 0.1085,
|
961 |
-
"step": 3400
|
962 |
-
},
|
963 |
-
{
|
964 |
-
"epoch": 0.3087811034980166,
|
965 |
-
"grad_norm": 0.7526177763938904,
|
966 |
-
"learning_rate": 3.130492738281031e-05,
|
967 |
-
"loss": 0.0989,
|
968 |
-
"step": 3425
|
969 |
-
},
|
970 |
-
{
|
971 |
-
"epoch": 0.3110349801658853,
|
972 |
-
"grad_norm": 18.201082229614258,
|
973 |
-
"learning_rate": 3.1187823407920194e-05,
|
974 |
-
"loss": 0.1785,
|
975 |
-
"step": 3450
|
976 |
-
},
|
977 |
-
{
|
978 |
-
"epoch": 0.31328885683375407,
|
979 |
-
"grad_norm": 1.3650627136230469,
|
980 |
-
"learning_rate": 3.107015850927967e-05,
|
981 |
-
"loss": 0.0578,
|
982 |
-
"step": 3475
|
983 |
-
},
|
984 |
-
{
|
985 |
-
"epoch": 0.3155427335016228,
|
986 |
-
"grad_norm": 0.20913846790790558,
|
987 |
-
"learning_rate": 3.095193858625219e-05,
|
988 |
-
"loss": 0.1555,
|
989 |
-
"step": 3500
|
990 |
-
},
|
991 |
-
{
|
992 |
-
"epoch": 0.3177966101694915,
|
993 |
-
"grad_norm": 0.17458997666835785,
|
994 |
-
"learning_rate": 3.083316956602842e-05,
|
995 |
-
"loss": 0.1053,
|
996 |
-
"step": 3525
|
997 |
-
},
|
998 |
-
{
|
999 |
-
"epoch": 0.32005048683736026,
|
1000 |
-
"grad_norm": 10.047566413879395,
|
1001 |
-
"learning_rate": 3.071385740332912e-05,
|
1002 |
-
"loss": 0.0686,
|
1003 |
-
"step": 3550
|
1004 |
-
},
|
1005 |
-
{
|
1006 |
-
"epoch": 0.322304363505229,
|
1007 |
-
"grad_norm": 0.1347886621952057,
|
1008 |
-
"learning_rate": 3.059400808010656e-05,
|
1009 |
-
"loss": 0.129,
|
1010 |
-
"step": 3575
|
1011 |
-
},
|
1012 |
-
{
|
1013 |
-
"epoch": 0.3245582401730977,
|
1014 |
-
"grad_norm": 0.3037494421005249,
|
1015 |
-
"learning_rate": 3.047362760524465e-05,
|
1016 |
-
"loss": 0.1341,
|
1017 |
-
"step": 3600
|
1018 |
-
},
|
1019 |
-
{
|
1020 |
-
"epoch": 0.32681211684096645,
|
1021 |
-
"grad_norm": 16.686107635498047,
|
1022 |
-
"learning_rate": 3.0352722014257606e-05,
|
1023 |
-
"loss": 0.1365,
|
1024 |
-
"step": 3625
|
1025 |
-
},
|
1026 |
-
{
|
1027 |
-
"epoch": 0.3290659935088352,
|
1028 |
-
"grad_norm": 0.15748658776283264,
|
1029 |
-
"learning_rate": 3.023129736898742e-05,
|
1030 |
-
"loss": 0.0753,
|
1031 |
-
"step": 3650
|
1032 |
-
},
|
1033 |
-
{
|
1034 |
-
"epoch": 0.33131987017670395,
|
1035 |
-
"grad_norm": 36.410057067871094,
|
1036 |
-
"learning_rate": 3.0109359757299877e-05,
|
1037 |
-
"loss": 0.141,
|
1038 |
-
"step": 3675
|
1039 |
-
},
|
1040 |
-
{
|
1041 |
-
"epoch": 0.33357374684457264,
|
1042 |
-
"grad_norm": 0.0776553526520729,
|
1043 |
-
"learning_rate": 2.9986915292779354e-05,
|
1044 |
-
"loss": 0.146,
|
1045 |
-
"step": 3700
|
1046 |
-
},
|
1047 |
-
{
|
1048 |
-
"epoch": 0.3358276235124414,
|
1049 |
-
"grad_norm": 59.83176803588867,
|
1050 |
-
"learning_rate": 2.9863970114422315e-05,
|
1051 |
-
"loss": 0.1407,
|
1052 |
-
"step": 3725
|
1053 |
-
},
|
1054 |
-
{
|
1055 |
-
"epoch": 0.33808150018031013,
|
1056 |
-
"grad_norm": 0.05585867166519165,
|
1057 |
-
"learning_rate": 2.974053038632949e-05,
|
1058 |
-
"loss": 0.1579,
|
1059 |
-
"step": 3750
|
1060 |
-
},
|
1061 |
-
{
|
1062 |
-
"epoch": 0.3403353768481789,
|
1063 |
-
"grad_norm": 0.16879096627235413,
|
1064 |
-
"learning_rate": 2.9616602297396865e-05,
|
1065 |
-
"loss": 0.1567,
|
1066 |
-
"step": 3775
|
1067 |
-
},
|
1068 |
-
{
|
1069 |
-
"epoch": 0.3425892535160476,
|
1070 |
-
"grad_norm": 22.54551887512207,
|
1071 |
-
"learning_rate": 2.949219206100535e-05,
|
1072 |
-
"loss": 0.1449,
|
1073 |
-
"step": 3800
|
1074 |
-
},
|
1075 |
-
{
|
1076 |
-
"epoch": 0.3448431301839163,
|
1077 |
-
"grad_norm": 0.030749494209885597,
|
1078 |
-
"learning_rate": 2.9367305914709267e-05,
|
1079 |
-
"loss": 0.0538,
|
1080 |
-
"step": 3825
|
1081 |
-
},
|
1082 |
-
{
|
1083 |
-
"epoch": 0.3470970068517851,
|
1084 |
-
"grad_norm": 1.6320780515670776,
|
1085 |
-
"learning_rate": 2.9241950119923642e-05,
|
1086 |
-
"loss": 0.1775,
|
1087 |
-
"step": 3850
|
1088 |
-
},
|
1089 |
-
{
|
1090 |
-
"epoch": 0.3493508835196538,
|
1091 |
-
"grad_norm": 0.3045240342617035,
|
1092 |
-
"learning_rate": 2.911613096161027e-05,
|
1093 |
-
"loss": 0.0438,
|
1094 |
-
"step": 3875
|
1095 |
-
},
|
1096 |
-
{
|
1097 |
-
"epoch": 0.3516047601875225,
|
1098 |
-
"grad_norm": 0.6519447565078735,
|
1099 |
-
"learning_rate": 2.8989854747962574e-05,
|
1100 |
-
"loss": 0.0807,
|
1101 |
-
"step": 3900
|
1102 |
-
},
|
1103 |
-
{
|
1104 |
-
"epoch": 0.35385863685539126,
|
1105 |
-
"grad_norm": 0.04941343143582344,
|
1106 |
-
"learning_rate": 2.8863127810089363e-05,
|
1107 |
-
"loss": 0.1084,
|
1108 |
-
"step": 3925
|
1109 |
-
},
|
1110 |
-
{
|
1111 |
-
"epoch": 0.35611251352326,
|
1112 |
-
"grad_norm": 0.13526397943496704,
|
1113 |
-
"learning_rate": 2.8735956501697398e-05,
|
1114 |
-
"loss": 0.1416,
|
1115 |
-
"step": 3950
|
1116 |
-
},
|
1117 |
-
{
|
1118 |
-
"epoch": 0.35836639019112876,
|
1119 |
-
"grad_norm": 0.05733851343393326,
|
1120 |
-
"learning_rate": 2.8608347198772856e-05,
|
1121 |
-
"loss": 0.0463,
|
1122 |
-
"step": 3975
|
1123 |
-
},
|
1124 |
-
{
|
1125 |
-
"epoch": 0.36062026685899745,
|
1126 |
-
"grad_norm": 0.20993520319461823,
|
1127 |
-
"learning_rate": 2.8480306299261618e-05,
|
1128 |
-
"loss": 0.2597,
|
1129 |
-
"step": 4000
|
1130 |
-
},
|
1131 |
-
{
|
1132 |
-
"epoch": 0.3628741435268662,
|
1133 |
-
"grad_norm": 0.28277480602264404,
|
1134 |
-
"learning_rate": 2.8351840222748524e-05,
|
1135 |
-
"loss": 0.1105,
|
1136 |
-
"step": 4025
|
1137 |
-
},
|
1138 |
-
{
|
1139 |
-
"epoch": 0.36512802019473495,
|
1140 |
-
"grad_norm": 0.19352126121520996,
|
1141 |
-
"learning_rate": 2.8222955410135512e-05,
|
1142 |
-
"loss": 0.072,
|
1143 |
-
"step": 4050
|
1144 |
-
},
|
1145 |
-
{
|
1146 |
-
"epoch": 0.3673818968626037,
|
1147 |
-
"grad_norm": 0.2709672152996063,
|
1148 |
-
"learning_rate": 2.8093658323318673e-05,
|
1149 |
-
"loss": 0.1663,
|
1150 |
-
"step": 4075
|
1151 |
-
},
|
1152 |
-
{
|
1153 |
-
"epoch": 0.3696357735304724,
|
1154 |
-
"grad_norm": 0.07498263567686081,
|
1155 |
-
"learning_rate": 2.796395544486432e-05,
|
1156 |
-
"loss": 0.1121,
|
1157 |
-
"step": 4100
|
1158 |
-
},
|
1159 |
-
{
|
1160 |
-
"epoch": 0.37188965019834114,
|
1161 |
-
"grad_norm": 0.17926153540611267,
|
1162 |
-
"learning_rate": 2.783385327768388e-05,
|
1163 |
-
"loss": 0.1044,
|
1164 |
-
"step": 4125
|
1165 |
-
},
|
1166 |
-
{
|
1167 |
-
"epoch": 0.3741435268662099,
|
1168 |
-
"grad_norm": 0.5393832921981812,
|
1169 |
-
"learning_rate": 2.770335834470796e-05,
|
1170 |
-
"loss": 0.1073,
|
1171 |
-
"step": 4150
|
1172 |
-
},
|
1173 |
-
{
|
1174 |
-
"epoch": 0.37639740353407863,
|
1175 |
-
"grad_norm": 0.2945524752140045,
|
1176 |
-
"learning_rate": 2.757247718855924e-05,
|
1177 |
-
"loss": 0.0903,
|
1178 |
-
"step": 4175
|
1179 |
-
},
|
1180 |
-
{
|
1181 |
-
"epoch": 0.3786512802019473,
|
1182 |
-
"grad_norm": 0.7488250136375427,
|
1183 |
-
"learning_rate": 2.7441216371224472e-05,
|
1184 |
-
"loss": 0.1258,
|
1185 |
-
"step": 4200
|
1186 |
-
},
|
1187 |
-
{
|
1188 |
-
"epoch": 0.3809051568698161,
|
1189 |
-
"grad_norm": 0.3737005293369293,
|
1190 |
-
"learning_rate": 2.7309582473725467e-05,
|
1191 |
-
"loss": 0.2835,
|
1192 |
-
"step": 4225
|
1193 |
-
},
|
1194 |
-
{
|
1195 |
-
"epoch": 0.3831590335376848,
|
1196 |
-
"grad_norm": 0.28525131940841675,
|
1197 |
-
"learning_rate": 2.717758209578917e-05,
|
1198 |
-
"loss": 0.138,
|
1199 |
-
"step": 4250
|
1200 |
-
},
|
1201 |
-
{
|
1202 |
-
"epoch": 0.38541291020555357,
|
1203 |
-
"grad_norm": 0.06608044356107712,
|
1204 |
-
"learning_rate": 2.7045221855516728e-05,
|
1205 |
-
"loss": 0.0891,
|
1206 |
-
"step": 4275
|
1207 |
-
},
|
1208 |
-
{
|
1209 |
-
"epoch": 0.38766678687342226,
|
1210 |
-
"grad_norm": 0.42144662141799927,
|
1211 |
-
"learning_rate": 2.6912508389051715e-05,
|
1212 |
-
"loss": 0.1733,
|
1213 |
-
"step": 4300
|
1214 |
-
},
|
1215 |
-
{
|
1216 |
-
"epoch": 0.389920663541291,
|
1217 |
-
"grad_norm": 0.10393067449331284,
|
1218 |
-
"learning_rate": 2.67794483502474e-05,
|
1219 |
-
"loss": 0.0267,
|
1220 |
-
"step": 4325
|
1221 |
-
},
|
1222 |
-
{
|
1223 |
-
"epoch": 0.39217454020915976,
|
1224 |
-
"grad_norm": 43.680458068847656,
|
1225 |
-
"learning_rate": 2.664604841033313e-05,
|
1226 |
-
"loss": 0.1411,
|
1227 |
-
"step": 4350
|
1228 |
-
},
|
1229 |
-
{
|
1230 |
-
"epoch": 0.3944284168770285,
|
1231 |
-
"grad_norm": 17.044269561767578,
|
1232 |
-
"learning_rate": 2.6512315257579905e-05,
|
1233 |
-
"loss": 0.205,
|
1234 |
-
"step": 4375
|
1235 |
-
},
|
1236 |
-
{
|
1237 |
-
"epoch": 0.3966822935448972,
|
1238 |
-
"grad_norm": 0.2856123149394989,
|
1239 |
-
"learning_rate": 2.6378255596964977e-05,
|
1240 |
-
"loss": 0.0624,
|
1241 |
-
"step": 4400
|
1242 |
-
},
|
1243 |
-
{
|
1244 |
-
"epoch": 0.39893617021276595,
|
1245 |
-
"grad_norm": 0.1673419326543808,
|
1246 |
-
"learning_rate": 2.6243876149835733e-05,
|
1247 |
-
"loss": 0.0594,
|
1248 |
-
"step": 4425
|
1249 |
-
},
|
1250 |
-
{
|
1251 |
-
"epoch": 0.4011900468806347,
|
1252 |
-
"grad_norm": 0.040798913687467575,
|
1253 |
-
"learning_rate": 2.6109183653572686e-05,
|
1254 |
-
"loss": 0.1538,
|
1255 |
-
"step": 4450
|
1256 |
-
},
|
1257 |
-
{
|
1258 |
-
"epoch": 0.40344392354850345,
|
1259 |
-
"grad_norm": 0.0503900982439518,
|
1260 |
-
"learning_rate": 2.597418486125169e-05,
|
1261 |
-
"loss": 0.0965,
|
1262 |
-
"step": 4475
|
1263 |
-
},
|
1264 |
-
{
|
1265 |
-
"epoch": 0.40569780021637214,
|
1266 |
-
"grad_norm": 15.682025909423828,
|
1267 |
-
"learning_rate": 2.5838886541305383e-05,
|
1268 |
-
"loss": 0.1365,
|
1269 |
-
"step": 4500
|
1270 |
-
},
|
1271 |
-
{
|
1272 |
-
"epoch": 0.4079516768842409,
|
1273 |
-
"grad_norm": 0.17503659427165985,
|
1274 |
-
"learning_rate": 2.570329547718378e-05,
|
1275 |
-
"loss": 0.1052,
|
1276 |
-
"step": 4525
|
1277 |
-
},
|
1278 |
-
{
|
1279 |
-
"epoch": 0.41020555355210964,
|
1280 |
-
"grad_norm": 0.03469785675406456,
|
1281 |
-
"learning_rate": 2.5567418467014225e-05,
|
1282 |
-
"loss": 0.1054,
|
1283 |
-
"step": 4550
|
1284 |
-
},
|
1285 |
-
{
|
1286 |
-
"epoch": 0.4124594302199784,
|
1287 |
-
"grad_norm": 13.422741889953613,
|
1288 |
-
"learning_rate": 2.543126232326054e-05,
|
1289 |
-
"loss": 0.1537,
|
1290 |
-
"step": 4575
|
1291 |
-
},
|
1292 |
-
{
|
1293 |
-
"epoch": 0.4147133068878471,
|
1294 |
-
"grad_norm": 34.65227508544922,
|
1295 |
-
"learning_rate": 2.529483387238145e-05,
|
1296 |
-
"loss": 0.1472,
|
1297 |
-
"step": 4600
|
1298 |
-
},
|
1299 |
-
{
|
1300 |
-
"epoch": 0.4169671835557158,
|
1301 |
-
"grad_norm": 3.792473793029785,
|
1302 |
-
"learning_rate": 2.515813995448835e-05,
|
1303 |
-
"loss": 0.1276,
|
1304 |
-
"step": 4625
|
1305 |
-
},
|
1306 |
-
{
|
1307 |
-
"epoch": 0.4192210602235846,
|
1308 |
-
"grad_norm": 0.3917059898376465,
|
1309 |
-
"learning_rate": 2.5021187423002336e-05,
|
1310 |
-
"loss": 0.1105,
|
1311 |
-
"step": 4650
|
1312 |
-
},
|
1313 |
-
{
|
1314 |
-
"epoch": 0.4214749368914533,
|
1315 |
-
"grad_norm": 21.639259338378906,
|
1316 |
-
"learning_rate": 2.4883983144310634e-05,
|
1317 |
-
"loss": 0.1714,
|
1318 |
-
"step": 4675
|
1319 |
-
},
|
1320 |
-
{
|
1321 |
-
"epoch": 0.423728813559322,
|
1322 |
-
"grad_norm": 0.9362630844116211,
|
1323 |
-
"learning_rate": 2.4746533997422303e-05,
|
1324 |
-
"loss": 0.0892,
|
1325 |
-
"step": 4700
|
1326 |
-
},
|
1327 |
-
{
|
1328 |
-
"epoch": 0.42598269022719076,
|
1329 |
-
"grad_norm": 0.06721477955579758,
|
1330 |
-
"learning_rate": 2.460884687362334e-05,
|
1331 |
-
"loss": 0.1947,
|
1332 |
-
"step": 4725
|
1333 |
-
},
|
1334 |
-
{
|
1335 |
-
"epoch": 0.4282365668950595,
|
1336 |
-
"grad_norm": 0.09544611722230911,
|
1337 |
-
"learning_rate": 2.447092867613122e-05,
|
1338 |
-
"loss": 0.1521,
|
1339 |
-
"step": 4750
|
1340 |
-
},
|
1341 |
-
{
|
1342 |
-
"epoch": 0.43049044356292826,
|
1343 |
-
"grad_norm": 0.4658750593662262,
|
1344 |
-
"learning_rate": 2.433278631974873e-05,
|
1345 |
-
"loss": 0.0727,
|
1346 |
-
"step": 4775
|
1347 |
-
},
|
1348 |
-
{
|
1349 |
-
"epoch": 0.43274432023079695,
|
1350 |
-
"grad_norm": 0.025953950360417366,
|
1351 |
-
"learning_rate": 2.419442673051732e-05,
|
1352 |
-
"loss": 0.1005,
|
1353 |
-
"step": 4800
|
1354 |
-
},
|
1355 |
-
{
|
1356 |
-
"epoch": 0.4349981968986657,
|
1357 |
-
"grad_norm": 0.43599462509155273,
|
1358 |
-
"learning_rate": 2.4055856845369836e-05,
|
1359 |
-
"loss": 0.1617,
|
1360 |
-
"step": 4825
|
1361 |
-
},
|
1362 |
-
{
|
1363 |
-
"epoch": 0.43725207356653445,
|
1364 |
-
"grad_norm": 0.29886549711227417,
|
1365 |
-
"learning_rate": 2.3917083611782738e-05,
|
1366 |
-
"loss": 0.0832,
|
1367 |
-
"step": 4850
|
1368 |
-
},
|
1369 |
-
{
|
1370 |
-
"epoch": 0.4395059502344032,
|
1371 |
-
"grad_norm": 56.6343994140625,
|
1372 |
-
"learning_rate": 2.3778113987427734e-05,
|
1373 |
-
"loss": 0.1425,
|
1374 |
-
"step": 4875
|
1375 |
-
},
|
1376 |
-
{
|
1377 |
-
"epoch": 0.4417598269022719,
|
1378 |
-
"grad_norm": 0.27752888202667236,
|
1379 |
-
"learning_rate": 2.363895493982302e-05,
|
1380 |
-
"loss": 0.1376,
|
1381 |
-
"step": 4900
|
1382 |
-
},
|
1383 |
-
{
|
1384 |
-
"epoch": 0.44401370357014064,
|
1385 |
-
"grad_norm": 45.0130615234375,
|
1386 |
-
"learning_rate": 2.349961344598386e-05,
|
1387 |
-
"loss": 0.0818,
|
1388 |
-
"step": 4925
|
1389 |
-
},
|
1390 |
-
{
|
1391 |
-
"epoch": 0.4462675802380094,
|
1392 |
-
"grad_norm": 3.2570018768310547,
|
1393 |
-
"learning_rate": 2.3360096492072853e-05,
|
1394 |
-
"loss": 0.0999,
|
1395 |
-
"step": 4950
|
1396 |
-
},
|
1397 |
-
{
|
1398 |
-
"epoch": 0.44852145690587814,
|
1399 |
-
"grad_norm": 25.19772720336914,
|
1400 |
-
"learning_rate": 2.322041107304961e-05,
|
1401 |
-
"loss": 0.1423,
|
1402 |
-
"step": 4975
|
1403 |
-
},
|
1404 |
-
{
|
1405 |
-
"epoch": 0.45077533357374683,
|
1406 |
-
"grad_norm": 196.02114868164062,
|
1407 |
-
"learning_rate": 2.3080564192320103e-05,
|
1408 |
-
"loss": 0.0712,
|
1409 |
-
"step": 5000
|
1410 |
-
},
|
1411 |
-
{
|
1412 |
-
"epoch": 0.4530292102416156,
|
1413 |
-
"grad_norm": 26.691713333129883,
|
1414 |
-
"learning_rate": 2.294056286138548e-05,
|
1415 |
-
"loss": 0.0922,
|
1416 |
-
"step": 5025
|
1417 |
-
},
|
1418 |
-
{
|
1419 |
-
"epoch": 0.4552830869094843,
|
1420 |
-
"grad_norm": 0.08442128449678421,
|
1421 |
-
"learning_rate": 2.2800414099490577e-05,
|
1422 |
-
"loss": 0.176,
|
1423 |
-
"step": 5050
|
1424 |
-
},
|
1425 |
-
{
|
1426 |
-
"epoch": 0.4575369635773531,
|
1427 |
-
"grad_norm": 68.13138580322266,
|
1428 |
-
"learning_rate": 2.266012493327197e-05,
|
1429 |
-
"loss": 0.0878,
|
1430 |
-
"step": 5075
|
1431 |
-
},
|
1432 |
-
{
|
1433 |
-
"epoch": 0.45979084024522177,
|
1434 |
-
"grad_norm": 94.9263687133789,
|
1435 |
-
"learning_rate": 2.251970239640569e-05,
|
1436 |
-
"loss": 0.1519,
|
1437 |
-
"step": 5100
|
1438 |
-
},
|
1439 |
-
{
|
1440 |
-
"epoch": 0.4620447169130905,
|
1441 |
-
"grad_norm": 0.05859652906656265,
|
1442 |
-
"learning_rate": 2.2379153529254536e-05,
|
1443 |
-
"loss": 0.1145,
|
1444 |
-
"step": 5125
|
1445 |
-
},
|
1446 |
-
{
|
1447 |
-
"epoch": 0.46429859358095926,
|
1448 |
-
"grad_norm": 75.00086212158203,
|
1449 |
-
"learning_rate": 2.2238485378515175e-05,
|
1450 |
-
"loss": 0.1751,
|
1451 |
-
"step": 5150
|
1452 |
-
},
|
1453 |
-
{
|
1454 |
-
"epoch": 0.466552470248828,
|
1455 |
-
"grad_norm": 0.6764459609985352,
|
1456 |
-
"learning_rate": 2.2097704996864766e-05,
|
1457 |
-
"loss": 0.0995,
|
1458 |
-
"step": 5175
|
1459 |
-
},
|
1460 |
-
{
|
1461 |
-
"epoch": 0.4688063469166967,
|
1462 |
-
"grad_norm": 0.047138020396232605,
|
1463 |
-
"learning_rate": 2.1956819442607395e-05,
|
1464 |
-
"loss": 0.0769,
|
1465 |
-
"step": 5200
|
1466 |
-
},
|
1467 |
-
{
|
1468 |
-
"epoch": 0.47106022358456545,
|
1469 |
-
"grad_norm": 3.8884665966033936,
|
1470 |
-
"learning_rate": 2.1815835779320183e-05,
|
1471 |
-
"loss": 0.1089,
|
1472 |
-
"step": 5225
|
1473 |
-
},
|
1474 |
-
{
|
1475 |
-
"epoch": 0.4733141002524342,
|
1476 |
-
"grad_norm": 14.387825012207031,
|
1477 |
-
"learning_rate": 2.1674761075499138e-05,
|
1478 |
-
"loss": 0.1347,
|
1479 |
-
"step": 5250
|
1480 |
-
},
|
1481 |
-
{
|
1482 |
-
"epoch": 0.47556797692030295,
|
1483 |
-
"grad_norm": 40.32550811767578,
|
1484 |
-
"learning_rate": 2.1533602404204775e-05,
|
1485 |
-
"loss": 0.1278,
|
1486 |
-
"step": 5275
|
1487 |
-
},
|
1488 |
-
{
|
1489 |
-
"epoch": 0.47782185358817164,
|
1490 |
-
"grad_norm": 0.21858257055282593,
|
1491 |
-
"learning_rate": 2.1392366842707495e-05,
|
1492 |
-
"loss": 0.1368,
|
1493 |
-
"step": 5300
|
1494 |
-
},
|
1495 |
-
{
|
1496 |
-
"epoch": 0.4800757302560404,
|
1497 |
-
"grad_norm": 3.796025514602661,
|
1498 |
-
"learning_rate": 2.1251061472132703e-05,
|
1499 |
-
"loss": 0.1291,
|
1500 |
-
"step": 5325
|
1501 |
-
},
|
1502 |
-
{
|
1503 |
-
"epoch": 0.48232960692390914,
|
1504 |
-
"grad_norm": 16.28775978088379,
|
1505 |
-
"learning_rate": 2.110969337710586e-05,
|
1506 |
-
"loss": 0.1295,
|
1507 |
-
"step": 5350
|
1508 |
-
},
|
1509 |
-
{
|
1510 |
-
"epoch": 0.48458348359177783,
|
1511 |
-
"grad_norm": 0.5657278895378113,
|
1512 |
-
"learning_rate": 2.0968269645397234e-05,
|
1513 |
-
"loss": 0.1399,
|
1514 |
-
"step": 5375
|
1515 |
-
},
|
1516 |
-
{
|
1517 |
-
"epoch": 0.4868373602596466,
|
1518 |
-
"grad_norm": 0.20052899420261383,
|
1519 |
-
"learning_rate": 2.0826797367566533e-05,
|
1520 |
-
"loss": 0.0837,
|
1521 |
-
"step": 5400
|
1522 |
-
},
|
1523 |
-
{
|
1524 |
-
"epoch": 0.4890912369275153,
|
1525 |
-
"grad_norm": 23.986915588378906,
|
1526 |
-
"learning_rate": 2.0685283636607432e-05,
|
1527 |
-
"loss": 0.0919,
|
1528 |
-
"step": 5425
|
1529 |
-
},
|
1530 |
-
{
|
1531 |
-
"epoch": 0.4913451135953841,
|
1532 |
-
"grad_norm": 0.024290593340992928,
|
1533 |
-
"learning_rate": 2.0543735547591934e-05,
|
1534 |
-
"loss": 0.0922,
|
1535 |
-
"step": 5450
|
1536 |
-
},
|
1537 |
-
{
|
1538 |
-
"epoch": 0.49359899026325277,
|
1539 |
-
"grad_norm": 19.735187530517578,
|
1540 |
-
"learning_rate": 2.0402160197314678e-05,
|
1541 |
-
"loss": 0.2404,
|
1542 |
-
"step": 5475
|
1543 |
-
},
|
1544 |
-
{
|
1545 |
-
"epoch": 0.4958528669311215,
|
1546 |
-
"grad_norm": 55.32980728149414,
|
1547 |
-
"learning_rate": 2.0260564683937053e-05,
|
1548 |
-
"loss": 0.1123,
|
1549 |
-
"step": 5500
|
1550 |
-
},
|
1551 |
-
{
|
1552 |
-
"epoch": 0.49810674359899026,
|
1553 |
-
"grad_norm": 53.07206344604492,
|
1554 |
-
"learning_rate": 2.0118956106631414e-05,
|
1555 |
-
"loss": 0.1266,
|
1556 |
-
"step": 5525
|
1557 |
-
},
|
1558 |
-
{
|
1559 |
-
"epoch": 0.500360620266859,
|
1560 |
-
"grad_norm": 0.10000480711460114,
|
1561 |
-
"learning_rate": 1.997734156522507e-05,
|
1562 |
-
"loss": 0.0627,
|
1563 |
-
"step": 5550
|
1564 |
-
},
|
1565 |
-
{
|
1566 |
-
"epoch": 0.5026144969347277,
|
1567 |
-
"grad_norm": 0.25646984577178955,
|
1568 |
-
"learning_rate": 1.9835728159844364e-05,
|
1569 |
-
"loss": 0.1062,
|
1570 |
-
"step": 5575
|
1571 |
-
},
|
1572 |
-
{
|
1573 |
-
"epoch": 0.5048683736025965,
|
1574 |
-
"grad_norm": 97.85262298583984,
|
1575 |
-
"learning_rate": 1.969412299055867e-05,
|
1576 |
-
"loss": 0.0901,
|
1577 |
-
"step": 5600
|
1578 |
-
},
|
1579 |
-
{
|
1580 |
-
"epoch": 0.5071222502704652,
|
1581 |
-
"grad_norm": 17.852195739746094,
|
1582 |
-
"learning_rate": 1.9552533157024462e-05,
|
1583 |
-
"loss": 0.1195,
|
1584 |
-
"step": 5625
|
1585 |
-
},
|
1586 |
-
{
|
1587 |
-
"epoch": 0.5093761269383339,
|
1588 |
-
"grad_norm": 0.01943562924861908,
|
1589 |
-
"learning_rate": 1.941096575812928e-05,
|
1590 |
-
"loss": 0.1197,
|
1591 |
-
"step": 5650
|
1592 |
-
},
|
1593 |
-
{
|
1594 |
-
"epoch": 0.5116300036062027,
|
1595 |
-
"grad_norm": 0.28579404950141907,
|
1596 |
-
"learning_rate": 1.9269427891635886e-05,
|
1597 |
-
"loss": 0.1335,
|
1598 |
-
"step": 5675
|
1599 |
-
},
|
1600 |
-
{
|
1601 |
-
"epoch": 0.5138838802740714,
|
1602 |
-
"grad_norm": 0.09043838083744049,
|
1603 |
-
"learning_rate": 1.912792665382638e-05,
|
1604 |
-
"loss": 0.0589,
|
1605 |
-
"step": 5700
|
1606 |
-
},
|
1607 |
-
{
|
1608 |
-
"epoch": 0.5161377569419401,
|
1609 |
-
"grad_norm": 0.26899954676628113,
|
1610 |
-
"learning_rate": 1.8986469139146406e-05,
|
1611 |
-
"loss": 0.1368,
|
1612 |
-
"step": 5725
|
1613 |
-
},
|
1614 |
-
{
|
1615 |
-
"epoch": 0.5183916336098089,
|
1616 |
-
"grad_norm": 0.019730588421225548,
|
1617 |
-
"learning_rate": 1.8845062439849437e-05,
|
1618 |
-
"loss": 0.1541,
|
1619 |
-
"step": 5750
|
1620 |
-
},
|
1621 |
-
{
|
1622 |
-
"epoch": 0.5206455102776776,
|
1623 |
-
"grad_norm": 0.5210295915603638,
|
1624 |
-
"learning_rate": 1.8703713645641253e-05,
|
1625 |
-
"loss": 0.1231,
|
1626 |
-
"step": 5775
|
1627 |
-
},
|
1628 |
-
{
|
1629 |
-
"epoch": 0.5228993869455464,
|
1630 |
-
"grad_norm": 64.45635223388672,
|
1631 |
-
"learning_rate": 1.8562429843324423e-05,
|
1632 |
-
"loss": 0.1226,
|
1633 |
-
"step": 5800
|
1634 |
-
},
|
1635 |
-
{
|
1636 |
-
"epoch": 0.5251532636134151,
|
1637 |
-
"grad_norm": 0.1178063303232193,
|
1638 |
-
"learning_rate": 1.8421218116443044e-05,
|
1639 |
-
"loss": 0.0728,
|
1640 |
-
"step": 5825
|
1641 |
-
},
|
1642 |
-
{
|
1643 |
-
"epoch": 0.5274071402812838,
|
1644 |
-
"grad_norm": 0.2532069981098175,
|
1645 |
-
"learning_rate": 1.8280085544927534e-05,
|
1646 |
-
"loss": 0.0745,
|
1647 |
-
"step": 5850
|
1648 |
-
},
|
1649 |
-
{
|
1650 |
-
"epoch": 0.5296610169491526,
|
1651 |
-
"grad_norm": 0.18521960079669952,
|
1652 |
-
"learning_rate": 1.8139039204739712e-05,
|
1653 |
-
"loss": 0.0579,
|
1654 |
-
"step": 5875
|
1655 |
-
},
|
1656 |
-
{
|
1657 |
-
"epoch": 0.5319148936170213,
|
1658 |
-
"grad_norm": 0.08436308056116104,
|
1659 |
-
"learning_rate": 1.7998086167518033e-05,
|
1660 |
-
"loss": 0.2131,
|
1661 |
-
"step": 5900
|
1662 |
-
},
|
1663 |
-
{
|
1664 |
-
"epoch": 0.53416877028489,
|
1665 |
-
"grad_norm": 0.11574628949165344,
|
1666 |
-
"learning_rate": 1.7857233500223018e-05,
|
1667 |
-
"loss": 0.1179,
|
1668 |
-
"step": 5925
|
1669 |
-
},
|
1670 |
-
{
|
1671 |
-
"epoch": 0.5364226469527588,
|
1672 |
-
"grad_norm": 0.30803030729293823,
|
1673 |
-
"learning_rate": 1.7716488264782915e-05,
|
1674 |
-
"loss": 0.1243,
|
1675 |
-
"step": 5950
|
1676 |
-
},
|
1677 |
-
{
|
1678 |
-
"epoch": 0.5386765236206275,
|
1679 |
-
"grad_norm": 0.37837812304496765,
|
1680 |
-
"learning_rate": 1.75758575177397e-05,
|
1681 |
-
"loss": 0.1289,
|
1682 |
-
"step": 5975
|
1683 |
-
},
|
1684 |
-
{
|
1685 |
-
"epoch": 0.5409304002884963,
|
1686 |
-
"grad_norm": 9.621341705322266,
|
1687 |
-
"learning_rate": 1.743534830989522e-05,
|
1688 |
-
"loss": 0.1628,
|
1689 |
-
"step": 6000
|
1690 |
-
},
|
1691 |
-
{
|
1692 |
-
"epoch": 0.543184276956365,
|
1693 |
-
"grad_norm": 0.15798382461071014,
|
1694 |
-
"learning_rate": 1.7294967685957733e-05,
|
1695 |
-
"loss": 0.0935,
|
1696 |
-
"step": 6025
|
1697 |
-
},
|
1698 |
-
{
|
1699 |
-
"epoch": 0.5454381536242336,
|
1700 |
-
"grad_norm": 8.106306076049805,
|
1701 |
-
"learning_rate": 1.7154722684188668e-05,
|
1702 |
-
"loss": 0.1203,
|
1703 |
-
"step": 6050
|
1704 |
-
},
|
1705 |
-
{
|
1706 |
-
"epoch": 0.5476920302921024,
|
1707 |
-
"grad_norm": 78.09608459472656,
|
1708 |
-
"learning_rate": 1.7014620336049774e-05,
|
1709 |
-
"loss": 0.0899,
|
1710 |
-
"step": 6075
|
1711 |
-
},
|
1712 |
-
{
|
1713 |
-
"epoch": 0.5499459069599711,
|
1714 |
-
"grad_norm": 0.09071781486272812,
|
1715 |
-
"learning_rate": 1.6874667665850583e-05,
|
1716 |
-
"loss": 0.1061,
|
1717 |
-
"step": 6100
|
1718 |
-
},
|
1719 |
-
{
|
1720 |
-
"epoch": 0.5521997836278398,
|
1721 |
-
"grad_norm": 14.566875457763672,
|
1722 |
-
"learning_rate": 1.673487169039623e-05,
|
1723 |
-
"loss": 0.1166,
|
1724 |
-
"step": 6125
|
1725 |
-
},
|
1726 |
-
{
|
1727 |
-
"epoch": 0.5544536602957086,
|
1728 |
-
"grad_norm": 0.03580047935247421,
|
1729 |
-
"learning_rate": 1.6595239418635616e-05,
|
1730 |
-
"loss": 0.1093,
|
1731 |
-
"step": 6150
|
1732 |
-
},
|
1733 |
-
{
|
1734 |
-
"epoch": 0.5567075369635773,
|
1735 |
-
"grad_norm": 15.507891654968262,
|
1736 |
-
"learning_rate": 1.6455777851310074e-05,
|
1737 |
-
"loss": 0.1161,
|
1738 |
-
"step": 6175
|
1739 |
-
},
|
1740 |
-
{
|
1741 |
-
"epoch": 0.5589614136314461,
|
1742 |
-
"grad_norm": 0.3939911425113678,
|
1743 |
-
"learning_rate": 1.6316493980602307e-05,
|
1744 |
-
"loss": 0.1262,
|
1745 |
-
"step": 6200
|
1746 |
-
},
|
1747 |
-
{
|
1748 |
-
"epoch": 0.5612152902993148,
|
1749 |
-
"grad_norm": 30.040477752685547,
|
1750 |
-
"learning_rate": 1.6177394789785855e-05,
|
1751 |
-
"loss": 0.0987,
|
1752 |
-
"step": 6225
|
1753 |
-
},
|
1754 |
-
{
|
1755 |
-
"epoch": 0.5634691669671835,
|
1756 |
-
"grad_norm": 0.21318642795085907,
|
1757 |
-
"learning_rate": 1.6038487252874964e-05,
|
1758 |
-
"loss": 0.0449,
|
1759 |
-
"step": 6250
|
1760 |
-
},
|
1761 |
-
{
|
1762 |
-
"epoch": 0.5657230436350523,
|
1763 |
-
"grad_norm": 0.08098239451646805,
|
1764 |
-
"learning_rate": 1.5899778334274908e-05,
|
1765 |
-
"loss": 0.0933,
|
1766 |
-
"step": 6275
|
1767 |
-
},
|
1768 |
-
{
|
1769 |
-
"epoch": 0.567976920302921,
|
1770 |
-
"grad_norm": 1.0314059257507324,
|
1771 |
-
"learning_rate": 1.5761274988432877e-05,
|
1772 |
-
"loss": 0.1268,
|
1773 |
-
"step": 6300
|
1774 |
-
},
|
1775 |
-
{
|
1776 |
-
"epoch": 0.5702307969707897,
|
1777 |
-
"grad_norm": 22.789411544799805,
|
1778 |
-
"learning_rate": 1.562298415948924e-05,
|
1779 |
-
"loss": 0.0803,
|
1780 |
-
"step": 6325
|
1781 |
-
},
|
1782 |
-
{
|
1783 |
-
"epoch": 0.5724846736386585,
|
1784 |
-
"grad_norm": 0.04815113916993141,
|
1785 |
-
"learning_rate": 1.5484912780929392e-05,
|
1786 |
-
"loss": 0.0464,
|
1787 |
-
"step": 6350
|
1788 |
-
},
|
1789 |
-
{
|
1790 |
-
"epoch": 0.5747385503065272,
|
1791 |
-
"grad_norm": 0.048456188291311264,
|
1792 |
-
"learning_rate": 1.534706777523619e-05,
|
1793 |
-
"loss": 0.1056,
|
1794 |
-
"step": 6375
|
1795 |
-
},
|
1796 |
-
{
|
1797 |
-
"epoch": 0.576992426974396,
|
1798 |
-
"grad_norm": 0.041415195912122726,
|
1799 |
-
"learning_rate": 1.5209456053542794e-05,
|
1800 |
-
"loss": 0.1252,
|
1801 |
-
"step": 6400
|
1802 |
-
},
|
1803 |
-
{
|
1804 |
-
"epoch": 0.5792463036422647,
|
1805 |
-
"grad_norm": 0.10412342101335526,
|
1806 |
-
"learning_rate": 1.5072084515286245e-05,
|
1807 |
-
"loss": 0.1177,
|
1808 |
-
"step": 6425
|
1809 |
-
},
|
1810 |
-
{
|
1811 |
-
"epoch": 0.5815001803101334,
|
1812 |
-
"grad_norm": 132.91761779785156,
|
1813 |
-
"learning_rate": 1.4934960047861489e-05,
|
1814 |
-
"loss": 0.1351,
|
1815 |
-
"step": 6450
|
1816 |
-
},
|
1817 |
-
{
|
1818 |
-
"epoch": 0.5837540569780022,
|
1819 |
-
"grad_norm": 0.47370755672454834,
|
1820 |
-
"learning_rate": 1.4798089526276074e-05,
|
1821 |
-
"loss": 0.187,
|
1822 |
-
"step": 6475
|
1823 |
-
},
|
1824 |
-
{
|
1825 |
-
"epoch": 0.5860079336458709,
|
1826 |
-
"grad_norm": 0.12512896955013275,
|
1827 |
-
"learning_rate": 1.4661479812805499e-05,
|
1828 |
-
"loss": 0.1448,
|
1829 |
-
"step": 6500
|
1830 |
-
},
|
1831 |
-
{
|
1832 |
-
"epoch": 0.5882618103137396,
|
1833 |
-
"grad_norm": 0.19954493641853333,
|
1834 |
-
"learning_rate": 1.4525137756649113e-05,
|
1835 |
-
"loss": 0.1572,
|
1836 |
-
"step": 6525
|
1837 |
-
},
|
1838 |
-
{
|
1839 |
-
"epoch": 0.5905156869816084,
|
1840 |
-
"grad_norm": 14.07094955444336,
|
1841 |
-
"learning_rate": 1.4389070193586721e-05,
|
1842 |
-
"loss": 0.0728,
|
1843 |
-
"step": 6550
|
1844 |
-
},
|
1845 |
-
{
|
1846 |
-
"epoch": 0.5927695636494771,
|
1847 |
-
"grad_norm": 0.07505636662244797,
|
1848 |
-
"learning_rate": 1.4253283945635907e-05,
|
1849 |
-
"loss": 0.1108,
|
1850 |
-
"step": 6575
|
1851 |
-
},
|
1852 |
-
{
|
1853 |
-
"epoch": 0.5950234403173459,
|
1854 |
-
"grad_norm": 0.37513038516044617,
|
1855 |
-
"learning_rate": 1.4117785820709926e-05,
|
1856 |
-
"loss": 0.1079,
|
1857 |
-
"step": 6600
|
1858 |
-
},
|
1859 |
-
{
|
1860 |
-
"epoch": 0.5972773169852146,
|
1861 |
-
"grad_norm": 54.23029708862305,
|
1862 |
-
"learning_rate": 1.3982582612276439e-05,
|
1863 |
-
"loss": 0.0967,
|
1864 |
-
"step": 6625
|
1865 |
-
},
|
1866 |
-
{
|
1867 |
-
"epoch": 0.5995311936530833,
|
1868 |
-
"grad_norm": 21.563617706298828,
|
1869 |
-
"learning_rate": 1.3847681099016872e-05,
|
1870 |
-
"loss": 0.0724,
|
1871 |
-
"step": 6650
|
1872 |
-
},
|
1873 |
-
{
|
1874 |
-
"epoch": 0.6017850703209521,
|
1875 |
-
"grad_norm": 0.016839193180203438,
|
1876 |
-
"learning_rate": 1.371308804448655e-05,
|
1877 |
-
"loss": 0.067,
|
1878 |
-
"step": 6675
|
1879 |
-
},
|
1880 |
-
{
|
1881 |
-
"epoch": 0.6040389469888208,
|
1882 |
-
"grad_norm": 0.04092521592974663,
|
1883 |
-
"learning_rate": 1.3578810196775639e-05,
|
1884 |
-
"loss": 0.0801,
|
1885 |
-
"step": 6700
|
1886 |
-
},
|
1887 |
-
{
|
1888 |
-
"epoch": 0.6062928236566895,
|
1889 |
-
"grad_norm": 64.4503402709961,
|
1890 |
-
"learning_rate": 1.3444854288170763e-05,
|
1891 |
-
"loss": 0.1357,
|
1892 |
-
"step": 6725
|
1893 |
-
},
|
1894 |
-
{
|
1895 |
-
"epoch": 0.6085467003245583,
|
1896 |
-
"grad_norm": 0.3482804298400879,
|
1897 |
-
"learning_rate": 1.3311227034817482e-05,
|
1898 |
-
"loss": 0.1528,
|
1899 |
-
"step": 6750
|
1900 |
-
},
|
1901 |
-
{
|
1902 |
-
"epoch": 0.610800576992427,
|
1903 |
-
"grad_norm": 17.13836097717285,
|
1904 |
-
"learning_rate": 1.3177935136383608e-05,
|
1905 |
-
"loss": 0.1839,
|
1906 |
-
"step": 6775
|
1907 |
-
},
|
1908 |
-
{
|
1909 |
-
"epoch": 0.6130544536602958,
|
1910 |
-
"grad_norm": 0.05015060678124428,
|
1911 |
-
"learning_rate": 1.3044985275723221e-05,
|
1912 |
-
"loss": 0.0678,
|
1913 |
-
"step": 6800
|
1914 |
-
},
|
1915 |
-
{
|
1916 |
-
"epoch": 0.6153083303281645,
|
1917 |
-
"grad_norm": 0.33494699001312256,
|
1918 |
-
"learning_rate": 1.2912384118541695e-05,
|
1919 |
-
"loss": 0.1702,
|
1920 |
-
"step": 6825
|
1921 |
-
},
|
1922 |
-
{
|
1923 |
-
"epoch": 0.6175622069960331,
|
1924 |
-
"grad_norm": 0.022038869559764862,
|
1925 |
-
"learning_rate": 1.2780138313061448e-05,
|
1926 |
-
"loss": 0.1024,
|
1927 |
-
"step": 6850
|
1928 |
-
},
|
1929 |
-
{
|
1930 |
-
"epoch": 0.619816083663902,
|
1931 |
-
"grad_norm": 0.5564080476760864,
|
1932 |
-
"learning_rate": 1.264825448968862e-05,
|
1933 |
-
"loss": 0.0643,
|
1934 |
-
"step": 6875
|
1935 |
-
},
|
1936 |
-
{
|
1937 |
-
"epoch": 0.6220699603317706,
|
1938 |
-
"grad_norm": 0.2769150733947754,
|
1939 |
-
"learning_rate": 1.2516739260680693e-05,
|
1940 |
-
"loss": 0.1608,
|
1941 |
-
"step": 6900
|
1942 |
-
},
|
1943 |
-
{
|
1944 |
-
"epoch": 0.6243238369996393,
|
1945 |
-
"grad_norm": 0.24671363830566406,
|
1946 |
-
"learning_rate": 1.2385599219814907e-05,
|
1947 |
-
"loss": 0.0939,
|
1948 |
-
"step": 6925
|
1949 |
-
},
|
1950 |
-
{
|
1951 |
-
"epoch": 0.6265777136675081,
|
1952 |
-
"grad_norm": 30.696514129638672,
|
1953 |
-
"learning_rate": 1.225484094205771e-05,
|
1954 |
-
"loss": 0.2044,
|
1955 |
-
"step": 6950
|
1956 |
-
},
|
1957 |
-
{
|
1958 |
-
"epoch": 0.6288315903353768,
|
1959 |
-
"grad_norm": 17.0674991607666,
|
1960 |
-
"learning_rate": 1.2124470983235117e-05,
|
1961 |
-
"loss": 0.1165,
|
1962 |
-
"step": 6975
|
1963 |
-
},
|
1964 |
-
{
|
1965 |
-
"epoch": 0.6310854670032456,
|
1966 |
-
"grad_norm": 19.00300407409668,
|
1967 |
-
"learning_rate": 1.1994495879703962e-05,
|
1968 |
-
"loss": 0.1166,
|
1969 |
-
"step": 7000
|
1970 |
-
},
|
1971 |
-
{
|
1972 |
-
"epoch": 0.6333393436711143,
|
1973 |
-
"grad_norm": 0.054979290813207626,
|
1974 |
-
"learning_rate": 1.186492214802426e-05,
|
1975 |
-
"loss": 0.1132,
|
1976 |
-
"step": 7025
|
1977 |
-
},
|
1978 |
-
{
|
1979 |
-
"epoch": 0.635593220338983,
|
1980 |
-
"grad_norm": 14.680729866027832,
|
1981 |
-
"learning_rate": 1.1735756284632442e-05,
|
1982 |
-
"loss": 0.141,
|
1983 |
-
"step": 7050
|
1984 |
-
},
|
1985 |
-
{
|
1986 |
-
"epoch": 0.6378470970068518,
|
1987 |
-
"grad_norm": 21.676219940185547,
|
1988 |
-
"learning_rate": 1.1607004765515628e-05,
|
1989 |
-
"loss": 0.0975,
|
1990 |
-
"step": 7075
|
1991 |
-
},
|
1992 |
-
{
|
1993 |
-
"epoch": 0.6401009736747205,
|
1994 |
-
"grad_norm": 9.62362003326416,
|
1995 |
-
"learning_rate": 1.1478674045887007e-05,
|
1996 |
-
"loss": 0.0414,
|
1997 |
-
"step": 7100
|
1998 |
-
},
|
1999 |
-
{
|
2000 |
-
"epoch": 0.6423548503425892,
|
2001 |
-
"grad_norm": 14.026162147521973,
|
2002 |
-
"learning_rate": 1.1350770559862125e-05,
|
2003 |
-
"loss": 0.1321,
|
2004 |
-
"step": 7125
|
2005 |
-
},
|
2006 |
-
{
|
2007 |
-
"epoch": 0.644608727010458,
|
2008 |
-
"grad_norm": 24.935630798339844,
|
2009 |
-
"learning_rate": 1.1223300720136305e-05,
|
2010 |
-
"loss": 0.1848,
|
2011 |
-
"step": 7150
|
2012 |
-
},
|
2013 |
-
{
|
2014 |
-
"epoch": 0.6468626036783267,
|
2015 |
-
"grad_norm": 0.6826255917549133,
|
2016 |
-
"learning_rate": 1.1096270917663195e-05,
|
2017 |
-
"loss": 0.1293,
|
2018 |
-
"step": 7175
|
2019 |
-
},
|
2020 |
-
{
|
2021 |
-
"epoch": 0.6491164803461954,
|
2022 |
-
"grad_norm": 0.3986845016479492,
|
2023 |
-
"learning_rate": 1.0969687521334275e-05,
|
2024 |
-
"loss": 0.1061,
|
2025 |
-
"step": 7200
|
2026 |
-
},
|
2027 |
-
{
|
2028 |
-
"epoch": 0.6513703570140642,
|
2029 |
-
"grad_norm": 0.09392484277486801,
|
2030 |
-
"learning_rate": 1.084355687765955e-05,
|
2031 |
-
"loss": 0.1398,
|
2032 |
-
"step": 7225
|
2033 |
-
},
|
2034 |
-
{
|
2035 |
-
"epoch": 0.6536242336819329,
|
2036 |
-
"grad_norm": 0.3693184554576874,
|
2037 |
-
"learning_rate": 1.0717885310449407e-05,
|
2038 |
-
"loss": 0.0785,
|
2039 |
-
"step": 7250
|
2040 |
-
},
|
2041 |
-
{
|
2042 |
-
"epoch": 0.6558781103498017,
|
2043 |
-
"grad_norm": 0.13649840652942657,
|
2044 |
-
"learning_rate": 1.0592679120497505e-05,
|
2045 |
-
"loss": 0.0854,
|
2046 |
-
"step": 7275
|
2047 |
-
},
|
2048 |
-
{
|
2049 |
-
"epoch": 0.6581319870176704,
|
2050 |
-
"grad_norm": 1.3029234409332275,
|
2051 |
-
"learning_rate": 1.0467944585264886e-05,
|
2052 |
-
"loss": 0.0822,
|
2053 |
-
"step": 7300
|
2054 |
-
},
|
2055 |
-
{
|
2056 |
-
"epoch": 0.6603858636855391,
|
2057 |
-
"grad_norm": 0.20036828517913818,
|
2058 |
-
"learning_rate": 1.0343687958565253e-05,
|
2059 |
-
"loss": 0.1049,
|
2060 |
-
"step": 7325
|
2061 |
-
},
|
2062 |
-
{
|
2063 |
-
"epoch": 0.6626397403534079,
|
2064 |
-
"grad_norm": 0.15013161301612854,
|
2065 |
-
"learning_rate": 1.0219915470251411e-05,
|
2066 |
-
"loss": 0.0885,
|
2067 |
-
"step": 7350
|
2068 |
-
},
|
2069 |
-
{
|
2070 |
-
"epoch": 0.6648936170212766,
|
2071 |
-
"grad_norm": 0.1293184757232666,
|
2072 |
-
"learning_rate": 1.0096633325902931e-05,
|
2073 |
-
"loss": 0.1193,
|
2074 |
-
"step": 7375
|
2075 |
-
},
|
2076 |
-
{
|
2077 |
-
"epoch": 0.6671474936891453,
|
2078 |
-
"grad_norm": 0.674525797367096,
|
2079 |
-
"learning_rate": 9.973847706515007e-06,
|
2080 |
-
"loss": 0.1117,
|
2081 |
-
"step": 7400
|
2082 |
-
},
|
2083 |
-
{
|
2084 |
-
"epoch": 0.6694013703570141,
|
2085 |
-
"grad_norm": 11.253371238708496,
|
2086 |
-
"learning_rate": 9.851564768188566e-06,
|
2087 |
-
"loss": 0.1461,
|
2088 |
-
"step": 7425
|
2089 |
-
},
|
2090 |
-
{
|
2091 |
-
"epoch": 0.6716552470248828,
|
2092 |
-
"grad_norm": 0.47691622376441956,
|
2093 |
-
"learning_rate": 9.729790641821652e-06,
|
2094 |
-
"loss": 0.1264,
|
2095 |
-
"step": 7450
|
2096 |
-
},
|
2097 |
-
{
|
2098 |
-
"epoch": 0.6739091236927516,
|
2099 |
-
"grad_norm": 116.1590347290039,
|
2100 |
-
"learning_rate": 9.608531432801953e-06,
|
2101 |
-
"loss": 0.1171,
|
2102 |
-
"step": 7475
|
2103 |
-
},
|
2104 |
-
{
|
2105 |
-
"epoch": 0.6761630003606203,
|
2106 |
-
"grad_norm": 2.279863119125366,
|
2107 |
-
"learning_rate": 9.48779322070081e-06,
|
2108 |
-
"loss": 0.099,
|
2109 |
-
"step": 7500
|
2110 |
-
},
|
2111 |
-
{
|
2112 |
-
"epoch": 0.678416877028489,
|
2113 |
-
"grad_norm": 1.397264003753662,
|
2114 |
-
"learning_rate": 9.367582058968325e-06,
|
2115 |
-
"loss": 0.0593,
|
2116 |
-
"step": 7525
|
2117 |
-
},
|
2118 |
-
{
|
2119 |
-
"epoch": 0.6806707536963578,
|
2120 |
-
"grad_norm": 0.4885930120944977,
|
2121 |
-
"learning_rate": 9.247903974629862e-06,
|
2122 |
-
"loss": 0.0856,
|
2123 |
-
"step": 7550
|
2124 |
-
},
|
2125 |
-
{
|
2126 |
-
"epoch": 0.6829246303642265,
|
2127 |
-
"grad_norm": 0.08295653015375137,
|
2128 |
-
"learning_rate": 9.128764967983929e-06,
|
2129 |
-
"loss": 0.1234,
|
2130 |
-
"step": 7575
|
2131 |
-
},
|
2132 |
-
{
|
2133 |
-
"epoch": 0.6851785070320952,
|
2134 |
-
"grad_norm": 1.3362579345703125,
|
2135 |
-
"learning_rate": 9.010171012301291e-06,
|
2136 |
-
"loss": 0.0777,
|
2137 |
-
"step": 7600
|
2138 |
-
},
|
2139 |
-
{
|
2140 |
-
"epoch": 0.687432383699964,
|
2141 |
-
"grad_norm": 0.3084631562232971,
|
2142 |
-
"learning_rate": 8.892128053525471e-06,
|
2143 |
-
"loss": 0.1273,
|
2144 |
-
"step": 7625
|
2145 |
-
},
|
2146 |
-
{
|
2147 |
-
"epoch": 0.6896862603678326,
|
2148 |
-
"grad_norm": 0.2445574402809143,
|
2149 |
-
"learning_rate": 8.77464200997471e-06,
|
2150 |
-
"loss": 0.1139,
|
2151 |
-
"step": 7650
|
2152 |
-
},
|
2153 |
-
{
|
2154 |
-
"epoch": 0.6919401370357015,
|
2155 |
-
"grad_norm": 85.26118469238281,
|
2156 |
-
"learning_rate": 8.657718772045176e-06,
|
2157 |
-
"loss": 0.09,
|
2158 |
-
"step": 7675
|
2159 |
-
},
|
2160 |
-
{
|
2161 |
-
"epoch": 0.6941940137035701,
|
2162 |
-
"grad_norm": 13.499275207519531,
|
2163 |
-
"learning_rate": 8.541364201915662e-06,
|
2164 |
-
"loss": 0.08,
|
2165 |
-
"step": 7700
|
2166 |
-
},
|
2167 |
-
{
|
2168 |
-
"epoch": 0.6964478903714388,
|
2169 |
-
"grad_norm": 0.7919896841049194,
|
2170 |
-
"learning_rate": 8.42558413325367e-06,
|
2171 |
-
"loss": 0.1191,
|
2172 |
-
"step": 7725
|
2173 |
-
},
|
2174 |
-
{
|
2175 |
-
"epoch": 0.6987017670393076,
|
2176 |
-
"grad_norm": 0.21591199934482574,
|
2177 |
-
"learning_rate": 8.310384370922935e-06,
|
2178 |
-
"loss": 0.0958,
|
2179 |
-
"step": 7750
|
2180 |
-
},
|
2181 |
-
{
|
2182 |
-
"epoch": 0.7009556437071763,
|
2183 |
-
"grad_norm": 46.4476203918457,
|
2184 |
-
"learning_rate": 8.195770690692373e-06,
|
2185 |
-
"loss": 0.1997,
|
2186 |
-
"step": 7775
|
2187 |
-
},
|
2188 |
-
{
|
2189 |
-
"epoch": 0.703209520375045,
|
2190 |
-
"grad_norm": 0.061485011130571365,
|
2191 |
-
"learning_rate": 8.081748838946516e-06,
|
2192 |
-
"loss": 0.0815,
|
2193 |
-
"step": 7800
|
2194 |
-
},
|
2195 |
-
{
|
2196 |
-
"epoch": 0.7054633970429138,
|
2197 |
-
"grad_norm": 11.417930603027344,
|
2198 |
-
"learning_rate": 7.96832453239739e-06,
|
2199 |
-
"loss": 0.1467,
|
2200 |
-
"step": 7825
|
2201 |
-
},
|
2202 |
-
{
|
2203 |
-
"epoch": 0.7077172737107825,
|
2204 |
-
"grad_norm": 0.6634320616722107,
|
2205 |
-
"learning_rate": 7.85550345779794e-06,
|
2206 |
-
"loss": 0.1075,
|
2207 |
-
"step": 7850
|
2208 |
-
},
|
2209 |
-
{
|
2210 |
-
"epoch": 0.7099711503786513,
|
2211 |
-
"grad_norm": 0.022482017055153847,
|
2212 |
-
"learning_rate": 7.743291271656818e-06,
|
2213 |
-
"loss": 0.1331,
|
2214 |
-
"step": 7875
|
2215 |
-
},
|
2216 |
-
{
|
2217 |
-
"epoch": 0.71222502704652,
|
2218 |
-
"grad_norm": 0.21041303873062134,
|
2219 |
-
"learning_rate": 7.631693599954901e-06,
|
2220 |
-
"loss": 0.1209,
|
2221 |
-
"step": 7900
|
2222 |
-
},
|
2223 |
-
{
|
2224 |
-
"epoch": 0.7144789037143887,
|
2225 |
-
"grad_norm": 0.19395408034324646,
|
2226 |
-
"learning_rate": 7.520716037863143e-06,
|
2227 |
-
"loss": 0.0381,
|
2228 |
-
"step": 7925
|
2229 |
-
},
|
2230 |
-
{
|
2231 |
-
"epoch": 0.7167327803822575,
|
2232 |
-
"grad_norm": 0.23180022835731506,
|
2233 |
-
"learning_rate": 7.4103641494620434e-06,
|
2234 |
-
"loss": 0.0033,
|
2235 |
-
"step": 7950
|
2236 |
-
},
|
2237 |
-
{
|
2238 |
-
"epoch": 0.7189866570501262,
|
2239 |
-
"grad_norm": 0.12363322079181671,
|
2240 |
-
"learning_rate": 7.300643467462758e-06,
|
2241 |
-
"loss": 0.0938,
|
2242 |
-
"step": 7975
|
2243 |
-
},
|
2244 |
-
{
|
2245 |
-
"epoch": 0.7212405337179949,
|
2246 |
-
"grad_norm": 0.027875280007719994,
|
2247 |
-
"learning_rate": 7.191559492929623e-06,
|
2248 |
-
"loss": 0.0755,
|
2249 |
-
"step": 8000
|
2250 |
-
},
|
2251 |
-
{
|
2252 |
-
"epoch": 0.7234944103858637,
|
2253 |
-
"grad_norm": 0.13949647545814514,
|
2254 |
-
"learning_rate": 7.083117695004389e-06,
|
2255 |
-
"loss": 0.0628,
|
2256 |
-
"step": 8025
|
2257 |
-
},
|
2258 |
-
{
|
2259 |
-
"epoch": 0.7257482870537324,
|
2260 |
-
"grad_norm": 0.3312737047672272,
|
2261 |
-
"learning_rate": 6.9753235106320084e-06,
|
2262 |
-
"loss": 0.1372,
|
2263 |
-
"step": 8050
|
2264 |
-
},
|
2265 |
-
{
|
2266 |
-
"epoch": 0.7280021637216012,
|
2267 |
-
"grad_norm": 1.7089711427688599,
|
2268 |
-
"learning_rate": 6.868182344288039e-06,
|
2269 |
-
"loss": 0.0803,
|
2270 |
-
"step": 8075
|
2271 |
-
},
|
2272 |
-
{
|
2273 |
-
"epoch": 0.7302560403894699,
|
2274 |
-
"grad_norm": 0.25711366534233093,
|
2275 |
-
"learning_rate": 6.761699567707685e-06,
|
2276 |
-
"loss": 0.1412,
|
2277 |
-
"step": 8100
|
2278 |
-
},
|
2279 |
-
{
|
2280 |
-
"epoch": 0.7325099170573386,
|
2281 |
-
"grad_norm": 4.719892501831055,
|
2282 |
-
"learning_rate": 6.655880519616469e-06,
|
2283 |
-
"loss": 0.0907,
|
2284 |
-
"step": 8125
|
2285 |
-
},
|
2286 |
-
{
|
2287 |
-
"epoch": 0.7347637937252074,
|
2288 |
-
"grad_norm": 0.087191142141819,
|
2289 |
-
"learning_rate": 6.550730505462566e-06,
|
2290 |
-
"loss": 0.0798,
|
2291 |
-
"step": 8150
|
2292 |
-
},
|
2293 |
-
{
|
2294 |
-
"epoch": 0.7370176703930761,
|
2295 |
-
"grad_norm": 0.10154922306537628,
|
2296 |
-
"learning_rate": 6.446254797150824e-06,
|
2297 |
-
"loss": 0.1265,
|
2298 |
-
"step": 8175
|
2299 |
-
},
|
2300 |
-
{
|
2301 |
-
"epoch": 0.7392715470609448,
|
2302 |
-
"grad_norm": 20.8658447265625,
|
2303 |
-
"learning_rate": 6.342458632778398e-06,
|
2304 |
-
"loss": 0.1207,
|
2305 |
-
"step": 8200
|
2306 |
-
},
|
2307 |
-
{
|
2308 |
-
"epoch": 0.7415254237288136,
|
2309 |
-
"grad_norm": 0.08694490045309067,
|
2310 |
-
"learning_rate": 6.239347216372171e-06,
|
2311 |
-
"loss": 0.1041,
|
2312 |
-
"step": 8225
|
2313 |
-
},
|
2314 |
-
{
|
2315 |
-
"epoch": 0.7437793003966823,
|
2316 |
-
"grad_norm": 1.5035579204559326,
|
2317 |
-
"learning_rate": 6.136925717627846e-06,
|
2318 |
-
"loss": 0.1157,
|
2319 |
-
"step": 8250
|
2320 |
-
},
|
2321 |
-
{
|
2322 |
-
"epoch": 0.7460331770645511,
|
2323 |
-
"grad_norm": 4.682013988494873,
|
2324 |
-
"learning_rate": 6.035199271650694e-06,
|
2325 |
-
"loss": 0.1308,
|
2326 |
-
"step": 8275
|
2327 |
-
},
|
2328 |
-
{
|
2329 |
-
"epoch": 0.7482870537324198,
|
2330 |
-
"grad_norm": 1.0564689636230469,
|
2331 |
-
"learning_rate": 5.934172978698172e-06,
|
2332 |
-
"loss": 0.0485,
|
2333 |
-
"step": 8300
|
2334 |
-
},
|
2335 |
-
{
|
2336 |
-
"epoch": 0.7505409304002885,
|
2337 |
-
"grad_norm": 0.13109451532363892,
|
2338 |
-
"learning_rate": 5.833851903924153e-06,
|
2339 |
-
"loss": 0.1207,
|
2340 |
-
"step": 8325
|
2341 |
-
},
|
2342 |
-
{
|
2343 |
-
"epoch": 0.7527948070681573,
|
2344 |
-
"grad_norm": 0.2509864568710327,
|
2345 |
-
"learning_rate": 5.7342410771250025e-06,
|
2346 |
-
"loss": 0.1318,
|
2347 |
-
"step": 8350
|
2348 |
-
},
|
2349 |
-
{
|
2350 |
-
"epoch": 0.755048683736026,
|
2351 |
-
"grad_norm": 0.8034769892692566,
|
2352 |
-
"learning_rate": 5.6353454924873916e-06,
|
2353 |
-
"loss": 0.1167,
|
2354 |
-
"step": 8375
|
2355 |
-
},
|
2356 |
-
{
|
2357 |
-
"epoch": 0.7573025604038947,
|
2358 |
-
"grad_norm": 0.0480596087872982,
|
2359 |
-
"learning_rate": 5.537170108337906e-06,
|
2360 |
-
"loss": 0.0985,
|
2361 |
-
"step": 8400
|
2362 |
-
},
|
2363 |
-
{
|
2364 |
-
"epoch": 0.7595564370717635,
|
2365 |
-
"grad_norm": 0.04104947671294212,
|
2366 |
-
"learning_rate": 5.439719846894449e-06,
|
2367 |
-
"loss": 0.1416,
|
2368 |
-
"step": 8425
|
2369 |
-
},
|
2370 |
-
{
|
2371 |
-
"epoch": 0.7618103137396322,
|
2372 |
-
"grad_norm": 0.134425550699234,
|
2373 |
-
"learning_rate": 5.342999594019456e-06,
|
2374 |
-
"loss": 0.0662,
|
2375 |
-
"step": 8450
|
2376 |
-
},
|
2377 |
-
{
|
2378 |
-
"epoch": 0.764064190407501,
|
2379 |
-
"grad_norm": 0.10297323763370514,
|
2380 |
-
"learning_rate": 5.2470141989749315e-06,
|
2381 |
-
"loss": 0.0952,
|
2382 |
-
"step": 8475
|
2383 |
-
},
|
2384 |
-
{
|
2385 |
-
"epoch": 0.7663180670753696,
|
2386 |
-
"grad_norm": 0.14851029217243195,
|
2387 |
-
"learning_rate": 5.151768474179324e-06,
|
2388 |
-
"loss": 0.1873,
|
2389 |
-
"step": 8500
|
2390 |
-
},
|
2391 |
-
{
|
2392 |
-
"epoch": 0.7685719437432383,
|
2393 |
-
"grad_norm": 11.838191986083984,
|
2394 |
-
"learning_rate": 5.057267194966249e-06,
|
2395 |
-
"loss": 0.1235,
|
2396 |
-
"step": 8525
|
2397 |
-
},
|
2398 |
-
{
|
2399 |
-
"epoch": 0.7708258204111071,
|
2400 |
-
"grad_norm": 0.02550826221704483,
|
2401 |
-
"learning_rate": 4.9635150993450505e-06,
|
2402 |
-
"loss": 0.1082,
|
2403 |
-
"step": 8550
|
2404 |
-
},
|
2405 |
-
{
|
2406 |
-
"epoch": 0.7730796970789758,
|
2407 |
-
"grad_norm": 0.041587576270103455,
|
2408 |
-
"learning_rate": 4.870516887763292e-06,
|
2409 |
-
"loss": 0.0939,
|
2410 |
-
"step": 8575
|
2411 |
-
},
|
2412 |
-
{
|
2413 |
-
"epoch": 0.7753335737468445,
|
2414 |
-
"grad_norm": 1.0598152875900269,
|
2415 |
-
"learning_rate": 4.778277222871033e-06,
|
2416 |
-
"loss": 0.1774,
|
2417 |
-
"step": 8600
|
2418 |
-
},
|
2419 |
-
{
|
2420 |
-
"epoch": 0.7775874504147133,
|
2421 |
-
"grad_norm": 1.3883267641067505,
|
2422 |
-
"learning_rate": 4.686800729287095e-06,
|
2423 |
-
"loss": 0.1326,
|
2424 |
-
"step": 8625
|
2425 |
-
},
|
2426 |
-
{
|
2427 |
-
"epoch": 0.779841327082582,
|
2428 |
-
"grad_norm": 0.03694922849535942,
|
2429 |
-
"learning_rate": 4.596091993367215e-06,
|
2430 |
-
"loss": 0.0745,
|
2431 |
-
"step": 8650
|
2432 |
-
},
|
2433 |
-
{
|
2434 |
-
"epoch": 0.7820952037504508,
|
2435 |
-
"grad_norm": 3.130089044570923,
|
2436 |
-
"learning_rate": 4.5061555629740325e-06,
|
2437 |
-
"loss": 0.1294,
|
2438 |
-
"step": 8675
|
2439 |
-
},
|
2440 |
-
{
|
2441 |
-
"epoch": 0.7843490804183195,
|
2442 |
-
"grad_norm": 0.1829773634672165,
|
2443 |
-
"learning_rate": 4.416995947249154e-06,
|
2444 |
-
"loss": 0.1027,
|
2445 |
-
"step": 8700
|
2446 |
-
},
|
2447 |
-
{
|
2448 |
-
"epoch": 0.7866029570861882,
|
2449 |
-
"grad_norm": 0.5552740097045898,
|
2450 |
-
"learning_rate": 4.3286176163870186e-06,
|
2451 |
-
"loss": 0.1147,
|
2452 |
-
"step": 8725
|
2453 |
-
},
|
2454 |
-
{
|
2455 |
-
"epoch": 0.788856833754057,
|
2456 |
-
"grad_norm": 13.672005653381348,
|
2457 |
-
"learning_rate": 4.241025001410801e-06,
|
2458 |
-
"loss": 0.124,
|
2459 |
-
"step": 8750
|
2460 |
-
},
|
2461 |
-
{
|
2462 |
-
"epoch": 0.7911107104219257,
|
2463 |
-
"grad_norm": 0.08975929021835327,
|
2464 |
-
"learning_rate": 4.154222493950248e-06,
|
2465 |
-
"loss": 0.1166,
|
2466 |
-
"step": 8775
|
2467 |
-
},
|
2468 |
-
{
|
2469 |
-
"epoch": 0.7933645870897944,
|
2470 |
-
"grad_norm": 0.079860158264637,
|
2471 |
-
"learning_rate": 4.0682144460215015e-06,
|
2472 |
-
"loss": 0.0722,
|
2473 |
-
"step": 8800
|
2474 |
-
},
|
2475 |
-
{
|
2476 |
-
"epoch": 0.7956184637576632,
|
2477 |
-
"grad_norm": 0.3616615831851959,
|
2478 |
-
"learning_rate": 3.983005169808891e-06,
|
2479 |
-
"loss": 0.0757,
|
2480 |
-
"step": 8825
|
2481 |
-
},
|
2482 |
-
{
|
2483 |
-
"epoch": 0.7978723404255319,
|
2484 |
-
"grad_norm": 0.37976741790771484,
|
2485 |
-
"learning_rate": 3.898598937448743e-06,
|
2486 |
-
"loss": 0.0707,
|
2487 |
-
"step": 8850
|
2488 |
-
},
|
2489 |
-
{
|
2490 |
-
"epoch": 0.8001262170934007,
|
2491 |
-
"grad_norm": 10.676128387451172,
|
2492 |
-
"learning_rate": 3.81499998081518e-06,
|
2493 |
-
"loss": 0.1153,
|
2494 |
-
"step": 8875
|
2495 |
-
},
|
2496 |
-
{
|
2497 |
-
"epoch": 0.8023800937612694,
|
2498 |
-
"grad_norm": 0.10521873086690903,
|
2499 |
-
"learning_rate": 3.7322124913079673e-06,
|
2500 |
-
"loss": 0.0695,
|
2501 |
-
"step": 8900
|
2502 |
-
},
|
2503 |
-
{
|
2504 |
-
"epoch": 0.8046339704291381,
|
2505 |
-
"grad_norm": 0.13010619580745697,
|
2506 |
-
"learning_rate": 3.650240619642331e-06,
|
2507 |
-
"loss": 0.1037,
|
2508 |
-
"step": 8925
|
2509 |
-
},
|
2510 |
-
{
|
2511 |
-
"epoch": 0.8068878470970069,
|
2512 |
-
"grad_norm": 28.179489135742188,
|
2513 |
-
"learning_rate": 3.5690884756408872e-06,
|
2514 |
-
"loss": 0.1263,
|
2515 |
-
"step": 8950
|
2516 |
-
},
|
2517 |
-
{
|
2518 |
-
"epoch": 0.8091417237648756,
|
2519 |
-
"grad_norm": 112.59784698486328,
|
2520 |
-
"learning_rate": 3.4887601280275973e-06,
|
2521 |
-
"loss": 0.0483,
|
2522 |
-
"step": 8975
|
2523 |
-
},
|
2524 |
-
{
|
2525 |
-
"epoch": 0.8113956004327443,
|
2526 |
-
"grad_norm": 25.88535499572754,
|
2527 |
-
"learning_rate": 3.409259604223716e-06,
|
2528 |
-
"loss": 0.0999,
|
2529 |
-
"step": 9000
|
2530 |
-
},
|
2531 |
-
{
|
2532 |
-
"epoch": 0.8136494771006131,
|
2533 |
-
"grad_norm": 0.05944238603115082,
|
2534 |
-
"learning_rate": 3.3305908901459282e-06,
|
2535 |
-
"loss": 0.0837,
|
2536 |
-
"step": 9025
|
2537 |
-
},
|
2538 |
-
{
|
2539 |
-
"epoch": 0.8159033537684818,
|
2540 |
-
"grad_norm": 0.20255301892757416,
|
2541 |
-
"learning_rate": 3.252757930006494e-06,
|
2542 |
-
"loss": 0.1378,
|
2543 |
-
"step": 9050
|
2544 |
-
},
|
2545 |
-
{
|
2546 |
-
"epoch": 0.8181572304363505,
|
2547 |
-
"grad_norm": 0.08721674978733063,
|
2548 |
-
"learning_rate": 3.1757646261154785e-06,
|
2549 |
-
"loss": 0.0539,
|
2550 |
-
"step": 9075
|
2551 |
-
},
|
2552 |
-
{
|
2553 |
-
"epoch": 0.8204111071042193,
|
2554 |
-
"grad_norm": 20.20242691040039,
|
2555 |
-
"learning_rate": 3.0996148386851033e-06,
|
2556 |
-
"loss": 0.1729,
|
2557 |
-
"step": 9100
|
2558 |
-
},
|
2559 |
-
{
|
2560 |
-
"epoch": 0.822664983772088,
|
2561 |
-
"grad_norm": 95.36360931396484,
|
2562 |
-
"learning_rate": 3.0243123856362364e-06,
|
2563 |
-
"loss": 0.0509,
|
2564 |
-
"step": 9125
|
2565 |
-
},
|
2566 |
-
{
|
2567 |
-
"epoch": 0.8249188604399568,
|
2568 |
-
"grad_norm": 0.11352234333753586,
|
2569 |
-
"learning_rate": 2.949861042406945e-06,
|
2570 |
-
"loss": 0.1085,
|
2571 |
-
"step": 9150
|
2572 |
-
},
|
2573 |
-
{
|
2574 |
-
"epoch": 0.8271727371078255,
|
2575 |
-
"grad_norm": 0.1542273312807083,
|
2576 |
-
"learning_rate": 2.8762645417632117e-06,
|
2577 |
-
"loss": 0.1072,
|
2578 |
-
"step": 9175
|
2579 |
-
},
|
2580 |
-
{
|
2581 |
-
"epoch": 0.8294266137756942,
|
2582 |
-
"grad_norm": 20.265731811523438,
|
2583 |
-
"learning_rate": 2.803526573611792e-06,
|
2584 |
-
"loss": 0.1221,
|
2585 |
-
"step": 9200
|
2586 |
-
},
|
2587 |
-
{
|
2588 |
-
"epoch": 0.831680490443563,
|
2589 |
-
"grad_norm": 17.208614349365234,
|
2590 |
-
"learning_rate": 2.731650784815205e-06,
|
2591 |
-
"loss": 0.1315,
|
2592 |
-
"step": 9225
|
2593 |
-
},
|
2594 |
-
{
|
2595 |
-
"epoch": 0.8339343671114317,
|
2596 |
-
"grad_norm": 0.11168497055768967,
|
2597 |
-
"learning_rate": 2.660640779008903e-06,
|
2598 |
-
"loss": 0.0845,
|
2599 |
-
"step": 9250
|
2600 |
-
},
|
2601 |
-
{
|
2602 |
-
"epoch": 0.8361882437793003,
|
2603 |
-
"grad_norm": 0.7374106645584106,
|
2604 |
-
"learning_rate": 2.5905001164205802e-06,
|
2605 |
-
"loss": 0.1493,
|
2606 |
-
"step": 9275
|
2607 |
-
},
|
2608 |
-
{
|
2609 |
-
"epoch": 0.8384421204471691,
|
2610 |
-
"grad_norm": 55.767250061035156,
|
2611 |
-
"learning_rate": 2.521232313691686e-06,
|
2612 |
-
"loss": 0.0589,
|
2613 |
-
"step": 9300
|
2614 |
-
},
|
2615 |
-
{
|
2616 |
-
"epoch": 0.8406959971150378,
|
2617 |
-
"grad_norm": 0.0631784126162529,
|
2618 |
-
"learning_rate": 2.4528408437011053e-06,
|
2619 |
-
"loss": 0.1091,
|
2620 |
-
"step": 9325
|
2621 |
-
},
|
2622 |
-
{
|
2623 |
-
"epoch": 0.8429498737829066,
|
2624 |
-
"grad_norm": 0.40094318985939026,
|
2625 |
-
"learning_rate": 2.3853291353910368e-06,
|
2626 |
-
"loss": 0.0734,
|
2627 |
-
"step": 9350
|
2628 |
-
},
|
2629 |
-
{
|
2630 |
-
"epoch": 0.8452037504507753,
|
2631 |
-
"grad_norm": 0.050913214683532715,
|
2632 |
-
"learning_rate": 2.3187005735950917e-06,
|
2633 |
-
"loss": 0.0873,
|
2634 |
-
"step": 9375
|
2635 |
-
},
|
2636 |
-
{
|
2637 |
-
"epoch": 0.847457627118644,
|
2638 |
-
"grad_norm": 0.302331805229187,
|
2639 |
-
"learning_rate": 2.2529584988685716e-06,
|
2640 |
-
"loss": 0.0812,
|
2641 |
-
"step": 9400
|
2642 |
-
},
|
2643 |
-
{
|
2644 |
-
"epoch": 0.8497115037865128,
|
2645 |
-
"grad_norm": 14.71583366394043,
|
2646 |
-
"learning_rate": 2.18810620732097e-06,
|
2647 |
-
"loss": 0.023,
|
2648 |
-
"step": 9425
|
2649 |
-
},
|
2650 |
-
{
|
2651 |
-
"epoch": 0.8519653804543815,
|
2652 |
-
"grad_norm": 0.033386293798685074,
|
2653 |
-
"learning_rate": 2.124146950450763e-06,
|
2654 |
-
"loss": 0.03,
|
2655 |
-
"step": 9450
|
2656 |
-
},
|
2657 |
-
{
|
2658 |
-
"epoch": 0.8542192571222502,
|
2659 |
-
"grad_norm": 18.24811363220215,
|
2660 |
-
"learning_rate": 2.0610839349823397e-06,
|
2661 |
-
"loss": 0.0981,
|
2662 |
-
"step": 9475
|
2663 |
-
},
|
2664 |
-
{
|
2665 |
-
"epoch": 0.856473133790119,
|
2666 |
-
"grad_norm": 17.850160598754883,
|
2667 |
-
"learning_rate": 1.9989203227052446e-06,
|
2668 |
-
"loss": 0.0695,
|
2669 |
-
"step": 9500
|
2670 |
-
},
|
2671 |
-
{
|
2672 |
-
"epoch": 0.8587270104579877,
|
2673 |
-
"grad_norm": 15.309844970703125,
|
2674 |
-
"learning_rate": 1.9376592303156695e-06,
|
2675 |
-
"loss": 0.1251,
|
2676 |
-
"step": 9525
|
2677 |
-
},
|
2678 |
-
{
|
2679 |
-
"epoch": 0.8609808871258565,
|
2680 |
-
"grad_norm": 0.05771295726299286,
|
2681 |
-
"learning_rate": 1.8773037292601716e-06,
|
2682 |
-
"loss": 0.0788,
|
2683 |
-
"step": 9550
|
2684 |
-
},
|
2685 |
-
{
|
2686 |
-
"epoch": 0.8632347637937252,
|
2687 |
-
"grad_norm": 0.12085491418838501,
|
2688 |
-
"learning_rate": 1.8178568455816891e-06,
|
2689 |
-
"loss": 0.139,
|
2690 |
-
"step": 9575
|
2691 |
-
},
|
2692 |
-
{
|
2693 |
-
"epoch": 0.8654886404615939,
|
2694 |
-
"grad_norm": 65.5085220336914,
|
2695 |
-
"learning_rate": 1.7593215597678193e-06,
|
2696 |
-
"loss": 0.1496,
|
2697 |
-
"step": 9600
|
2698 |
-
},
|
2699 |
-
{
|
2700 |
-
"epoch": 0.8677425171294627,
|
2701 |
-
"grad_norm": 0.34007498621940613,
|
2702 |
-
"learning_rate": 1.7017008066013963e-06,
|
2703 |
-
"loss": 0.0861,
|
2704 |
-
"step": 9625
|
2705 |
-
},
|
2706 |
-
{
|
2707 |
-
"epoch": 0.8699963937973314,
|
2708 |
-
"grad_norm": 0.053930457681417465,
|
2709 |
-
"learning_rate": 1.6449974750133346e-06,
|
2710 |
-
"loss": 0.211,
|
2711 |
-
"step": 9650
|
2712 |
-
},
|
2713 |
-
{
|
2714 |
-
"epoch": 0.8722502704652001,
|
2715 |
-
"grad_norm": 0.12199834734201431,
|
2716 |
-
"learning_rate": 1.589214407937807e-06,
|
2717 |
-
"loss": 0.1641,
|
2718 |
-
"step": 9675
|
2719 |
-
},
|
2720 |
-
{
|
2721 |
-
"epoch": 0.8745041471330689,
|
2722 |
-
"grad_norm": 0.06262707710266113,
|
2723 |
-
"learning_rate": 1.5343544021696866e-06,
|
2724 |
-
"loss": 0.0992,
|
2725 |
-
"step": 9700
|
2726 |
-
},
|
2727 |
-
{
|
2728 |
-
"epoch": 0.8767580238009376,
|
2729 |
-
"grad_norm": 19.91883659362793,
|
2730 |
-
"learning_rate": 1.480420208224338e-06,
|
2731 |
-
"loss": 0.0709,
|
2732 |
-
"step": 9725
|
2733 |
-
},
|
2734 |
-
{
|
2735 |
-
"epoch": 0.8790119004688064,
|
2736 |
-
"grad_norm": 0.03692913427948952,
|
2737 |
-
"learning_rate": 1.4274145301997088e-06,
|
2738 |
-
"loss": 0.1296,
|
2739 |
-
"step": 9750
|
2740 |
-
},
|
2741 |
-
{
|
2742 |
-
"epoch": 0.8812657771366751,
|
2743 |
-
"grad_norm": 16.19684410095215,
|
2744 |
-
"learning_rate": 1.3753400256407656e-06,
|
2745 |
-
"loss": 0.0858,
|
2746 |
-
"step": 9775
|
2747 |
-
},
|
2748 |
-
{
|
2749 |
-
"epoch": 0.8835196538045438,
|
2750 |
-
"grad_norm": 212.5207977294922,
|
2751 |
-
"learning_rate": 1.3241993054062375e-06,
|
2752 |
-
"loss": 0.134,
|
2753 |
-
"step": 9800
|
2754 |
-
},
|
2755 |
-
{
|
2756 |
-
"epoch": 0.8857735304724126,
|
2757 |
-
"grad_norm": 0.016040222719311714,
|
2758 |
-
"learning_rate": 1.2739949335377077e-06,
|
2759 |
-
"loss": 0.1047,
|
2760 |
-
"step": 9825
|
2761 |
-
},
|
2762 |
-
{
|
2763 |
-
"epoch": 0.8880274071402813,
|
2764 |
-
"grad_norm": 0.08780741691589355,
|
2765 |
-
"learning_rate": 1.2247294271311017e-06,
|
2766 |
-
"loss": 0.0581,
|
2767 |
-
"step": 9850
|
2768 |
-
},
|
2769 |
-
{
|
2770 |
-
"epoch": 0.89028128380815,
|
2771 |
-
"grad_norm": 0.24466896057128906,
|
2772 |
-
"learning_rate": 1.1764052562104444e-06,
|
2773 |
-
"loss": 0.0927,
|
2774 |
-
"step": 9875
|
2775 |
-
},
|
2776 |
-
{
|
2777 |
-
"epoch": 0.8925351604760188,
|
2778 |
-
"grad_norm": 1.2905555963516235,
|
2779 |
-
"learning_rate": 1.1290248436040274e-06,
|
2780 |
-
"loss": 0.094,
|
2781 |
-
"step": 9900
|
2782 |
-
},
|
2783 |
-
{
|
2784 |
-
"epoch": 0.8947890371438875,
|
2785 |
-
"grad_norm": 0.01533698383718729,
|
2786 |
-
"learning_rate": 1.082590564822963e-06,
|
2787 |
-
"loss": 0.114,
|
2788 |
-
"step": 9925
|
2789 |
-
},
|
2790 |
-
{
|
2791 |
-
"epoch": 0.8970429138117563,
|
2792 |
-
"grad_norm": 0.10088209062814713,
|
2793 |
-
"learning_rate": 1.0371047479420547e-06,
|
2794 |
-
"loss": 0.1122,
|
2795 |
-
"step": 9950
|
2796 |
-
},
|
2797 |
-
{
|
2798 |
-
"epoch": 0.899296790479625,
|
2799 |
-
"grad_norm": 7.259499549865723,
|
2800 |
-
"learning_rate": 9.925696734830858e-07,
|
2801 |
-
"loss": 0.0938,
|
2802 |
-
"step": 9975
|
2803 |
-
},
|
2804 |
-
{
|
2805 |
-
"epoch": 0.9015506671474937,
|
2806 |
-
"grad_norm": 1.123653531074524,
|
2807 |
-
"learning_rate": 9.489875743004795e-07,
|
2808 |
-
"loss": 0.0856,
|
2809 |
-
"step": 10000
|
2810 |
-
},
|
2811 |
-
{
|
2812 |
-
"epoch": 0.9038045438153625,
|
2813 |
-
"grad_norm": 0.11663197726011276,
|
2814 |
-
"learning_rate": 9.063606354693521e-07,
|
2815 |
-
"loss": 0.0974,
|
2816 |
-
"step": 10025
|
2817 |
-
},
|
2818 |
-
{
|
2819 |
-
"epoch": 0.9060584204832312,
|
2820 |
-
"grad_norm": 0.07199467718601227,
|
2821 |
-
"learning_rate": 8.646909941759563e-07,
|
2822 |
-
"loss": 0.1364,
|
2823 |
-
"step": 10050
|
2824 |
-
},
|
2825 |
-
{
|
2826 |
-
"epoch": 0.9083122971510998,
|
2827 |
-
"grad_norm": 0.43899527192115784,
|
2828 |
-
"learning_rate": 8.239807396105326e-07,
|
2829 |
-
"loss": 0.0741,
|
2830 |
-
"step": 10075
|
2831 |
-
},
|
2832 |
-
{
|
2833 |
-
"epoch": 0.9105661738189686,
|
2834 |
-
"grad_norm": 0.6985141634941101,
|
2835 |
-
"learning_rate": 7.842319128625608e-07,
|
2836 |
-
"loss": 0.0063,
|
2837 |
-
"step": 10100
|
2838 |
-
},
|
2839 |
-
{
|
2840 |
-
"epoch": 0.9128200504868373,
|
2841 |
-
"grad_norm": 0.19404803216457367,
|
2842 |
-
"learning_rate": 7.454465068184347e-07,
|
2843 |
-
"loss": 0.079,
|
2844 |
-
"step": 10125
|
2845 |
-
},
|
2846 |
-
{
|
2847 |
-
"epoch": 0.9150739271547061,
|
2848 |
-
"grad_norm": 23.735002517700195,
|
2849 |
-
"learning_rate": 7.076264660615196e-07,
|
2850 |
-
"loss": 0.1429,
|
2851 |
-
"step": 10150
|
2852 |
-
},
|
2853 |
-
{
|
2854 |
-
"epoch": 0.9173278038225748,
|
2855 |
-
"grad_norm": 0.049147795885801315,
|
2856 |
-
"learning_rate": 6.70773686774695e-07,
|
2857 |
-
"loss": 0.0823,
|
2858 |
-
"step": 10175
|
2859 |
-
},
|
2860 |
-
{
|
2861 |
-
"epoch": 0.9195816804904435,
|
2862 |
-
"grad_norm": 36.494075775146484,
|
2863 |
-
"learning_rate": 6.348900166452555e-07,
|
2864 |
-
"loss": 0.0828,
|
2865 |
-
"step": 10200
|
2866 |
-
},
|
2867 |
-
{
|
2868 |
-
"epoch": 0.9218355571583123,
|
2869 |
-
"grad_norm": 9.039645195007324,
|
2870 |
-
"learning_rate": 5.999772547722748e-07,
|
2871 |
-
"loss": 0.1286,
|
2872 |
-
"step": 10225
|
2873 |
-
},
|
2874 |
-
{
|
2875 |
-
"epoch": 0.924089433826181,
|
2876 |
-
"grad_norm": 0.5087681412696838,
|
2877 |
-
"learning_rate": 5.660371515764306e-07,
|
2878 |
-
"loss": 0.0959,
|
2879 |
-
"step": 10250
|
2880 |
-
},
|
2881 |
-
{
|
2882 |
-
"epoch": 0.9263433104940497,
|
2883 |
-
"grad_norm": 0.03386404365301132,
|
2884 |
-
"learning_rate": 5.330714087122113e-07,
|
2885 |
-
"loss": 0.0676,
|
2886 |
-
"step": 10275
|
2887 |
-
},
|
2888 |
-
{
|
2889 |
-
"epoch": 0.9285971871619185,
|
2890 |
-
"grad_norm": 0.08505436778068542,
|
2891 |
-
"learning_rate": 5.010816789826223e-07,
|
2892 |
-
"loss": 0.1809,
|
2893 |
-
"step": 10300
|
2894 |
-
},
|
2895 |
-
{
|
2896 |
-
"epoch": 0.9308510638297872,
|
2897 |
-
"grad_norm": 0.23430904746055603,
|
2898 |
-
"learning_rate": 4.7006956625630596e-07,
|
2899 |
-
"loss": 0.0842,
|
2900 |
-
"step": 10325
|
2901 |
-
},
|
2902 |
-
{
|
2903 |
-
"epoch": 0.933104940497656,
|
2904 |
-
"grad_norm": 0.33207935094833374,
|
2905 |
-
"learning_rate": 4.4003662538714133e-07,
|
2906 |
-
"loss": 0.0891,
|
2907 |
-
"step": 10350
|
2908 |
-
},
|
2909 |
-
{
|
2910 |
-
"epoch": 0.9353588171655247,
|
2911 |
-
"grad_norm": 0.046333521604537964,
|
2912 |
-
"learning_rate": 4.1098436213627303e-07,
|
2913 |
-
"loss": 0.1337,
|
2914 |
-
"step": 10375
|
2915 |
-
},
|
2916 |
-
{
|
2917 |
-
"epoch": 0.9376126938333934,
|
2918 |
-
"grad_norm": 0.08626014739274979,
|
2919 |
-
"learning_rate": 3.829142330966318e-07,
|
2920 |
-
"loss": 0.1029,
|
2921 |
-
"step": 10400
|
2922 |
-
},
|
2923 |
-
{
|
2924 |
-
"epoch": 0.9398665705012622,
|
2925 |
-
"grad_norm": 0.9835343956947327,
|
2926 |
-
"learning_rate": 3.5582764561989504e-07,
|
2927 |
-
"loss": 0.0736,
|
2928 |
-
"step": 10425
|
2929 |
-
},
|
2930 |
-
{
|
2931 |
-
"epoch": 0.9421204471691309,
|
2932 |
-
"grad_norm": 0.13581854104995728,
|
2933 |
-
"learning_rate": 3.297259577459322e-07,
|
2934 |
-
"loss": 0.0854,
|
2935 |
-
"step": 10450
|
2936 |
-
},
|
2937 |
-
{
|
2938 |
-
"epoch": 0.9443743238369996,
|
2939 |
-
"grad_norm": 0.10793433338403702,
|
2940 |
-
"learning_rate": 3.046104781347148e-07,
|
2941 |
-
"loss": 0.1951,
|
2942 |
-
"step": 10475
|
2943 |
-
},
|
2944 |
-
{
|
2945 |
-
"epoch": 0.9466282005048684,
|
2946 |
-
"grad_norm": 0.17822891473770142,
|
2947 |
-
"learning_rate": 2.8048246600070217e-07,
|
2948 |
-
"loss": 0.0125,
|
2949 |
-
"step": 10500
|
2950 |
-
},
|
2951 |
-
{
|
2952 |
-
"epoch": 0.9488820771727371,
|
2953 |
-
"grad_norm": 0.1914602369070053,
|
2954 |
-
"learning_rate": 2.573431310497143e-07,
|
2955 |
-
"loss": 0.1082,
|
2956 |
-
"step": 10525
|
2957 |
-
},
|
2958 |
-
{
|
2959 |
-
"epoch": 0.9511359538406059,
|
2960 |
-
"grad_norm": 0.38527971506118774,
|
2961 |
-
"learning_rate": 2.3519363341826696e-07,
|
2962 |
-
"loss": 0.1474,
|
2963 |
-
"step": 10550
|
2964 |
-
},
|
2965 |
-
{
|
2966 |
-
"epoch": 0.9533898305084746,
|
2967 |
-
"grad_norm": 0.04099535197019577,
|
2968 |
-
"learning_rate": 2.1403508361542924e-07,
|
2969 |
-
"loss": 0.1188,
|
2970 |
-
"step": 10575
|
2971 |
-
},
|
2972 |
-
{
|
2973 |
-
"epoch": 0.9556437071763433,
|
2974 |
-
"grad_norm": 0.1044796034693718,
|
2975 |
-
"learning_rate": 1.9386854246712383e-07,
|
2976 |
-
"loss": 0.106,
|
2977 |
-
"step": 10600
|
2978 |
-
},
|
2979 |
-
{
|
2980 |
-
"epoch": 0.9578975838442121,
|
2981 |
-
"grad_norm": 17.672197341918945,
|
2982 |
-
"learning_rate": 1.7469502106294721e-07,
|
2983 |
-
"loss": 0.0919,
|
2984 |
-
"step": 10625
|
2985 |
-
},
|
2986 |
-
{
|
2987 |
-
"epoch": 0.9601514605120808,
|
2988 |
-
"grad_norm": 0.023873858153820038,
|
2989 |
-
"learning_rate": 1.5651548070548362e-07,
|
2990 |
-
"loss": 0.0265,
|
2991 |
-
"step": 10650
|
2992 |
-
},
|
2993 |
-
{
|
2994 |
-
"epoch": 0.9624053371799495,
|
2995 |
-
"grad_norm": 189.225830078125,
|
2996 |
-
"learning_rate": 1.3933083286210348e-07,
|
2997 |
-
"loss": 0.051,
|
2998 |
-
"step": 10675
|
2999 |
-
},
|
3000 |
-
{
|
3001 |
-
"epoch": 0.9646592138478183,
|
3002 |
-
"grad_norm": 0.07408806681632996,
|
3003 |
-
"learning_rate": 1.2314193911926231e-07,
|
3004 |
-
"loss": 0.1615,
|
3005 |
-
"step": 10700
|
3006 |
-
},
|
3007 |
-
{
|
3008 |
-
"epoch": 0.966913090515687,
|
3009 |
-
"grad_norm": 0.47920337319374084,
|
3010 |
-
"learning_rate": 1.0794961113930192e-07,
|
3011 |
-
"loss": 0.1026,
|
3012 |
-
"step": 10725
|
3013 |
-
},
|
3014 |
-
{
|
3015 |
-
"epoch": 0.9691669671835557,
|
3016 |
-
"grad_norm": 99.75161743164062,
|
3017 |
-
"learning_rate": 9.375461061976732e-08,
|
3018 |
-
"loss": 0.0521,
|
3019 |
-
"step": 10750
|
3020 |
-
},
|
3021 |
-
{
|
3022 |
-
"epoch": 0.9714208438514245,
|
3023 |
-
"grad_norm": 0.2004551738500595,
|
3024 |
-
"learning_rate": 8.055764925520848e-08,
|
3025 |
-
"loss": 0.1462,
|
3026 |
-
"step": 10775
|
3027 |
-
},
|
3028 |
-
{
|
3029 |
-
"epoch": 0.9736747205192932,
|
3030 |
-
"grad_norm": 23.04667854309082,
|
3031 |
-
"learning_rate": 6.83593887014955e-08,
|
3032 |
-
"loss": 0.1151,
|
3033 |
-
"step": 10800
|
3034 |
-
},
|
3035 |
-
{
|
3036 |
-
"epoch": 0.975928597187162,
|
3037 |
-
"grad_norm": 0.1808137744665146,
|
3038 |
-
"learning_rate": 5.716044054264958e-08,
|
3039 |
-
"loss": 0.1206,
|
3040 |
-
"step": 10825
|
3041 |
-
},
|
3042 |
-
{
|
3043 |
-
"epoch": 0.9781824738550307,
|
3044 |
-
"grad_norm": 0.16548039019107819,
|
3045 |
-
"learning_rate": 4.6961366260189765e-08,
|
3046 |
-
"loss": 0.0911,
|
3047 |
-
"step": 10850
|
3048 |
-
},
|
3049 |
-
{
|
3050 |
-
"epoch": 0.9804363505228993,
|
3051 |
-
"grad_norm": 0.2627655267715454,
|
3052 |
-
"learning_rate": 3.776267720495996e-08,
|
3053 |
-
"loss": 0.0777,
|
3054 |
-
"step": 10875
|
3055 |
-
},
|
3056 |
-
{
|
3057 |
-
"epoch": 0.9826902271907682,
|
3058 |
-
"grad_norm": 0.16840891540050507,
|
3059 |
-
"learning_rate": 2.9564834571509382e-08,
|
3060 |
-
"loss": 0.1089,
|
3061 |
-
"step": 10900
|
3062 |
-
},
|
3063 |
-
{
|
3064 |
-
"epoch": 0.9849441038586368,
|
3065 |
-
"grad_norm": 144.99005126953125,
|
3066 |
-
"learning_rate": 2.23682493749644e-08,
|
3067 |
-
"loss": 0.1439,
|
3068 |
-
"step": 10925
|
3069 |
-
},
|
3070 |
-
{
|
3071 |
-
"epoch": 0.9871979805265055,
|
3072 |
-
"grad_norm": 0.636849045753479,
|
3073 |
-
"learning_rate": 1.617328243041838e-08,
|
3074 |
-
"loss": 0.0694,
|
3075 |
-
"step": 10950
|
3076 |
-
},
|
3077 |
-
{
|
3078 |
-
"epoch": 0.9894518571943743,
|
3079 |
-
"grad_norm": 0.046460844576358795,
|
3080 |
-
"learning_rate": 1.0980244334843903e-08,
|
3081 |
-
"loss": 0.096,
|
3082 |
-
"step": 10975
|
3083 |
-
},
|
3084 |
-
{
|
3085 |
-
"epoch": 0.991705733862243,
|
3086 |
-
"grad_norm": 12.102967262268066,
|
3087 |
-
"learning_rate": 6.789395451520797e-09,
|
3088 |
-
"loss": 0.1351,
|
3089 |
-
"step": 11000
|
3090 |
-
},
|
3091 |
-
{
|
3092 |
-
"epoch": 0.9939596105301118,
|
3093 |
-
"grad_norm": 0.4296446144580841,
|
3094 |
-
"learning_rate": 3.600945896986563e-09,
|
3095 |
-
"loss": 0.1455,
|
3096 |
-
"step": 11025
|
3097 |
-
},
|
3098 |
-
{
|
3099 |
-
"epoch": 0.9962134871979805,
|
3100 |
-
"grad_norm": 0.07985327392816544,
|
3101 |
-
"learning_rate": 1.4150555304914826e-09,
|
3102 |
-
"loss": 0.073,
|
3103 |
-
"step": 11050
|
3104 |
-
},
|
3105 |
-
{
|
3106 |
-
"epoch": 0.9984673638658492,
|
3107 |
-
"grad_norm": 99.93488311767578,
|
3108 |
-
"learning_rate": 2.3183394598724762e-10,
|
3109 |
-
"loss": 0.2491,
|
3110 |
-
"step": 11075
|
3111 |
-
},
|
3112 |
-
{
|
3113 |
-
"epoch": 1.0,
|
3114 |
-
"eval_accuracy": 0.9676605652933187,
|
3115 |
-
"eval_auc": 0.9917595865231005,
|
3116 |
-
"eval_f1": 0.9706055261610818,
|
3117 |
-
"eval_loss": 0.16565780341625214,
|
3118 |
-
"eval_precision": 0.9654970760233919,
|
3119 |
-
"eval_recall": 0.9757683215130024,
|
3120 |
-
"eval_runtime": 2178.8382,
|
3121 |
-
"eval_samples_per_second": 7.096,
|
3122 |
-
"eval_steps_per_second": 0.296,
|
3123 |
-
"step": 11092
|
3124 |
-
}
|
3125 |
-
],
|
3126 |
-
"logging_steps": 25,
|
3127 |
-
"max_steps": 11092,
|
3128 |
-
"num_input_tokens_seen": 0,
|
3129 |
-
"num_train_epochs": 1,
|
3130 |
-
"save_steps": 500,
|
3131 |
-
"stateful_callbacks": {
|
3132 |
-
"EarlyStoppingCallback": {
|
3133 |
-
"args": {
|
3134 |
-
"early_stopping_patience": 5,
|
3135 |
-
"early_stopping_threshold": 0.01
|
3136 |
-
},
|
3137 |
-
"attributes": {
|
3138 |
-
"early_stopping_patience_counter": 0
|
3139 |
-
}
|
3140 |
-
},
|
3141 |
-
"TrainerControl": {
|
3142 |
-
"args": {
|
3143 |
-
"should_epoch_stop": false,
|
3144 |
-
"should_evaluate": false,
|
3145 |
-
"should_log": false,
|
3146 |
-
"should_save": true,
|
3147 |
-
"should_training_stop": true
|
3148 |
-
},
|
3149 |
-
"attributes": {}
|
3150 |
-
}
|
3151 |
-
},
|
3152 |
-
"total_flos": 3.064790733409773e+19,
|
3153 |
-
"train_batch_size": 12,
|
3154 |
-
"trial_name": null,
|
3155 |
-
"trial_params": null
|
3156 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
{checkpoint-11092 β checkpoint-20331}/config.json
RENAMED
File without changes
|
{checkpoint-11092 β checkpoint-20331}/model.safetensors
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 780886280
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df3f551fce6d4b7e8725592fd5a46f06990747e9592e7d82ea904c7881fd5c4f
|
3 |
size 780886280
|
{checkpoint-11092 β checkpoint-20331}/optimizer.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1562042973
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7345e457b17e1d826226eca831ba6234eff16ac4eb4dbd53d6505fee08bfa438
|
3 |
size 1562042973
|
{checkpoint-11092 β checkpoint-20331}/rng_state.pth
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1cefffd54ff03c7fa15e5b942d60b545988afd3dd13ba57423d428189e224208
|
3 |
size 14244
|
{checkpoint-11092 β checkpoint-20331}/scheduler.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ff20343cfef47b8ed69f169b5043c84c27e794c90dcdec71f53e52cb44c52547
|
3 |
size 1064
|
checkpoint-20331/trainer_state.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
{checkpoint-11092 β checkpoint-20331}/training_args.bin
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6fff517acbdded1210f1897a9e7f965d2f8b7505cb4a3980ffcda72566d8efd0
|
3 |
size 5304
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 780886280
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df3f551fce6d4b7e8725592fd5a46f06990747e9592e7d82ea904c7881fd5c4f
|
3 |
size 780886280
|
runs/{Dec09_07-00-37_e11b5028ca4b/events.out.tfevents.1733727645.e11b5028ca4b.82.0 β Dec17_07-54-46_31406a8e7e2f/events.out.tfevents.1734422095.31406a8e7e2f.61.0}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:83b223540e4ac7f4e1f3a224b9f5b0e06ebc3540b1d8e44bd8b144d4e236796b
|
3 |
+
size 178672
|
runs/{Dec09_07-00-37_e11b5028ca4b/events.out.tfevents.1733763045.e11b5028ca4b.82.1 β Dec17_07-54-46_31406a8e7e2f/events.out.tfevents.1734490506.31406a8e7e2f.61.1}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:09f5edf6df6b886c20c6bb8bee518536092c89de0d9a828b170316aa8111fa58
|
3 |
+
size 617
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5304
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6fff517acbdded1210f1897a9e7f965d2f8b7505cb4a3980ffcda72566d8efd0
|
3 |
size 5304
|
training_params.json
CHANGED
@@ -11,11 +11,11 @@
|
|
11 |
"scheduler": "cosine",
|
12 |
"weight_decay": 0.001,
|
13 |
"max_grad_norm": 1.0,
|
14 |
-
"seed":
|
15 |
"train_split": "train",
|
16 |
"valid_split": "test",
|
17 |
"logging_steps": -1,
|
18 |
-
"project_name": "autotrain-ai-image-detect-
|
19 |
"auto_find_batch_size": true,
|
20 |
"mixed_precision": "bf16",
|
21 |
"save_total_limit": 1,
|
|
|
11 |
"scheduler": "cosine",
|
12 |
"weight_decay": 0.001,
|
13 |
"max_grad_norm": 1.0,
|
14 |
+
"seed": 3216,
|
15 |
"train_split": "train",
|
16 |
"valid_split": "test",
|
17 |
"logging_steps": -1,
|
18 |
+
"project_name": "autotrain-ai-image-detect-20241217-0215",
|
19 |
"auto_find_batch_size": true,
|
20 |
"mixed_precision": "bf16",
|
21 |
"save_total_limit": 1,
|