update project
Browse files- trainer_state.json +0 -322
trainer_state.json
DELETED
@@ -1,322 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": null,
|
3 |
-
"best_model_checkpoint": null,
|
4 |
-
"epoch": 1.0,
|
5 |
-
"global_step": 5148,
|
6 |
-
"is_hyper_param_search": false,
|
7 |
-
"is_local_process_zero": true,
|
8 |
-
"is_world_process_zero": true,
|
9 |
-
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.02,
|
12 |
-
"learning_rate": 9.955322455322455e-05,
|
13 |
-
"loss": 4.3405,
|
14 |
-
"step": 100
|
15 |
-
},
|
16 |
-
{
|
17 |
-
"epoch": 0.04,
|
18 |
-
"learning_rate": 9.906759906759907e-05,
|
19 |
-
"loss": 3.3274,
|
20 |
-
"step": 200
|
21 |
-
},
|
22 |
-
{
|
23 |
-
"epoch": 0.06,
|
24 |
-
"learning_rate": 9.858197358197358e-05,
|
25 |
-
"loss": 3.3008,
|
26 |
-
"step": 300
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 0.08,
|
30 |
-
"learning_rate": 9.80963480963481e-05,
|
31 |
-
"loss": 3.2864,
|
32 |
-
"step": 400
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 0.1,
|
36 |
-
"learning_rate": 9.761072261072261e-05,
|
37 |
-
"loss": 3.2596,
|
38 |
-
"step": 500
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 0.12,
|
42 |
-
"learning_rate": 9.712509712509713e-05,
|
43 |
-
"loss": 3.2282,
|
44 |
-
"step": 600
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.14,
|
48 |
-
"learning_rate": 9.663947163947164e-05,
|
49 |
-
"loss": 3.2515,
|
50 |
-
"step": 700
|
51 |
-
},
|
52 |
-
{
|
53 |
-
"epoch": 0.16,
|
54 |
-
"learning_rate": 9.615384615384617e-05,
|
55 |
-
"loss": 3.24,
|
56 |
-
"step": 800
|
57 |
-
},
|
58 |
-
{
|
59 |
-
"epoch": 0.17,
|
60 |
-
"learning_rate": 9.566822066822068e-05,
|
61 |
-
"loss": 3.2679,
|
62 |
-
"step": 900
|
63 |
-
},
|
64 |
-
{
|
65 |
-
"epoch": 0.19,
|
66 |
-
"learning_rate": 9.51825951825952e-05,
|
67 |
-
"loss": 3.2357,
|
68 |
-
"step": 1000
|
69 |
-
},
|
70 |
-
{
|
71 |
-
"epoch": 0.21,
|
72 |
-
"learning_rate": 9.469696969696971e-05,
|
73 |
-
"loss": 3.2323,
|
74 |
-
"step": 1100
|
75 |
-
},
|
76 |
-
{
|
77 |
-
"epoch": 0.23,
|
78 |
-
"learning_rate": 9.421134421134422e-05,
|
79 |
-
"loss": 3.2238,
|
80 |
-
"step": 1200
|
81 |
-
},
|
82 |
-
{
|
83 |
-
"epoch": 0.25,
|
84 |
-
"learning_rate": 9.372571872571873e-05,
|
85 |
-
"loss": 3.1898,
|
86 |
-
"step": 1300
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.27,
|
90 |
-
"learning_rate": 9.324009324009324e-05,
|
91 |
-
"loss": 3.2354,
|
92 |
-
"step": 1400
|
93 |
-
},
|
94 |
-
{
|
95 |
-
"epoch": 0.29,
|
96 |
-
"learning_rate": 9.275446775446775e-05,
|
97 |
-
"loss": 3.1935,
|
98 |
-
"step": 1500
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 0.31,
|
102 |
-
"learning_rate": 9.226884226884227e-05,
|
103 |
-
"loss": 3.2049,
|
104 |
-
"step": 1600
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"epoch": 0.33,
|
108 |
-
"learning_rate": 9.178321678321678e-05,
|
109 |
-
"loss": 3.1813,
|
110 |
-
"step": 1700
|
111 |
-
},
|
112 |
-
{
|
113 |
-
"epoch": 0.35,
|
114 |
-
"learning_rate": 9.12975912975913e-05,
|
115 |
-
"loss": 3.1799,
|
116 |
-
"step": 1800
|
117 |
-
},
|
118 |
-
{
|
119 |
-
"epoch": 0.37,
|
120 |
-
"learning_rate": 9.081196581196581e-05,
|
121 |
-
"loss": 3.1848,
|
122 |
-
"step": 1900
|
123 |
-
},
|
124 |
-
{
|
125 |
-
"epoch": 0.39,
|
126 |
-
"learning_rate": 9.032634032634033e-05,
|
127 |
-
"loss": 3.2137,
|
128 |
-
"step": 2000
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 0.41,
|
132 |
-
"learning_rate": 8.984071484071484e-05,
|
133 |
-
"loss": 3.186,
|
134 |
-
"step": 2100
|
135 |
-
},
|
136 |
-
{
|
137 |
-
"epoch": 0.43,
|
138 |
-
"learning_rate": 8.935508935508936e-05,
|
139 |
-
"loss": 3.1948,
|
140 |
-
"step": 2200
|
141 |
-
},
|
142 |
-
{
|
143 |
-
"epoch": 0.45,
|
144 |
-
"learning_rate": 8.886946386946387e-05,
|
145 |
-
"loss": 3.1564,
|
146 |
-
"step": 2300
|
147 |
-
},
|
148 |
-
{
|
149 |
-
"epoch": 0.47,
|
150 |
-
"learning_rate": 8.83838383838384e-05,
|
151 |
-
"loss": 3.173,
|
152 |
-
"step": 2400
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 0.49,
|
156 |
-
"learning_rate": 8.789821289821291e-05,
|
157 |
-
"loss": 3.1713,
|
158 |
-
"step": 2500
|
159 |
-
},
|
160 |
-
{
|
161 |
-
"epoch": 0.51,
|
162 |
-
"learning_rate": 8.741258741258743e-05,
|
163 |
-
"loss": 3.1565,
|
164 |
-
"step": 2600
|
165 |
-
},
|
166 |
-
{
|
167 |
-
"epoch": 0.52,
|
168 |
-
"learning_rate": 8.692696192696194e-05,
|
169 |
-
"loss": 3.19,
|
170 |
-
"step": 2700
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.54,
|
174 |
-
"learning_rate": 8.644133644133644e-05,
|
175 |
-
"loss": 3.1557,
|
176 |
-
"step": 2800
|
177 |
-
},
|
178 |
-
{
|
179 |
-
"epoch": 0.56,
|
180 |
-
"learning_rate": 8.595571095571096e-05,
|
181 |
-
"loss": 3.1731,
|
182 |
-
"step": 2900
|
183 |
-
},
|
184 |
-
{
|
185 |
-
"epoch": 0.58,
|
186 |
-
"learning_rate": 8.547008547008547e-05,
|
187 |
-
"loss": 3.1801,
|
188 |
-
"step": 3000
|
189 |
-
},
|
190 |
-
{
|
191 |
-
"epoch": 0.6,
|
192 |
-
"learning_rate": 8.498445998445998e-05,
|
193 |
-
"loss": 3.1607,
|
194 |
-
"step": 3100
|
195 |
-
},
|
196 |
-
{
|
197 |
-
"epoch": 0.62,
|
198 |
-
"learning_rate": 8.44988344988345e-05,
|
199 |
-
"loss": 3.1505,
|
200 |
-
"step": 3200
|
201 |
-
},
|
202 |
-
{
|
203 |
-
"epoch": 0.64,
|
204 |
-
"learning_rate": 8.401320901320901e-05,
|
205 |
-
"loss": 3.163,
|
206 |
-
"step": 3300
|
207 |
-
},
|
208 |
-
{
|
209 |
-
"epoch": 0.66,
|
210 |
-
"learning_rate": 8.352758352758353e-05,
|
211 |
-
"loss": 3.1526,
|
212 |
-
"step": 3400
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 0.68,
|
216 |
-
"learning_rate": 8.304195804195804e-05,
|
217 |
-
"loss": 3.1508,
|
218 |
-
"step": 3500
|
219 |
-
},
|
220 |
-
{
|
221 |
-
"epoch": 0.7,
|
222 |
-
"learning_rate": 8.255633255633256e-05,
|
223 |
-
"loss": 3.1673,
|
224 |
-
"step": 3600
|
225 |
-
},
|
226 |
-
{
|
227 |
-
"epoch": 0.72,
|
228 |
-
"learning_rate": 8.207070707070707e-05,
|
229 |
-
"loss": 3.1417,
|
230 |
-
"step": 3700
|
231 |
-
},
|
232 |
-
{
|
233 |
-
"epoch": 0.74,
|
234 |
-
"learning_rate": 8.158508158508159e-05,
|
235 |
-
"loss": 3.1682,
|
236 |
-
"step": 3800
|
237 |
-
},
|
238 |
-
{
|
239 |
-
"epoch": 0.76,
|
240 |
-
"learning_rate": 8.10994560994561e-05,
|
241 |
-
"loss": 3.1461,
|
242 |
-
"step": 3900
|
243 |
-
},
|
244 |
-
{
|
245 |
-
"epoch": 0.78,
|
246 |
-
"learning_rate": 8.061383061383063e-05,
|
247 |
-
"loss": 3.1476,
|
248 |
-
"step": 4000
|
249 |
-
},
|
250 |
-
{
|
251 |
-
"epoch": 0.8,
|
252 |
-
"learning_rate": 8.012820512820514e-05,
|
253 |
-
"loss": 3.1342,
|
254 |
-
"step": 4100
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 0.82,
|
258 |
-
"learning_rate": 7.964257964257966e-05,
|
259 |
-
"loss": 3.1473,
|
260 |
-
"step": 4200
|
261 |
-
},
|
262 |
-
{
|
263 |
-
"epoch": 0.84,
|
264 |
-
"learning_rate": 7.915695415695417e-05,
|
265 |
-
"loss": 3.1348,
|
266 |
-
"step": 4300
|
267 |
-
},
|
268 |
-
{
|
269 |
-
"epoch": 0.85,
|
270 |
-
"learning_rate": 7.867132867132867e-05,
|
271 |
-
"loss": 3.1287,
|
272 |
-
"step": 4400
|
273 |
-
},
|
274 |
-
{
|
275 |
-
"epoch": 0.87,
|
276 |
-
"learning_rate": 7.818570318570319e-05,
|
277 |
-
"loss": 3.1278,
|
278 |
-
"step": 4500
|
279 |
-
},
|
280 |
-
{
|
281 |
-
"epoch": 0.89,
|
282 |
-
"learning_rate": 7.77000777000777e-05,
|
283 |
-
"loss": 3.1179,
|
284 |
-
"step": 4600
|
285 |
-
},
|
286 |
-
{
|
287 |
-
"epoch": 0.91,
|
288 |
-
"learning_rate": 7.721445221445221e-05,
|
289 |
-
"loss": 3.1255,
|
290 |
-
"step": 4700
|
291 |
-
},
|
292 |
-
{
|
293 |
-
"epoch": 0.93,
|
294 |
-
"learning_rate": 7.672882672882673e-05,
|
295 |
-
"loss": 3.121,
|
296 |
-
"step": 4800
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 0.95,
|
300 |
-
"learning_rate": 7.624320124320124e-05,
|
301 |
-
"loss": 3.1312,
|
302 |
-
"step": 4900
|
303 |
-
},
|
304 |
-
{
|
305 |
-
"epoch": 0.97,
|
306 |
-
"learning_rate": 7.575757575757576e-05,
|
307 |
-
"loss": 3.0984,
|
308 |
-
"step": 5000
|
309 |
-
},
|
310 |
-
{
|
311 |
-
"epoch": 0.99,
|
312 |
-
"learning_rate": 7.528166278166279e-05,
|
313 |
-
"loss": 3.1285,
|
314 |
-
"step": 5100
|
315 |
-
}
|
316 |
-
],
|
317 |
-
"max_steps": 20592,
|
318 |
-
"num_train_epochs": 4,
|
319 |
-
"total_flos": 2.8530438570084663e+19,
|
320 |
-
"trial_name": null,
|
321 |
-
"trial_params": null
|
322 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|