eac123 commited on
Commit
ab166f2
·
verified ·
1 Parent(s): c5d5af6

Upload results/rec1m4-seed1_history.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. results/rec1m4-seed1_history.json +2238 -0
results/rec1m4-seed1_history.json ADDED
@@ -0,0 +1,2238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "loss": [
3
+ {
4
+ "step": 0,
5
+ "loss": 4.3096,
6
+ "lr": 1e-06,
7
+ "iters": 8,
8
+ "steps_per_s": 0.0,
9
+ "flops_cum": 173983924224.0
10
+ },
11
+ {
12
+ "step": 100,
13
+ "loss": 4.2214,
14
+ "lr": 0.000101,
15
+ "iters": 5,
16
+ "steps_per_s": 9.71,
17
+ "flops_cum": 19044180688896.0
18
+ },
19
+ {
20
+ "step": 200,
21
+ "loss": 4.1688,
22
+ "lr": 0.000201,
23
+ "iters": 10,
24
+ "steps_per_s": 9.6,
25
+ "flops_cum": 38065886724096.0
26
+ },
27
+ {
28
+ "step": 300,
29
+ "loss": 4.1611,
30
+ "lr": 0.000301,
31
+ "iters": 9,
32
+ "steps_per_s": 9.6,
33
+ "flops_cum": 57001016033280.0
34
+ },
35
+ {
36
+ "step": 400,
37
+ "loss": 4.1632,
38
+ "lr": 0.00040100000000000004,
39
+ "iters": 10,
40
+ "steps_per_s": 9.62,
41
+ "flops_cum": 75914501160960.0
42
+ },
43
+ {
44
+ "step": 500,
45
+ "loss": 4.1693,
46
+ "lr": 0.000501,
47
+ "iters": 7,
48
+ "steps_per_s": 10.11,
49
+ "flops_cum": 93810709757952.0
50
+ },
51
+ {
52
+ "step": 600,
53
+ "loss": 4.1607,
54
+ "lr": 0.000601,
55
+ "iters": 9,
56
+ "steps_per_s": 10.2,
57
+ "flops_cum": 111641985810432.0
58
+ },
59
+ {
60
+ "step": 700,
61
+ "loss": 4.1544,
62
+ "lr": 0.000701,
63
+ "iters": 8,
64
+ "steps_per_s": 10.03,
65
+ "flops_cum": 129689703677952.0
66
+ },
67
+ {
68
+ "step": 800,
69
+ "loss": 4.1381,
70
+ "lr": 0.0008010000000000001,
71
+ "iters": 12,
72
+ "steps_per_s": 9.87,
73
+ "flops_cum": 148083728449536.0
74
+ },
75
+ {
76
+ "step": 900,
77
+ "loss": 4.137,
78
+ "lr": 0.000901,
79
+ "iters": 9,
80
+ "steps_per_s": 9.95,
81
+ "flops_cum": 166261311406080.0
82
+ },
83
+ {
84
+ "step": 1000,
85
+ "loss": 4.1346,
86
+ "lr": 0.001,
87
+ "iters": 11,
88
+ "steps_per_s": 9.65,
89
+ "flops_cum": 185109863989248.0
90
+ },
91
+ {
92
+ "step": 1100,
93
+ "loss": 4.1328,
94
+ "lr": 0.0009999571637870036,
95
+ "iters": 10,
96
+ "steps_per_s": 9.82,
97
+ "flops_cum": 203568821305344.0
98
+ },
99
+ {
100
+ "step": 1200,
101
+ "loss": 4.1113,
102
+ "lr": 0.0009998286624877785,
103
+ "iters": 12,
104
+ "steps_per_s": 10.1,
105
+ "flops_cum": 221551606628352.0
106
+ },
107
+ {
108
+ "step": 1300,
109
+ "loss": 4.1324,
110
+ "lr": 0.0009996145181203615,
111
+ "iters": 7,
112
+ "steps_per_s": 9.75,
113
+ "flops_cum": 240032208125952.0
114
+ },
115
+ {
116
+ "step": 1400,
117
+ "loss": 4.1551,
118
+ "lr": 0.0009993147673772868,
119
+ "iters": 12,
120
+ "steps_per_s": 9.14,
121
+ "flops_cum": 259962969784320.0
122
+ },
123
+ {
124
+ "step": 1500,
125
+ "loss": 4.0853,
126
+ "lr": 0.0009989294616193018,
127
+ "iters": 9,
128
+ "steps_per_s": 9.9,
129
+ "flops_cum": 278205485285376.0
130
+ },
131
+ {
132
+ "step": 1600,
133
+ "loss": 4.1232,
134
+ "lr": 0.000998458666866564,
135
+ "iters": 6,
136
+ "steps_per_s": 9.79,
137
+ "flops_cum": 296664442601472.0
138
+ },
139
+ {
140
+ "step": 1700,
141
+ "loss": 4.1169,
142
+ "lr": 0.0009979024637873308,
143
+ "iters": 9,
144
+ "steps_per_s": 9.7,
145
+ "flops_cum": 315404774277120.0
146
+ },
147
+ {
148
+ "step": 1800,
149
+ "loss": 4.0996,
150
+ "lr": 0.0009972609476841367,
151
+ "iters": 8,
152
+ "steps_per_s": 9.75,
153
+ "flops_cum": 333971952500736.0
154
+ },
155
+ {
156
+ "step": 1900,
157
+ "loss": 4.0982,
158
+ "lr": 0.0009965342284774632,
159
+ "iters": 6,
160
+ "steps_per_s": 9.75,
161
+ "flops_cum": 352539130724352.0
162
+ },
163
+ {
164
+ "step": 2000,
165
+ "loss": 4.0751,
166
+ "lr": 0.0009957224306869053,
167
+ "iters": 9,
168
+ "steps_per_s": 9.78,
169
+ "flops_cum": 371127953129472.0
170
+ },
171
+ {
172
+ "step": 2100,
173
+ "loss": 4.1442,
174
+ "lr": 0.0009948256934098352,
175
+ "iters": 8,
176
+ "steps_per_s": 9.8,
177
+ "flops_cum": 389608554627072.0
178
+ },
179
+ {
180
+ "step": 2200,
181
+ "loss": 4.1066,
182
+ "lr": 0.0009938441702975688,
183
+ "iters": 5,
184
+ "steps_per_s": 9.88,
185
+ "flops_cum": 407894358491136.0
186
+ },
187
+ {
188
+ "step": 2300,
189
+ "loss": 4.0979,
190
+ "lr": 0.0009927780295290389,
191
+ "iters": 7,
192
+ "steps_per_s": 9.95,
193
+ "flops_cum": 426136873992192.0
194
+ },
195
+ {
196
+ "step": 2400,
197
+ "loss": 4.0698,
198
+ "lr": 0.0009916274537819774,
199
+ "iters": 7,
200
+ "steps_per_s": 9.45,
201
+ "flops_cum": 445353377660928.0
202
+ },
203
+ {
204
+ "step": 2500,
205
+ "loss": 4.1077,
206
+ "lr": 0.0009903926402016153,
207
+ "iters": 10,
208
+ "steps_per_s": 8.48,
209
+ "flops_cum": 464396727877632.0
210
+ },
211
+ {
212
+ "step": 2600,
213
+ "loss": 4.0873,
214
+ "lr": 0.0009890738003669028,
215
+ "iters": 8,
216
+ "steps_per_s": 9.39,
217
+ "flops_cum": 483613231546368.0
218
+ },
219
+ {
220
+ "step": 2700,
221
+ "loss": 4.0645,
222
+ "lr": 0.0009876711602542563,
223
+ "iters": 7,
224
+ "steps_per_s": 9.47,
225
+ "flops_cum": 502699870126080.0
226
+ },
227
+ {
228
+ "step": 2800,
229
+ "loss": 4.0396,
230
+ "lr": 0.0009861849601988384,
231
+ "iters": 9,
232
+ "steps_per_s": 9.9,
233
+ "flops_cum": 521028962353152.0
234
+ },
235
+ {
236
+ "step": 2900,
237
+ "loss": 4.1011,
238
+ "lr": 0.0009846154548533773,
239
+ "iters": 10,
240
+ "steps_per_s": 9.2,
241
+ "flops_cum": 540743282196480.0
242
+ },
243
+ {
244
+ "step": 3000,
245
+ "loss": 4.1045,
246
+ "lr": 0.0009829629131445341,
247
+ "iters": 10,
248
+ "steps_per_s": 9.87,
249
+ "flops_cum": 559094018605056.0
250
+ },
251
+ {
252
+ "step": 3100,
253
+ "loss": 4.0893,
254
+ "lr": 0.0009812276182268236,
255
+ "iters": 8,
256
+ "steps_per_s": 9.45,
257
+ "flops_cum": 578245589729280.0
258
+ },
259
+ {
260
+ "step": 3200,
261
+ "loss": 4.0718,
262
+ "lr": 0.0009794098674340967,
263
+ "iters": 12,
264
+ "steps_per_s": 10.0,
265
+ "flops_cum": 596401528504320.0
266
+ },
267
+ {
268
+ "step": 3300,
269
+ "loss": 4.0982,
270
+ "lr": 0.0009775099722285933,
271
+ "iters": 12,
272
+ "steps_per_s": 9.69,
273
+ "flops_cum": 615098571816960.0
274
+ },
275
+ {
276
+ "step": 3400,
277
+ "loss": 4.0642,
278
+ "lr": 0.0009755282581475768,
279
+ "iters": 9,
280
+ "steps_per_s": 9.99,
281
+ "flops_cum": 633167933865984.0
282
+ },
283
+ {
284
+ "step": 3500,
285
+ "loss": 4.0398,
286
+ "lr": 0.0009734650647475529,
287
+ "iters": 11,
288
+ "steps_per_s": 9.8,
289
+ "flops_cum": 651713467908096.0
290
+ },
291
+ {
292
+ "step": 3600,
293
+ "loss": 4.0802,
294
+ "lr": 0.0009713207455460893,
295
+ "iters": 11,
296
+ "steps_per_s": 10.02,
297
+ "flops_cum": 669782829957120.0
298
+ },
299
+ {
300
+ "step": 3700,
301
+ "loss": 4.0767,
302
+ "lr": 0.0009690956679612422,
303
+ "iters": 6,
304
+ "steps_per_s": 10.31,
305
+ "flops_cum": 687246154924032.0
306
+ },
307
+ {
308
+ "step": 3800,
309
+ "loss": 4.0734,
310
+ "lr": 0.0009667902132486009,
311
+ "iters": 5,
312
+ "steps_per_s": 9.64,
313
+ "flops_cum": 706008130781184.0
314
+ },
315
+ {
316
+ "step": 3900,
317
+ "loss": 4.0607,
318
+ "lr": 0.0009644047764359622,
319
+ "iters": 12,
320
+ "steps_per_s": 9.9,
321
+ "flops_cum": 724423799734272.0
322
+ },
323
+ {
324
+ "step": 4000,
325
+ "loss": 4.0633,
326
+ "lr": 0.0009619397662556434,
327
+ "iters": 7,
328
+ "steps_per_s": 9.4,
329
+ "flops_cum": 743618659221504.0
330
+ },
331
+ {
332
+ "step": 4100,
333
+ "loss": 4.0773,
334
+ "lr": 0.0009593956050744492,
335
+ "iters": 11,
336
+ "steps_per_s": 10.05,
337
+ "flops_cum": 761688021270528.0
338
+ },
339
+ {
340
+ "step": 4200,
341
+ "loss": 4.0238,
342
+ "lr": 0.0009567727288213005,
343
+ "iters": 8,
344
+ "steps_per_s": 9.29,
345
+ "flops_cum": 781077678391296.0
346
+ },
347
+ {
348
+ "step": 4300,
349
+ "loss": 4.0107,
350
+ "lr": 0.0009540715869125407,
351
+ "iters": 5,
352
+ "steps_per_s": 9.8,
353
+ "flops_cum": 799536635707392.0
354
+ },
355
+ {
356
+ "step": 4400,
357
+ "loss": 4.0983,
358
+ "lr": 0.0009512926421749304,
359
+ "iters": 8,
360
+ "steps_per_s": 10.04,
361
+ "flops_cum": 817627641937920.0
362
+ },
363
+ {
364
+ "step": 4500,
365
+ "loss": 4.0443,
366
+ "lr": 0.0009484363707663442,
367
+ "iters": 8,
368
+ "steps_per_s": 9.33,
369
+ "flops_cum": 837038943240192.0
370
+ },
371
+ {
372
+ "step": 4600,
373
+ "loss": 3.9835,
374
+ "lr": 0.0009455032620941839,
375
+ "iters": 6,
376
+ "steps_per_s": 9.55,
377
+ "flops_cum": 855930784186368.0
378
+ },
379
+ {
380
+ "step": 4700,
381
+ "loss": 4.0161,
382
+ "lr": 0.0009424938187315209,
383
+ "iters": 10,
384
+ "steps_per_s": 9.62,
385
+ "flops_cum": 874822625132544.0
386
+ },
387
+ {
388
+ "step": 4800,
389
+ "loss": 4.0353,
390
+ "lr": 0.0009394085563309827,
391
+ "iters": 5,
392
+ "steps_per_s": 9.61,
393
+ "flops_cum": 893584600989696.0
394
+ },
395
+ {
396
+ "step": 4900,
397
+ "loss": 4.0178,
398
+ "lr": 0.0009362480035363986,
399
+ "iters": 6,
400
+ "steps_per_s": 9.73,
401
+ "flops_cum": 912216711757824.0
402
+ },
403
+ {
404
+ "step": 5000,
405
+ "loss": 4.0757,
406
+ "lr": 0.0009330127018922195,
407
+ "iters": 8,
408
+ "steps_per_s": 8.71,
409
+ "flops_cum": 930697313255424.0
410
+ },
411
+ {
412
+ "step": 5100,
413
+ "loss": 4.1215,
414
+ "lr": 0.0009297032057507264,
415
+ "iters": 6,
416
+ "steps_per_s": 9.66,
417
+ "flops_cum": 949416000749568.0
418
+ },
419
+ {
420
+ "step": 5200,
421
+ "loss": 4.041,
422
+ "lr": 0.0009263200821770461,
423
+ "iters": 10,
424
+ "steps_per_s": 9.7,
425
+ "flops_cum": 968177976606720.0
426
+ },
427
+ {
428
+ "step": 5300,
429
+ "loss": 4.0051,
430
+ "lr": 0.0009228639108519867,
431
+ "iters": 8,
432
+ "steps_per_s": 10.37,
433
+ "flops_cum": 985619657392128.0
434
+ },
435
+ {
436
+ "step": 5400,
437
+ "loss": 4.0267,
438
+ "lr": 0.0009193352839727121,
439
+ "iters": 11,
440
+ "steps_per_s": 10.23,
441
+ "flops_cum": 1003364356718592.0
442
+ },
443
+ {
444
+ "step": 5500,
445
+ "loss": 3.9641,
446
+ "lr": 0.0009157348061512727,
447
+ "iters": 11,
448
+ "steps_per_s": 9.82,
449
+ "flops_cum": 1021823314034688.0
450
+ },
451
+ {
452
+ "step": 5600,
453
+ "loss": 3.9565,
454
+ "lr": 0.0009120630943110077,
455
+ "iters": 10,
456
+ "steps_per_s": 9.83,
457
+ "flops_cum": 1040238982987776.0
458
+ },
459
+ {
460
+ "step": 5700,
461
+ "loss": 3.9953,
462
+ "lr": 0.0009083207775808396,
463
+ "iters": 8,
464
+ "steps_per_s": 9.44,
465
+ "flops_cum": 1059455486656512.0
466
+ },
467
+ {
468
+ "step": 5800,
469
+ "loss": 3.9517,
470
+ "lr": 0.0009045084971874737,
471
+ "iters": 10,
472
+ "steps_per_s": 9.8,
473
+ "flops_cum": 1078001020698624.0
474
+ },
475
+ {
476
+ "step": 5900,
477
+ "loss": 4.0184,
478
+ "lr": 0.0009006269063455304,
479
+ "iters": 10,
480
+ "steps_per_s": 9.96,
481
+ "flops_cum": 1096221892018176.0
482
+ },
483
+ {
484
+ "step": 6000,
485
+ "loss": 3.9712,
486
+ "lr": 0.0008966766701456176,
487
+ "iters": 5,
488
+ "steps_per_s": 10.08,
489
+ "flops_cum": 1114096456433664.0
490
+ },
491
+ {
492
+ "step": 6100,
493
+ "loss": 4.0342,
494
+ "lr": 0.0008926584654403724,
495
+ "iters": 10,
496
+ "steps_per_s": 9.82,
497
+ "flops_cum": 1132620346294272.0
498
+ },
499
+ {
500
+ "step": 6200,
501
+ "loss": 4.0009,
502
+ "lr": 0.0008885729807284854,
503
+ "iters": 6,
504
+ "steps_per_s": 9.54,
505
+ "flops_cum": 1151555475603456.0
506
+ },
507
+ {
508
+ "step": 6300,
509
+ "loss": 3.9161,
510
+ "lr": 0.0008844209160367298,
511
+ "iters": 11,
512
+ "steps_per_s": 9.78,
513
+ "flops_cum": 1170165942190080.0
514
+ },
515
+ {
516
+ "step": 6400,
517
+ "loss": 3.9813,
518
+ "lr": 0.0008802029828000156,
519
+ "iters": 11,
520
+ "steps_per_s": 10.31,
521
+ "flops_cum": 1187737488064512.0
522
+ },
523
+ {
524
+ "step": 6500,
525
+ "loss": 3.9394,
526
+ "lr": 0.0008759199037394887,
527
+ "iters": 8,
528
+ "steps_per_s": 9.94,
529
+ "flops_cum": 1205936715202560.0
530
+ },
531
+ {
532
+ "step": 6600,
533
+ "loss": 3.9852,
534
+ "lr": 0.0008715724127386971,
535
+ "iters": 11,
536
+ "steps_per_s": 10.05,
537
+ "flops_cum": 1224006077251584.0
538
+ },
539
+ {
540
+ "step": 6700,
541
+ "loss": 3.9975,
542
+ "lr": 0.0008671612547178428,
543
+ "iters": 11,
544
+ "steps_per_s": 9.94,
545
+ "flops_cum": 1242248592752640.0
546
+ },
547
+ {
548
+ "step": 6800,
549
+ "loss": 3.9671,
550
+ "lr": 0.0008626871855061438,
551
+ "iters": 5,
552
+ "steps_per_s": 9.48,
553
+ "flops_cum": 1261248654606336.0
554
+ },
555
+ {
556
+ "step": 6900,
557
+ "loss": 3.9461,
558
+ "lr": 0.0008581509717123273,
559
+ "iters": 9,
560
+ "steps_per_s": 9.7,
561
+ "flops_cum": 1279967342100480.0
562
+ },
563
+ {
564
+ "step": 7000,
565
+ "loss": 3.9293,
566
+ "lr": 0.0008535533905932737,
567
+ "iters": 9,
568
+ "steps_per_s": 10.42,
569
+ "flops_cum": 1297365734522880.0
570
+ },
571
+ {
572
+ "step": 7100,
573
+ "loss": 3.9151,
574
+ "lr": 0.0008488952299208401,
575
+ "iters": 10,
576
+ "steps_per_s": 9.6,
577
+ "flops_cum": 1316257575469056.0
578
+ },
579
+ {
580
+ "step": 7200,
581
+ "loss": 3.9181,
582
+ "lr": 0.000844177287846877,
583
+ "iters": 8,
584
+ "steps_per_s": 9.78,
585
+ "flops_cum": 1334738176966656.0
586
+ },
587
+ {
588
+ "step": 7300,
589
+ "loss": 3.9867,
590
+ "lr": 0.0008394003727664709,
591
+ "iters": 5,
592
+ "steps_per_s": 10.0,
593
+ "flops_cum": 1352785894834176.0
594
+ },
595
+ {
596
+ "step": 7400,
597
+ "loss": 3.8648,
598
+ "lr": 0.0008345653031794292,
599
+ "iters": 8,
600
+ "steps_per_s": 10.01,
601
+ "flops_cum": 1370920189427712.0
602
+ },
603
+ {
604
+ "step": 7500,
605
+ "loss": 3.863,
606
+ "lr": 0.0008296729075500344,
607
+ "iters": 10,
608
+ "steps_per_s": 8.7,
609
+ "flops_cum": 1389444079288320.0
610
+ },
611
+ {
612
+ "step": 7600,
613
+ "loss": 3.8088,
614
+ "lr": 0.0008247240241650918,
615
+ "iters": 12,
616
+ "steps_per_s": 9.47,
617
+ "flops_cum": 1408574006231040.0
618
+ },
619
+ {
620
+ "step": 7700,
621
+ "loss": 3.8433,
622
+ "lr": 0.0008197195009902923,
623
+ "iters": 11,
624
+ "steps_per_s": 9.6,
625
+ "flops_cum": 1427444202995712.0
626
+ },
627
+ {
628
+ "step": 7800,
629
+ "loss": 3.7611,
630
+ "lr": 0.0008146601955249188,
631
+ "iters": 5,
632
+ "steps_per_s": 10.2,
633
+ "flops_cum": 1445102325596160.0
634
+ },
635
+ {
636
+ "step": 7900,
637
+ "loss": 3.8347,
638
+ "lr": 0.0008095469746549171,
639
+ "iters": 7,
640
+ "steps_per_s": 10.09,
641
+ "flops_cum": 1463085110919168.0
642
+ },
643
+ {
644
+ "step": 8000,
645
+ "loss": 3.7817,
646
+ "lr": 0.0008043807145043603,
647
+ "iters": 5,
648
+ "steps_per_s": 9.88,
649
+ "flops_cum": 1481370914783232.0
650
+ },
651
+ {
652
+ "step": 8100,
653
+ "loss": 3.7839,
654
+ "lr": 0.0007991623002853296,
655
+ "iters": 7,
656
+ "steps_per_s": 9.63,
657
+ "flops_cum": 1500176179003392.0
658
+ },
659
+ {
660
+ "step": 8200,
661
+ "loss": 3.5984,
662
+ "lr": 0.0007938926261462366,
663
+ "iters": 8,
664
+ "steps_per_s": 9.87,
665
+ "flops_cum": 1518548559593472.0
666
+ },
667
+ {
668
+ "step": 8300,
669
+ "loss": 3.6649,
670
+ "lr": 0.0007885725950186169,
671
+ "iters": 7,
672
+ "steps_per_s": 10.08,
673
+ "flops_cum": 1536488056553472.0
674
+ },
675
+ {
676
+ "step": 8400,
677
+ "loss": 3.7404,
678
+ "lr": 0.0007832031184624164,
679
+ "iters": 5,
680
+ "steps_per_s": 10.09,
681
+ "flops_cum": 1554405909331968.0
682
+ },
683
+ {
684
+ "step": 8500,
685
+ "loss": 3.5434,
686
+ "lr": 0.0007777851165098011,
687
+ "iters": 11,
688
+ "steps_per_s": 10.09,
689
+ "flops_cum": 1572475271380992.0
690
+ },
691
+ {
692
+ "step": 8600,
693
+ "loss": 3.5683,
694
+ "lr": 0.0007723195175075137,
695
+ "iters": 8,
696
+ "steps_per_s": 9.82,
697
+ "flops_cum": 1590869296152576.0
698
+ },
699
+ {
700
+ "step": 8700,
701
+ "loss": 3.5158,
702
+ "lr": 0.0007668072579578058,
703
+ "iters": 12,
704
+ "steps_per_s": 9.85,
705
+ "flops_cum": 1609284965105664.0
706
+ },
707
+ {
708
+ "step": 8800,
709
+ "loss": 3.3258,
710
+ "lr": 0.0007612492823579744,
711
+ "iters": 7,
712
+ "steps_per_s": 9.81,
713
+ "flops_cum": 1627635701514240.0
714
+ },
715
+ {
716
+ "step": 8900,
717
+ "loss": 3.0295,
718
+ "lr": 0.0007556465430385259,
719
+ "iters": 9,
720
+ "steps_per_s": 9.72,
721
+ "flops_cum": 1646289456463872.0
722
+ },
723
+ {
724
+ "step": 9000,
725
+ "loss": 3.0879,
726
+ "lr": 0.00075,
727
+ "iters": 7,
728
+ "steps_per_s": 9.71,
729
+ "flops_cum": 1664899923050496.0
730
+ },
731
+ {
732
+ "step": 9100,
733
+ "loss": 2.7951,
734
+ "lr": 0.0007443106207484776,
735
+ "iters": 11,
736
+ "steps_per_s": 9.17,
737
+ "flops_cum": 1684700819619840.0
738
+ },
739
+ {
740
+ "step": 9200,
741
+ "loss": 2.9678,
742
+ "lr": 0.0007385793801298042,
743
+ "iters": 7,
744
+ "steps_per_s": 10.28,
745
+ "flops_cum": 1702250721312768.0
746
+ },
747
+ {
748
+ "step": 9300,
749
+ "loss": 3.0122,
750
+ "lr": 0.0007328072601625557,
751
+ "iters": 6,
752
+ "steps_per_s": 9.93,
753
+ "flops_cum": 1720449948450816.0
754
+ },
755
+ {
756
+ "step": 9400,
757
+ "loss": 2.6483,
758
+ "lr": 0.0007269952498697733,
759
+ "iters": 11,
760
+ "steps_per_s": 9.73,
761
+ "flops_cum": 1739125347581952.0
762
+ },
763
+ {
764
+ "step": 9500,
765
+ "loss": 2.6587,
766
+ "lr": 0.0007211443451095007,
767
+ "iters": 12,
768
+ "steps_per_s": 9.67,
769
+ "flops_cum": 1757887323439104.0
770
+ },
771
+ {
772
+ "step": 9600,
773
+ "loss": 2.6181,
774
+ "lr": 0.0007152555484041476,
775
+ "iters": 8,
776
+ "steps_per_s": 9.42,
777
+ "flops_cum": 1777038894563328.0
778
+ },
779
+ {
780
+ "step": 9700,
781
+ "loss": 2.9213,
782
+ "lr": 0.0007093298687687141,
783
+ "iters": 6,
784
+ "steps_per_s": 9.7,
785
+ "flops_cum": 1795671005331456.0
786
+ },
787
+ {
788
+ "step": 9800,
789
+ "loss": 2.8777,
790
+ "lr": 0.0007033683215379002,
791
+ "iters": 6,
792
+ "steps_per_s": 10.1,
793
+ "flops_cum": 1813588858109952.0
794
+ },
795
+ {
796
+ "step": 9900,
797
+ "loss": 2.4354,
798
+ "lr": 0.0006973719281921336,
799
+ "iters": 10,
800
+ "steps_per_s": 9.48,
801
+ "flops_cum": 1832762073415680.0
802
+ },
803
+ {
804
+ "step": 10000,
805
+ "loss": 2.7083,
806
+ "lr": 0.000691341716182545,
807
+ "iters": 6,
808
+ "steps_per_s": 8.67,
809
+ "flops_cum": 1851199386550272.0
810
+ },
811
+ {
812
+ "step": 10100,
813
+ "loss": 3.1843,
814
+ "lr": 0.0006852787187549182,
815
+ "iters": 5,
816
+ "steps_per_s": 9.63,
817
+ "flops_cum": 1870004650770432.0
818
+ },
819
+ {
820
+ "step": 10200,
821
+ "loss": 2.9582,
822
+ "lr": 0.0006791839747726501,
823
+ "iters": 5,
824
+ "steps_per_s": 9.68,
825
+ "flops_cum": 1888701694083072.0
826
+ },
827
+ {
828
+ "step": 10300,
829
+ "loss": 2.688,
830
+ "lr": 0.0006730585285387465,
831
+ "iters": 7,
832
+ "steps_per_s": 9.43,
833
+ "flops_cum": 1907961486114816.0
834
+ },
835
+ {
836
+ "step": 10400,
837
+ "loss": 2.4627,
838
+ "lr": 0.0006669034296168854,
839
+ "iters": 8,
840
+ "steps_per_s": 9.81,
841
+ "flops_cum": 1926420443430912.0
842
+ },
843
+ {
844
+ "step": 10500,
845
+ "loss": 2.7117,
846
+ "lr": 0.0006607197326515808,
847
+ "iters": 7,
848
+ "steps_per_s": 9.47,
849
+ "flops_cum": 1945507082010624.0
850
+ },
851
+ {
852
+ "step": 10600,
853
+ "loss": 2.2933,
854
+ "lr": 0.0006545084971874737,
855
+ "iters": 11,
856
+ "steps_per_s": 9.87,
857
+ "flops_cum": 1963944395145216.0
858
+ },
859
+ {
860
+ "step": 10700,
861
+ "loss": 2.6093,
862
+ "lr": 0.0006482707874877854,
863
+ "iters": 7,
864
+ "steps_per_s": 9.66,
865
+ "flops_cum": 1982641438457856.0
866
+ },
867
+ {
868
+ "step": 10800,
869
+ "loss": 2.1905,
870
+ "lr": 0.0006420076723519614,
871
+ "iters": 10,
872
+ "steps_per_s": 9.77,
873
+ "flops_cum": 2001230260862976.0
874
+ },
875
+ {
876
+ "step": 10900,
877
+ "loss": 2.1571,
878
+ "lr": 0.0006357202249325371,
879
+ "iters": 11,
880
+ "steps_per_s": 10.04,
881
+ "flops_cum": 2019277978730496.0
882
+ },
883
+ {
884
+ "step": 11000,
885
+ "loss": 2.2005,
886
+ "lr": 0.0006294095225512603,
887
+ "iters": 10,
888
+ "steps_per_s": 9.76,
889
+ "flops_cum": 2037823512772608.0
890
+ },
891
+ {
892
+ "step": 11100,
893
+ "loss": 2.0736,
894
+ "lr": 0.0006230766465144965,
895
+ "iters": 9,
896
+ "steps_per_s": 9.39,
897
+ "flops_cum": 2057061660622848.0
898
+ },
899
+ {
900
+ "step": 11200,
901
+ "loss": 2.5497,
902
+ "lr": 0.0006167226819279528,
903
+ "iters": 7,
904
+ "steps_per_s": 9.83,
905
+ "flops_cum": 2075434041212928.0
906
+ },
907
+ {
908
+ "step": 11300,
909
+ "loss": 2.1066,
910
+ "lr": 0.0006103487175107507,
911
+ "iters": 11,
912
+ "steps_per_s": 9.68,
913
+ "flops_cum": 2094217661251584.0
914
+ },
915
+ {
916
+ "step": 11400,
917
+ "loss": 2.196,
918
+ "lr": 0.0006039558454088796,
919
+ "iters": 10,
920
+ "steps_per_s": 9.51,
921
+ "flops_cum": 2113196078923776.0
922
+ },
923
+ {
924
+ "step": 11500,
925
+ "loss": 2.7455,
926
+ "lr": 0.0005975451610080642,
927
+ "iters": 5,
928
+ "steps_per_s": 9.44,
929
+ "flops_cum": 2132282717503488.0
930
+ },
931
+ {
932
+ "step": 11600,
933
+ "loss": 2.1884,
934
+ "lr": 0.0005911177627460738,
935
+ "iters": 9,
936
+ "steps_per_s": 9.94,
937
+ "flops_cum": 2150546877186048.0
938
+ },
939
+ {
940
+ "step": 11700,
941
+ "loss": 2.2152,
942
+ "lr": 0.0005846747519245122,
943
+ "iters": 9,
944
+ "steps_per_s": 10.1,
945
+ "flops_cum": 2168443085783040.0
946
+ },
947
+ {
948
+ "step": 11800,
949
+ "loss": 2.8085,
950
+ "lr": 0.0005782172325201155,
951
+ "iters": 5,
952
+ "steps_per_s": 9.7,
953
+ "flops_cum": 2187031908188160.0
954
+ },
955
+ {
956
+ "step": 11900,
957
+ "loss": 2.8641,
958
+ "lr": 0.0005717463109955896,
959
+ "iters": 5,
960
+ "steps_per_s": 9.98,
961
+ "flops_cum": 2205144558600192.0
962
+ },
963
+ {
964
+ "step": 12000,
965
+ "loss": 2.5605,
966
+ "lr": 0.000565263096110026,
967
+ "iters": 7,
968
+ "steps_per_s": 9.8,
969
+ "flops_cum": 2223668448460800.0
970
+ },
971
+ {
972
+ "step": 12100,
973
+ "loss": 2.0566,
974
+ "lr": 0.0005587686987289189,
975
+ "iters": 9,
976
+ "steps_per_s": 9.59,
977
+ "flops_cum": 2242581933588480.0
978
+ },
979
+ {
980
+ "step": 12200,
981
+ "loss": 2.1377,
982
+ "lr": 0.0005522642316338268,
983
+ "iters": 9,
984
+ "steps_per_s": 9.79,
985
+ "flops_cum": 2261084179267584.0
986
+ },
987
+ {
988
+ "step": 12300,
989
+ "loss": 2.1445,
990
+ "lr": 0.0005457508093317013,
991
+ "iters": 11,
992
+ "steps_per_s": 9.44,
993
+ "flops_cum": 2280257394573312.0
994
+ },
995
+ {
996
+ "step": 12400,
997
+ "loss": 2.8746,
998
+ "lr": 0.0005392295478639225,
999
+ "iters": 5,
1000
+ "steps_per_s": 9.87,
1001
+ "flops_cum": 2298521554255872.0
1002
+ },
1003
+ {
1004
+ "step": 12500,
1005
+ "loss": 2.0834,
1006
+ "lr": 0.0005327015646150716,
1007
+ "iters": 11,
1008
+ "steps_per_s": 8.31,
1009
+ "flops_cum": 2318062720647168.0
1010
+ },
1011
+ {
1012
+ "step": 12600,
1013
+ "loss": 2.2612,
1014
+ "lr": 0.000526167978121472,
1015
+ "iters": 8,
1016
+ "steps_per_s": 9.93,
1017
+ "flops_cum": 2336261947785216.0
1018
+ },
1019
+ {
1020
+ "step": 12700,
1021
+ "loss": 2.0504,
1022
+ "lr": 0.0005196299078795343,
1023
+ "iters": 10,
1024
+ "steps_per_s": 9.43,
1025
+ "flops_cum": 2355521739816960.0
1026
+ },
1027
+ {
1028
+ "step": 12800,
1029
+ "loss": 1.9142,
1030
+ "lr": 0.0005130884741539367,
1031
+ "iters": 12,
1032
+ "steps_per_s": 9.48,
1033
+ "flops_cum": 2374673310941184.0
1034
+ },
1035
+ {
1036
+ "step": 12900,
1037
+ "loss": 2.009,
1038
+ "lr": 0.0005065447977856722,
1039
+ "iters": 9,
1040
+ "steps_per_s": 9.57,
1041
+ "flops_cum": 2393543507705856.0
1042
+ },
1043
+ {
1044
+ "step": 13000,
1045
+ "loss": 2.4994,
1046
+ "lr": 0.0005,
1047
+ "iters": 6,
1048
+ "steps_per_s": 9.55,
1049
+ "flops_cum": 2412435348652032.0
1050
+ },
1051
+ {
1052
+ "step": 13100,
1053
+ "loss": 2.049,
1054
+ "lr": 0.0004934552022143279,
1055
+ "iters": 8,
1056
+ "steps_per_s": 9.98,
1057
+ "flops_cum": 2430569643245568.0
1058
+ },
1059
+ {
1060
+ "step": 13200,
1061
+ "loss": 2.0328,
1062
+ "lr": 0.0004869115258460635,
1063
+ "iters": 12,
1064
+ "steps_per_s": 10.03,
1065
+ "flops_cum": 2448703937839104.0
1066
+ },
1067
+ {
1068
+ "step": 13300,
1069
+ "loss": 2.7702,
1070
+ "lr": 0.00048037009212046586,
1071
+ "iters": 5,
1072
+ "steps_per_s": 9.55,
1073
+ "flops_cum": 2467530846240768.0
1074
+ },
1075
+ {
1076
+ "step": 13400,
1077
+ "loss": 2.4413,
1078
+ "lr": 0.0004738320218785281,
1079
+ "iters": 7,
1080
+ "steps_per_s": 9.89,
1081
+ "flops_cum": 2485859938467840.0
1082
+ },
1083
+ {
1084
+ "step": 13500,
1085
+ "loss": 2.2181,
1086
+ "lr": 0.00046729843538492847,
1087
+ "iters": 8,
1088
+ "steps_per_s": 9.99,
1089
+ "flops_cum": 2503972588879872.0
1090
+ },
1091
+ {
1092
+ "step": 13600,
1093
+ "loss": 2.6502,
1094
+ "lr": 0.0004607704521360776,
1095
+ "iters": 5,
1096
+ "steps_per_s": 9.67,
1097
+ "flops_cum": 2522647988011008.0
1098
+ },
1099
+ {
1100
+ "step": 13700,
1101
+ "loss": 2.5721,
1102
+ "lr": 0.00045424919066829885,
1103
+ "iters": 6,
1104
+ "steps_per_s": 9.83,
1105
+ "flops_cum": 2541085301145600.0
1106
+ },
1107
+ {
1108
+ "step": 13800,
1109
+ "loss": 2.7172,
1110
+ "lr": 0.00044773576836617336,
1111
+ "iters": 5,
1112
+ "steps_per_s": 9.65,
1113
+ "flops_cum": 2559847277002752.0
1114
+ },
1115
+ {
1116
+ "step": 13900,
1117
+ "loss": 2.3191,
1118
+ "lr": 0.00044123130127108126,
1119
+ "iters": 7,
1120
+ "steps_per_s": 9.73,
1121
+ "flops_cum": 2578501031952384.0
1122
+ },
1123
+ {
1124
+ "step": 14000,
1125
+ "loss": 2.2618,
1126
+ "lr": 0.00043473690388997434,
1127
+ "iters": 7,
1128
+ "steps_per_s": 9.52,
1129
+ "flops_cum": 2597522737987584.0
1130
+ },
1131
+ {
1132
+ "step": 14100,
1133
+ "loss": 1.9426,
1134
+ "lr": 0.0004282536890044104,
1135
+ "iters": 12,
1136
+ "steps_per_s": 9.62,
1137
+ "flops_cum": 2616414578933760.0
1138
+ },
1139
+ {
1140
+ "step": 14200,
1141
+ "loss": 2.4823,
1142
+ "lr": 0.0004217827674798845,
1143
+ "iters": 7,
1144
+ "steps_per_s": 10.03,
1145
+ "flops_cum": 2634375720075264.0
1146
+ },
1147
+ {
1148
+ "step": 14300,
1149
+ "loss": 1.967,
1150
+ "lr": 0.0004153252480754877,
1151
+ "iters": 11,
1152
+ "steps_per_s": 9.22,
1153
+ "flops_cum": 2654068395737088.0
1154
+ },
1155
+ {
1156
+ "step": 14400,
1157
+ "loss": 2.5707,
1158
+ "lr": 0.00040888223725392626,
1159
+ "iters": 6,
1160
+ "steps_per_s": 9.61,
1161
+ "flops_cum": 2672830371594240.0
1162
+ },
1163
+ {
1164
+ "step": 14500,
1165
+ "loss": 1.9909,
1166
+ "lr": 0.00040245483899193594,
1167
+ "iters": 9,
1168
+ "steps_per_s": 9.45,
1169
+ "flops_cum": 2692046875262976.0
1170
+ },
1171
+ {
1172
+ "step": 14600,
1173
+ "loss": 2.8218,
1174
+ "lr": 0.0003960441545911204,
1175
+ "iters": 5,
1176
+ "steps_per_s": 9.77,
1177
+ "flops_cum": 2710527476760576.0
1178
+ },
1179
+ {
1180
+ "step": 14700,
1181
+ "loss": 2.584,
1182
+ "lr": 0.0003896512824892495,
1183
+ "iters": 5,
1184
+ "steps_per_s": 9.85,
1185
+ "flops_cum": 2728899857350656.0
1186
+ },
1187
+ {
1188
+ "step": 14800,
1189
+ "loss": 2.2358,
1190
+ "lr": 0.00038327731807204744,
1191
+ "iters": 8,
1192
+ "steps_per_s": 9.64,
1193
+ "flops_cum": 2747726765752320.0
1194
+ },
1195
+ {
1196
+ "step": 14900,
1197
+ "loss": 2.1283,
1198
+ "lr": 0.0003769233534855035,
1199
+ "iters": 8,
1200
+ "steps_per_s": 9.62,
1201
+ "flops_cum": 2766553674153984.0
1202
+ },
1203
+ {
1204
+ "step": 15000,
1205
+ "loss": 1.9897,
1206
+ "lr": 0.0003705904774487396,
1207
+ "iters": 12,
1208
+ "steps_per_s": 8.91,
1209
+ "flops_cum": 2784579747840000.0
1210
+ },
1211
+ {
1212
+ "step": 15100,
1213
+ "loss": 1.8867,
1214
+ "lr": 0.0003642797750674629,
1215
+ "iters": 10,
1216
+ "steps_per_s": 9.46,
1217
+ "flops_cum": 2803688030601216.0
1218
+ },
1219
+ {
1220
+ "step": 15200,
1221
+ "loss": 1.9314,
1222
+ "lr": 0.0003579923276480387,
1223
+ "iters": 9,
1224
+ "steps_per_s": 9.76,
1225
+ "flops_cum": 2822190276280320.0
1226
+ },
1227
+ {
1228
+ "step": 15300,
1229
+ "loss": 1.8862,
1230
+ "lr": 0.0003517292125122146,
1231
+ "iters": 11,
1232
+ "steps_per_s": 9.41,
1233
+ "flops_cum": 2841428424130560.0
1234
+ },
1235
+ {
1236
+ "step": 15400,
1237
+ "loss": 2.533,
1238
+ "lr": 0.00034549150281252633,
1239
+ "iters": 6,
1240
+ "steps_per_s": 9.96,
1241
+ "flops_cum": 2859519430361088.0
1242
+ },
1243
+ {
1244
+ "step": 15500,
1245
+ "loss": 2.3289,
1246
+ "lr": 0.0003392802673484193,
1247
+ "iters": 7,
1248
+ "steps_per_s": 9.96,
1249
+ "flops_cum": 2877675369136128.0
1250
+ },
1251
+ {
1252
+ "step": 15600,
1253
+ "loss": 2.079,
1254
+ "lr": 0.00033309657038311456,
1255
+ "iters": 9,
1256
+ "steps_per_s": 9.68,
1257
+ "flops_cum": 2896394056630272.0
1258
+ },
1259
+ {
1260
+ "step": 15700,
1261
+ "loss": 1.9916,
1262
+ "lr": 0.0003269414714612534,
1263
+ "iters": 10,
1264
+ "steps_per_s": 9.83,
1265
+ "flops_cum": 2914788081401856.0
1266
+ },
1267
+ {
1268
+ "step": 15800,
1269
+ "loss": 2.7287,
1270
+ "lr": 0.00032081602522734986,
1271
+ "iters": 5,
1272
+ "steps_per_s": 9.74,
1273
+ "flops_cum": 2933290327080960.0
1274
+ },
1275
+ {
1276
+ "step": 15900,
1277
+ "loss": 2.147,
1278
+ "lr": 0.00031472128124508187,
1279
+ "iters": 7,
1280
+ "steps_per_s": 9.33,
1281
+ "flops_cum": 2952679984201728.0
1282
+ },
1283
+ {
1284
+ "step": 16000,
1285
+ "loss": 2.2722,
1286
+ "lr": 0.0003086582838174551,
1287
+ "iters": 6,
1288
+ "steps_per_s": 9.7,
1289
+ "flops_cum": 2971290450788352.0
1290
+ },
1291
+ {
1292
+ "step": 16100,
1293
+ "loss": 2.5841,
1294
+ "lr": 0.00030262807180786645,
1295
+ "iters": 5,
1296
+ "steps_per_s": 10.05,
1297
+ "flops_cum": 2989251591929856.0
1298
+ },
1299
+ {
1300
+ "step": 16200,
1301
+ "loss": 2.3754,
1302
+ "lr": 0.0002966316784621,
1303
+ "iters": 7,
1304
+ "steps_per_s": 9.86,
1305
+ "flops_cum": 3007623972519936.0
1306
+ },
1307
+ {
1308
+ "step": 16300,
1309
+ "loss": 2.2585,
1310
+ "lr": 0.0002906701312312861,
1311
+ "iters": 7,
1312
+ "steps_per_s": 9.81,
1313
+ "flops_cum": 3026061285654528.0
1314
+ },
1315
+ {
1316
+ "step": 16400,
1317
+ "loss": 1.9229,
1318
+ "lr": 0.0002847444515958523,
1319
+ "iters": 12,
1320
+ "steps_per_s": 9.73,
1321
+ "flops_cum": 3044758328967168.0
1322
+ },
1323
+ {
1324
+ "step": 16500,
1325
+ "loss": 2.247,
1326
+ "lr": 0.00027885565489049947,
1327
+ "iters": 6,
1328
+ "steps_per_s": 9.86,
1329
+ "flops_cum": 3062979200286720.0
1330
+ },
1331
+ {
1332
+ "step": 16600,
1333
+ "loss": 2.6724,
1334
+ "lr": 0.00027300475013022663,
1335
+ "iters": 5,
1336
+ "steps_per_s": 9.89,
1337
+ "flops_cum": 3081221715787776.0
1338
+ },
1339
+ {
1340
+ "step": 16700,
1341
+ "loss": 1.9804,
1342
+ "lr": 0.0002671927398374443,
1343
+ "iters": 10,
1344
+ "steps_per_s": 9.61,
1345
+ "flops_cum": 3100135200915456.0
1346
+ },
1347
+ {
1348
+ "step": 16800,
1349
+ "loss": 2.2886,
1350
+ "lr": 0.00026142061987019576,
1351
+ "iters": 7,
1352
+ "steps_per_s": 10.3,
1353
+ "flops_cum": 3117641814245376.0
1354
+ },
1355
+ {
1356
+ "step": 16900,
1357
+ "loss": 2.1789,
1358
+ "lr": 0.0002556893792515227,
1359
+ "iters": 7,
1360
+ "steps_per_s": 9.68,
1361
+ "flops_cum": 3136317213376512.0
1362
+ },
1363
+ {
1364
+ "step": 17000,
1365
+ "loss": 2.4573,
1366
+ "lr": 0.0002500000000000001,
1367
+ "iters": 6,
1368
+ "steps_per_s": 9.6,
1369
+ "flops_cum": 3155122477596672.0
1370
+ },
1371
+ {
1372
+ "step": 17100,
1373
+ "loss": 2.6159,
1374
+ "lr": 0.00024435345696147403,
1375
+ "iters": 5,
1376
+ "steps_per_s": 10.22,
1377
+ "flops_cum": 3172802244378624.0
1378
+ },
1379
+ {
1380
+ "step": 17200,
1381
+ "loss": 2.2826,
1382
+ "lr": 0.00023875071764202561,
1383
+ "iters": 7,
1384
+ "steps_per_s": 9.85,
1385
+ "flops_cum": 3191196269150208.0
1386
+ },
1387
+ {
1388
+ "step": 17300,
1389
+ "loss": 2.2721,
1390
+ "lr": 0.00023319274204219425,
1391
+ "iters": 6,
1392
+ "steps_per_s": 10.06,
1393
+ "flops_cum": 3209135766110208.0
1394
+ },
1395
+ {
1396
+ "step": 17400,
1397
+ "loss": 2.613,
1398
+ "lr": 0.00022768048249248646,
1399
+ "iters": 5,
1400
+ "steps_per_s": 9.98,
1401
+ "flops_cum": 3227226772340736.0
1402
+ },
1403
+ {
1404
+ "step": 17500,
1405
+ "loss": 2.6948,
1406
+ "lr": 0.00022221488349019903,
1407
+ "iters": 5,
1408
+ "steps_per_s": 8.83,
1409
+ "flops_cum": 3245339422752768.0
1410
+ },
1411
+ {
1412
+ "step": 17600,
1413
+ "loss": 2.3915,
1414
+ "lr": 0.0002167968815375837,
1415
+ "iters": 6,
1416
+ "steps_per_s": 9.84,
1417
+ "flops_cum": 3263711803342848.0
1418
+ },
1419
+ {
1420
+ "step": 17700,
1421
+ "loss": 2.6084,
1422
+ "lr": 0.00021142740498138323,
1423
+ "iters": 5,
1424
+ "steps_per_s": 9.48,
1425
+ "flops_cum": 3282733509378048.0
1426
+ },
1427
+ {
1428
+ "step": 17800,
1429
+ "loss": 1.8803,
1430
+ "lr": 0.00020610737385376348,
1431
+ "iters": 9,
1432
+ "steps_per_s": 9.87,
1433
+ "flops_cum": 3301127534149632.0
1434
+ },
1435
+ {
1436
+ "step": 17900,
1437
+ "loss": 1.9358,
1438
+ "lr": 0.00020083769971467047,
1439
+ "iters": 10,
1440
+ "steps_per_s": 9.78,
1441
+ "flops_cum": 3319651424010240.0
1442
+ },
1443
+ {
1444
+ "step": 18000,
1445
+ "loss": 2.0779,
1446
+ "lr": 0.00019561928549563967,
1447
+ "iters": 8,
1448
+ "steps_per_s": 9.76,
1449
+ "flops_cum": 3338132025507840.0
1450
+ },
1451
+ {
1452
+ "step": 18100,
1453
+ "loss": 1.9217,
1454
+ "lr": 0.00019045302534508295,
1455
+ "iters": 12,
1456
+ "steps_per_s": 9.98,
1457
+ "flops_cum": 3356331252645888.0
1458
+ },
1459
+ {
1460
+ "step": 18200,
1461
+ "loss": 1.9165,
1462
+ "lr": 0.00018533980447508135,
1463
+ "iters": 12,
1464
+ "steps_per_s": 9.58,
1465
+ "flops_cum": 3375223093592064.0
1466
+ },
1467
+ {
1468
+ "step": 18300,
1469
+ "loss": 2.6252,
1470
+ "lr": 0.00018028049900970766,
1471
+ "iters": 5,
1472
+ "steps_per_s": 9.84,
1473
+ "flops_cum": 3393465609093120.0
1474
+ },
1475
+ {
1476
+ "step": 18400,
1477
+ "loss": 1.8428,
1478
+ "lr": 0.00017527597583490823,
1479
+ "iters": 10,
1480
+ "steps_per_s": 9.82,
1481
+ "flops_cum": 3412011143135232.0
1482
+ },
1483
+ {
1484
+ "step": 18500,
1485
+ "loss": 1.9012,
1486
+ "lr": 0.00017032709244996558,
1487
+ "iters": 9,
1488
+ "steps_per_s": 9.62,
1489
+ "flops_cum": 3430816407355392.0
1490
+ },
1491
+ {
1492
+ "step": 18600,
1493
+ "loss": 2.4853,
1494
+ "lr": 0.00016543469682057105,
1495
+ "iters": 5,
1496
+ "steps_per_s": 9.84,
1497
+ "flops_cum": 3449102211219456.0
1498
+ },
1499
+ {
1500
+ "step": 18700,
1501
+ "loss": 2.2982,
1502
+ "lr": 0.0001605996272335291,
1503
+ "iters": 6,
1504
+ "steps_per_s": 9.72,
1505
+ "flops_cum": 3467734321987584.0
1506
+ },
1507
+ {
1508
+ "step": 18800,
1509
+ "loss": 2.2198,
1510
+ "lr": 0.00015582271215312294,
1511
+ "iters": 7,
1512
+ "steps_per_s": 9.48,
1513
+ "flops_cum": 3486842604748800.0
1514
+ },
1515
+ {
1516
+ "step": 18900,
1517
+ "loss": 2.2655,
1518
+ "lr": 0.00015110477007916002,
1519
+ "iters": 7,
1520
+ "steps_per_s": 10.08,
1521
+ "flops_cum": 3504782101708800.0
1522
+ },
1523
+ {
1524
+ "step": 19000,
1525
+ "loss": 2.5617,
1526
+ "lr": 0.00014644660940672628,
1527
+ "iters": 5,
1528
+ "steps_per_s": 9.85,
1529
+ "flops_cum": 3523132838117376.0
1530
+ },
1531
+ {
1532
+ "step": 19100,
1533
+ "loss": 1.9135,
1534
+ "lr": 0.00014184902828767287,
1535
+ "iters": 9,
1536
+ "steps_per_s": 10.2,
1537
+ "flops_cum": 3540942469988352.0
1538
+ },
1539
+ {
1540
+ "step": 19200,
1541
+ "loss": 1.8071,
1542
+ "lr": 0.0001373128144938563,
1543
+ "iters": 10,
1544
+ "steps_per_s": 9.74,
1545
+ "flops_cum": 3559552936574976.0
1546
+ },
1547
+ {
1548
+ "step": 19300,
1549
+ "loss": 1.8479,
1550
+ "lr": 0.00013283874528215734,
1551
+ "iters": 10,
1552
+ "steps_per_s": 10.17,
1553
+ "flops_cum": 3577340924264448.0
1554
+ },
1555
+ {
1556
+ "step": 19400,
1557
+ "loss": 2.3403,
1558
+ "lr": 0.00012842758726130281,
1559
+ "iters": 6,
1560
+ "steps_per_s": 9.7,
1561
+ "flops_cum": 3595908102488064.0
1562
+ },
1563
+ {
1564
+ "step": 19500,
1565
+ "loss": 1.8826,
1566
+ "lr": 0.00012408009626051135,
1567
+ "iters": 11,
1568
+ "steps_per_s": 10.05,
1569
+ "flops_cum": 3613977464537088.0
1570
+ },
1571
+ {
1572
+ "step": 19600,
1573
+ "loss": 1.9027,
1574
+ "lr": 0.00011979701719998454,
1575
+ "iters": 10,
1576
+ "steps_per_s": 10.04,
1577
+ "flops_cum": 3632003538223104.0
1578
+ },
1579
+ {
1580
+ "step": 19700,
1581
+ "loss": 1.883,
1582
+ "lr": 0.00011557908396327027,
1583
+ "iters": 12,
1584
+ "steps_per_s": 9.5,
1585
+ "flops_cum": 3651068532621312.0
1586
+ },
1587
+ {
1588
+ "step": 19800,
1589
+ "loss": 1.8821,
1590
+ "lr": 0.00011142701927151455,
1591
+ "iters": 10,
1592
+ "steps_per_s": 9.77,
1593
+ "flops_cum": 3669527489937408.0
1594
+ },
1595
+ {
1596
+ "step": 19900,
1597
+ "loss": 2.0984,
1598
+ "lr": 0.00010734153455962764,
1599
+ "iters": 8,
1600
+ "steps_per_s": 10.24,
1601
+ "flops_cum": 3687163968356352.0
1602
+ },
1603
+ {
1604
+ "step": 20000,
1605
+ "loss": 2.4296,
1606
+ "lr": 0.00010332332985438247,
1607
+ "iters": 5,
1608
+ "steps_per_s": 8.74,
1609
+ "flops_cum": 3705449772220416.0
1610
+ },
1611
+ {
1612
+ "step": 20100,
1613
+ "loss": 2.578,
1614
+ "lr": 9.937309365446973e-05,
1615
+ "iters": 5,
1616
+ "steps_per_s": 9.73,
1617
+ "flops_cum": 3724038594625536.0
1618
+ },
1619
+ {
1620
+ "step": 20200,
1621
+ "loss": 2.4069,
1622
+ "lr": 9.549150281252633e-05,
1623
+ "iters": 5,
1624
+ "steps_per_s": 9.79,
1625
+ "flops_cum": 3742540840304640.0
1626
+ },
1627
+ {
1628
+ "step": 20300,
1629
+ "loss": 1.8465,
1630
+ "lr": 9.167922241916055e-05,
1631
+ "iters": 9,
1632
+ "steps_per_s": 10.0,
1633
+ "flops_cum": 3760718423261184.0
1634
+ },
1635
+ {
1636
+ "step": 20400,
1637
+ "loss": 1.905,
1638
+ "lr": 8.793690568899215e-05,
1639
+ "iters": 10,
1640
+ "steps_per_s": 9.23,
1641
+ "flops_cum": 3780302878015488.0
1642
+ },
1643
+ {
1644
+ "step": 20500,
1645
+ "loss": 1.8215,
1646
+ "lr": 8.426519384872733e-05,
1647
+ "iters": 12,
1648
+ "steps_per_s": 10.24,
1649
+ "flops_cum": 3798004288978944.0
1650
+ },
1651
+ {
1652
+ "step": 20600,
1653
+ "loss": 1.8479,
1654
+ "lr": 8.066471602728804e-05,
1655
+ "iters": 11,
1656
+ "steps_per_s": 9.59,
1657
+ "flops_cum": 3816874485743616.0
1658
+ },
1659
+ {
1660
+ "step": 20700,
1661
+ "loss": 2.4496,
1662
+ "lr": 7.71360891480134e-05,
1663
+ "iters": 5,
1664
+ "steps_per_s": 9.97,
1665
+ "flops_cum": 3834943847792640.0
1666
+ },
1667
+ {
1668
+ "step": 20800,
1669
+ "loss": 1.8314,
1670
+ "lr": 7.367991782295391e-05,
1671
+ "iters": 12,
1672
+ "steps_per_s": 9.86,
1673
+ "flops_cum": 3853424449290240.0
1674
+ },
1675
+ {
1676
+ "step": 20900,
1677
+ "loss": 1.8225,
1678
+ "lr": 7.029679424927366e-05,
1679
+ "iters": 11,
1680
+ "steps_per_s": 10.01,
1681
+ "flops_cum": 3871515455520768.0
1682
+ },
1683
+ {
1684
+ "step": 21000,
1685
+ "loss": 2.5524,
1686
+ "lr": 6.698729810778065e-05,
1687
+ "iters": 5,
1688
+ "steps_per_s": 9.4,
1689
+ "flops_cum": 3890645382463488.0
1690
+ },
1691
+ {
1692
+ "step": 21100,
1693
+ "loss": 1.7767,
1694
+ "lr": 6.375199646360142e-05,
1695
+ "iters": 10,
1696
+ "steps_per_s": 9.77,
1697
+ "flops_cum": 3909190916505600.0
1698
+ },
1699
+ {
1700
+ "step": 21200,
1701
+ "loss": 2.5879,
1702
+ "lr": 6.059144366901737e-05,
1703
+ "iters": 5,
1704
+ "steps_per_s": 9.96,
1705
+ "flops_cum": 3927303566917632.0
1706
+ },
1707
+ {
1708
+ "step": 21300,
1709
+ "loss": 1.9993,
1710
+ "lr": 5.750618126847912e-05,
1711
+ "iters": 8,
1712
+ "steps_per_s": 9.62,
1713
+ "flops_cum": 3946173763682304.0
1714
+ },
1715
+ {
1716
+ "step": 21400,
1717
+ "loss": 1.8638,
1718
+ "lr": 5.449673790581611e-05,
1719
+ "iters": 11,
1720
+ "steps_per_s": 9.96,
1721
+ "flops_cum": 3964394635001856.0
1722
+ },
1723
+ {
1724
+ "step": 21500,
1725
+ "loss": 1.9277,
1726
+ "lr": 5.1563629233655876e-05,
1727
+ "iters": 12,
1728
+ "steps_per_s": 9.65,
1729
+ "flops_cum": 3983178255040512.0
1730
+ },
1731
+ {
1732
+ "step": 21600,
1733
+ "loss": 1.8451,
1734
+ "lr": 4.87073578250698e-05,
1735
+ "iters": 12,
1736
+ "steps_per_s": 9.82,
1737
+ "flops_cum": 4001615568175104.0
1738
+ },
1739
+ {
1740
+ "step": 21700,
1741
+ "loss": 2.1708,
1742
+ "lr": 4.592841308745932e-05,
1743
+ "iters": 8,
1744
+ "steps_per_s": 9.28,
1745
+ "flops_cum": 4021070157840384.0
1746
+ },
1747
+ {
1748
+ "step": 21800,
1749
+ "loss": 1.8548,
1750
+ "lr": 4.322727117869951e-05,
1751
+ "iters": 9,
1752
+ "steps_per_s": 9.96,
1753
+ "flops_cum": 4039291029159936.0
1754
+ },
1755
+ {
1756
+ "step": 21900,
1757
+ "loss": 1.9048,
1758
+ "lr": 4.06043949255509e-05,
1759
+ "iters": 10,
1760
+ "steps_per_s": 9.99,
1761
+ "flops_cum": 4057403679571968.0
1762
+ },
1763
+ {
1764
+ "step": 22000,
1765
+ "loss": 1.9439,
1766
+ "lr": 3.806023374435663e-05,
1767
+ "iters": 11,
1768
+ "steps_per_s": 9.51,
1769
+ "flops_cum": 4076447029788672.0
1770
+ },
1771
+ {
1772
+ "step": 22100,
1773
+ "loss": 1.813,
1774
+ "lr": 3.559522356403788e-05,
1775
+ "iters": 9,
1776
+ "steps_per_s": 9.91,
1777
+ "flops_cum": 4094689545289728.0
1778
+ },
1779
+ {
1780
+ "step": 22200,
1781
+ "loss": 1.8323,
1782
+ "lr": 3.3209786751399184e-05,
1783
+ "iters": 10,
1784
+ "steps_per_s": 10.03,
1785
+ "flops_cum": 4112758907338752.0
1786
+ },
1787
+ {
1788
+ "step": 22300,
1789
+ "loss": 1.8636,
1790
+ "lr": 3.0904332038757974e-05,
1791
+ "iters": 11,
1792
+ "steps_per_s": 9.64,
1793
+ "flops_cum": 4131564171558912.0
1794
+ },
1795
+ {
1796
+ "step": 22400,
1797
+ "loss": 2.3475,
1798
+ "lr": 2.8679254453910786e-05,
1799
+ "iters": 6,
1800
+ "steps_per_s": 9.73,
1801
+ "flops_cum": 4150088061419520.0
1802
+ },
1803
+ {
1804
+ "step": 22500,
1805
+ "loss": 2.1483,
1806
+ "lr": 2.653493525244721e-05,
1807
+ "iters": 8,
1808
+ "steps_per_s": 8.77,
1809
+ "flops_cum": 4168417153646592.0
1810
+ },
1811
+ {
1812
+ "step": 22600,
1813
+ "loss": 1.8919,
1814
+ "lr": 2.4471741852423235e-05,
1815
+ "iters": 11,
1816
+ "steps_per_s": 10.06,
1817
+ "flops_cum": 4186464871514112.0
1818
+ },
1819
+ {
1820
+ "step": 22700,
1821
+ "loss": 2.1376,
1822
+ "lr": 2.2490027771406685e-05,
1823
+ "iters": 8,
1824
+ "steps_per_s": 10.11,
1825
+ "flops_cum": 4204339435929600.0
1826
+ },
1827
+ {
1828
+ "step": 22800,
1829
+ "loss": 2.5045,
1830
+ "lr": 2.0590132565903473e-05,
1831
+ "iters": 5,
1832
+ "steps_per_s": 9.81,
1833
+ "flops_cum": 4222733460701184.0
1834
+ },
1835
+ {
1836
+ "step": 22900,
1837
+ "loss": 1.9733,
1838
+ "lr": 1.8772381773176416e-05,
1839
+ "iters": 8,
1840
+ "steps_per_s": 9.97,
1841
+ "flops_cum": 4240932687839232.0
1842
+ },
1843
+ {
1844
+ "step": 23000,
1845
+ "loss": 1.8414,
1846
+ "lr": 1.70370868554659e-05,
1847
+ "iters": 12,
1848
+ "steps_per_s": 9.8,
1849
+ "flops_cum": 4259456577699840.0
1850
+ },
1851
+ {
1852
+ "step": 23100,
1853
+ "loss": 2.3108,
1854
+ "lr": 1.538454514662285e-05,
1855
+ "iters": 6,
1856
+ "steps_per_s": 10.07,
1857
+ "flops_cum": 4277331142115328.0
1858
+ },
1859
+ {
1860
+ "step": 23200,
1861
+ "loss": 2.156,
1862
+ "lr": 1.3815039801161721e-05,
1863
+ "iters": 7,
1864
+ "steps_per_s": 10.16,
1865
+ "flops_cum": 4295162418167808.0
1866
+ },
1867
+ {
1868
+ "step": 23300,
1869
+ "loss": 2.0757,
1870
+ "lr": 1.232883974574367e-05,
1871
+ "iters": 8,
1872
+ "steps_per_s": 9.78,
1873
+ "flops_cum": 4313707952209920.0
1874
+ },
1875
+ {
1876
+ "step": 23400,
1877
+ "loss": 1.8186,
1878
+ "lr": 1.0926199633097156e-05,
1879
+ "iters": 11,
1880
+ "steps_per_s": 10.18,
1881
+ "flops_cum": 4331539228262400.0
1882
+ },
1883
+ {
1884
+ "step": 23500,
1885
+ "loss": 1.8452,
1886
+ "lr": 9.607359798384786e-06,
1887
+ "iters": 12,
1888
+ "steps_per_s": 9.71,
1889
+ "flops_cum": 4350171339030528.0
1890
+ },
1891
+ {
1892
+ "step": 23600,
1893
+ "loss": 2.0267,
1894
+ "lr": 8.372546218022748e-06,
1895
+ "iters": 8,
1896
+ "steps_per_s": 9.93,
1897
+ "flops_cum": 4368370566168576.0
1898
+ },
1899
+ {
1900
+ "step": 23700,
1901
+ "loss": 2.4369,
1902
+ "lr": 7.221970470961125e-06,
1903
+ "iters": 5,
1904
+ "steps_per_s": 9.61,
1905
+ "flops_cum": 4387154186207232.0
1906
+ },
1907
+ {
1908
+ "step": 23800,
1909
+ "loss": 2.3796,
1910
+ "lr": 6.15582970243117e-06,
1911
+ "iters": 6,
1912
+ "steps_per_s": 10.0,
1913
+ "flops_cum": 4405288480800768.0
1914
+ },
1915
+ {
1916
+ "step": 23900,
1917
+ "loss": 1.8084,
1918
+ "lr": 5.174306590164879e-06,
1919
+ "iters": 10,
1920
+ "steps_per_s": 10.01,
1921
+ "flops_cum": 4423444419575808.0
1922
+ },
1923
+ {
1924
+ "step": 24000,
1925
+ "loss": 2.25,
1926
+ "lr": 4.277569313094809e-06,
1927
+ "iters": 7,
1928
+ "steps_per_s": 9.66,
1929
+ "flops_cum": 4442163107069952.0
1930
+ },
1931
+ {
1932
+ "step": 24100,
1933
+ "loss": 1.8961,
1934
+ "lr": 3.4657715225368535e-06,
1935
+ "iters": 10,
1936
+ "steps_per_s": 9.34,
1937
+ "flops_cum": 4461596052553728.0
1938
+ },
1939
+ {
1940
+ "step": 24200,
1941
+ "loss": 1.8762,
1942
+ "lr": 2.739052315863355e-06,
1943
+ "iters": 10,
1944
+ "steps_per_s": 9.57,
1945
+ "flops_cum": 4480509537681408.0
1946
+ },
1947
+ {
1948
+ "step": 24300,
1949
+ "loss": 2.7098,
1950
+ "lr": 2.097536212669171e-06,
1951
+ "iters": 5,
1952
+ "steps_per_s": 9.47,
1953
+ "flops_cum": 4499552887898112.0
1954
+ },
1955
+ {
1956
+ "step": 24400,
1957
+ "loss": 1.8893,
1958
+ "lr": 1.541333133436018e-06,
1959
+ "iters": 10,
1960
+ "steps_per_s": 9.76,
1961
+ "flops_cum": 4518184998666240.0
1962
+ },
1963
+ {
1964
+ "step": 24500,
1965
+ "loss": 1.8306,
1966
+ "lr": 1.0705383806982606e-06,
1967
+ "iters": 10,
1968
+ "steps_per_s": 9.38,
1969
+ "flops_cum": 4537509723242496.0
1970
+ },
1971
+ {
1972
+ "step": 24600,
1973
+ "loss": 1.8545,
1974
+ "lr": 6.852326227130834e-07,
1975
+ "iters": 12,
1976
+ "steps_per_s": 9.53,
1977
+ "flops_cum": 4556553073459200.0
1978
+ },
1979
+ {
1980
+ "step": 24700,
1981
+ "loss": 2.4365,
1982
+ "lr": 3.854818796385495e-07,
1983
+ "iters": 5,
1984
+ "steps_per_s": 9.86,
1985
+ "flops_cum": 4574773944778752.0
1986
+ },
1987
+ {
1988
+ "step": 24800,
1989
+ "loss": 1.8731,
1990
+ "lr": 1.7133751222137007e-07,
1991
+ "iters": 9,
1992
+ "steps_per_s": 9.97,
1993
+ "flops_cum": 4593016460279808.0
1994
+ },
1995
+ {
1996
+ "step": 24900,
1997
+ "loss": 2.5486,
1998
+ "lr": 4.283621299649987e-08,
1999
+ "iters": 5,
2000
+ "steps_per_s": 9.84,
2001
+ "flops_cum": 4611345552506880.0
2002
+ },
2003
+ {
2004
+ "step": 24999,
2005
+ "loss": 1.8157,
2006
+ "lr": 4.283682464301109e-12,
2007
+ "iters": 11,
2008
+ "steps_per_s": 9.46,
2009
+ "flops_cum": 4630626158247936.0
2010
+ }
2011
+ ],
2012
+ "id_acc": [
2013
+ {
2014
+ "step": 2500,
2015
+ "mean": 0.03968749847263098,
2016
+ "per_k": {
2017
+ "1": 0.022499999031424522,
2018
+ "2": 0.03749999776482582,
2019
+ "3": 0.03749999776482582,
2020
+ "4": 0.05249999836087227,
2021
+ "5": 0.03500000014901161,
2022
+ "6": 0.04999999701976776,
2023
+ "7": 0.029999999329447746,
2024
+ "8": 0.05249999836087227
2025
+ }
2026
+ },
2027
+ {
2028
+ "step": 5000,
2029
+ "mean": 0.04968749824911356,
2030
+ "per_k": {
2031
+ "1": 0.042500000447034836,
2032
+ "2": 0.044999998062849045,
2033
+ "3": 0.042500000447034836,
2034
+ "4": 0.03749999776482582,
2035
+ "5": 0.03749999776482582,
2036
+ "6": 0.07499999552965164,
2037
+ "7": 0.05249999836087227,
2038
+ "8": 0.06499999761581421
2039
+ }
2040
+ },
2041
+ {
2042
+ "step": 7500,
2043
+ "mean": 0.07437499705702066,
2044
+ "per_k": {
2045
+ "1": 0.19999998807907104,
2046
+ "2": 0.08249999582767487,
2047
+ "3": 0.027499999850988388,
2048
+ "4": 0.057499997317790985,
2049
+ "5": 0.05249999836087227,
2050
+ "6": 0.0625,
2051
+ "7": 0.05249999836087227,
2052
+ "8": 0.05999999865889549
2053
+ }
2054
+ },
2055
+ {
2056
+ "step": 10000,
2057
+ "mean": 0.26312499679625034,
2058
+ "per_k": {
2059
+ "1": 0.3050000071525574,
2060
+ "2": 0.27250000834465027,
2061
+ "3": 0.26499998569488525,
2062
+ "4": 0.23999999463558197,
2063
+ "5": 0.25,
2064
+ "6": 0.2824999988079071,
2065
+ "7": 0.22999998927116394,
2066
+ "8": 0.25999999046325684
2067
+ }
2068
+ },
2069
+ {
2070
+ "step": 12500,
2071
+ "mean": 0.2574999947100878,
2072
+ "per_k": {
2073
+ "1": 0.2800000011920929,
2074
+ "2": 0.29249998927116394,
2075
+ "3": 0.22999998927116394,
2076
+ "4": 0.21249999105930328,
2077
+ "5": 0.23499999940395355,
2078
+ "6": 0.2474999874830246,
2079
+ "7": 0.3050000071525574,
2080
+ "8": 0.2574999928474426
2081
+ }
2082
+ },
2083
+ {
2084
+ "step": 15000,
2085
+ "mean": 0.2568749934434891,
2086
+ "per_k": {
2087
+ "1": 0.21249999105930328,
2088
+ "2": 0.2800000011920929,
2089
+ "3": 0.26499998569488525,
2090
+ "4": 0.23749999701976776,
2091
+ "5": 0.25,
2092
+ "6": 0.22749999165534973,
2093
+ "7": 0.2874999940395355,
2094
+ "8": 0.29499998688697815
2095
+ }
2096
+ },
2097
+ {
2098
+ "step": 17500,
2099
+ "mean": 0.2540624961256981,
2100
+ "per_k": {
2101
+ "1": 0.1850000023841858,
2102
+ "2": 0.2449999898672104,
2103
+ "3": 0.23999999463558197,
2104
+ "4": 0.29249998927116394,
2105
+ "5": 0.2800000011920929,
2106
+ "6": 0.2175000011920929,
2107
+ "7": 0.29999998211860657,
2108
+ "8": 0.27250000834465027
2109
+ }
2110
+ },
2111
+ {
2112
+ "step": 20000,
2113
+ "mean": 0.26062499545514584,
2114
+ "per_k": {
2115
+ "1": 0.23999999463558197,
2116
+ "2": 0.2800000011920929,
2117
+ "3": 0.20250000059604645,
2118
+ "4": 0.25999999046325684,
2119
+ "5": 0.23749999701976776,
2120
+ "6": 0.25999999046325684,
2121
+ "7": 0.3125,
2122
+ "8": 0.29249998927116394
2123
+ }
2124
+ },
2125
+ {
2126
+ "step": 22500,
2127
+ "mean": 0.2559374924749136,
2128
+ "per_k": {
2129
+ "1": 0.23499999940395355,
2130
+ "2": 0.2549999952316284,
2131
+ "3": 0.2199999988079071,
2132
+ "4": 0.2549999952316284,
2133
+ "5": 0.22749999165534973,
2134
+ "6": 0.26749998331069946,
2135
+ "7": 0.3174999952316284,
2136
+ "8": 0.26999998092651367
2137
+ }
2138
+ },
2139
+ {
2140
+ "step": 25000,
2141
+ "mean": 0.2649999912828207,
2142
+ "per_k": {
2143
+ "1": 0.2824999988079071,
2144
+ "2": 0.28999999165534973,
2145
+ "3": 0.22999998927116394,
2146
+ "4": 0.2524999976158142,
2147
+ "5": 0.25999999046325684,
2148
+ "6": 0.24249999225139618,
2149
+ "7": 0.26999998092651367,
2150
+ "8": 0.29249998927116394
2151
+ }
2152
+ }
2153
+ ],
2154
+ "ood": [
2155
+ {
2156
+ "step": 2500,
2157
+ "acc": {
2158
+ "16": 0.04749999940395355,
2159
+ "32": 0.04999999701976776,
2160
+ "64": 0.029999999329447746
2161
+ }
2162
+ },
2163
+ {
2164
+ "step": 5000,
2165
+ "acc": {
2166
+ "16": 0.044999998062849045,
2167
+ "32": 0.06749999523162842,
2168
+ "64": 0.029999999329447746
2169
+ }
2170
+ },
2171
+ {
2172
+ "step": 7500,
2173
+ "acc": {
2174
+ "16": 0.06749999523162842,
2175
+ "32": 0.027499999850988388,
2176
+ "64": 0.017500000074505806
2177
+ }
2178
+ },
2179
+ {
2180
+ "step": 10000,
2181
+ "acc": {
2182
+ "16": 0.10749999433755875,
2183
+ "32": 0.07000000029802322,
2184
+ "64": 0.03749999776482582
2185
+ }
2186
+ },
2187
+ {
2188
+ "step": 12500,
2189
+ "acc": {
2190
+ "16": 0.06499999761581421,
2191
+ "32": 0.03999999910593033,
2192
+ "64": 0.02499999850988388
2193
+ }
2194
+ },
2195
+ {
2196
+ "step": 15000,
2197
+ "acc": {
2198
+ "16": 0.07499999552965164,
2199
+ "32": 0.054999999701976776,
2200
+ "64": 0.02499999850988388
2201
+ }
2202
+ },
2203
+ {
2204
+ "step": 17500,
2205
+ "acc": {
2206
+ "16": 0.05249999836087227,
2207
+ "32": 0.027499999850988388,
2208
+ "64": 0.02499999850988388
2209
+ }
2210
+ },
2211
+ {
2212
+ "step": 20000,
2213
+ "acc": {
2214
+ "16": 0.044999998062849045,
2215
+ "32": 0.03999999910593033,
2216
+ "64": 0.019999999552965164
2217
+ }
2218
+ },
2219
+ {
2220
+ "step": 22500,
2221
+ "acc": {
2222
+ "16": 0.044999998062849045,
2223
+ "32": 0.03999999910593033,
2224
+ "64": 0.019999999552965164
2225
+ }
2226
+ },
2227
+ {
2228
+ "step": 25000,
2229
+ "acc": {
2230
+ "16": 0.05249999836087227,
2231
+ "32": 0.03749999776482582,
2232
+ "64": 0.029999999329447746
2233
+ }
2234
+ }
2235
+ ],
2236
+ "iters": [],
2237
+ "flops_cum": 4630626158247936.0
2238
+ }