eac123 commited on
Commit
a754306
·
verified ·
1 Parent(s): bfa6c07

Upload results/rec1aligned-seed1_history.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. results/rec1aligned-seed1_history.json +2238 -0
results/rec1aligned-seed1_history.json ADDED
@@ -0,0 +1,2238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "loss": [
3
+ {
4
+ "step": 0,
5
+ "loss": 4.3164,
6
+ "lr": 1e-06,
7
+ "iters": 4,
8
+ "steps_per_s": 0.0,
9
+ "flops_cum": 87407198208.0
10
+ },
11
+ {
12
+ "step": 100,
13
+ "loss": 4.2973,
14
+ "lr": 0.000101,
15
+ "iters": 1,
16
+ "steps_per_s": 18.89,
17
+ "flops_cum": 10299931361280.0
18
+ },
19
+ {
20
+ "step": 200,
21
+ "loss": 4.172,
22
+ "lr": 0.000201,
23
+ "iters": 6,
24
+ "steps_per_s": 18.66,
25
+ "flops_cum": 20663964794880.0
26
+ },
27
+ {
28
+ "step": 300,
29
+ "loss": 4.1684,
30
+ "lr": 0.000301,
31
+ "iters": 5,
32
+ "steps_per_s": 18.65,
33
+ "flops_cum": 30941421502464.0
34
+ },
35
+ {
36
+ "step": 400,
37
+ "loss": 4.1718,
38
+ "lr": 0.00040100000000000004,
39
+ "iters": 6,
40
+ "steps_per_s": 18.74,
41
+ "flops_cum": 41197234028544.0
42
+ },
43
+ {
44
+ "step": 500,
45
+ "loss": 4.162,
46
+ "lr": 0.000501,
47
+ "iters": 3,
48
+ "steps_per_s": 20.62,
49
+ "flops_cum": 50435770023936.0
50
+ },
51
+ {
52
+ "step": 600,
53
+ "loss": 4.1583,
54
+ "lr": 0.000601,
55
+ "iters": 5,
56
+ "steps_per_s": 20.92,
57
+ "flops_cum": 59609373474816.0
58
+ },
59
+ {
60
+ "step": 700,
61
+ "loss": 4.1481,
62
+ "lr": 0.000701,
63
+ "iters": 4,
64
+ "steps_per_s": 20.34,
65
+ "flops_cum": 68999418740736.0
66
+ },
67
+ {
68
+ "step": 800,
69
+ "loss": 4.1476,
70
+ "lr": 0.0008010000000000001,
71
+ "iters": 8,
72
+ "steps_per_s": 19.84,
73
+ "flops_cum": 78735770910720.0
74
+ },
75
+ {
76
+ "step": 900,
77
+ "loss": 4.1479,
78
+ "lr": 0.000901,
79
+ "iters": 5,
80
+ "steps_per_s": 19.99,
81
+ "flops_cum": 88255681265664.0
82
+ },
83
+ {
84
+ "step": 1000,
85
+ "loss": 4.1388,
86
+ "lr": 0.001,
87
+ "iters": 7,
88
+ "steps_per_s": 18.83,
89
+ "flops_cum": 98446561247232.0
90
+ },
91
+ {
92
+ "step": 1100,
93
+ "loss": 4.1351,
94
+ "lr": 0.0009999571637870036,
95
+ "iters": 6,
96
+ "steps_per_s": 19.46,
97
+ "flops_cum": 108247845961728.0
98
+ },
99
+ {
100
+ "step": 1200,
101
+ "loss": 4.137,
102
+ "lr": 0.0009998286624877785,
103
+ "iters": 8,
104
+ "steps_per_s": 20.6,
105
+ "flops_cum": 117572958683136.0
106
+ },
107
+ {
108
+ "step": 1300,
109
+ "loss": 4.1355,
110
+ "lr": 0.0009996145181203615,
111
+ "iters": 3,
112
+ "steps_per_s": 19.32,
113
+ "flops_cum": 127395887579136.0
114
+ },
115
+ {
116
+ "step": 1400,
117
+ "loss": 4.1004,
118
+ "lr": 0.0009993147673772868,
119
+ "iters": 8,
120
+ "steps_per_s": 17.1,
121
+ "flops_cum": 138668976635904.0
122
+ },
123
+ {
124
+ "step": 1500,
125
+ "loss": 4.1117,
126
+ "lr": 0.0009989294616193018,
127
+ "iters": 5,
128
+ "steps_per_s": 19.81,
129
+ "flops_cum": 148253819535360.0
130
+ },
131
+ {
132
+ "step": 1600,
133
+ "loss": 4.1484,
134
+ "lr": 0.000998458666866564,
135
+ "iters": 2,
136
+ "steps_per_s": 19.44,
137
+ "flops_cum": 158055104249856.0
138
+ },
139
+ {
140
+ "step": 1700,
141
+ "loss": 4.1332,
142
+ "lr": 0.0009979024637873308,
143
+ "iters": 5,
144
+ "steps_per_s": 19.05,
145
+ "flops_cum": 168137763323904.0
146
+ },
147
+ {
148
+ "step": 1800,
149
+ "loss": 4.1046,
150
+ "lr": 0.0009972609476841367,
151
+ "iters": 4,
152
+ "steps_per_s": 19.26,
153
+ "flops_cum": 178047268945920.0
154
+ },
155
+ {
156
+ "step": 1900,
157
+ "loss": 4.109,
158
+ "lr": 0.0009965342284774632,
159
+ "iters": 2,
160
+ "steps_per_s": 19.18,
161
+ "flops_cum": 187956774567936.0
162
+ },
163
+ {
164
+ "step": 2000,
165
+ "loss": 4.1057,
166
+ "lr": 0.0009957224306869053,
167
+ "iters": 5,
168
+ "steps_per_s": 19.34,
169
+ "flops_cum": 197887924371456.0
170
+ },
171
+ {
172
+ "step": 2100,
173
+ "loss": 4.1303,
174
+ "lr": 0.0009948256934098352,
175
+ "iters": 4,
176
+ "steps_per_s": 19.43,
177
+ "flops_cum": 207710853267456.0
178
+ },
179
+ {
180
+ "step": 2200,
181
+ "loss": 4.1385,
182
+ "lr": 0.0009938441702975688,
183
+ "iters": 1,
184
+ "steps_per_s": 19.77,
185
+ "flops_cum": 217338984529920.0
186
+ },
187
+ {
188
+ "step": 2300,
189
+ "loss": 4.0868,
190
+ "lr": 0.0009927780295290389,
191
+ "iters": 3,
192
+ "steps_per_s": 20.03,
193
+ "flops_cum": 226923827429376.0
194
+ },
195
+ {
196
+ "step": 2400,
197
+ "loss": 4.0876,
198
+ "lr": 0.0009916274537819774,
199
+ "iters": 3,
200
+ "steps_per_s": 18.12,
201
+ "flops_cum": 237482658496512.0
202
+ },
203
+ {
204
+ "step": 2500,
205
+ "loss": 4.1084,
206
+ "lr": 0.0009903926402016153,
207
+ "iters": 6,
208
+ "steps_per_s": 15.67,
209
+ "flops_cum": 247868336111616.0
210
+ },
211
+ {
212
+ "step": 2600,
213
+ "loss": 4.1035,
214
+ "lr": 0.0009890738003669028,
215
+ "iters": 4,
216
+ "steps_per_s": 17.85,
217
+ "flops_cum": 258427167178752.0
218
+ },
219
+ {
220
+ "step": 2700,
221
+ "loss": 4.0927,
222
+ "lr": 0.0009876711602542563,
223
+ "iters": 3,
224
+ "steps_per_s": 18.26,
225
+ "flops_cum": 268856133156864.0
226
+ },
227
+ {
228
+ "step": 2800,
229
+ "loss": 4.0986,
230
+ "lr": 0.0009861849601988384,
231
+ "iters": 5,
232
+ "steps_per_s": 19.79,
233
+ "flops_cum": 278527552782336.0
234
+ },
235
+ {
236
+ "step": 2900,
237
+ "loss": 4.0873,
238
+ "lr": 0.0009846154548533773,
239
+ "iters": 6,
240
+ "steps_per_s": 17.31,
241
+ "flops_cum": 289584200024064.0
242
+ },
243
+ {
244
+ "step": 3000,
245
+ "loss": 4.1057,
246
+ "lr": 0.0009829629131445341,
247
+ "iters": 6,
248
+ "steps_per_s": 19.73,
249
+ "flops_cum": 299277263831040.0
250
+ },
251
+ {
252
+ "step": 3100,
253
+ "loss": 4.1074,
254
+ "lr": 0.0009812276182268236,
255
+ "iters": 4,
256
+ "steps_per_s": 18.13,
257
+ "flops_cum": 309771162353664.0
258
+ },
259
+ {
260
+ "step": 3200,
261
+ "loss": 4.0956,
262
+ "lr": 0.0009794098674340967,
263
+ "iters": 8,
264
+ "steps_per_s": 20.27,
265
+ "flops_cum": 319269428527104.0
266
+ },
267
+ {
268
+ "step": 3300,
269
+ "loss": 4.1119,
270
+ "lr": 0.0009775099722285933,
271
+ "iters": 8,
272
+ "steps_per_s": 19.02,
273
+ "flops_cum": 329308799238144.0
274
+ },
275
+ {
276
+ "step": 3400,
277
+ "loss": 4.0833,
278
+ "lr": 0.0009755282581475768,
279
+ "iters": 5,
280
+ "steps_per_s": 20.15,
281
+ "flops_cum": 338720488685568.0
282
+ },
283
+ {
284
+ "step": 3500,
285
+ "loss": 4.1036,
286
+ "lr": 0.0009734650647475529,
287
+ "iters": 7,
288
+ "steps_per_s": 19.4,
289
+ "flops_cum": 348608350126080.0
290
+ },
291
+ {
292
+ "step": 3600,
293
+ "loss": 4.0491,
294
+ "lr": 0.0009713207455460893,
295
+ "iters": 7,
296
+ "steps_per_s": 20.32,
297
+ "flops_cum": 358020039573504.0
298
+ },
299
+ {
300
+ "step": 3700,
301
+ "loss": 4.0949,
302
+ "lr": 0.0009690956679612422,
303
+ "iters": 2,
304
+ "steps_per_s": 21.5,
305
+ "flops_cum": 366825691938816.0
306
+ },
307
+ {
308
+ "step": 3800,
309
+ "loss": 4.1456,
310
+ "lr": 0.0009667902132486009,
311
+ "iters": 1,
312
+ "steps_per_s": 18.88,
313
+ "flops_cum": 376929995194368.0
314
+ },
315
+ {
316
+ "step": 3900,
317
+ "loss": 4.0586,
318
+ "lr": 0.0009644047764359622,
319
+ "iters": 8,
320
+ "steps_per_s": 19.83,
321
+ "flops_cum": 386687991545856.0
322
+ },
323
+ {
324
+ "step": 4000,
325
+ "loss": 4.0989,
326
+ "lr": 0.0009619397662556434,
327
+ "iters": 3,
328
+ "steps_per_s": 17.98,
329
+ "flops_cum": 397225178431488.0
330
+ },
331
+ {
332
+ "step": 4100,
333
+ "loss": 4.0982,
334
+ "lr": 0.0009593956050744492,
335
+ "iters": 7,
336
+ "steps_per_s": 20.42,
337
+ "flops_cum": 406636867878912.0
338
+ },
339
+ {
340
+ "step": 4200,
341
+ "loss": 4.0379,
342
+ "lr": 0.0009567727288213005,
343
+ "iters": 4,
344
+ "steps_per_s": 17.71,
345
+ "flops_cum": 417368852398080.0
346
+ },
347
+ {
348
+ "step": 4300,
349
+ "loss": 4.099,
350
+ "lr": 0.0009540715869125407,
351
+ "iters": 1,
352
+ "steps_per_s": 19.39,
353
+ "flops_cum": 427170137112576.0
354
+ },
355
+ {
356
+ "step": 4400,
357
+ "loss": 4.0749,
358
+ "lr": 0.0009512926421749304,
359
+ "iters": 4,
360
+ "steps_per_s": 20.39,
361
+ "flops_cum": 436603470741504.0
362
+ },
363
+ {
364
+ "step": 4500,
365
+ "loss": 4.0495,
366
+ "lr": 0.0009484363707663442,
367
+ "iters": 4,
368
+ "steps_per_s": 17.73,
369
+ "flops_cum": 447357099442176.0
370
+ },
371
+ {
372
+ "step": 4600,
373
+ "loss": 4.0607,
374
+ "lr": 0.0009455032620941839,
375
+ "iters": 2,
376
+ "steps_per_s": 18.61,
377
+ "flops_cum": 457591267786752.0
378
+ },
379
+ {
380
+ "step": 4700,
381
+ "loss": 4.0566,
382
+ "lr": 0.0009424938187315209,
383
+ "iters": 6,
384
+ "steps_per_s": 18.76,
385
+ "flops_cum": 467825436131328.0
386
+ },
387
+ {
388
+ "step": 4800,
389
+ "loss": 4.0209,
390
+ "lr": 0.0009394085563309827,
391
+ "iters": 1,
392
+ "steps_per_s": 18.74,
393
+ "flops_cum": 477929739386880.0
394
+ },
395
+ {
396
+ "step": 4900,
397
+ "loss": 4.066,
398
+ "lr": 0.0009362480035363986,
399
+ "iters": 2,
400
+ "steps_per_s": 19.21,
401
+ "flops_cum": 487904177553408.0
402
+ },
403
+ {
404
+ "step": 5000,
405
+ "loss": 4.0701,
406
+ "lr": 0.0009330127018922195,
407
+ "iters": 4,
408
+ "steps_per_s": 16.46,
409
+ "flops_cum": 497727106449408.0
410
+ },
411
+ {
412
+ "step": 5100,
413
+ "loss": 4.0869,
414
+ "lr": 0.0009297032057507264,
415
+ "iters": 2,
416
+ "steps_per_s": 18.93,
417
+ "flops_cum": 507788121341952.0
418
+ },
419
+ {
420
+ "step": 5200,
421
+ "loss": 4.0414,
422
+ "lr": 0.0009263200821770461,
423
+ "iters": 6,
424
+ "steps_per_s": 19.05,
425
+ "flops_cum": 517892424597504.0
426
+ },
427
+ {
428
+ "step": 5300,
429
+ "loss": 4.0914,
430
+ "lr": 0.0009228639108519867,
431
+ "iters": 4,
432
+ "steps_per_s": 21.63,
433
+ "flops_cum": 526676432781312.0
434
+ },
435
+ {
436
+ "step": 5400,
437
+ "loss": 3.9958,
438
+ "lr": 0.0009193352839727121,
439
+ "iters": 7,
440
+ "steps_per_s": 21.16,
441
+ "flops_cum": 535763459506176.0
442
+ },
443
+ {
444
+ "step": 5500,
445
+ "loss": 4.009,
446
+ "lr": 0.0009157348061512727,
447
+ "iters": 7,
448
+ "steps_per_s": 19.52,
449
+ "flops_cum": 545564744220672.0
450
+ },
451
+ {
452
+ "step": 5600,
453
+ "loss": 4.0409,
454
+ "lr": 0.0009120630943110077,
455
+ "iters": 6,
456
+ "steps_per_s": 19.54,
457
+ "flops_cum": 555322740572160.0
458
+ },
459
+ {
460
+ "step": 5700,
461
+ "loss": 4.0367,
462
+ "lr": 0.0009083207775808396,
463
+ "iters": 4,
464
+ "steps_per_s": 18.03,
465
+ "flops_cum": 565881571639296.0
466
+ },
467
+ {
468
+ "step": 5800,
469
+ "loss": 4.0512,
470
+ "lr": 0.0009045084971874737,
471
+ "iters": 6,
472
+ "steps_per_s": 19.41,
473
+ "flops_cum": 575769433079808.0
474
+ },
475
+ {
476
+ "step": 5900,
477
+ "loss": 4.0534,
478
+ "lr": 0.0009006269063455304,
479
+ "iters": 6,
480
+ "steps_per_s": 19.98,
481
+ "flops_cum": 585332631797760.0
482
+ },
483
+ {
484
+ "step": 6000,
485
+ "loss": 4.0895,
486
+ "lr": 0.0008966766701456176,
487
+ "iters": 1,
488
+ "steps_per_s": 20.51,
489
+ "flops_cum": 594549523611648.0
490
+ },
491
+ {
492
+ "step": 6100,
493
+ "loss": 4.0255,
494
+ "lr": 0.0008926584654403724,
495
+ "iters": 6,
496
+ "steps_per_s": 19.53,
497
+ "flops_cum": 604415740870656.0
498
+ },
499
+ {
500
+ "step": 6200,
501
+ "loss": 4.0339,
502
+ "lr": 0.0008885729807284854,
503
+ "iters": 2,
504
+ "steps_per_s": 18.45,
505
+ "flops_cum": 614693197578240.0
506
+ },
507
+ {
508
+ "step": 6300,
509
+ "loss": 3.9972,
510
+ "lr": 0.0008844209160367298,
511
+ "iters": 7,
512
+ "steps_per_s": 19.36,
513
+ "flops_cum": 624645991563264.0
514
+ },
515
+ {
516
+ "step": 6400,
517
+ "loss": 4.0448,
518
+ "lr": 0.0008802029828000156,
519
+ "iters": 7,
520
+ "steps_per_s": 21.48,
521
+ "flops_cum": 633559864836096.0
522
+ },
523
+ {
524
+ "step": 6500,
525
+ "loss": 3.9994,
526
+ "lr": 0.0008759199037394887,
527
+ "iters": 4,
528
+ "steps_per_s": 19.9,
529
+ "flops_cum": 643101419372544.0
530
+ },
531
+ {
532
+ "step": 6600,
533
+ "loss": 3.9968,
534
+ "lr": 0.0008715724127386971,
535
+ "iters": 7,
536
+ "steps_per_s": 20.42,
537
+ "flops_cum": 652513108819968.0
538
+ },
539
+ {
540
+ "step": 6700,
541
+ "loss": 4.0514,
542
+ "lr": 0.0008671612547178428,
543
+ "iters": 7,
544
+ "steps_per_s": 19.91,
545
+ "flops_cum": 662097951719424.0
546
+ },
547
+ {
548
+ "step": 6800,
549
+ "loss": 4.0594,
550
+ "lr": 0.0008626871855061438,
551
+ "iters": 1,
552
+ "steps_per_s": 18.29,
553
+ "flops_cum": 672440340971520.0
554
+ },
555
+ {
556
+ "step": 6900,
557
+ "loss": 4.0321,
558
+ "lr": 0.0008581509717123273,
559
+ "iters": 5,
560
+ "steps_per_s": 19.16,
561
+ "flops_cum": 682501355864064.0
562
+ },
563
+ {
564
+ "step": 7000,
565
+ "loss": 3.9821,
566
+ "lr": 0.0008535533905932737,
567
+ "iters": 5,
568
+ "steps_per_s": 21.88,
569
+ "flops_cum": 691242075684864.0
570
+ },
571
+ {
572
+ "step": 7100,
573
+ "loss": 3.9897,
574
+ "lr": 0.0008488952299208401,
575
+ "iters": 6,
576
+ "steps_per_s": 18.68,
577
+ "flops_cum": 701476244029440.0
578
+ },
579
+ {
580
+ "step": 7200,
581
+ "loss": 4.0442,
582
+ "lr": 0.000844177287846877,
583
+ "iters": 4,
584
+ "steps_per_s": 19.4,
585
+ "flops_cum": 711299172925440.0
586
+ },
587
+ {
588
+ "step": 7300,
589
+ "loss": 4.0825,
590
+ "lr": 0.0008394003727664709,
591
+ "iters": 1,
592
+ "steps_per_s": 20.24,
593
+ "flops_cum": 720689218191360.0
594
+ },
595
+ {
596
+ "step": 7400,
597
+ "loss": 4.029,
598
+ "lr": 0.0008345653031794292,
599
+ "iters": 4,
600
+ "steps_per_s": 20.27,
601
+ "flops_cum": 730165840183296.0
602
+ },
603
+ {
604
+ "step": 7500,
605
+ "loss": 3.988,
606
+ "lr": 0.0008296729075500344,
607
+ "iters": 6,
608
+ "steps_per_s": 16.39,
609
+ "flops_cum": 740032057442304.0
610
+ },
611
+ {
612
+ "step": 7600,
613
+ "loss": 4.0141,
614
+ "lr": 0.0008247240241650918,
615
+ "iters": 8,
616
+ "steps_per_s": 18.29,
617
+ "flops_cum": 750504311783424.0
618
+ },
619
+ {
620
+ "step": 7700,
621
+ "loss": 3.9687,
622
+ "lr": 0.0008197195009902923,
623
+ "iters": 7,
624
+ "steps_per_s": 18.65,
625
+ "flops_cum": 760716835946496.0
626
+ },
627
+ {
628
+ "step": 7800,
629
+ "loss": 4.0807,
630
+ "lr": 0.0008146601955249188,
631
+ "iters": 1,
632
+ "steps_per_s": 20.93,
633
+ "flops_cum": 769717285945344.0
634
+ },
635
+ {
636
+ "step": 7900,
637
+ "loss": 3.965,
638
+ "lr": 0.0008095469746549171,
639
+ "iters": 3,
640
+ "steps_per_s": 20.57,
641
+ "flops_cum": 779042398666752.0
642
+ },
643
+ {
644
+ "step": 8000,
645
+ "loss": 4.023,
646
+ "lr": 0.0008043807145043603,
647
+ "iters": 1,
648
+ "steps_per_s": 19.75,
649
+ "flops_cum": 788670529929216.0
650
+ },
651
+ {
652
+ "step": 8100,
653
+ "loss": 4.0174,
654
+ "lr": 0.0007991623002853296,
655
+ "iters": 3,
656
+ "steps_per_s": 18.91,
657
+ "flops_cum": 798818121547776.0
658
+ },
659
+ {
660
+ "step": 8200,
661
+ "loss": 3.9885,
662
+ "lr": 0.0007938926261462366,
663
+ "iters": 4,
664
+ "steps_per_s": 19.68,
665
+ "flops_cum": 808532829536256.0
666
+ },
667
+ {
668
+ "step": 8300,
669
+ "loss": 3.9904,
670
+ "lr": 0.0007885725950186169,
671
+ "iters": 3,
672
+ "steps_per_s": 20.58,
673
+ "flops_cum": 817814653894656.0
674
+ },
675
+ {
676
+ "step": 8400,
677
+ "loss": 4.0478,
678
+ "lr": 0.0007832031184624164,
679
+ "iters": 1,
680
+ "steps_per_s": 20.57,
681
+ "flops_cum": 827074834071552.0
682
+ },
683
+ {
684
+ "step": 8500,
685
+ "loss": 3.9514,
686
+ "lr": 0.0007777851165098011,
687
+ "iters": 7,
688
+ "steps_per_s": 20.52,
689
+ "flops_cum": 836486523518976.0
690
+ },
691
+ {
692
+ "step": 8600,
693
+ "loss": 3.9587,
694
+ "lr": 0.0007723195175075137,
695
+ "iters": 4,
696
+ "steps_per_s": 19.5,
697
+ "flops_cum": 846222875688960.0
698
+ },
699
+ {
700
+ "step": 8700,
701
+ "loss": 4.0136,
702
+ "lr": 0.0007668072579578058,
703
+ "iters": 8,
704
+ "steps_per_s": 19.77,
705
+ "flops_cum": 855980872040448.0
706
+ },
707
+ {
708
+ "step": 8800,
709
+ "loss": 3.9401,
710
+ "lr": 0.0007612492823579744,
711
+ "iters": 3,
712
+ "steps_per_s": 19.52,
713
+ "flops_cum": 865673935847424.0
714
+ },
715
+ {
716
+ "step": 8900,
717
+ "loss": 3.9703,
718
+ "lr": 0.0007556465430385259,
719
+ "iters": 5,
720
+ "steps_per_s": 19.17,
721
+ "flops_cum": 875670018195456.0
722
+ },
723
+ {
724
+ "step": 9000,
725
+ "loss": 3.9521,
726
+ "lr": 0.00075,
727
+ "iters": 3,
728
+ "steps_per_s": 19.11,
729
+ "flops_cum": 885622812180480.0
730
+ },
731
+ {
732
+ "step": 9100,
733
+ "loss": 3.9471,
734
+ "lr": 0.0007443106207484776,
735
+ "iters": 7,
736
+ "steps_per_s": 17.24,
737
+ "flops_cum": 896766036148224.0
738
+ },
739
+ {
740
+ "step": 9200,
741
+ "loss": 3.9586,
742
+ "lr": 0.0007385793801298042,
743
+ "iters": 3,
744
+ "steps_per_s": 21.31,
745
+ "flops_cum": 905658265239552.0
746
+ },
747
+ {
748
+ "step": 9300,
749
+ "loss": 3.9576,
750
+ "lr": 0.0007328072601625557,
751
+ "iters": 2,
752
+ "steps_per_s": 19.96,
753
+ "flops_cum": 915199819776000.0
754
+ },
755
+ {
756
+ "step": 9400,
757
+ "loss": 4.0036,
758
+ "lr": 0.0007269952498697733,
759
+ "iters": 7,
760
+ "steps_per_s": 19.26,
761
+ "flops_cum": 925217546305536.0
762
+ },
763
+ {
764
+ "step": 9500,
765
+ "loss": 3.9218,
766
+ "lr": 0.0007211443451095007,
767
+ "iters": 8,
768
+ "steps_per_s": 18.93,
769
+ "flops_cum": 935321849561088.0
770
+ },
771
+ {
772
+ "step": 9600,
773
+ "loss": 3.994,
774
+ "lr": 0.0007152555484041476,
775
+ "iters": 4,
776
+ "steps_per_s": 18.06,
777
+ "flops_cum": 945815748083712.0
778
+ },
779
+ {
780
+ "step": 9700,
781
+ "loss": 3.9517,
782
+ "lr": 0.0007093298687687141,
783
+ "iters": 2,
784
+ "steps_per_s": 19.08,
785
+ "flops_cum": 955790186250240.0
786
+ },
787
+ {
788
+ "step": 9800,
789
+ "loss": 3.9612,
790
+ "lr": 0.0007033683215379002,
791
+ "iters": 2,
792
+ "steps_per_s": 20.61,
793
+ "flops_cum": 965050366427136.0
794
+ },
795
+ {
796
+ "step": 9900,
797
+ "loss": 3.9698,
798
+ "lr": 0.0006973719281921336,
799
+ "iters": 6,
800
+ "steps_per_s": 18.3,
801
+ "flops_cum": 975565909131264.0
802
+ },
803
+ {
804
+ "step": 10000,
805
+ "loss": 3.9438,
806
+ "lr": 0.000691341716182545,
807
+ "iters": 2,
808
+ "steps_per_s": 16.36,
809
+ "flops_cum": 985345549664256.0
810
+ },
811
+ {
812
+ "step": 10100,
813
+ "loss": 4.0785,
814
+ "lr": 0.0006852787187549182,
815
+ "iters": 1,
816
+ "steps_per_s": 18.77,
817
+ "flops_cum": 995493141282816.0
818
+ },
819
+ {
820
+ "step": 10200,
821
+ "loss": 4.0234,
822
+ "lr": 0.0006791839747726501,
823
+ "iters": 1,
824
+ "steps_per_s": 18.99,
825
+ "flops_cum": 1005532511993856.0
826
+ },
827
+ {
828
+ "step": 10300,
829
+ "loss": 3.9488,
830
+ "lr": 0.0006730585285387465,
831
+ "iters": 3,
832
+ "steps_per_s": 18.03,
833
+ "flops_cum": 1016134631424000.0
834
+ },
835
+ {
836
+ "step": 10400,
837
+ "loss": 3.9066,
838
+ "lr": 0.0006669034296168854,
839
+ "iters": 4,
840
+ "steps_per_s": 19.53,
841
+ "flops_cum": 1025935916138496.0
842
+ },
843
+ {
844
+ "step": 10500,
845
+ "loss": 3.9816,
846
+ "lr": 0.0006607197326515808,
847
+ "iters": 3,
848
+ "steps_per_s": 18.29,
849
+ "flops_cum": 1036364882116608.0
850
+ },
851
+ {
852
+ "step": 10600,
853
+ "loss": 3.9927,
854
+ "lr": 0.0006545084971874737,
855
+ "iters": 7,
856
+ "steps_per_s": 19.67,
857
+ "flops_cum": 1046144522649600.0
858
+ },
859
+ {
860
+ "step": 10700,
861
+ "loss": 3.876,
862
+ "lr": 0.0006482707874877854,
863
+ "iters": 3,
864
+ "steps_per_s": 18.86,
865
+ "flops_cum": 1056183893360640.0
866
+ },
867
+ {
868
+ "step": 10800,
869
+ "loss": 3.8905,
870
+ "lr": 0.0006420076723519614,
871
+ "iters": 6,
872
+ "steps_per_s": 19.37,
873
+ "flops_cum": 1066115043164160.0
874
+ },
875
+ {
876
+ "step": 10900,
877
+ "loss": 3.8685,
878
+ "lr": 0.0006357202249325371,
879
+ "iters": 7,
880
+ "steps_per_s": 20.35,
881
+ "flops_cum": 1075505088430080.0
882
+ },
883
+ {
884
+ "step": 11000,
885
+ "loss": 3.8743,
886
+ "lr": 0.0006294095225512603,
887
+ "iters": 6,
888
+ "steps_per_s": 19.26,
889
+ "flops_cum": 1085392949870592.0
890
+ },
891
+ {
892
+ "step": 11100,
893
+ "loss": 3.8232,
894
+ "lr": 0.0006230766465144965,
895
+ "iters": 5,
896
+ "steps_per_s": 18.03,
897
+ "flops_cum": 1095973425119232.0
898
+ },
899
+ {
900
+ "step": 11200,
901
+ "loss": 3.7956,
902
+ "lr": 0.0006167226819279528,
903
+ "iters": 3,
904
+ "steps_per_s": 19.6,
905
+ "flops_cum": 1105688133107712.0
906
+ },
907
+ {
908
+ "step": 11300,
909
+ "loss": 3.8314,
910
+ "lr": 0.0006103487175107507,
911
+ "iters": 7,
912
+ "steps_per_s": 19.04,
913
+ "flops_cum": 1115814080544768.0
914
+ },
915
+ {
916
+ "step": 11400,
917
+ "loss": 3.8648,
918
+ "lr": 0.0006039558454088796,
919
+ "iters": 6,
920
+ "steps_per_s": 18.46,
921
+ "flops_cum": 1126134825615360.0
922
+ },
923
+ {
924
+ "step": 11500,
925
+ "loss": 4.0732,
926
+ "lr": 0.0005975451610080642,
927
+ "iters": 1,
928
+ "steps_per_s": 18.16,
929
+ "flops_cum": 1136563791593472.0
930
+ },
931
+ {
932
+ "step": 11600,
933
+ "loss": 3.8564,
934
+ "lr": 0.0005911177627460738,
935
+ "iters": 5,
936
+ "steps_per_s": 20.02,
937
+ "flops_cum": 1146170278674432.0
938
+ },
939
+ {
940
+ "step": 11700,
941
+ "loss": 3.9098,
942
+ "lr": 0.0005846747519245122,
943
+ "iters": 5,
944
+ "steps_per_s": 20.67,
945
+ "flops_cum": 1155408814669824.0
946
+ },
947
+ {
948
+ "step": 11800,
949
+ "loss": 3.9583,
950
+ "lr": 0.0005782172325201155,
951
+ "iters": 1,
952
+ "steps_per_s": 19.07,
953
+ "flops_cum": 1165339964473344.0
954
+ },
955
+ {
956
+ "step": 11900,
957
+ "loss": 4.0088,
958
+ "lr": 0.0005717463109955896,
959
+ "iters": 1,
960
+ "steps_per_s": 20.21,
961
+ "flops_cum": 1174794942283776.0
962
+ },
963
+ {
964
+ "step": 12000,
965
+ "loss": 3.886,
966
+ "lr": 0.000565263096110026,
967
+ "iters": 3,
968
+ "steps_per_s": 19.4,
969
+ "flops_cum": 1184661159542784.0
970
+ },
971
+ {
972
+ "step": 12100,
973
+ "loss": 3.7801,
974
+ "lr": 0.0005587686987289189,
975
+ "iters": 5,
976
+ "steps_per_s": 18.68,
977
+ "flops_cum": 1194916972068864.0
978
+ },
979
+ {
980
+ "step": 12200,
981
+ "loss": 3.8216,
982
+ "lr": 0.0005522642316338268,
983
+ "iters": 5,
984
+ "steps_per_s": 19.36,
985
+ "flops_cum": 1204761545146368.0
986
+ },
987
+ {
988
+ "step": 12300,
989
+ "loss": 3.7635,
990
+ "lr": 0.0005457508093317013,
991
+ "iters": 7,
992
+ "steps_per_s": 18.2,
993
+ "flops_cum": 1215277087850496.0
994
+ },
995
+ {
996
+ "step": 12400,
997
+ "loss": 4.0494,
998
+ "lr": 0.0005392295478639225,
999
+ "iters": 1,
1000
+ "steps_per_s": 19.66,
1001
+ "flops_cum": 1224883574931456.0
1002
+ },
1003
+ {
1004
+ "step": 12500,
1005
+ "loss": 3.793,
1006
+ "lr": 0.0005327015646150716,
1007
+ "iters": 7,
1008
+ "steps_per_s": 15.14,
1009
+ "flops_cum": 1235767068721152.0
1010
+ },
1011
+ {
1012
+ "step": 12600,
1013
+ "loss": 3.9155,
1014
+ "lr": 0.000526167978121472,
1015
+ "iters": 4,
1016
+ "steps_per_s": 19.92,
1017
+ "flops_cum": 1245308623257600.0
1018
+ },
1019
+ {
1020
+ "step": 12700,
1021
+ "loss": 3.6494,
1022
+ "lr": 0.0005196299078795343,
1023
+ "iters": 6,
1024
+ "steps_per_s": 18.04,
1025
+ "flops_cum": 1255910742687744.0
1026
+ },
1027
+ {
1028
+ "step": 12800,
1029
+ "loss": 3.6399,
1030
+ "lr": 0.0005130884741539367,
1031
+ "iters": 8,
1032
+ "steps_per_s": 18.3,
1033
+ "flops_cum": 1266404641210368.0
1034
+ },
1035
+ {
1036
+ "step": 12900,
1037
+ "loss": 3.5974,
1038
+ "lr": 0.0005065447977856722,
1039
+ "iters": 5,
1040
+ "steps_per_s": 18.58,
1041
+ "flops_cum": 1276617165373440.0
1042
+ },
1043
+ {
1044
+ "step": 13000,
1045
+ "loss": 3.7325,
1046
+ "lr": 0.0005,
1047
+ "iters": 2,
1048
+ "steps_per_s": 18.55,
1049
+ "flops_cum": 1286851333718016.0
1050
+ },
1051
+ {
1052
+ "step": 13100,
1053
+ "loss": 3.7241,
1054
+ "lr": 0.0004934552022143279,
1055
+ "iters": 4,
1056
+ "steps_per_s": 20.21,
1057
+ "flops_cum": 1296327955709952.0
1058
+ },
1059
+ {
1060
+ "step": 13200,
1061
+ "loss": 3.6788,
1062
+ "lr": 0.0004869115258460635,
1063
+ "iters": 8,
1064
+ "steps_per_s": 20.27,
1065
+ "flops_cum": 1305804577701888.0
1066
+ },
1067
+ {
1068
+ "step": 13300,
1069
+ "loss": 3.9883,
1070
+ "lr": 0.00048037009212046586,
1071
+ "iters": 1,
1072
+ "steps_per_s": 18.5,
1073
+ "flops_cum": 1315973813501952.0
1074
+ },
1075
+ {
1076
+ "step": 13400,
1077
+ "loss": 3.7072,
1078
+ "lr": 0.0004738320218785281,
1079
+ "iters": 3,
1080
+ "steps_per_s": 19.81,
1081
+ "flops_cum": 1325645233127424.0
1082
+ },
1083
+ {
1084
+ "step": 13500,
1085
+ "loss": 3.5947,
1086
+ "lr": 0.00046729843538492847,
1087
+ "iters": 4,
1088
+ "steps_per_s": 20.23,
1089
+ "flops_cum": 1335100210937856.0
1090
+ },
1091
+ {
1092
+ "step": 13600,
1093
+ "loss": 4.0122,
1094
+ "lr": 0.0004607704521360776,
1095
+ "iters": 1,
1096
+ "steps_per_s": 18.95,
1097
+ "flops_cum": 1345117937467392.0
1098
+ },
1099
+ {
1100
+ "step": 13700,
1101
+ "loss": 3.7538,
1102
+ "lr": 0.00045424919066829885,
1103
+ "iters": 2,
1104
+ "steps_per_s": 19.59,
1105
+ "flops_cum": 1354897578000384.0
1106
+ },
1107
+ {
1108
+ "step": 13800,
1109
+ "loss": 3.9598,
1110
+ "lr": 0.00044773576836617336,
1111
+ "iters": 1,
1112
+ "steps_per_s": 18.88,
1113
+ "flops_cum": 1365001881255936.0
1114
+ },
1115
+ {
1116
+ "step": 13900,
1117
+ "loss": 3.6141,
1118
+ "lr": 0.00044123130127108126,
1119
+ "iters": 3,
1120
+ "steps_per_s": 19.16,
1121
+ "flops_cum": 1374997963603968.0
1122
+ },
1123
+ {
1124
+ "step": 14000,
1125
+ "loss": 3.5233,
1126
+ "lr": 0.00043473690388997434,
1127
+ "iters": 3,
1128
+ "steps_per_s": 18.39,
1129
+ "flops_cum": 1385361997037568.0
1130
+ },
1131
+ {
1132
+ "step": 14100,
1133
+ "loss": 3.3622,
1134
+ "lr": 0.0004282536890044104,
1135
+ "iters": 8,
1136
+ "steps_per_s": 18.8,
1137
+ "flops_cum": 1395596165382144.0
1138
+ },
1139
+ {
1140
+ "step": 14200,
1141
+ "loss": 3.5642,
1142
+ "lr": 0.0004217827674798845,
1143
+ "iters": 3,
1144
+ "steps_per_s": 20.32,
1145
+ "flops_cum": 1404899633922048.0
1146
+ },
1147
+ {
1148
+ "step": 14300,
1149
+ "loss": 3.2818,
1150
+ "lr": 0.0004153252480754877,
1151
+ "iters": 7,
1152
+ "steps_per_s": 17.4,
1153
+ "flops_cum": 1415934636982272.0
1154
+ },
1155
+ {
1156
+ "step": 14400,
1157
+ "loss": 3.7256,
1158
+ "lr": 0.00040888223725392626,
1159
+ "iters": 2,
1160
+ "steps_per_s": 18.72,
1161
+ "flops_cum": 1426038940237824.0
1162
+ },
1163
+ {
1164
+ "step": 14500,
1165
+ "loss": 3.3756,
1166
+ "lr": 0.00040245483899193594,
1167
+ "iters": 5,
1168
+ "steps_per_s": 18.18,
1169
+ "flops_cum": 1436597771304960.0
1170
+ },
1171
+ {
1172
+ "step": 14600,
1173
+ "loss": 3.9495,
1174
+ "lr": 0.0003960441545911204,
1175
+ "iters": 1,
1176
+ "steps_per_s": 19.3,
1177
+ "flops_cum": 1446420700200960.0
1178
+ },
1179
+ {
1180
+ "step": 14700,
1181
+ "loss": 3.9462,
1182
+ "lr": 0.0003896512824892495,
1183
+ "iters": 1,
1184
+ "steps_per_s": 19.69,
1185
+ "flops_cum": 1456135408189440.0
1186
+ },
1187
+ {
1188
+ "step": 14800,
1189
+ "loss": 3.251,
1190
+ "lr": 0.00038327731807204744,
1191
+ "iters": 4,
1192
+ "steps_per_s": 18.85,
1193
+ "flops_cum": 1466304643989504.0
1194
+ },
1195
+ {
1196
+ "step": 14900,
1197
+ "loss": 3.2128,
1198
+ "lr": 0.0003769233534855035,
1199
+ "iters": 4,
1200
+ "steps_per_s": 18.73,
1201
+ "flops_cum": 1476473879789568.0
1202
+ },
1203
+ {
1204
+ "step": 15000,
1205
+ "loss": 3.1312,
1206
+ "lr": 0.0003705904774487396,
1207
+ "iters": 8,
1208
+ "steps_per_s": 17.17,
1209
+ "flops_cum": 1485842280873984.0
1210
+ },
1211
+ {
1212
+ "step": 15100,
1213
+ "loss": 3.009,
1214
+ "lr": 0.0003642797750674629,
1215
+ "iters": 6,
1216
+ "steps_per_s": 18.18,
1217
+ "flops_cum": 1496292891033600.0
1218
+ },
1219
+ {
1220
+ "step": 15200,
1221
+ "loss": 3.0092,
1222
+ "lr": 0.0003579923276480387,
1223
+ "iters": 5,
1224
+ "steps_per_s": 19.37,
1225
+ "flops_cum": 1506137464111104.0
1226
+ },
1227
+ {
1228
+ "step": 15300,
1229
+ "loss": 2.7843,
1230
+ "lr": 0.0003517292125122146,
1231
+ "iters": 7,
1232
+ "steps_per_s": 18.11,
1233
+ "flops_cum": 1516717939359744.0
1234
+ },
1235
+ {
1236
+ "step": 15400,
1237
+ "loss": 3.6386,
1238
+ "lr": 0.00034549150281252633,
1239
+ "iters": 2,
1240
+ "steps_per_s": 20.03,
1241
+ "flops_cum": 1526151272988672.0
1242
+ },
1243
+ {
1244
+ "step": 15500,
1245
+ "loss": 3.4773,
1246
+ "lr": 0.0003392802673484193,
1247
+ "iters": 3,
1248
+ "steps_per_s": 20.12,
1249
+ "flops_cum": 1535649539162112.0
1250
+ },
1251
+ {
1252
+ "step": 15600,
1253
+ "loss": 3.085,
1254
+ "lr": 0.00033309657038311456,
1255
+ "iters": 5,
1256
+ "steps_per_s": 19.05,
1257
+ "flops_cum": 1545710554054656.0
1258
+ },
1259
+ {
1260
+ "step": 15700,
1261
+ "loss": 3.0234,
1262
+ "lr": 0.0003269414714612534,
1263
+ "iters": 6,
1264
+ "steps_per_s": 19.65,
1265
+ "flops_cum": 1555446906224640.0
1266
+ },
1267
+ {
1268
+ "step": 15800,
1269
+ "loss": 3.9224,
1270
+ "lr": 0.00032081602522734986,
1271
+ "iters": 1,
1272
+ "steps_per_s": 19.21,
1273
+ "flops_cum": 1565291479302144.0
1274
+ },
1275
+ {
1276
+ "step": 15900,
1277
+ "loss": 3.3305,
1278
+ "lr": 0.00031472128124508187,
1279
+ "iters": 3,
1280
+ "steps_per_s": 17.82,
1281
+ "flops_cum": 1576023463821312.0
1282
+ },
1283
+ {
1284
+ "step": 16000,
1285
+ "loss": 3.5702,
1286
+ "lr": 0.0003086582838174551,
1287
+ "iters": 2,
1288
+ "steps_per_s": 19.16,
1289
+ "flops_cum": 1585976257806336.0
1290
+ },
1291
+ {
1292
+ "step": 16100,
1293
+ "loss": 3.9055,
1294
+ "lr": 0.00030262807180786645,
1295
+ "iters": 1,
1296
+ "steps_per_s": 20.46,
1297
+ "flops_cum": 1595279726346240.0
1298
+ },
1299
+ {
1300
+ "step": 16200,
1301
+ "loss": 3.3701,
1302
+ "lr": 0.0002966316784621,
1303
+ "iters": 3,
1304
+ "steps_per_s": 19.73,
1305
+ "flops_cum": 1604994434334720.0
1306
+ },
1307
+ {
1308
+ "step": 16300,
1309
+ "loss": 3.355,
1310
+ "lr": 0.0002906701312312861,
1311
+ "iters": 3,
1312
+ "steps_per_s": 19.5,
1313
+ "flops_cum": 1614774074867712.0
1314
+ },
1315
+ {
1316
+ "step": 16400,
1317
+ "loss": 2.6766,
1318
+ "lr": 0.0002847444515958523,
1319
+ "iters": 8,
1320
+ "steps_per_s": 19.21,
1321
+ "flops_cum": 1624813445578752.0
1322
+ },
1323
+ {
1324
+ "step": 16500,
1325
+ "loss": 3.6282,
1326
+ "lr": 0.00027885565489049947,
1327
+ "iters": 2,
1328
+ "steps_per_s": 19.76,
1329
+ "flops_cum": 1634376644296704.0
1330
+ },
1331
+ {
1332
+ "step": 16600,
1333
+ "loss": 3.9924,
1334
+ "lr": 0.00027300475013022663,
1335
+ "iters": 1,
1336
+ "steps_per_s": 19.87,
1337
+ "flops_cum": 1643961487196160.0
1338
+ },
1339
+ {
1340
+ "step": 16700,
1341
+ "loss": 2.9149,
1342
+ "lr": 0.0002671927398374443,
1343
+ "iters": 6,
1344
+ "steps_per_s": 18.8,
1345
+ "flops_cum": 1654217299722240.0
1346
+ },
1347
+ {
1348
+ "step": 16800,
1349
+ "loss": 3.2221,
1350
+ "lr": 0.00026142061987019576,
1351
+ "iters": 3,
1352
+ "steps_per_s": 21.44,
1353
+ "flops_cum": 1663066240450560.0
1354
+ },
1355
+ {
1356
+ "step": 16900,
1357
+ "loss": 3.3278,
1358
+ "lr": 0.0002556893792515227,
1359
+ "iters": 3,
1360
+ "steps_per_s": 19.05,
1361
+ "flops_cum": 1673083966980096.0
1362
+ },
1363
+ {
1364
+ "step": 17000,
1365
+ "loss": 3.666,
1366
+ "lr": 0.0002500000000000001,
1367
+ "iters": 2,
1368
+ "steps_per_s": 18.8,
1369
+ "flops_cum": 1683231558598656.0
1370
+ },
1371
+ {
1372
+ "step": 17100,
1373
+ "loss": 3.9606,
1374
+ "lr": 0.00024435345696147403,
1375
+ "iters": 1,
1376
+ "steps_per_s": 21.14,
1377
+ "flops_cum": 1692253652779008.0
1378
+ },
1379
+ {
1380
+ "step": 17200,
1381
+ "loss": 3.4525,
1382
+ "lr": 0.00023875071764202561,
1383
+ "iters": 3,
1384
+ "steps_per_s": 19.69,
1385
+ "flops_cum": 1701990004948992.0
1386
+ },
1387
+ {
1388
+ "step": 17300,
1389
+ "loss": 3.438,
1390
+ "lr": 0.00023319274204219425,
1391
+ "iters": 2,
1392
+ "steps_per_s": 20.53,
1393
+ "flops_cum": 1711271829307392.0
1394
+ },
1395
+ {
1396
+ "step": 17400,
1397
+ "loss": 3.9758,
1398
+ "lr": 0.00022768048249248646,
1399
+ "iters": 1,
1400
+ "steps_per_s": 20.16,
1401
+ "flops_cum": 1720705162936320.0
1402
+ },
1403
+ {
1404
+ "step": 17500,
1405
+ "loss": 3.9489,
1406
+ "lr": 0.00022221488349019903,
1407
+ "iters": 1,
1408
+ "steps_per_s": 16.92,
1409
+ "flops_cum": 1730160140746752.0
1410
+ },
1411
+ {
1412
+ "step": 17600,
1413
+ "loss": 3.5673,
1414
+ "lr": 0.0002167968815375837,
1415
+ "iters": 2,
1416
+ "steps_per_s": 19.46,
1417
+ "flops_cum": 1739874848735232.0
1418
+ },
1419
+ {
1420
+ "step": 17700,
1421
+ "loss": 3.8691,
1422
+ "lr": 0.00021142740498138323,
1423
+ "iters": 1,
1424
+ "steps_per_s": 18.4,
1425
+ "flops_cum": 1750238882168832.0
1426
+ },
1427
+ {
1428
+ "step": 17800,
1429
+ "loss": 3.003,
1430
+ "lr": 0.00020610737385376348,
1431
+ "iters": 5,
1432
+ "steps_per_s": 19.77,
1433
+ "flops_cum": 1759975234338816.0
1434
+ },
1435
+ {
1436
+ "step": 17900,
1437
+ "loss": 2.7005,
1438
+ "lr": 0.00020083769971467047,
1439
+ "iters": 6,
1440
+ "steps_per_s": 19.38,
1441
+ "flops_cum": 1769841451597824.0
1442
+ },
1443
+ {
1444
+ "step": 18000,
1445
+ "loss": 2.9404,
1446
+ "lr": 0.00019561928549563967,
1447
+ "iters": 4,
1448
+ "steps_per_s": 19.35,
1449
+ "flops_cum": 1779664380493824.0
1450
+ },
1451
+ {
1452
+ "step": 18100,
1453
+ "loss": 2.6544,
1454
+ "lr": 0.00019045302534508295,
1455
+ "iters": 8,
1456
+ "steps_per_s": 20.17,
1457
+ "flops_cum": 1789205935030272.0
1458
+ },
1459
+ {
1460
+ "step": 18200,
1461
+ "loss": 2.6968,
1462
+ "lr": 0.00018533980447508135,
1463
+ "iters": 8,
1464
+ "steps_per_s": 18.59,
1465
+ "flops_cum": 1799440103374848.0
1466
+ },
1467
+ {
1468
+ "step": 18300,
1469
+ "loss": 3.9863,
1470
+ "lr": 0.00018028049900970766,
1471
+ "iters": 1,
1472
+ "steps_per_s": 19.71,
1473
+ "flops_cum": 1809024946274304.0
1474
+ },
1475
+ {
1476
+ "step": 18400,
1477
+ "loss": 2.6604,
1478
+ "lr": 0.00017527597583490823,
1479
+ "iters": 6,
1480
+ "steps_per_s": 19.48,
1481
+ "flops_cum": 1818912807714816.0
1482
+ },
1483
+ {
1484
+ "step": 18500,
1485
+ "loss": 2.8787,
1486
+ "lr": 0.00017032709244996558,
1487
+ "iters": 5,
1488
+ "steps_per_s": 18.77,
1489
+ "flops_cum": 1829060399333376.0
1490
+ },
1491
+ {
1492
+ "step": 18600,
1493
+ "loss": 3.8638,
1494
+ "lr": 0.00016543469682057105,
1495
+ "iters": 1,
1496
+ "steps_per_s": 19.69,
1497
+ "flops_cum": 1838688530595840.0
1498
+ },
1499
+ {
1500
+ "step": 18700,
1501
+ "loss": 3.4884,
1502
+ "lr": 0.0001605996272335291,
1503
+ "iters": 2,
1504
+ "steps_per_s": 19.16,
1505
+ "flops_cum": 1848662968762368.0
1506
+ },
1507
+ {
1508
+ "step": 18800,
1509
+ "loss": 3.2018,
1510
+ "lr": 0.00015582271215312294,
1511
+ "iters": 3,
1512
+ "steps_per_s": 18.3,
1513
+ "flops_cum": 1859113578921984.0
1514
+ },
1515
+ {
1516
+ "step": 18900,
1517
+ "loss": 3.1893,
1518
+ "lr": 0.00015110477007916002,
1519
+ "iters": 3,
1520
+ "steps_per_s": 20.59,
1521
+ "flops_cum": 1868395403280384.0
1522
+ },
1523
+ {
1524
+ "step": 19000,
1525
+ "loss": 3.8906,
1526
+ "lr": 0.00014644660940672628,
1527
+ "iters": 1,
1528
+ "steps_per_s": 19.66,
1529
+ "flops_cum": 1878088467087360.0
1530
+ },
1531
+ {
1532
+ "step": 19100,
1533
+ "loss": 3.037,
1534
+ "lr": 0.00014184902828767287,
1535
+ "iters": 5,
1536
+ "steps_per_s": 21.02,
1537
+ "flops_cum": 1887240426356736.0
1538
+ },
1539
+ {
1540
+ "step": 19200,
1541
+ "loss": 2.7155,
1542
+ "lr": 0.0001373128144938563,
1543
+ "iters": 6,
1544
+ "steps_per_s": 19.19,
1545
+ "flops_cum": 1897193220341760.0
1546
+ },
1547
+ {
1548
+ "step": 19300,
1549
+ "loss": 2.5989,
1550
+ "lr": 0.00013283874528215734,
1551
+ "iters": 6,
1552
+ "steps_per_s": 20.9,
1553
+ "flops_cum": 1906323535429632.0
1554
+ },
1555
+ {
1556
+ "step": 19400,
1557
+ "loss": 3.5312,
1558
+ "lr": 0.00012842758726130281,
1559
+ "iters": 2,
1560
+ "steps_per_s": 19.12,
1561
+ "flops_cum": 1916233041051648.0
1562
+ },
1563
+ {
1564
+ "step": 19500,
1565
+ "loss": 2.5595,
1566
+ "lr": 0.00012408009626051135,
1567
+ "iters": 7,
1568
+ "steps_per_s": 20.43,
1569
+ "flops_cum": 1925644730499072.0
1570
+ },
1571
+ {
1572
+ "step": 19600,
1573
+ "loss": 2.7189,
1574
+ "lr": 0.00011979701719998454,
1575
+ "iters": 6,
1576
+ "steps_per_s": 20.37,
1577
+ "flops_cum": 1935013131583488.0
1578
+ },
1579
+ {
1580
+ "step": 19700,
1581
+ "loss": 2.5329,
1582
+ "lr": 0.00011557908396327027,
1583
+ "iters": 8,
1584
+ "steps_per_s": 18.4,
1585
+ "flops_cum": 1945420453380096.0
1586
+ },
1587
+ {
1588
+ "step": 19800,
1589
+ "loss": 2.8145,
1590
+ "lr": 0.00011142701927151455,
1591
+ "iters": 6,
1592
+ "steps_per_s": 19.44,
1593
+ "flops_cum": 1955221738094592.0
1594
+ },
1595
+ {
1596
+ "step": 19900,
1597
+ "loss": 3.0458,
1598
+ "lr": 0.00010734153455962764,
1599
+ "iters": 4,
1600
+ "steps_per_s": 21.16,
1601
+ "flops_cum": 1964200543911936.0
1602
+ },
1603
+ {
1604
+ "step": 20000,
1605
+ "loss": 3.8803,
1606
+ "lr": 0.00010332332985438247,
1607
+ "iters": 1,
1608
+ "steps_per_s": 16.6,
1609
+ "flops_cum": 1973828675174400.0
1610
+ },
1611
+ {
1612
+ "step": 20100,
1613
+ "loss": 3.9402,
1614
+ "lr": 9.937309365446973e-05,
1615
+ "iters": 1,
1616
+ "steps_per_s": 19.23,
1617
+ "flops_cum": 1983759824977920.0
1618
+ },
1619
+ {
1620
+ "step": 20200,
1621
+ "loss": 3.8446,
1622
+ "lr": 9.549150281252633e-05,
1623
+ "iters": 1,
1624
+ "steps_per_s": 19.41,
1625
+ "flops_cum": 1993604398055424.0
1626
+ },
1627
+ {
1628
+ "step": 20300,
1629
+ "loss": 2.8567,
1630
+ "lr": 9.167922241916055e-05,
1631
+ "iters": 5,
1632
+ "steps_per_s": 20.23,
1633
+ "flops_cum": 2003124308410368.0
1634
+ },
1635
+ {
1636
+ "step": 20400,
1637
+ "loss": 2.6889,
1638
+ "lr": 8.793690568899215e-05,
1639
+ "iters": 6,
1640
+ "steps_per_s": 17.5,
1641
+ "flops_cum": 2014051090563072.0
1642
+ },
1643
+ {
1644
+ "step": 20500,
1645
+ "loss": 2.3658,
1646
+ "lr": 8.426519384872733e-05,
1647
+ "iters": 8,
1648
+ "steps_per_s": 21.23,
1649
+ "flops_cum": 2023094828924928.0
1650
+ },
1651
+ {
1652
+ "step": 20600,
1653
+ "loss": 2.5851,
1654
+ "lr": 8.066471602728804e-05,
1655
+ "iters": 7,
1656
+ "steps_per_s": 18.66,
1657
+ "flops_cum": 2033307353088000.0
1658
+ },
1659
+ {
1660
+ "step": 20700,
1661
+ "loss": 3.7993,
1662
+ "lr": 7.71360891480134e-05,
1663
+ "iters": 1,
1664
+ "steps_per_s": 20.08,
1665
+ "flops_cum": 2042719042535424.0
1666
+ },
1667
+ {
1668
+ "step": 20800,
1669
+ "loss": 2.4288,
1670
+ "lr": 7.367991782295391e-05,
1671
+ "iters": 8,
1672
+ "steps_per_s": 19.69,
1673
+ "flops_cum": 2052541971431424.0
1674
+ },
1675
+ {
1676
+ "step": 20900,
1677
+ "loss": 2.4809,
1678
+ "lr": 7.029679424927366e-05,
1679
+ "iters": 7,
1680
+ "steps_per_s": 20.17,
1681
+ "flops_cum": 2061975305060352.0
1682
+ },
1683
+ {
1684
+ "step": 21000,
1685
+ "loss": 3.8681,
1686
+ "lr": 6.698729810778065e-05,
1687
+ "iters": 1,
1688
+ "steps_per_s": 18.07,
1689
+ "flops_cum": 2072447559401472.0
1690
+ },
1691
+ {
1692
+ "step": 21100,
1693
+ "loss": 2.6276,
1694
+ "lr": 6.375199646360142e-05,
1695
+ "iters": 6,
1696
+ "steps_per_s": 19.45,
1697
+ "flops_cum": 2082335420841984.0
1698
+ },
1699
+ {
1700
+ "step": 21200,
1701
+ "loss": 3.8835,
1702
+ "lr": 6.059144366901737e-05,
1703
+ "iters": 1,
1704
+ "steps_per_s": 20.03,
1705
+ "flops_cum": 2091790398652416.0
1706
+ },
1707
+ {
1708
+ "step": 21300,
1709
+ "loss": 2.8633,
1710
+ "lr": 5.750618126847912e-05,
1711
+ "iters": 4,
1712
+ "steps_per_s": 18.78,
1713
+ "flops_cum": 2102002922815488.0
1714
+ },
1715
+ {
1716
+ "step": 21400,
1717
+ "loss": 2.6484,
1718
+ "lr": 5.449673790581611e-05,
1719
+ "iters": 7,
1720
+ "steps_per_s": 20.05,
1721
+ "flops_cum": 2111566121533440.0
1722
+ },
1723
+ {
1724
+ "step": 21500,
1725
+ "loss": 2.6082,
1726
+ "lr": 5.1563629233655876e-05,
1727
+ "iters": 8,
1728
+ "steps_per_s": 18.85,
1729
+ "flops_cum": 2121692068970496.0
1730
+ },
1731
+ {
1732
+ "step": 21600,
1733
+ "loss": 2.5068,
1734
+ "lr": 4.87073578250698e-05,
1735
+ "iters": 8,
1736
+ "steps_per_s": 19.5,
1737
+ "flops_cum": 2131471709503488.0
1738
+ },
1739
+ {
1740
+ "step": 21700,
1741
+ "loss": 2.9866,
1742
+ "lr": 4.592841308745932e-05,
1743
+ "iters": 4,
1744
+ "steps_per_s": 17.56,
1745
+ "flops_cum": 2142268626567168.0
1746
+ },
1747
+ {
1748
+ "step": 21800,
1749
+ "loss": 2.8883,
1750
+ "lr": 4.322727117869951e-05,
1751
+ "iters": 5,
1752
+ "steps_per_s": 20.02,
1753
+ "flops_cum": 2151831825285120.0
1754
+ },
1755
+ {
1756
+ "step": 21900,
1757
+ "loss": 2.8526,
1758
+ "lr": 4.06043949255509e-05,
1759
+ "iters": 6,
1760
+ "steps_per_s": 20.19,
1761
+ "flops_cum": 2161286803095552.0
1762
+ },
1763
+ {
1764
+ "step": 22000,
1765
+ "loss": 2.615,
1766
+ "lr": 3.806023374435663e-05,
1767
+ "iters": 7,
1768
+ "steps_per_s": 18.41,
1769
+ "flops_cum": 2171672480710656.0
1770
+ },
1771
+ {
1772
+ "step": 22100,
1773
+ "loss": 2.7584,
1774
+ "lr": 3.559522356403788e-05,
1775
+ "iters": 5,
1776
+ "steps_per_s": 19.85,
1777
+ "flops_cum": 2181257323610112.0
1778
+ },
1779
+ {
1780
+ "step": 22200,
1781
+ "loss": 2.7618,
1782
+ "lr": 3.3209786751399184e-05,
1783
+ "iters": 6,
1784
+ "steps_per_s": 20.33,
1785
+ "flops_cum": 2190669013057536.0
1786
+ },
1787
+ {
1788
+ "step": 22300,
1789
+ "loss": 2.5266,
1790
+ "lr": 3.0904332038757974e-05,
1791
+ "iters": 7,
1792
+ "steps_per_s": 18.83,
1793
+ "flops_cum": 2200816604676096.0
1794
+ },
1795
+ {
1796
+ "step": 22400,
1797
+ "loss": 3.5668,
1798
+ "lr": 2.8679254453910786e-05,
1799
+ "iters": 2,
1800
+ "steps_per_s": 19.17,
1801
+ "flops_cum": 2210682821935104.0
1802
+ },
1803
+ {
1804
+ "step": 22500,
1805
+ "loss": 2.9036,
1806
+ "lr": 2.653493525244721e-05,
1807
+ "iters": 4,
1808
+ "steps_per_s": 16.65,
1809
+ "flops_cum": 2220354241560576.0
1810
+ },
1811
+ {
1812
+ "step": 22600,
1813
+ "loss": 2.6906,
1814
+ "lr": 2.4471741852423235e-05,
1815
+ "iters": 7,
1816
+ "steps_per_s": 20.46,
1817
+ "flops_cum": 2229744286826496.0
1818
+ },
1819
+ {
1820
+ "step": 22700,
1821
+ "loss": 3.0348,
1822
+ "lr": 2.2490027771406685e-05,
1823
+ "iters": 4,
1824
+ "steps_per_s": 20.6,
1825
+ "flops_cum": 2238961178640384.0
1826
+ },
1827
+ {
1828
+ "step": 22800,
1829
+ "loss": 3.9221,
1830
+ "lr": 2.0590132565903473e-05,
1831
+ "iters": 1,
1832
+ "steps_per_s": 19.49,
1833
+ "flops_cum": 2248697530810368.0
1834
+ },
1835
+ {
1836
+ "step": 22900,
1837
+ "loss": 2.8979,
1838
+ "lr": 1.8772381773176416e-05,
1839
+ "iters": 4,
1840
+ "steps_per_s": 20.12,
1841
+ "flops_cum": 2258239085346816.0
1842
+ },
1843
+ {
1844
+ "step": 23000,
1845
+ "loss": 2.537,
1846
+ "lr": 1.70370868554659e-05,
1847
+ "iters": 8,
1848
+ "steps_per_s": 19.49,
1849
+ "flops_cum": 2268105302605824.0
1850
+ },
1851
+ {
1852
+ "step": 23100,
1853
+ "loss": 3.3298,
1854
+ "lr": 1.538454514662285e-05,
1855
+ "iters": 2,
1856
+ "steps_per_s": 20.47,
1857
+ "flops_cum": 2277322194419712.0
1858
+ },
1859
+ {
1860
+ "step": 23200,
1861
+ "loss": 3.3401,
1862
+ "lr": 1.3815039801161721e-05,
1863
+ "iters": 3,
1864
+ "steps_per_s": 20.88,
1865
+ "flops_cum": 2286495797870592.0
1866
+ },
1867
+ {
1868
+ "step": 23300,
1869
+ "loss": 2.9028,
1870
+ "lr": 1.232883974574367e-05,
1871
+ "iters": 4,
1872
+ "steps_per_s": 19.35,
1873
+ "flops_cum": 2296383659311104.0
1874
+ },
1875
+ {
1876
+ "step": 23400,
1877
+ "loss": 2.3637,
1878
+ "lr": 1.0926199633097156e-05,
1879
+ "iters": 7,
1880
+ "steps_per_s": 20.93,
1881
+ "flops_cum": 2305557262761984.0
1882
+ },
1883
+ {
1884
+ "step": 23500,
1885
+ "loss": 2.4128,
1886
+ "lr": 9.607359798384786e-06,
1887
+ "iters": 8,
1888
+ "steps_per_s": 19.17,
1889
+ "flops_cum": 2315531700928512.0
1890
+ },
1891
+ {
1892
+ "step": 23600,
1893
+ "loss": 2.9314,
1894
+ "lr": 8.372546218022748e-06,
1895
+ "iters": 4,
1896
+ "steps_per_s": 19.86,
1897
+ "flops_cum": 2325073255464960.0
1898
+ },
1899
+ {
1900
+ "step": 23700,
1901
+ "loss": 3.8145,
1902
+ "lr": 7.221970470961125e-06,
1903
+ "iters": 1,
1904
+ "steps_per_s": 18.77,
1905
+ "flops_cum": 2335199202902016.0
1906
+ },
1907
+ {
1908
+ "step": 23800,
1909
+ "loss": 3.5397,
1910
+ "lr": 6.15582970243117e-06,
1911
+ "iters": 2,
1912
+ "steps_per_s": 20.19,
1913
+ "flops_cum": 2344675824893952.0
1914
+ },
1915
+ {
1916
+ "step": 23900,
1917
+ "loss": 2.6324,
1918
+ "lr": 5.174306590164879e-06,
1919
+ "iters": 6,
1920
+ "steps_per_s": 20.27,
1921
+ "flops_cum": 2354174091067392.0
1922
+ },
1923
+ {
1924
+ "step": 24000,
1925
+ "loss": 3.2826,
1926
+ "lr": 4.277569313094809e-06,
1927
+ "iters": 3,
1928
+ "steps_per_s": 18.88,
1929
+ "flops_cum": 2364235105959936.0
1930
+ },
1931
+ {
1932
+ "step": 24100,
1933
+ "loss": 2.674,
1934
+ "lr": 3.4657715225368535e-06,
1935
+ "iters": 6,
1936
+ "steps_per_s": 17.81,
1937
+ "flops_cum": 2375010378842112.0
1938
+ },
1939
+ {
1940
+ "step": 24200,
1941
+ "loss": 2.7635,
1942
+ "lr": 2.739052315863355e-06,
1943
+ "iters": 6,
1944
+ "steps_per_s": 18.6,
1945
+ "flops_cum": 2385266191368192.0
1946
+ },
1947
+ {
1948
+ "step": 24300,
1949
+ "loss": 3.9369,
1950
+ "lr": 2.097536212669171e-06,
1951
+ "iters": 1,
1952
+ "steps_per_s": 18.26,
1953
+ "flops_cum": 2395651868983296.0
1954
+ },
1955
+ {
1956
+ "step": 24400,
1957
+ "loss": 2.6477,
1958
+ "lr": 1.541333133436018e-06,
1959
+ "iters": 6,
1960
+ "steps_per_s": 19.32,
1961
+ "flops_cum": 2405626307149824.0
1962
+ },
1963
+ {
1964
+ "step": 24500,
1965
+ "loss": 2.5986,
1966
+ "lr": 1.0705383806982606e-06,
1967
+ "iters": 6,
1968
+ "steps_per_s": 17.88,
1969
+ "flops_cum": 2416293359124480.0
1970
+ },
1971
+ {
1972
+ "step": 24600,
1973
+ "loss": 2.5269,
1974
+ "lr": 6.852326227130834e-07,
1975
+ "iters": 8,
1976
+ "steps_per_s": 18.44,
1977
+ "flops_cum": 2426679036739584.0
1978
+ },
1979
+ {
1980
+ "step": 24700,
1981
+ "loss": 3.8773,
1982
+ "lr": 3.854818796385495e-07,
1983
+ "iters": 1,
1984
+ "steps_per_s": 19.71,
1985
+ "flops_cum": 2436242235457536.0
1986
+ },
1987
+ {
1988
+ "step": 24800,
1989
+ "loss": 2.8471,
1990
+ "lr": 1.7133751222137007e-07,
1991
+ "iters": 5,
1992
+ "steps_per_s": 20.03,
1993
+ "flops_cum": 2445827078356992.0
1994
+ },
1995
+ {
1996
+ "step": 24900,
1997
+ "loss": 3.8757,
1998
+ "lr": 4.283621299649987e-08,
1999
+ "iters": 1,
2000
+ "steps_per_s": 19.61,
2001
+ "flops_cum": 2455498497982464.0
2002
+ },
2003
+ {
2004
+ "step": 24999,
2005
+ "loss": 2.5893,
2006
+ "lr": 4.283682464301109e-12,
2007
+ "iters": 7,
2008
+ "steps_per_s": 17.98,
2009
+ "flops_cum": 2466208007847936.0
2010
+ }
2011
+ ],
2012
+ "id_acc": [
2013
+ {
2014
+ "step": 2500,
2015
+ "mean": 0.03906249883584678,
2016
+ "per_k": {
2017
+ "1": 0.03500000014901161,
2018
+ "2": 0.03999999910593033,
2019
+ "3": 0.019999999552965164,
2020
+ "4": 0.057499997317790985,
2021
+ "5": 0.042500000447034836,
2022
+ "6": 0.04999999701976776,
2023
+ "7": 0.017500000074505806,
2024
+ "8": 0.04999999701976776
2025
+ }
2026
+ },
2027
+ {
2028
+ "step": 5000,
2029
+ "mean": 0.045312499394640326,
2030
+ "per_k": {
2031
+ "1": 0.04749999940395355,
2032
+ "2": 0.03999999910593033,
2033
+ "3": 0.042500000447034836,
2034
+ "4": 0.042500000447034836,
2035
+ "5": 0.029999999329447746,
2036
+ "6": 0.04999999701976776,
2037
+ "7": 0.04749999940395355,
2038
+ "8": 0.0625
2039
+ }
2040
+ },
2041
+ {
2042
+ "step": 7500,
2043
+ "mean": 0.052499998826533556,
2044
+ "per_k": {
2045
+ "1": 0.05249999836087227,
2046
+ "2": 0.042500000447034836,
2047
+ "3": 0.044999998062849045,
2048
+ "4": 0.054999999701976776,
2049
+ "5": 0.05249999836087227,
2050
+ "6": 0.07249999791383743,
2051
+ "7": 0.03749999776482582,
2052
+ "8": 0.0625
2053
+ }
2054
+ },
2055
+ {
2056
+ "step": 10000,
2057
+ "mean": 0.0637499988079071,
2058
+ "per_k": {
2059
+ "1": 0.042500000447034836,
2060
+ "2": 0.07000000029802322,
2061
+ "3": 0.04749999940395355,
2062
+ "4": 0.04749999940395355,
2063
+ "5": 0.05999999865889549,
2064
+ "6": 0.08999999612569809,
2065
+ "7": 0.06499999761581421,
2066
+ "8": 0.08749999850988388
2067
+ }
2068
+ },
2069
+ {
2070
+ "step": 12500,
2071
+ "mean": 0.06968749826774001,
2072
+ "per_k": {
2073
+ "1": 0.04999999701976776,
2074
+ "2": 0.044999998062849045,
2075
+ "3": 0.05999999865889549,
2076
+ "4": 0.05249999836087227,
2077
+ "5": 0.07999999821186066,
2078
+ "6": 0.07750000059604645,
2079
+ "7": 0.10999999940395355,
2080
+ "8": 0.08249999582767487
2081
+ }
2082
+ },
2083
+ {
2084
+ "step": 15000,
2085
+ "mean": 0.08531249780207872,
2086
+ "per_k": {
2087
+ "1": 0.04999999701976776,
2088
+ "2": 0.04999999701976776,
2089
+ "3": 0.06499999761581421,
2090
+ "4": 0.06749999523162842,
2091
+ "5": 0.08500000089406967,
2092
+ "6": 0.07750000059604645,
2093
+ "7": 0.1525000035762787,
2094
+ "8": 0.13499999046325684
2095
+ }
2096
+ },
2097
+ {
2098
+ "step": 17500,
2099
+ "mean": 0.09749999735504389,
2100
+ "per_k": {
2101
+ "1": 0.054999999701976776,
2102
+ "2": 0.04999999701976776,
2103
+ "3": 0.057499997317790985,
2104
+ "4": 0.07999999821186066,
2105
+ "5": 0.11499999463558197,
2106
+ "6": 0.08999999612569809,
2107
+ "7": 0.1599999964237213,
2108
+ "8": 0.17249999940395355
2109
+ }
2110
+ },
2111
+ {
2112
+ "step": 20000,
2113
+ "mean": 0.10812499793246388,
2114
+ "per_k": {
2115
+ "1": 0.057499997317790985,
2116
+ "2": 0.07249999791383743,
2117
+ "3": 0.05249999836087227,
2118
+ "4": 0.08500000089406967,
2119
+ "5": 0.125,
2120
+ "6": 0.08999999612569809,
2121
+ "7": 0.17000000178813934,
2122
+ "8": 0.21249999105930328
2123
+ }
2124
+ },
2125
+ {
2126
+ "step": 22500,
2127
+ "mean": 0.10406249901279807,
2128
+ "per_k": {
2129
+ "1": 0.05249999836087227,
2130
+ "2": 0.054999999701976776,
2131
+ "3": 0.054999999701976776,
2132
+ "4": 0.08500000089406967,
2133
+ "5": 0.10499999672174454,
2134
+ "6": 0.11999999731779099,
2135
+ "7": 0.1875,
2136
+ "8": 0.17249999940395355
2137
+ }
2138
+ },
2139
+ {
2140
+ "step": 25000,
2141
+ "mean": 0.09968749713152647,
2142
+ "per_k": {
2143
+ "1": 0.06749999523162842,
2144
+ "2": 0.03999999910593033,
2145
+ "3": 0.07499999552965164,
2146
+ "4": 0.07249999791383743,
2147
+ "5": 0.10999999940395355,
2148
+ "6": 0.11499999463558197,
2149
+ "7": 0.17499999701976776,
2150
+ "8": 0.14249999821186066
2151
+ }
2152
+ }
2153
+ ],
2154
+ "ood": [
2155
+ {
2156
+ "step": 2500,
2157
+ "acc": {
2158
+ "16": 0.027499999850988388,
2159
+ "32": 0.02499999850988388,
2160
+ "64": 0.029999999329447746
2161
+ }
2162
+ },
2163
+ {
2164
+ "step": 5000,
2165
+ "acc": {
2166
+ "16": 0.04749999940395355,
2167
+ "32": 0.032499998807907104,
2168
+ "64": 0.02499999850988388
2169
+ }
2170
+ },
2171
+ {
2172
+ "step": 7500,
2173
+ "acc": {
2174
+ "16": 0.057499997317790985,
2175
+ "32": 0.027499999850988388,
2176
+ "64": 0.03500000014901161
2177
+ }
2178
+ },
2179
+ {
2180
+ "step": 10000,
2181
+ "acc": {
2182
+ "16": 0.029999999329447746,
2183
+ "32": 0.042500000447034836,
2184
+ "64": 0.029999999329447746
2185
+ }
2186
+ },
2187
+ {
2188
+ "step": 12500,
2189
+ "acc": {
2190
+ "16": 0.07249999791383743,
2191
+ "32": 0.032499998807907104,
2192
+ "64": 0.02499999850988388
2193
+ }
2194
+ },
2195
+ {
2196
+ "step": 15000,
2197
+ "acc": {
2198
+ "16": 0.07249999791383743,
2199
+ "32": 0.019999999552965164,
2200
+ "64": 0.03500000014901161
2201
+ }
2202
+ },
2203
+ {
2204
+ "step": 17500,
2205
+ "acc": {
2206
+ "16": 0.03999999910593033,
2207
+ "32": 0.044999998062849045,
2208
+ "64": 0.042500000447034836
2209
+ }
2210
+ },
2211
+ {
2212
+ "step": 20000,
2213
+ "acc": {
2214
+ "16": 0.042500000447034836,
2215
+ "32": 0.03749999776482582,
2216
+ "64": 0.017500000074505806
2217
+ }
2218
+ },
2219
+ {
2220
+ "step": 22500,
2221
+ "acc": {
2222
+ "16": 0.05249999836087227,
2223
+ "32": 0.017500000074505806,
2224
+ "64": 0.022499999031424522
2225
+ }
2226
+ },
2227
+ {
2228
+ "step": 25000,
2229
+ "acc": {
2230
+ "16": 0.042500000447034836,
2231
+ "32": 0.032499998807907104,
2232
+ "64": 0.027499999850988388
2233
+ }
2234
+ }
2235
+ ],
2236
+ "iters": [],
2237
+ "flops_cum": 2466208007847936.0
2238
+ }