eac123 commited on
Commit
76b4b9e
·
verified ·
1 Parent(s): cb60673

Upload results/rec1aligned-seed0_history.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. results/rec1aligned-seed0_history.json +2238 -0
results/rec1aligned-seed0_history.json ADDED
@@ -0,0 +1,2238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "loss": [
3
+ {
4
+ "step": 0,
5
+ "loss": 4.4968,
6
+ "lr": 1e-06,
7
+ "iters": 7,
8
+ "steps_per_s": 0.0,
9
+ "flops_cum": 152339742720.0
10
+ },
11
+ {
12
+ "step": 100,
13
+ "loss": 4.2236,
14
+ "lr": 0.000101,
15
+ "iters": 8,
16
+ "steps_per_s": 22.47,
17
+ "flops_cum": 9109501378560.0
18
+ },
19
+ {
20
+ "step": 200,
21
+ "loss": 4.1624,
22
+ "lr": 0.000201,
23
+ "iters": 4,
24
+ "steps_per_s": 21.75,
25
+ "flops_cum": 18131595558912.0
26
+ },
27
+ {
28
+ "step": 300,
29
+ "loss": 4.1767,
30
+ "lr": 0.000301,
31
+ "iters": 1,
32
+ "steps_per_s": 20.03,
33
+ "flops_cum": 27889591910400.0
34
+ },
35
+ {
36
+ "step": 400,
37
+ "loss": 4.1823,
38
+ "lr": 0.00040100000000000004,
39
+ "iters": 3,
40
+ "steps_per_s": 21.91,
41
+ "flops_cum": 36890041909248.0
42
+ },
43
+ {
44
+ "step": 500,
45
+ "loss": 4.1778,
46
+ "lr": 0.000501,
47
+ "iters": 6,
48
+ "steps_per_s": 20.29,
49
+ "flops_cum": 46583105716224.0
50
+ },
51
+ {
52
+ "step": 600,
53
+ "loss": 4.1599,
54
+ "lr": 0.000601,
55
+ "iters": 3,
56
+ "steps_per_s": 20.23,
57
+ "flops_cum": 56189592797184.0
58
+ },
59
+ {
60
+ "step": 700,
61
+ "loss": 4.1526,
62
+ "lr": 0.000701,
63
+ "iters": 2,
64
+ "steps_per_s": 21.03,
65
+ "flops_cum": 65428128792576.0
66
+ },
67
+ {
68
+ "step": 800,
69
+ "loss": 4.1515,
70
+ "lr": 0.0008010000000000001,
71
+ "iters": 4,
72
+ "steps_per_s": 20.37,
73
+ "flops_cum": 75012971692032.0
74
+ },
75
+ {
76
+ "step": 900,
77
+ "loss": 4.1595,
78
+ "lr": 0.000901,
79
+ "iters": 1,
80
+ "steps_per_s": 21.03,
81
+ "flops_cum": 84208219324416.0
82
+ },
83
+ {
84
+ "step": 1000,
85
+ "loss": 4.1358,
86
+ "lr": 0.001,
87
+ "iters": 7,
88
+ "steps_per_s": 20.97,
89
+ "flops_cum": 93576620408832.0
90
+ },
91
+ {
92
+ "step": 1100,
93
+ "loss": 4.148,
94
+ "lr": 0.0009999571637870036,
95
+ "iters": 2,
96
+ "steps_per_s": 20.57,
97
+ "flops_cum": 102901733130240.0
98
+ },
99
+ {
100
+ "step": 1200,
101
+ "loss": 4.1603,
102
+ "lr": 0.0009998286624877785,
103
+ "iters": 4,
104
+ "steps_per_s": 20.45,
105
+ "flops_cum": 112421643485184.0
106
+ },
107
+ {
108
+ "step": 1300,
109
+ "loss": 4.1415,
110
+ "lr": 0.0009996145181203615,
111
+ "iters": 4,
112
+ "steps_per_s": 19.51,
113
+ "flops_cum": 122331149107200.0
114
+ },
115
+ {
116
+ "step": 1400,
117
+ "loss": 4.1368,
118
+ "lr": 0.0009993147673772868,
119
+ "iters": 3,
120
+ "steps_per_s": 21.08,
121
+ "flops_cum": 131461464195072.0
122
+ },
123
+ {
124
+ "step": 1500,
125
+ "loss": 4.1426,
126
+ "lr": 0.0009989294616193018,
127
+ "iters": 5,
128
+ "steps_per_s": 19.41,
129
+ "flops_cum": 141435902361600.0
130
+ },
131
+ {
132
+ "step": 1600,
133
+ "loss": 4.1368,
134
+ "lr": 0.000998458666866564,
135
+ "iters": 4,
136
+ "steps_per_s": 18.33,
137
+ "flops_cum": 151929800884224.0
138
+ },
139
+ {
140
+ "step": 1700,
141
+ "loss": 4.1346,
142
+ "lr": 0.0009979024637873308,
143
+ "iters": 7,
144
+ "steps_per_s": 19.83,
145
+ "flops_cum": 161709441417216.0
146
+ },
147
+ {
148
+ "step": 1800,
149
+ "loss": 4.1504,
150
+ "lr": 0.0009972609476841367,
151
+ "iters": 8,
152
+ "steps_per_s": 19.51,
153
+ "flops_cum": 171575658676224.0
154
+ },
155
+ {
156
+ "step": 1900,
157
+ "loss": 4.1199,
158
+ "lr": 0.0009965342284774632,
159
+ "iters": 3,
160
+ "steps_per_s": 20.15,
161
+ "flops_cum": 181052280668160.0
162
+ },
163
+ {
164
+ "step": 2000,
165
+ "loss": 4.1709,
166
+ "lr": 0.0009957224306869053,
167
+ "iters": 5,
168
+ "steps_per_s": 21.2,
169
+ "flops_cum": 190160951574528.0
170
+ },
171
+ {
172
+ "step": 2100,
173
+ "loss": 4.099,
174
+ "lr": 0.0009948256934098352,
175
+ "iters": 2,
176
+ "steps_per_s": 18.88,
177
+ "flops_cum": 200286899011584.0
178
+ },
179
+ {
180
+ "step": 2200,
181
+ "loss": 4.129,
182
+ "lr": 0.0009938441702975688,
183
+ "iters": 3,
184
+ "steps_per_s": 21.5,
185
+ "flops_cum": 209244060647424.0
186
+ },
187
+ {
188
+ "step": 2300,
189
+ "loss": 4.12,
190
+ "lr": 0.0009927780295290389,
191
+ "iters": 4,
192
+ "steps_per_s": 20.86,
193
+ "flops_cum": 218460952461312.0
194
+ },
195
+ {
196
+ "step": 2400,
197
+ "loss": 4.1226,
198
+ "lr": 0.0009916274537819774,
199
+ "iters": 1,
200
+ "steps_per_s": 20.76,
201
+ "flops_cum": 227656200093696.0
202
+ },
203
+ {
204
+ "step": 2500,
205
+ "loss": 4.1144,
206
+ "lr": 0.0009903926402016153,
207
+ "iters": 7,
208
+ "steps_per_s": 15.75,
209
+ "flops_cum": 238063521890304.0
210
+ },
211
+ {
212
+ "step": 2600,
213
+ "loss": 4.0884,
214
+ "lr": 0.0009890738003669028,
215
+ "iters": 2,
216
+ "steps_per_s": 20.01,
217
+ "flops_cum": 247561788063744.0
218
+ },
219
+ {
220
+ "step": 2700,
221
+ "loss": 4.1036,
222
+ "lr": 0.0009876711602542563,
223
+ "iters": 4,
224
+ "steps_per_s": 19.59,
225
+ "flops_cum": 257384716959744.0
226
+ },
227
+ {
228
+ "step": 2800,
229
+ "loss": 4.137,
230
+ "lr": 0.0009861849601988384,
231
+ "iters": 1,
232
+ "steps_per_s": 19.55,
233
+ "flops_cum": 267142713311232.0
234
+ },
235
+ {
236
+ "step": 2900,
237
+ "loss": 4.1281,
238
+ "lr": 0.0009846154548533773,
239
+ "iters": 2,
240
+ "steps_per_s": 20.25,
241
+ "flops_cum": 276619335303168.0
242
+ },
243
+ {
244
+ "step": 3000,
245
+ "loss": 4.0741,
246
+ "lr": 0.0009829629131445341,
247
+ "iters": 8,
248
+ "steps_per_s": 17.82,
249
+ "flops_cum": 287481184911360.0
250
+ },
251
+ {
252
+ "step": 3100,
253
+ "loss": 4.1497,
254
+ "lr": 0.0009812276182268236,
255
+ "iters": 1,
256
+ "steps_per_s": 18.63,
257
+ "flops_cum": 297650420711424.0
258
+ },
259
+ {
260
+ "step": 3200,
261
+ "loss": 4.0708,
262
+ "lr": 0.0009794098674340967,
263
+ "iters": 5,
264
+ "steps_per_s": 19.39,
265
+ "flops_cum": 307603214696448.0
266
+ },
267
+ {
268
+ "step": 3300,
269
+ "loss": 4.0821,
270
+ "lr": 0.0009775099722285933,
271
+ "iters": 5,
272
+ "steps_per_s": 19.59,
273
+ "flops_cum": 317361211047936.0
274
+ },
275
+ {
276
+ "step": 3400,
277
+ "loss": 4.0902,
278
+ "lr": 0.0009755282581475768,
279
+ "iters": 3,
280
+ "steps_per_s": 18.16,
281
+ "flops_cum": 327876753752064.0
282
+ },
283
+ {
284
+ "step": 3500,
285
+ "loss": 4.1018,
286
+ "lr": 0.0009734650647475529,
287
+ "iters": 5,
288
+ "steps_per_s": 19.06,
289
+ "flops_cum": 337959412826112.0
290
+ },
291
+ {
292
+ "step": 3600,
293
+ "loss": 4.0462,
294
+ "lr": 0.0009713207455460893,
295
+ "iters": 2,
296
+ "steps_per_s": 20.14,
297
+ "flops_cum": 347414390636544.0
298
+ },
299
+ {
300
+ "step": 3700,
301
+ "loss": 4.0654,
302
+ "lr": 0.0009690956679612422,
303
+ "iters": 3,
304
+ "steps_per_s": 18.36,
305
+ "flops_cum": 357865000796160.0
306
+ },
307
+ {
308
+ "step": 3800,
309
+ "loss": 4.0881,
310
+ "lr": 0.0009667902132486009,
311
+ "iters": 7,
312
+ "steps_per_s": 19.71,
313
+ "flops_cum": 367644641329152.0
314
+ },
315
+ {
316
+ "step": 3900,
317
+ "loss": 4.1025,
318
+ "lr": 0.0009644047764359622,
319
+ "iters": 5,
320
+ "steps_per_s": 20.44,
321
+ "flops_cum": 376948109869056.0
322
+ },
323
+ {
324
+ "step": 4000,
325
+ "loss": 4.0951,
326
+ "lr": 0.0009619397662556434,
327
+ "iters": 3,
328
+ "steps_per_s": 16.8,
329
+ "flops_cum": 388307775651840.0
330
+ },
331
+ {
332
+ "step": 4100,
333
+ "loss": 4.0831,
334
+ "lr": 0.0009593956050744492,
335
+ "iters": 6,
336
+ "steps_per_s": 19.36,
337
+ "flops_cum": 398238925455360.0
338
+ },
339
+ {
340
+ "step": 4200,
341
+ "loss": 4.1035,
342
+ "lr": 0.0009567727288213005,
343
+ "iters": 3,
344
+ "steps_per_s": 21.46,
345
+ "flops_cum": 407109510365184.0
346
+ },
347
+ {
348
+ "step": 4300,
349
+ "loss": 4.0991,
350
+ "lr": 0.0009540715869125407,
351
+ "iters": 4,
352
+ "steps_per_s": 18.34,
353
+ "flops_cum": 417538476343296.0
354
+ },
355
+ {
356
+ "step": 4400,
357
+ "loss": 4.0795,
358
+ "lr": 0.0009512926421749304,
359
+ "iters": 3,
360
+ "steps_per_s": 19.47,
361
+ "flops_cum": 427361405239296.0
362
+ },
363
+ {
364
+ "step": 4500,
365
+ "loss": 4.0499,
366
+ "lr": 0.0009484363707663442,
367
+ "iters": 4,
368
+ "steps_per_s": 20.39,
369
+ "flops_cum": 436751450505216.0
370
+ },
371
+ {
372
+ "step": 4600,
373
+ "loss": 4.0936,
374
+ "lr": 0.0009455032620941839,
375
+ "iters": 4,
376
+ "steps_per_s": 21.43,
377
+ "flops_cum": 445665323778048.0
378
+ },
379
+ {
380
+ "step": 4700,
381
+ "loss": 4.091,
382
+ "lr": 0.0009424938187315209,
383
+ "iters": 5,
384
+ "steps_per_s": 19.58,
385
+ "flops_cum": 455444964311040.0
386
+ },
387
+ {
388
+ "step": 4800,
389
+ "loss": 4.1009,
390
+ "lr": 0.0009394085563309827,
391
+ "iters": 4,
392
+ "steps_per_s": 18.19,
393
+ "flops_cum": 465938862833664.0
394
+ },
395
+ {
396
+ "step": 4900,
397
+ "loss": 4.0964,
398
+ "lr": 0.0009362480035363986,
399
+ "iters": 8,
400
+ "steps_per_s": 18.65,
401
+ "flops_cum": 476259607904256.0
402
+ },
403
+ {
404
+ "step": 5000,
405
+ "loss": 4.0828,
406
+ "lr": 0.0009330127018922195,
407
+ "iters": 3,
408
+ "steps_per_s": 17.13,
409
+ "flops_cum": 485476499718144.0
410
+ },
411
+ {
412
+ "step": 5100,
413
+ "loss": 4.0779,
414
+ "lr": 0.0009297032057507264,
415
+ "iters": 8,
416
+ "steps_per_s": 20.15,
417
+ "flops_cum": 495039698436096.0
418
+ },
419
+ {
420
+ "step": 5200,
421
+ "loss": 4.0546,
422
+ "lr": 0.0009263200821770461,
423
+ "iters": 3,
424
+ "steps_per_s": 20.02,
425
+ "flops_cum": 504494676246528.0
426
+ },
427
+ {
428
+ "step": 5300,
429
+ "loss": 4.061,
430
+ "lr": 0.0009228639108519867,
431
+ "iters": 8,
432
+ "steps_per_s": 19.82,
433
+ "flops_cum": 514231028416512.0
434
+ },
435
+ {
436
+ "step": 5400,
437
+ "loss": 4.0849,
438
+ "lr": 0.0009193352839727121,
439
+ "iters": 3,
440
+ "steps_per_s": 18.39,
441
+ "flops_cum": 524530129305600.0
442
+ },
443
+ {
444
+ "step": 5500,
445
+ "loss": 4.0939,
446
+ "lr": 0.0009157348061512727,
447
+ "iters": 5,
448
+ "steps_per_s": 21.67,
449
+ "flops_cum": 533400714215424.0
450
+ },
451
+ {
452
+ "step": 5600,
453
+ "loss": 4.049,
454
+ "lr": 0.0009120630943110077,
455
+ "iters": 4,
456
+ "steps_per_s": 18.54,
457
+ "flops_cum": 543699815104512.0
458
+ },
459
+ {
460
+ "step": 5700,
461
+ "loss": 4.0777,
462
+ "lr": 0.0009083207775808396,
463
+ "iters": 3,
464
+ "steps_per_s": 18.68,
465
+ "flops_cum": 553912339267584.0
466
+ },
467
+ {
468
+ "step": 5800,
469
+ "loss": 4.0612,
470
+ "lr": 0.0009045084971874737,
471
+ "iters": 8,
472
+ "steps_per_s": 21.91,
473
+ "flops_cum": 562717991632896.0
474
+ },
475
+ {
476
+ "step": 5900,
477
+ "loss": 4.0492,
478
+ "lr": 0.0009006269063455304,
479
+ "iters": 6,
480
+ "steps_per_s": 19.81,
481
+ "flops_cum": 572324478713856.0
482
+ },
483
+ {
484
+ "step": 6000,
485
+ "loss": 4.0491,
486
+ "lr": 0.0008966766701456176,
487
+ "iters": 8,
488
+ "steps_per_s": 20.12,
489
+ "flops_cum": 581866033250304.0
490
+ },
491
+ {
492
+ "step": 6100,
493
+ "loss": 4.0309,
494
+ "lr": 0.0008926584654403724,
495
+ "iters": 6,
496
+ "steps_per_s": 19.72,
497
+ "flops_cum": 591515808694272.0
498
+ },
499
+ {
500
+ "step": 6200,
501
+ "loss": 4.0181,
502
+ "lr": 0.0008885729807284854,
503
+ "iters": 2,
504
+ "steps_per_s": 19.14,
505
+ "flops_cum": 601425314316288.0
506
+ },
507
+ {
508
+ "step": 6300,
509
+ "loss": 4.0572,
510
+ "lr": 0.0008844209160367298,
511
+ "iters": 4,
512
+ "steps_per_s": 18.81,
513
+ "flops_cum": 611637838479360.0
514
+ },
515
+ {
516
+ "step": 6400,
517
+ "loss": 4.0495,
518
+ "lr": 0.0008802029828000156,
519
+ "iters": 5,
520
+ "steps_per_s": 19.06,
521
+ "flops_cum": 621677209190400.0
522
+ },
523
+ {
524
+ "step": 6500,
525
+ "loss": 4.1307,
526
+ "lr": 0.0008759199037394887,
527
+ "iters": 7,
528
+ "steps_per_s": 19.18,
529
+ "flops_cum": 631673291538432.0
530
+ },
531
+ {
532
+ "step": 6600,
533
+ "loss": 4.0637,
534
+ "lr": 0.0008715724127386971,
535
+ "iters": 6,
536
+ "steps_per_s": 19.01,
537
+ "flops_cum": 641712662249472.0
538
+ },
539
+ {
540
+ "step": 6700,
541
+ "loss": 4.0774,
542
+ "lr": 0.0008671612547178428,
543
+ "iters": 8,
544
+ "steps_per_s": 18.44,
545
+ "flops_cum": 652098339864576.0
546
+ },
547
+ {
548
+ "step": 6800,
549
+ "loss": 4.0351,
550
+ "lr": 0.0008626871855061438,
551
+ "iters": 4,
552
+ "steps_per_s": 18.37,
553
+ "flops_cum": 662419084935168.0
554
+ },
555
+ {
556
+ "step": 6900,
557
+ "loss": 4.017,
558
+ "lr": 0.0008581509717123273,
559
+ "iters": 7,
560
+ "steps_per_s": 19.22,
561
+ "flops_cum": 672415167283200.0
562
+ },
563
+ {
564
+ "step": 7000,
565
+ "loss": 4.0311,
566
+ "lr": 0.0008535533905932737,
567
+ "iters": 4,
568
+ "steps_per_s": 19.38,
569
+ "flops_cum": 682216451997696.0
570
+ },
571
+ {
572
+ "step": 7100,
573
+ "loss": 4.0468,
574
+ "lr": 0.0008488952299208401,
575
+ "iters": 6,
576
+ "steps_per_s": 19.42,
577
+ "flops_cum": 692082669256704.0
578
+ },
579
+ {
580
+ "step": 7200,
581
+ "loss": 4.1043,
582
+ "lr": 0.000844177287846877,
583
+ "iters": 1,
584
+ "steps_per_s": 19.48,
585
+ "flops_cum": 701819021426688.0
586
+ },
587
+ {
588
+ "step": 7300,
589
+ "loss": 4.0615,
590
+ "lr": 0.0008394003727664709,
591
+ "iters": 8,
592
+ "steps_per_s": 17.81,
593
+ "flops_cum": 712680871034880.0
594
+ },
595
+ {
596
+ "step": 7400,
597
+ "loss": 4.0447,
598
+ "lr": 0.0008345653031794292,
599
+ "iters": 4,
600
+ "steps_per_s": 18.99,
601
+ "flops_cum": 722676953382912.0
602
+ },
603
+ {
604
+ "step": 7500,
605
+ "loss": 3.9877,
606
+ "lr": 0.0008296729075500344,
607
+ "iters": 2,
608
+ "steps_per_s": 15.79,
609
+ "flops_cum": 732911121727488.0
610
+ },
611
+ {
612
+ "step": 7600,
613
+ "loss": 4.0334,
614
+ "lr": 0.0008247240241650918,
615
+ "iters": 4,
616
+ "steps_per_s": 19.95,
617
+ "flops_cum": 742517608808448.0
618
+ },
619
+ {
620
+ "step": 7700,
621
+ "loss": 4.0311,
622
+ "lr": 0.0008197195009902923,
623
+ "iters": 7,
624
+ "steps_per_s": 21.0,
625
+ "flops_cum": 751669568077824.0
626
+ },
627
+ {
628
+ "step": 7800,
629
+ "loss": 4.0932,
630
+ "lr": 0.0008146601955249188,
631
+ "iters": 1,
632
+ "steps_per_s": 20.2,
633
+ "flops_cum": 761016324980736.0
634
+ },
635
+ {
636
+ "step": 7900,
637
+ "loss": 4.031,
638
+ "lr": 0.0008095469746549171,
639
+ "iters": 8,
640
+ "steps_per_s": 20.45,
641
+ "flops_cum": 770492946972672.0
642
+ },
643
+ {
644
+ "step": 8000,
645
+ "loss": 4.018,
646
+ "lr": 0.0008043807145043603,
647
+ "iters": 2,
648
+ "steps_per_s": 18.59,
649
+ "flops_cum": 780640538591232.0
650
+ },
651
+ {
652
+ "step": 8100,
653
+ "loss": 4.0136,
654
+ "lr": 0.0007991623002853296,
655
+ "iters": 4,
656
+ "steps_per_s": 20.32,
657
+ "flops_cum": 790073872220160.0
658
+ },
659
+ {
660
+ "step": 8200,
661
+ "loss": 4.0234,
662
+ "lr": 0.0007938926261462366,
663
+ "iters": 8,
664
+ "steps_per_s": 19.68,
665
+ "flops_cum": 799853512753152.0
666
+ },
667
+ {
668
+ "step": 8300,
669
+ "loss": 4.0548,
670
+ "lr": 0.0007885725950186169,
671
+ "iters": 2,
672
+ "steps_per_s": 19.67,
673
+ "flops_cum": 809459999834112.0
674
+ },
675
+ {
676
+ "step": 8400,
677
+ "loss": 3.9545,
678
+ "lr": 0.0007832031184624164,
679
+ "iters": 6,
680
+ "steps_per_s": 19.64,
681
+ "flops_cum": 819261284548608.0
682
+ },
683
+ {
684
+ "step": 8500,
685
+ "loss": 3.9897,
686
+ "lr": 0.0007777851165098011,
687
+ "iters": 4,
688
+ "steps_per_s": 19.14,
689
+ "flops_cum": 829192434352128.0
690
+ },
691
+ {
692
+ "step": 8600,
693
+ "loss": 3.9551,
694
+ "lr": 0.0007723195175075137,
695
+ "iters": 5,
696
+ "steps_per_s": 20.89,
697
+ "flops_cum": 838344393621504.0
698
+ },
699
+ {
700
+ "step": 8700,
701
+ "loss": 4.026,
702
+ "lr": 0.0007668072579578058,
703
+ "iters": 7,
704
+ "steps_per_s": 19.63,
705
+ "flops_cum": 848102389972992.0
706
+ },
707
+ {
708
+ "step": 8800,
709
+ "loss": 4.0672,
710
+ "lr": 0.0007612492823579744,
711
+ "iters": 2,
712
+ "steps_per_s": 19.4,
713
+ "flops_cum": 857838742142976.0
714
+ },
715
+ {
716
+ "step": 8900,
717
+ "loss": 4.0454,
718
+ "lr": 0.0007556465430385259,
719
+ "iters": 8,
720
+ "steps_per_s": 19.24,
721
+ "flops_cum": 867878112854016.0
722
+ },
723
+ {
724
+ "step": 9000,
725
+ "loss": 4.033,
726
+ "lr": 0.00075,
727
+ "iters": 5,
728
+ "steps_per_s": 19.92,
729
+ "flops_cum": 877398023208960.0
730
+ },
731
+ {
732
+ "step": 9100,
733
+ "loss": 3.9884,
734
+ "lr": 0.0007443106207484776,
735
+ "iters": 3,
736
+ "steps_per_s": 20.09,
737
+ "flops_cum": 886874645200896.0
738
+ },
739
+ {
740
+ "step": 9200,
741
+ "loss": 4.0537,
742
+ "lr": 0.0007385793801298042,
743
+ "iters": 4,
744
+ "steps_per_s": 18.01,
745
+ "flops_cum": 897498408812544.0
746
+ },
747
+ {
748
+ "step": 9300,
749
+ "loss": 4.1042,
750
+ "lr": 0.0007328072601625557,
751
+ "iters": 1,
752
+ "steps_per_s": 18.68,
753
+ "flops_cum": 907667644612608.0
754
+ },
755
+ {
756
+ "step": 9400,
757
+ "loss": 4.0827,
758
+ "lr": 0.0007269952498697733,
759
+ "iters": 1,
760
+ "steps_per_s": 19.42,
761
+ "flops_cum": 917490573508608.0
762
+ },
763
+ {
764
+ "step": 9500,
765
+ "loss": 4.0842,
766
+ "lr": 0.0007211443451095007,
767
+ "iters": 1,
768
+ "steps_per_s": 18.42,
769
+ "flops_cum": 927854606942208.0
770
+ },
771
+ {
772
+ "step": 9600,
773
+ "loss": 3.986,
774
+ "lr": 0.0007152555484041476,
775
+ "iters": 4,
776
+ "steps_per_s": 19.96,
777
+ "flops_cum": 937482738204672.0
778
+ },
779
+ {
780
+ "step": 9700,
781
+ "loss": 4.0275,
782
+ "lr": 0.0007093298687687141,
783
+ "iters": 2,
784
+ "steps_per_s": 19.17,
785
+ "flops_cum": 947392243826688.0
786
+ },
787
+ {
788
+ "step": 9800,
789
+ "loss": 4.0827,
790
+ "lr": 0.0007033683215379002,
791
+ "iters": 1,
792
+ "steps_per_s": 20.43,
793
+ "flops_cum": 956717356548096.0
794
+ },
795
+ {
796
+ "step": 9900,
797
+ "loss": 4.0087,
798
+ "lr": 0.0006973719281921336,
799
+ "iters": 5,
800
+ "steps_per_s": 20.41,
801
+ "flops_cum": 966150690177024.0
802
+ },
803
+ {
804
+ "step": 10000,
805
+ "loss": 4.0149,
806
+ "lr": 0.000691341716182545,
807
+ "iters": 7,
808
+ "steps_per_s": 16.44,
809
+ "flops_cum": 975973619073024.0
810
+ },
811
+ {
812
+ "step": 10100,
813
+ "loss": 4.0795,
814
+ "lr": 0.0006852787187549182,
815
+ "iters": 1,
816
+ "steps_per_s": 19.85,
817
+ "flops_cum": 985471885246464.0
818
+ },
819
+ {
820
+ "step": 10200,
821
+ "loss": 4.0236,
822
+ "lr": 0.0006791839747726501,
823
+ "iters": 3,
824
+ "steps_per_s": 21.24,
825
+ "flops_cum": 994493979426816.0
826
+ },
827
+ {
828
+ "step": 10300,
829
+ "loss": 4.0097,
830
+ "lr": 0.0006730585285387465,
831
+ "iters": 2,
832
+ "steps_per_s": 19.69,
833
+ "flops_cum": 1004187043233792.0
834
+ },
835
+ {
836
+ "step": 10400,
837
+ "loss": 4.0156,
838
+ "lr": 0.0006669034296168854,
839
+ "iters": 5,
840
+ "steps_per_s": 18.79,
841
+ "flops_cum": 1014399567396864.0
842
+ },
843
+ {
844
+ "step": 10500,
845
+ "loss": 4.0114,
846
+ "lr": 0.0006607197326515808,
847
+ "iters": 6,
848
+ "steps_per_s": 19.34,
849
+ "flops_cum": 1024287428837376.0
850
+ },
851
+ {
852
+ "step": 10600,
853
+ "loss": 4.0905,
854
+ "lr": 0.0006545084971874737,
855
+ "iters": 1,
856
+ "steps_per_s": 19.87,
857
+ "flops_cum": 1033807339192320.0
858
+ },
859
+ {
860
+ "step": 10700,
861
+ "loss": 4.0435,
862
+ "lr": 0.0006482707874877854,
863
+ "iters": 5,
864
+ "steps_per_s": 19.28,
865
+ "flops_cum": 1043781777358848.0
866
+ },
867
+ {
868
+ "step": 10800,
869
+ "loss": 3.9855,
870
+ "lr": 0.0006420076723519614,
871
+ "iters": 6,
872
+ "steps_per_s": 19.95,
873
+ "flops_cum": 1053366620258304.0
874
+ },
875
+ {
876
+ "step": 10900,
877
+ "loss": 4.0336,
878
+ "lr": 0.0006357202249325371,
879
+ "iters": 7,
880
+ "steps_per_s": 20.45,
881
+ "flops_cum": 1062735021342720.0
882
+ },
883
+ {
884
+ "step": 11000,
885
+ "loss": 3.987,
886
+ "lr": 0.0006294095225512603,
887
+ "iters": 8,
888
+ "steps_per_s": 19.87,
889
+ "flops_cum": 1072363152605184.0
890
+ },
891
+ {
892
+ "step": 11100,
893
+ "loss": 4.1052,
894
+ "lr": 0.0006230766465144965,
895
+ "iters": 1,
896
+ "steps_per_s": 18.93,
897
+ "flops_cum": 1082337590771712.0
898
+ },
899
+ {
900
+ "step": 11200,
901
+ "loss": 3.9834,
902
+ "lr": 0.0006167226819279528,
903
+ "iters": 2,
904
+ "steps_per_s": 19.49,
905
+ "flops_cum": 1092160519667712.0
906
+ },
907
+ {
908
+ "step": 11300,
909
+ "loss": 4.0464,
910
+ "lr": 0.0006103487175107507,
911
+ "iters": 1,
912
+ "steps_per_s": 18.32,
913
+ "flops_cum": 1102546197282816.0
914
+ },
915
+ {
916
+ "step": 11400,
917
+ "loss": 3.9094,
918
+ "lr": 0.0006039558454088796,
919
+ "iters": 4,
920
+ "steps_per_s": 19.47,
921
+ "flops_cum": 1112412414541824.0
922
+ },
923
+ {
924
+ "step": 11500,
925
+ "loss": 3.9789,
926
+ "lr": 0.0005975451610080642,
927
+ "iters": 6,
928
+ "steps_per_s": 18.82,
929
+ "flops_cum": 1122603294523392.0
930
+ },
931
+ {
932
+ "step": 11600,
933
+ "loss": 3.9771,
934
+ "lr": 0.0005911177627460738,
935
+ "iters": 2,
936
+ "steps_per_s": 20.91,
937
+ "flops_cum": 1131647032885248.0
938
+ },
939
+ {
940
+ "step": 11700,
941
+ "loss": 3.9294,
942
+ "lr": 0.0005846747519245122,
943
+ "iters": 3,
944
+ "steps_per_s": 18.87,
945
+ "flops_cum": 1141772980322304.0
946
+ },
947
+ {
948
+ "step": 11800,
949
+ "loss": 4.0137,
950
+ "lr": 0.0005782172325201155,
951
+ "iters": 3,
952
+ "steps_per_s": 18.9,
953
+ "flops_cum": 1151855639396352.0
954
+ },
955
+ {
956
+ "step": 11900,
957
+ "loss": 3.9521,
958
+ "lr": 0.0005717463109955896,
959
+ "iters": 4,
960
+ "steps_per_s": 18.6,
961
+ "flops_cum": 1162111451922432.0
962
+ },
963
+ {
964
+ "step": 12000,
965
+ "loss": 3.8624,
966
+ "lr": 0.000565263096110026,
967
+ "iters": 2,
968
+ "steps_per_s": 20.3,
969
+ "flops_cum": 1171479853006848.0
970
+ },
971
+ {
972
+ "step": 12100,
973
+ "loss": 3.988,
974
+ "lr": 0.0005587686987289189,
975
+ "iters": 4,
976
+ "steps_per_s": 18.36,
977
+ "flops_cum": 1181908818984960.0
978
+ },
979
+ {
980
+ "step": 12200,
981
+ "loss": 3.9801,
982
+ "lr": 0.0005522642316338268,
983
+ "iters": 4,
984
+ "steps_per_s": 18.84,
985
+ "flops_cum": 1192034766422016.0
986
+ },
987
+ {
988
+ "step": 12300,
989
+ "loss": 4.0225,
990
+ "lr": 0.0005457508093317013,
991
+ "iters": 1,
992
+ "steps_per_s": 18.61,
993
+ "flops_cum": 1202225646403584.0
994
+ },
995
+ {
996
+ "step": 12400,
997
+ "loss": 3.9025,
998
+ "lr": 0.0005392295478639225,
999
+ "iters": 4,
1000
+ "steps_per_s": 21.01,
1001
+ "flops_cum": 1211355961491456.0
1002
+ },
1003
+ {
1004
+ "step": 12500,
1005
+ "loss": 4.0461,
1006
+ "lr": 0.0005327015646150716,
1007
+ "iters": 1,
1008
+ "steps_per_s": 16.65,
1009
+ "flops_cum": 1220940804390912.0
1010
+ },
1011
+ {
1012
+ "step": 12600,
1013
+ "loss": 3.9524,
1014
+ "lr": 0.000526167978121472,
1015
+ "iters": 8,
1016
+ "steps_per_s": 19.85,
1017
+ "flops_cum": 1230677156560896.0
1018
+ },
1019
+ {
1020
+ "step": 12700,
1021
+ "loss": 3.8961,
1022
+ "lr": 0.0005196299078795343,
1023
+ "iters": 8,
1024
+ "steps_per_s": 19.88,
1025
+ "flops_cum": 1240283643641856.0
1026
+ },
1027
+ {
1028
+ "step": 12800,
1029
+ "loss": 4.0374,
1030
+ "lr": 0.0005130884741539367,
1031
+ "iters": 1,
1032
+ "steps_per_s": 21.35,
1033
+ "flops_cum": 1249110940188672.0
1034
+ },
1035
+ {
1036
+ "step": 12900,
1037
+ "loss": 3.9113,
1038
+ "lr": 0.0005065447977856722,
1039
+ "iters": 2,
1040
+ "steps_per_s": 19.98,
1041
+ "flops_cum": 1258695783088128.0
1042
+ },
1043
+ {
1044
+ "step": 13000,
1045
+ "loss": 3.9186,
1046
+ "lr": 0.0005,
1047
+ "iters": 8,
1048
+ "steps_per_s": 18.38,
1049
+ "flops_cum": 1269211325792256.0
1050
+ },
1051
+ {
1052
+ "step": 13100,
1053
+ "loss": 3.8851,
1054
+ "lr": 0.0004934552022143279,
1055
+ "iters": 8,
1056
+ "steps_per_s": 20.15,
1057
+ "flops_cum": 1278666303602688.0
1058
+ },
1059
+ {
1060
+ "step": 13200,
1061
+ "loss": 3.8464,
1062
+ "lr": 0.0004869115258460635,
1063
+ "iters": 8,
1064
+ "steps_per_s": 19.07,
1065
+ "flops_cum": 1288684030132224.0
1066
+ },
1067
+ {
1068
+ "step": 13300,
1069
+ "loss": 3.8931,
1070
+ "lr": 0.00048037009212046586,
1071
+ "iters": 8,
1072
+ "steps_per_s": 20.19,
1073
+ "flops_cum": 1298139007942656.0
1074
+ },
1075
+ {
1076
+ "step": 13400,
1077
+ "loss": 3.8284,
1078
+ "lr": 0.0004738320218785281,
1079
+ "iters": 3,
1080
+ "steps_per_s": 19.66,
1081
+ "flops_cum": 1307745495023616.0
1082
+ },
1083
+ {
1084
+ "step": 13500,
1085
+ "loss": 4.061,
1086
+ "lr": 0.00046729843538492847,
1087
+ "iters": 1,
1088
+ "steps_per_s": 19.81,
1089
+ "flops_cum": 1317351982104576.0
1090
+ },
1091
+ {
1092
+ "step": 13600,
1093
+ "loss": 3.7919,
1094
+ "lr": 0.0004607704521360776,
1095
+ "iters": 4,
1096
+ "steps_per_s": 18.7,
1097
+ "flops_cum": 1327607794630656.0
1098
+ },
1099
+ {
1100
+ "step": 13700,
1101
+ "loss": 3.891,
1102
+ "lr": 0.00045424919066829885,
1103
+ "iters": 5,
1104
+ "steps_per_s": 19.3,
1105
+ "flops_cum": 1337495656071168.0
1106
+ },
1107
+ {
1108
+ "step": 13800,
1109
+ "loss": 3.8389,
1110
+ "lr": 0.00044773576836617336,
1111
+ "iters": 4,
1112
+ "steps_per_s": 21.26,
1113
+ "flops_cum": 1346452817707008.0
1114
+ },
1115
+ {
1116
+ "step": 13900,
1117
+ "loss": 3.7805,
1118
+ "lr": 0.00044123130127108126,
1119
+ "iters": 3,
1120
+ "steps_per_s": 19.16,
1121
+ "flops_cum": 1356383967510528.0
1122
+ },
1123
+ {
1124
+ "step": 14000,
1125
+ "loss": 3.8142,
1126
+ "lr": 0.00043473690388997434,
1127
+ "iters": 8,
1128
+ "steps_per_s": 21.91,
1129
+ "flops_cum": 1365189619875840.0
1130
+ },
1131
+ {
1132
+ "step": 14100,
1133
+ "loss": 3.8412,
1134
+ "lr": 0.0004282536890044104,
1135
+ "iters": 8,
1136
+ "steps_per_s": 21.36,
1137
+ "flops_cum": 1374146781511680.0
1138
+ },
1139
+ {
1140
+ "step": 14200,
1141
+ "loss": 3.696,
1142
+ "lr": 0.0004217827674798845,
1143
+ "iters": 4,
1144
+ "steps_per_s": 19.61,
1145
+ "flops_cum": 1383796556955648.0
1146
+ },
1147
+ {
1148
+ "step": 14300,
1149
+ "loss": 3.7604,
1150
+ "lr": 0.0004153252480754877,
1151
+ "iters": 7,
1152
+ "steps_per_s": 18.98,
1153
+ "flops_cum": 1393922504392704.0
1154
+ },
1155
+ {
1156
+ "step": 14400,
1157
+ "loss": 3.8397,
1158
+ "lr": 0.00040888223725392626,
1159
+ "iters": 4,
1160
+ "steps_per_s": 18.49,
1161
+ "flops_cum": 1404178316918784.0
1162
+ },
1163
+ {
1164
+ "step": 14500,
1165
+ "loss": 3.8553,
1166
+ "lr": 0.00040245483899193594,
1167
+ "iters": 2,
1168
+ "steps_per_s": 19.33,
1169
+ "flops_cum": 1414022889996288.0
1170
+ },
1171
+ {
1172
+ "step": 14600,
1173
+ "loss": 4.0137,
1174
+ "lr": 0.0003960441545911204,
1175
+ "iters": 1,
1176
+ "steps_per_s": 19.49,
1177
+ "flops_cum": 1423802530529280.0
1178
+ },
1179
+ {
1180
+ "step": 14700,
1181
+ "loss": 4.0253,
1182
+ "lr": 0.0003896512824892495,
1183
+ "iters": 1,
1184
+ "steps_per_s": 20.59,
1185
+ "flops_cum": 1433062710706176.0
1186
+ },
1187
+ {
1188
+ "step": 14800,
1189
+ "loss": 3.7679,
1190
+ "lr": 0.00038327731807204744,
1191
+ "iters": 8,
1192
+ "steps_per_s": 19.14,
1193
+ "flops_cum": 1443167013961728.0
1194
+ },
1195
+ {
1196
+ "step": 14900,
1197
+ "loss": 3.7208,
1198
+ "lr": 0.0003769233534855035,
1199
+ "iters": 7,
1200
+ "steps_per_s": 18.64,
1201
+ "flops_cum": 1453401182306304.0
1202
+ },
1203
+ {
1204
+ "step": 15000,
1205
+ "loss": 3.7844,
1206
+ "lr": 0.0003705904774487396,
1207
+ "iters": 2,
1208
+ "steps_per_s": 15.24,
1209
+ "flops_cum": 1464003301736448.0
1210
+ },
1211
+ {
1212
+ "step": 15100,
1213
+ "loss": 3.7567,
1214
+ "lr": 0.0003642797750674629,
1215
+ "iters": 2,
1216
+ "steps_per_s": 19.59,
1217
+ "flops_cum": 1473739653906432.0
1218
+ },
1219
+ {
1220
+ "step": 15200,
1221
+ "loss": 3.749,
1222
+ "lr": 0.0003579923276480387,
1223
+ "iters": 3,
1224
+ "steps_per_s": 18.6,
1225
+ "flops_cum": 1484017110614016.0
1226
+ },
1227
+ {
1228
+ "step": 15300,
1229
+ "loss": 3.6559,
1230
+ "lr": 0.0003517292125122146,
1231
+ "iters": 8,
1232
+ "steps_per_s": 19.48,
1233
+ "flops_cum": 1493904972054528.0
1234
+ },
1235
+ {
1236
+ "step": 15400,
1237
+ "loss": 3.714,
1238
+ "lr": 0.00034549150281252633,
1239
+ "iters": 3,
1240
+ "steps_per_s": 20.49,
1241
+ "flops_cum": 1503143508049920.0
1242
+ },
1243
+ {
1244
+ "step": 15500,
1245
+ "loss": 3.6398,
1246
+ "lr": 0.0003392802673484193,
1247
+ "iters": 7,
1248
+ "steps_per_s": 20.82,
1249
+ "flops_cum": 1512382044045312.0
1250
+ },
1251
+ {
1252
+ "step": 15600,
1253
+ "loss": 4.004,
1254
+ "lr": 0.00033309657038311456,
1255
+ "iters": 1,
1256
+ "steps_per_s": 20.77,
1257
+ "flops_cum": 1521469070770176.0
1258
+ },
1259
+ {
1260
+ "step": 15700,
1261
+ "loss": 3.5206,
1262
+ "lr": 0.0003269414714612534,
1263
+ "iters": 5,
1264
+ "steps_per_s": 19.33,
1265
+ "flops_cum": 1531400220573696.0
1266
+ },
1267
+ {
1268
+ "step": 15800,
1269
+ "loss": 3.4684,
1270
+ "lr": 0.00032081602522734986,
1271
+ "iters": 7,
1272
+ "steps_per_s": 18.29,
1273
+ "flops_cum": 1541872474914816.0
1274
+ },
1275
+ {
1276
+ "step": 15900,
1277
+ "loss": 3.6148,
1278
+ "lr": 0.00031472128124508187,
1279
+ "iters": 4,
1280
+ "steps_per_s": 19.29,
1281
+ "flops_cum": 1551695403810816.0
1282
+ },
1283
+ {
1284
+ "step": 16000,
1285
+ "loss": 3.4992,
1286
+ "lr": 0.0003086582838174551,
1287
+ "iters": 6,
1288
+ "steps_per_s": 20.86,
1289
+ "flops_cum": 1560890651443200.0
1290
+ },
1291
+ {
1292
+ "step": 16100,
1293
+ "loss": 3.5494,
1294
+ "lr": 0.00030262807180786645,
1295
+ "iters": 8,
1296
+ "steps_per_s": 18.53,
1297
+ "flops_cum": 1571211396513792.0
1298
+ },
1299
+ {
1300
+ "step": 16200,
1301
+ "loss": 3.4088,
1302
+ "lr": 0.0002966316784621,
1303
+ "iters": 6,
1304
+ "steps_per_s": 19.76,
1305
+ "flops_cum": 1580839527776256.0
1306
+ },
1307
+ {
1308
+ "step": 16300,
1309
+ "loss": 3.5107,
1310
+ "lr": 0.0002906701312312861,
1311
+ "iters": 6,
1312
+ "steps_per_s": 19.61,
1313
+ "flops_cum": 1590575879946240.0
1314
+ },
1315
+ {
1316
+ "step": 16400,
1317
+ "loss": 3.4532,
1318
+ "lr": 0.0002847444515958523,
1319
+ "iters": 7,
1320
+ "steps_per_s": 18.88,
1321
+ "flops_cum": 1600680183201792.0
1322
+ },
1323
+ {
1324
+ "step": 16500,
1325
+ "loss": 3.7564,
1326
+ "lr": 0.00027885565489049947,
1327
+ "iters": 2,
1328
+ "steps_per_s": 19.93,
1329
+ "flops_cum": 1610156805193728.0
1330
+ },
1331
+ {
1332
+ "step": 16600,
1333
+ "loss": 3.5701,
1334
+ "lr": 0.00027300475013022663,
1335
+ "iters": 3,
1336
+ "steps_per_s": 18.02,
1337
+ "flops_cum": 1620758924623872.0
1338
+ },
1339
+ {
1340
+ "step": 16700,
1341
+ "loss": 3.6816,
1342
+ "lr": 0.0002671927398374443,
1343
+ "iters": 2,
1344
+ "steps_per_s": 19.47,
1345
+ "flops_cum": 1630538565156864.0
1346
+ },
1347
+ {
1348
+ "step": 16800,
1349
+ "loss": 3.3537,
1350
+ "lr": 0.00026142061987019576,
1351
+ "iters": 6,
1352
+ "steps_per_s": 20.61,
1353
+ "flops_cum": 1639885322059776.0
1354
+ },
1355
+ {
1356
+ "step": 16900,
1357
+ "loss": 3.7755,
1358
+ "lr": 0.0002556893792515227,
1359
+ "iters": 2,
1360
+ "steps_per_s": 20.88,
1361
+ "flops_cum": 1648950704603136.0
1362
+ },
1363
+ {
1364
+ "step": 17000,
1365
+ "loss": 3.37,
1366
+ "lr": 0.0002500000000000001,
1367
+ "iters": 5,
1368
+ "steps_per_s": 20.24,
1369
+ "flops_cum": 1658427326595072.0
1370
+ },
1371
+ {
1372
+ "step": 17100,
1373
+ "loss": 3.7081,
1374
+ "lr": 0.00024435345696147403,
1375
+ "iters": 2,
1376
+ "steps_per_s": 20.77,
1377
+ "flops_cum": 1667579285864448.0
1378
+ },
1379
+ {
1380
+ "step": 17200,
1381
+ "loss": 3.4859,
1382
+ "lr": 0.00023875071764202561,
1383
+ "iters": 3,
1384
+ "steps_per_s": 19.45,
1385
+ "flops_cum": 1677402214760448.0
1386
+ },
1387
+ {
1388
+ "step": 17300,
1389
+ "loss": 3.1944,
1390
+ "lr": 0.00023319274204219425,
1391
+ "iters": 5,
1392
+ "steps_per_s": 20.89,
1393
+ "flops_cum": 1686575818211328.0
1394
+ },
1395
+ {
1396
+ "step": 17400,
1397
+ "loss": 3.4698,
1398
+ "lr": 0.00022768048249248646,
1399
+ "iters": 3,
1400
+ "steps_per_s": 18.29,
1401
+ "flops_cum": 1696961495826432.0
1402
+ },
1403
+ {
1404
+ "step": 17500,
1405
+ "loss": 2.9826,
1406
+ "lr": 0.00022221488349019903,
1407
+ "iters": 6,
1408
+ "steps_per_s": 17.54,
1409
+ "flops_cum": 1706070166732800.0
1410
+ },
1411
+ {
1412
+ "step": 17600,
1413
+ "loss": 3.0757,
1414
+ "lr": 0.0002167968815375837,
1415
+ "iters": 8,
1416
+ "steps_per_s": 19.18,
1417
+ "flops_cum": 1716044604899328.0
1418
+ },
1419
+ {
1420
+ "step": 17700,
1421
+ "loss": 3.1987,
1422
+ "lr": 0.00021142740498138323,
1423
+ "iters": 5,
1424
+ "steps_per_s": 17.8,
1425
+ "flops_cum": 1726711656873984.0
1426
+ },
1427
+ {
1428
+ "step": 17800,
1429
+ "loss": 3.4683,
1430
+ "lr": 0.00020610737385376348,
1431
+ "iters": 3,
1432
+ "steps_per_s": 19.79,
1433
+ "flops_cum": 1736318143954944.0
1434
+ },
1435
+ {
1436
+ "step": 17900,
1437
+ "loss": 3.2041,
1438
+ "lr": 0.00020083769971467047,
1439
+ "iters": 4,
1440
+ "steps_per_s": 19.54,
1441
+ "flops_cum": 1746097784487936.0
1442
+ },
1443
+ {
1444
+ "step": 18000,
1445
+ "loss": 3.9534,
1446
+ "lr": 0.00019561928549563967,
1447
+ "iters": 1,
1448
+ "steps_per_s": 20.37,
1449
+ "flops_cum": 1755422897209344.0
1450
+ },
1451
+ {
1452
+ "step": 18100,
1453
+ "loss": 3.7774,
1454
+ "lr": 0.00019045302534508295,
1455
+ "iters": 2,
1456
+ "steps_per_s": 19.19,
1457
+ "flops_cum": 1765397335375872.0
1458
+ },
1459
+ {
1460
+ "step": 18200,
1461
+ "loss": 3.1896,
1462
+ "lr": 0.00018533980447508135,
1463
+ "iters": 5,
1464
+ "steps_per_s": 19.18,
1465
+ "flops_cum": 1775393417723904.0
1466
+ },
1467
+ {
1468
+ "step": 18300,
1469
+ "loss": 3.018,
1470
+ "lr": 0.00018028049900970766,
1471
+ "iters": 6,
1472
+ "steps_per_s": 18.82,
1473
+ "flops_cum": 1785541009342464.0
1474
+ },
1475
+ {
1476
+ "step": 18400,
1477
+ "loss": 3.9561,
1478
+ "lr": 0.00017527597583490823,
1479
+ "iters": 1,
1480
+ "steps_per_s": 19.37,
1481
+ "flops_cum": 1795299005693952.0
1482
+ },
1483
+ {
1484
+ "step": 18500,
1485
+ "loss": 3.9769,
1486
+ "lr": 0.00017032709244996558,
1487
+ "iters": 1,
1488
+ "steps_per_s": 20.66,
1489
+ "flops_cum": 1804537541689344.0
1490
+ },
1491
+ {
1492
+ "step": 18600,
1493
+ "loss": 2.8201,
1494
+ "lr": 0.00016543469682057105,
1495
+ "iters": 7,
1496
+ "steps_per_s": 21.03,
1497
+ "flops_cum": 1813711145140224.0
1498
+ },
1499
+ {
1500
+ "step": 18700,
1501
+ "loss": 2.8076,
1502
+ "lr": 0.0001605996272335291,
1503
+ "iters": 7,
1504
+ "steps_per_s": 19.29,
1505
+ "flops_cum": 1823620650762240.0
1506
+ },
1507
+ {
1508
+ "step": 18800,
1509
+ "loss": 3.1355,
1510
+ "lr": 0.00015582271215312294,
1511
+ "iters": 5,
1512
+ "steps_per_s": 17.82,
1513
+ "flops_cum": 1834287702736896.0
1514
+ },
1515
+ {
1516
+ "step": 18900,
1517
+ "loss": 2.8,
1518
+ "lr": 0.00015110477007916002,
1519
+ "iters": 8,
1520
+ "steps_per_s": 19.81,
1521
+ "flops_cum": 1843980766543872.0
1522
+ },
1523
+ {
1524
+ "step": 19000,
1525
+ "loss": 3.8893,
1526
+ "lr": 0.00014644660940672628,
1527
+ "iters": 1,
1528
+ "steps_per_s": 18.5,
1529
+ "flops_cum": 1854171646525440.0
1530
+ },
1531
+ {
1532
+ "step": 19100,
1533
+ "loss": 3.3706,
1534
+ "lr": 0.00014184902828767287,
1535
+ "iters": 3,
1536
+ "steps_per_s": 20.52,
1537
+ "flops_cum": 1863496759246848.0
1538
+ },
1539
+ {
1540
+ "step": 19200,
1541
+ "loss": 2.838,
1542
+ "lr": 0.0001373128144938563,
1543
+ "iters": 7,
1544
+ "steps_per_s": 20.22,
1545
+ "flops_cum": 1873016669601792.0
1546
+ },
1547
+ {
1548
+ "step": 19300,
1549
+ "loss": 3.6294,
1550
+ "lr": 0.00013283874528215734,
1551
+ "iters": 2,
1552
+ "steps_per_s": 19.51,
1553
+ "flops_cum": 1882709733408768.0
1554
+ },
1555
+ {
1556
+ "step": 19400,
1557
+ "loss": 3.1529,
1558
+ "lr": 0.00012842758726130281,
1559
+ "iters": 4,
1560
+ "steps_per_s": 18.36,
1561
+ "flops_cum": 1893138699386880.0
1562
+ },
1563
+ {
1564
+ "step": 19500,
1565
+ "loss": 3.5123,
1566
+ "lr": 0.00012408009626051135,
1567
+ "iters": 2,
1568
+ "steps_per_s": 22.02,
1569
+ "flops_cum": 1901771198300160.0
1570
+ },
1571
+ {
1572
+ "step": 19600,
1573
+ "loss": 3.3627,
1574
+ "lr": 0.00011979701719998454,
1575
+ "iters": 3,
1576
+ "steps_per_s": 20.46,
1577
+ "flops_cum": 1911117955203072.0
1578
+ },
1579
+ {
1580
+ "step": 19700,
1581
+ "loss": 3.1757,
1582
+ "lr": 0.00011557908396327027,
1583
+ "iters": 5,
1584
+ "steps_per_s": 20.79,
1585
+ "flops_cum": 1920334847016960.0
1586
+ },
1587
+ {
1588
+ "step": 19800,
1589
+ "loss": 2.9972,
1590
+ "lr": 0.00011142701927151455,
1591
+ "iters": 6,
1592
+ "steps_per_s": 20.06,
1593
+ "flops_cum": 1929854757371904.0
1594
+ },
1595
+ {
1596
+ "step": 19900,
1597
+ "loss": 3.3672,
1598
+ "lr": 0.00010734153455962764,
1599
+ "iters": 3,
1600
+ "steps_per_s": 19.47,
1601
+ "flops_cum": 1939612753723392.0
1602
+ },
1603
+ {
1604
+ "step": 20000,
1605
+ "loss": 3.2789,
1606
+ "lr": 0.00010332332985438247,
1607
+ "iters": 4,
1608
+ "steps_per_s": 15.97,
1609
+ "flops_cum": 1949760345341952.0
1610
+ },
1611
+ {
1612
+ "step": 20100,
1613
+ "loss": 3.9212,
1614
+ "lr": 9.937309365446973e-05,
1615
+ "iters": 1,
1616
+ "steps_per_s": 19.36,
1617
+ "flops_cum": 1959561630056448.0
1618
+ },
1619
+ {
1620
+ "step": 20200,
1621
+ "loss": 3.4885,
1622
+ "lr": 9.549150281252633e-05,
1623
+ "iters": 3,
1624
+ "steps_per_s": 21.31,
1625
+ "flops_cum": 1968562080055296.0
1626
+ },
1627
+ {
1628
+ "step": 20300,
1629
+ "loss": 2.9657,
1630
+ "lr": 9.167922241916055e-05,
1631
+ "iters": 7,
1632
+ "steps_per_s": 18.66,
1633
+ "flops_cum": 1978861180944384.0
1634
+ },
1635
+ {
1636
+ "step": 20400,
1637
+ "loss": 3.2816,
1638
+ "lr": 8.793690568899215e-05,
1639
+ "iters": 3,
1640
+ "steps_per_s": 20.43,
1641
+ "flops_cum": 1988121361121280.0
1642
+ },
1643
+ {
1644
+ "step": 20500,
1645
+ "loss": 3.9725,
1646
+ "lr": 8.426519384872733e-05,
1647
+ "iters": 1,
1648
+ "steps_per_s": 19.01,
1649
+ "flops_cum": 1998139087650816.0
1650
+ },
1651
+ {
1652
+ "step": 20600,
1653
+ "loss": 2.837,
1654
+ "lr": 8.066471602728804e-05,
1655
+ "iters": 7,
1656
+ "steps_per_s": 20.11,
1657
+ "flops_cum": 2007745574731776.0
1658
+ },
1659
+ {
1660
+ "step": 20700,
1661
+ "loss": 3.3961,
1662
+ "lr": 7.71360891480134e-05,
1663
+ "iters": 3,
1664
+ "steps_per_s": 19.65,
1665
+ "flops_cum": 2017373705994240.0
1666
+ },
1667
+ {
1668
+ "step": 20800,
1669
+ "loss": 2.7765,
1670
+ "lr": 7.367991782295391e-05,
1671
+ "iters": 7,
1672
+ "steps_per_s": 19.52,
1673
+ "flops_cum": 2027218279071744.0
1674
+ },
1675
+ {
1676
+ "step": 20900,
1677
+ "loss": 3.0458,
1678
+ "lr": 7.029679424927366e-05,
1679
+ "iters": 5,
1680
+ "steps_per_s": 19.9,
1681
+ "flops_cum": 2036781477789696.0
1682
+ },
1683
+ {
1684
+ "step": 21000,
1685
+ "loss": 2.7421,
1686
+ "lr": 6.698729810778065e-05,
1687
+ "iters": 7,
1688
+ "steps_per_s": 18.7,
1689
+ "flops_cum": 2047015646134272.0
1690
+ },
1691
+ {
1692
+ "step": 21100,
1693
+ "loss": 3.1867,
1694
+ "lr": 6.375199646360142e-05,
1695
+ "iters": 4,
1696
+ "steps_per_s": 18.86,
1697
+ "flops_cum": 2057076661026816.0
1698
+ },
1699
+ {
1700
+ "step": 21200,
1701
+ "loss": 2.7443,
1702
+ "lr": 6.059144366901737e-05,
1703
+ "iters": 7,
1704
+ "steps_per_s": 18.94,
1705
+ "flops_cum": 2067180964282368.0
1706
+ },
1707
+ {
1708
+ "step": 21300,
1709
+ "loss": 3.9399,
1710
+ "lr": 5.750618126847912e-05,
1711
+ "iters": 1,
1712
+ "steps_per_s": 20.01,
1713
+ "flops_cum": 2076614297911296.0
1714
+ },
1715
+ {
1716
+ "step": 21400,
1717
+ "loss": 3.2749,
1718
+ "lr": 5.449673790581611e-05,
1719
+ "iters": 4,
1720
+ "steps_per_s": 19.32,
1721
+ "flops_cum": 2086523803533312.0
1722
+ },
1723
+ {
1724
+ "step": 21500,
1725
+ "loss": 3.2295,
1726
+ "lr": 5.1563629233655876e-05,
1727
+ "iters": 4,
1728
+ "steps_per_s": 18.16,
1729
+ "flops_cum": 2097039346237440.0
1730
+ },
1731
+ {
1732
+ "step": 21600,
1733
+ "loss": 3.4402,
1734
+ "lr": 4.87073578250698e-05,
1735
+ "iters": 3,
1736
+ "steps_per_s": 19.86,
1737
+ "flops_cum": 2106624189136896.0
1738
+ },
1739
+ {
1740
+ "step": 21700,
1741
+ "loss": 3.6534,
1742
+ "lr": 4.592841308745932e-05,
1743
+ "iters": 2,
1744
+ "steps_per_s": 20.35,
1745
+ "flops_cum": 2115992590221312.0
1746
+ },
1747
+ {
1748
+ "step": 21800,
1749
+ "loss": 2.7617,
1750
+ "lr": 4.322727117869951e-05,
1751
+ "iters": 6,
1752
+ "steps_per_s": 18.32,
1753
+ "flops_cum": 2126486488743936.0
1754
+ },
1755
+ {
1756
+ "step": 21900,
1757
+ "loss": 3.8589,
1758
+ "lr": 4.06043949255509e-05,
1759
+ "iters": 1,
1760
+ "steps_per_s": 18.86,
1761
+ "flops_cum": 2136504215273472.0
1762
+ },
1763
+ {
1764
+ "step": 22000,
1765
+ "loss": 2.9378,
1766
+ "lr": 3.806023374435663e-05,
1767
+ "iters": 5,
1768
+ "steps_per_s": 18.97,
1769
+ "flops_cum": 2146651806892032.0
1770
+ },
1771
+ {
1772
+ "step": 22100,
1773
+ "loss": 2.7566,
1774
+ "lr": 3.559522356403788e-05,
1775
+ "iters": 7,
1776
+ "steps_per_s": 21.53,
1777
+ "flops_cum": 2155565680164864.0
1778
+ },
1779
+ {
1780
+ "step": 22200,
1781
+ "loss": 3.1214,
1782
+ "lr": 3.3209786751399184e-05,
1783
+ "iters": 4,
1784
+ "steps_per_s": 18.8,
1785
+ "flops_cum": 2165648339238912.0
1786
+ },
1787
+ {
1788
+ "step": 22300,
1789
+ "loss": 2.8964,
1790
+ "lr": 3.0904332038757974e-05,
1791
+ "iters": 6,
1792
+ "steps_per_s": 19.41,
1793
+ "flops_cum": 2175514556497920.0
1794
+ },
1795
+ {
1796
+ "step": 22400,
1797
+ "loss": 3.8919,
1798
+ "lr": 2.8679254453910786e-05,
1799
+ "iters": 1,
1800
+ "steps_per_s": 18.56,
1801
+ "flops_cum": 2185727080660992.0
1802
+ },
1803
+ {
1804
+ "step": 22500,
1805
+ "loss": 2.7172,
1806
+ "lr": 2.653493525244721e-05,
1807
+ "iters": 8,
1808
+ "steps_per_s": 15.12,
1809
+ "flops_cum": 2196653862813696.0
1810
+ },
1811
+ {
1812
+ "step": 22600,
1813
+ "loss": 3.16,
1814
+ "lr": 2.4471741852423235e-05,
1815
+ "iters": 4,
1816
+ "steps_per_s": 17.91,
1817
+ "flops_cum": 2207234338062336.0
1818
+ },
1819
+ {
1820
+ "step": 22700,
1821
+ "loss": 2.7317,
1822
+ "lr": 2.2490027771406685e-05,
1823
+ "iters": 8,
1824
+ "steps_per_s": 18.71,
1825
+ "flops_cum": 2217511794769920.0
1826
+ },
1827
+ {
1828
+ "step": 22800,
1829
+ "loss": 3.6204,
1830
+ "lr": 2.0590132565903473e-05,
1831
+ "iters": 2,
1832
+ "steps_per_s": 19.66,
1833
+ "flops_cum": 2227096637669376.0
1834
+ },
1835
+ {
1836
+ "step": 22900,
1837
+ "loss": 3.0106,
1838
+ "lr": 1.8772381773176416e-05,
1839
+ "iters": 6,
1840
+ "steps_per_s": 19.08,
1841
+ "flops_cum": 2237157652561920.0
1842
+ },
1843
+ {
1844
+ "step": 23000,
1845
+ "loss": 3.1273,
1846
+ "lr": 1.70370868554659e-05,
1847
+ "iters": 5,
1848
+ "steps_per_s": 18.23,
1849
+ "flops_cum": 2247608262721536.0
1850
+ },
1851
+ {
1852
+ "step": 23100,
1853
+ "loss": 3.2552,
1854
+ "lr": 1.538454514662285e-05,
1855
+ "iters": 4,
1856
+ "steps_per_s": 21.66,
1857
+ "flops_cum": 2256392270905344.0
1858
+ },
1859
+ {
1860
+ "step": 23200,
1861
+ "loss": 3.5441,
1862
+ "lr": 1.3815039801161721e-05,
1863
+ "iters": 3,
1864
+ "steps_per_s": 20.01,
1865
+ "flops_cum": 2265933825441792.0
1866
+ },
1867
+ {
1868
+ "step": 23300,
1869
+ "loss": 3.1145,
1870
+ "lr": 1.232883974574367e-05,
1871
+ "iters": 5,
1872
+ "steps_per_s": 20.44,
1873
+ "flops_cum": 2275323870707712.0
1874
+ },
1875
+ {
1876
+ "step": 23400,
1877
+ "loss": 3.1055,
1878
+ "lr": 1.0926199633097156e-05,
1879
+ "iters": 4,
1880
+ "steps_per_s": 19.58,
1881
+ "flops_cum": 2285060222877696.0
1882
+ },
1883
+ {
1884
+ "step": 23500,
1885
+ "loss": 3.5739,
1886
+ "lr": 9.607359798384786e-06,
1887
+ "iters": 2,
1888
+ "steps_per_s": 18.86,
1889
+ "flops_cum": 2295142881951744.0
1890
+ },
1891
+ {
1892
+ "step": 23600,
1893
+ "loss": 3.3183,
1894
+ "lr": 8.372546218022748e-06,
1895
+ "iters": 4,
1896
+ "steps_per_s": 17.81,
1897
+ "flops_cum": 2305874866470912.0
1898
+ },
1899
+ {
1900
+ "step": 23700,
1901
+ "loss": 2.7759,
1902
+ "lr": 7.221970470961125e-06,
1903
+ "iters": 6,
1904
+ "steps_per_s": 20.0,
1905
+ "flops_cum": 2315459709370368.0
1906
+ },
1907
+ {
1908
+ "step": 23800,
1909
+ "loss": 3.2027,
1910
+ "lr": 6.15582970243117e-06,
1911
+ "iters": 4,
1912
+ "steps_per_s": 20.78,
1913
+ "flops_cum": 2324611668639744.0
1914
+ },
1915
+ {
1916
+ "step": 23900,
1917
+ "loss": 2.5994,
1918
+ "lr": 5.174306590164879e-06,
1919
+ "iters": 8,
1920
+ "steps_per_s": 20.61,
1921
+ "flops_cum": 2333958425542656.0
1922
+ },
1923
+ {
1924
+ "step": 24000,
1925
+ "loss": 2.6356,
1926
+ "lr": 4.277569313094809e-06,
1927
+ "iters": 8,
1928
+ "steps_per_s": 18.83,
1929
+ "flops_cum": 2344084372979712.0
1930
+ },
1931
+ {
1932
+ "step": 24100,
1933
+ "loss": 3.0904,
1934
+ "lr": 3.4657715225368535e-06,
1935
+ "iters": 4,
1936
+ "steps_per_s": 19.87,
1937
+ "flops_cum": 2353604283334656.0
1938
+ },
1939
+ {
1940
+ "step": 24200,
1941
+ "loss": 3.6824,
1942
+ "lr": 2.739052315863355e-06,
1943
+ "iters": 2,
1944
+ "steps_per_s": 19.89,
1945
+ "flops_cum": 2363167482052608.0
1946
+ },
1947
+ {
1948
+ "step": 24300,
1949
+ "loss": 2.6988,
1950
+ "lr": 2.097536212669171e-06,
1951
+ "iters": 8,
1952
+ "steps_per_s": 18.86,
1953
+ "flops_cum": 2373401650397184.0
1954
+ },
1955
+ {
1956
+ "step": 24400,
1957
+ "loss": 2.5732,
1958
+ "lr": 1.541333133436018e-06,
1959
+ "iters": 8,
1960
+ "steps_per_s": 18.94,
1961
+ "flops_cum": 2383484309471232.0
1962
+ },
1963
+ {
1964
+ "step": 24500,
1965
+ "loss": 3.9331,
1966
+ "lr": 1.0705383806982606e-06,
1967
+ "iters": 1,
1968
+ "steps_per_s": 19.87,
1969
+ "flops_cum": 2392960931463168.0
1970
+ },
1971
+ {
1972
+ "step": 24600,
1973
+ "loss": 2.7363,
1974
+ "lr": 6.852326227130834e-07,
1975
+ "iters": 8,
1976
+ "steps_per_s": 19.25,
1977
+ "flops_cum": 2403000302174208.0
1978
+ },
1979
+ {
1980
+ "step": 24700,
1981
+ "loss": 2.8153,
1982
+ "lr": 3.854818796385495e-07,
1983
+ "iters": 6,
1984
+ "steps_per_s": 19.42,
1985
+ "flops_cum": 2412779942707200.0
1986
+ },
1987
+ {
1988
+ "step": 24800,
1989
+ "loss": 3.2063,
1990
+ "lr": 1.7133751222137007e-07,
1991
+ "iters": 4,
1992
+ "steps_per_s": 19.68,
1993
+ "flops_cum": 2422451362332672.0
1994
+ },
1995
+ {
1996
+ "step": 24900,
1997
+ "loss": 3.7407,
1998
+ "lr": 4.283621299649987e-08,
1999
+ "iters": 2,
2000
+ "steps_per_s": 20.56,
2001
+ "flops_cum": 2431689898328064.0
2002
+ },
2003
+ {
2004
+ "step": 24999,
2005
+ "loss": 3.3425,
2006
+ "lr": 4.283682464301109e-12,
2007
+ "iters": 3,
2008
+ "steps_per_s": 20.03,
2009
+ "flops_cum": 2441230622392320.0
2010
+ }
2011
+ ],
2012
+ "id_acc": [
2013
+ {
2014
+ "step": 2500,
2015
+ "mean": 0.026249999180436134,
2016
+ "per_k": {
2017
+ "1": 0.022499999031424522,
2018
+ "2": 0.02499999850988388,
2019
+ "3": 0.03500000014901161,
2020
+ "4": 0.03749999776482582,
2021
+ "5": 0.027499999850988388,
2022
+ "6": 0.017500000074505806,
2023
+ "7": 0.022499999031424522,
2024
+ "8": 0.022499999031424522
2025
+ }
2026
+ },
2027
+ {
2028
+ "step": 5000,
2029
+ "mean": 0.043749998323619366,
2030
+ "per_k": {
2031
+ "1": 0.03999999910593033,
2032
+ "2": 0.02499999850988388,
2033
+ "3": 0.03500000014901161,
2034
+ "4": 0.044999998062849045,
2035
+ "5": 0.04999999701976776,
2036
+ "6": 0.04999999701976776,
2037
+ "7": 0.05999999865889549,
2038
+ "8": 0.044999998062849045
2039
+ }
2040
+ },
2041
+ {
2042
+ "step": 7500,
2043
+ "mean": 0.04468749836087227,
2044
+ "per_k": {
2045
+ "1": 0.032499998807907104,
2046
+ "2": 0.03749999776482582,
2047
+ "3": 0.04999999701976776,
2048
+ "4": 0.054999999701976776,
2049
+ "5": 0.044999998062849045,
2050
+ "6": 0.05249999836087227,
2051
+ "7": 0.03999999910593033,
2052
+ "8": 0.044999998062849045
2053
+ }
2054
+ },
2055
+ {
2056
+ "step": 10000,
2057
+ "mean": 0.05249999836087227,
2058
+ "per_k": {
2059
+ "1": 0.03749999776482582,
2060
+ "2": 0.07000000029802322,
2061
+ "3": 0.03749999776482582,
2062
+ "4": 0.04999999701976776,
2063
+ "5": 0.07249999791383743,
2064
+ "6": 0.05999999865889549,
2065
+ "7": 0.06499999761581421,
2066
+ "8": 0.027499999850988388
2067
+ }
2068
+ },
2069
+ {
2070
+ "step": 12500,
2071
+ "mean": 0.05687499837949872,
2072
+ "per_k": {
2073
+ "1": 0.03999999910593033,
2074
+ "2": 0.05249999836087227,
2075
+ "3": 0.04749999940395355,
2076
+ "4": 0.057499997317790985,
2077
+ "5": 0.05999999865889549,
2078
+ "6": 0.0625,
2079
+ "7": 0.07499999552965164,
2080
+ "8": 0.05999999865889549
2081
+ }
2082
+ },
2083
+ {
2084
+ "step": 15000,
2085
+ "mean": 0.07656249776482582,
2086
+ "per_k": {
2087
+ "1": 0.054999999701976776,
2088
+ "2": 0.04749999940395355,
2089
+ "3": 0.04999999701976776,
2090
+ "4": 0.09999999403953552,
2091
+ "5": 0.08249999582767487,
2092
+ "6": 0.0625,
2093
+ "7": 0.08500000089406967,
2094
+ "8": 0.12999999523162842
2095
+ }
2096
+ },
2097
+ {
2098
+ "step": 17500,
2099
+ "mean": 0.0924999974668026,
2100
+ "per_k": {
2101
+ "1": 0.03500000014901161,
2102
+ "2": 0.044999998062849045,
2103
+ "3": 0.07000000029802322,
2104
+ "4": 0.06499999761581421,
2105
+ "5": 0.1224999949336052,
2106
+ "6": 0.07999999821186066,
2107
+ "7": 0.125,
2108
+ "8": 0.19749999046325684
2109
+ }
2110
+ },
2111
+ {
2112
+ "step": 20000,
2113
+ "mean": 0.09218749729916453,
2114
+ "per_k": {
2115
+ "1": 0.04999999701976776,
2116
+ "2": 0.03999999910593033,
2117
+ "3": 0.042500000447034836,
2118
+ "4": 0.07999999821186066,
2119
+ "5": 0.125,
2120
+ "6": 0.08500000089406967,
2121
+ "7": 0.14749999344348907,
2122
+ "8": 0.16749998927116394
2123
+ }
2124
+ },
2125
+ {
2126
+ "step": 22500,
2127
+ "mean": 0.08718749857507646,
2128
+ "per_k": {
2129
+ "1": 0.04999999701976776,
2130
+ "2": 0.022499999031424522,
2131
+ "3": 0.04749999940395355,
2132
+ "4": 0.08249999582767487,
2133
+ "5": 0.10249999910593033,
2134
+ "6": 0.09999999403953552,
2135
+ "7": 0.14000000059604645,
2136
+ "8": 0.1525000035762787
2137
+ }
2138
+ },
2139
+ {
2140
+ "step": 25000,
2141
+ "mean": 0.09624999854713678,
2142
+ "per_k": {
2143
+ "1": 0.03999999910593033,
2144
+ "2": 0.03999999910593033,
2145
+ "3": 0.08500000089406967,
2146
+ "4": 0.07499999552965164,
2147
+ "5": 0.11749999970197678,
2148
+ "6": 0.11499999463558197,
2149
+ "7": 0.14249999821186066,
2150
+ "8": 0.1550000011920929
2151
+ }
2152
+ }
2153
+ ],
2154
+ "ood": [
2155
+ {
2156
+ "step": 2500,
2157
+ "acc": {
2158
+ "16": 0.032499998807907104,
2159
+ "32": 0.02499999850988388,
2160
+ "64": 0.017500000074505806
2161
+ }
2162
+ },
2163
+ {
2164
+ "step": 5000,
2165
+ "acc": {
2166
+ "16": 0.04999999701976776,
2167
+ "32": 0.02499999850988388,
2168
+ "64": 0.022499999031424522
2169
+ }
2170
+ },
2171
+ {
2172
+ "step": 7500,
2173
+ "acc": {
2174
+ "16": 0.042500000447034836,
2175
+ "32": 0.03500000014901161,
2176
+ "64": 0.044999998062849045
2177
+ }
2178
+ },
2179
+ {
2180
+ "step": 10000,
2181
+ "acc": {
2182
+ "16": 0.07499999552965164,
2183
+ "32": 0.032499998807907104,
2184
+ "64": 0.027499999850988388
2185
+ }
2186
+ },
2187
+ {
2188
+ "step": 12500,
2189
+ "acc": {
2190
+ "16": 0.042500000447034836,
2191
+ "32": 0.04749999940395355,
2192
+ "64": 0.019999999552965164
2193
+ }
2194
+ },
2195
+ {
2196
+ "step": 15000,
2197
+ "acc": {
2198
+ "16": 0.08749999850988388,
2199
+ "32": 0.029999999329447746,
2200
+ "64": 0.022499999031424522
2201
+ }
2202
+ },
2203
+ {
2204
+ "step": 17500,
2205
+ "acc": {
2206
+ "16": 0.08999999612569809,
2207
+ "32": 0.02499999850988388,
2208
+ "64": 0.019999999552965164
2209
+ }
2210
+ },
2211
+ {
2212
+ "step": 20000,
2213
+ "acc": {
2214
+ "16": 0.042500000447034836,
2215
+ "32": 0.02499999850988388,
2216
+ "64": 0.019999999552965164
2217
+ }
2218
+ },
2219
+ {
2220
+ "step": 22500,
2221
+ "acc": {
2222
+ "16": 0.03999999910593033,
2223
+ "32": 0.029999999329447746,
2224
+ "64": 0.007499999832361937
2225
+ }
2226
+ },
2227
+ {
2228
+ "step": 25000,
2229
+ "acc": {
2230
+ "16": 0.04749999940395355,
2231
+ "32": 0.02499999850988388,
2232
+ "64": 0.01249999925494194
2233
+ }
2234
+ }
2235
+ ],
2236
+ "iters": [],
2237
+ "flops_cum": 2441230622392320.0
2238
+ }