Hugging Face's logo

eac123
/
ptr-recurrent-depth

vanilla-pytorch
ml-intern
research
recurrent-depth
length-generalization
pointer-chasing
Model card Files
xet
 Community
eac123 commited on
Commit
a07e11f
·
verified ·
1 Parent(s): edf178f

Upload results/lr0.0003-rec1x4_history.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. results/lr0.0003-rec1x4_history.json +302 -0
results/lr0.0003-rec1x4_history.json ADDED
@@ -0,0 +1,302 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "loss": [
3
+ {
4
+ "step": 0,
5
+ "loss": 4.3035,
6
+ "lr": 3e-07,
7
+ "iters": 4,
8
+ "steps_per_s": 0.0,
9
+ "flops_cum": 87407198208.0
10
+ },
11
+ {
12
+ "step": 100,
13
+ "loss": 4.2337,
14
+ "lr": 3.0299999999999998e-05,
15
+ "iters": 4,
16
+ "steps_per_s": 21.84,
17
+ "flops_cum": 8828127019008.0
18
+ },
19
+ {
20
+ "step": 200,
21
+ "loss": 4.1791,
22
+ "lr": 6.0299999999999995e-05,
23
+ "iters": 4,
24
+ "steps_per_s": 21.61,
25
+ "flops_cum": 17568846839808.0
26
+ },
27
+ {
28
+ "step": 300,
29
+ "loss": 4.1806,
30
+ "lr": 9.029999999999999e-05,
31
+ "iters": 4,
32
+ "steps_per_s": 21.59,
33
+ "flops_cum": 26309566660608.0
34
+ },
35
+ {
36
+ "step": 400,
37
+ "loss": 4.1627,
38
+ "lr": 0.0001203,
39
+ "iters": 4,
40
+ "steps_per_s": 21.56,
41
+ "flops_cum": 35050286481408.0
42
+ },
43
+ {
44
+ "step": 500,
45
+ "loss": 4.1601,
46
+ "lr": 0.0001503,
47
+ "iters": 4,
48
+ "steps_per_s": 21.55,
49
+ "flops_cum": 43791006302208.0
50
+ },
51
+ {
52
+ "step": 600,
53
+ "loss": 4.1692,
54
+ "lr": 0.0001803,
55
+ "iters": 4,
56
+ "steps_per_s": 21.53,
57
+ "flops_cum": 52531726123008.0
58
+ },
59
+ {
60
+ "step": 700,
61
+ "loss": 4.1646,
62
+ "lr": 0.0002103,
63
+ "iters": 4,
64
+ "steps_per_s": 21.56,
65
+ "flops_cum": 61272445943808.0
66
+ },
67
+ {
68
+ "step": 800,
69
+ "loss": 4.1573,
70
+ "lr": 0.00024029999999999999,
71
+ "iters": 4,
72
+ "steps_per_s": 21.52,
73
+ "flops_cum": 70013165764608.0
74
+ },
75
+ {
76
+ "step": 900,
77
+ "loss": 4.1607,
78
+ "lr": 0.0002703,
79
+ "iters": 4,
80
+ "steps_per_s": 21.5,
81
+ "flops_cum": 78753885585408.0
82
+ },
83
+ {
84
+ "step": 1000,
85
+ "loss": 4.147,
86
+ "lr": 0.0003,
87
+ "iters": 4,
88
+ "steps_per_s": 21.47,
89
+ "flops_cum": 87494605406208.0
90
+ },
91
+ {
92
+ "step": 1100,
93
+ "loss": 4.1478,
94
+ "lr": 0.00029815325108927063,
95
+ "iters": 4,
96
+ "steps_per_s": 21.5,
97
+ "flops_cum": 96235325227008.0
98
+ },
99
+ {
100
+ "step": 1200,
101
+ "loss": 4.1503,
102
+ "lr": 0.00029265847744427303,
103
+ "iters": 4,
104
+ "steps_per_s": 21.5,
105
+ "flops_cum": 104976045047808.0
106
+ },
107
+ {
108
+ "step": 1300,
109
+ "loss": 4.1672,
110
+ "lr": 0.00028365097862825513,
111
+ "iters": 4,
112
+ "steps_per_s": 21.47,
113
+ "flops_cum": 113716764868608.0
114
+ },
115
+ {
116
+ "step": 1400,
117
+ "loss": 4.1367,
118
+ "lr": 0.0002713525491562421,
119
+ "iters": 4,
120
+ "steps_per_s": 21.5,
121
+ "flops_cum": 122457484689408.0
122
+ },
123
+ {
124
+ "step": 1500,
125
+ "loss": 4.1212,
126
+ "lr": 0.00025606601717798207,
127
+ "iters": 4,
128
+ "steps_per_s": 21.48,
129
+ "flops_cum": 131198204510208.0
130
+ },
131
+ {
132
+ "step": 1600,
133
+ "loss": 4.1083,
134
+ "lr": 0.00023816778784387094,
135
+ "iters": 4,
136
+ "steps_per_s": 21.5,
137
+ "flops_cum": 139938924331008.0
138
+ },
139
+ {
140
+ "step": 1700,
141
+ "loss": 4.0927,
142
+ "lr": 0.00021809857496093199,
143
+ "iters": 4,
144
+ "steps_per_s": 21.46,
145
+ "flops_cum": 148679644151808.0
146
+ },
147
+ {
148
+ "step": 1800,
149
+ "loss": 4.0778,
150
+ "lr": 0.0001963525491562421,
151
+ "iters": 4,
152
+ "steps_per_s": 21.47,
153
+ "flops_cum": 157420363972608.0
154
+ },
155
+ {
156
+ "step": 1900,
157
+ "loss": 4.0498,
158
+ "lr": 0.00017346516975603462,
159
+ "iters": 4,
160
+ "steps_per_s": 21.45,
161
+ "flops_cum": 166161083793408.0
162
+ },
163
+ {
164
+ "step": 2000,
165
+ "loss": 4.0827,
166
+ "lr": 0.00015,
167
+ "iters": 4,
168
+ "steps_per_s": 21.47,
169
+ "flops_cum": 174901803614208.0
170
+ },
171
+ {
172
+ "step": 2100,
173
+ "loss": 4.0476,
174
+ "lr": 0.00012653483024396533,
175
+ "iters": 4,
176
+ "steps_per_s": 21.48,
177
+ "flops_cum": 183642523435008.0
178
+ },
179
+ {
180
+ "step": 2200,
181
+ "loss": 4.0813,
182
+ "lr": 0.0001036474508437579,
183
+ "iters": 4,
184
+ "steps_per_s": 21.48,
185
+ "flops_cum": 192383243255808.0
186
+ },
187
+ {
188
+ "step": 2300,
189
+ "loss": 4.0611,
190
+ "lr": 8.190142503906798e-05,
191
+ "iters": 4,
192
+ "steps_per_s": 21.44,
193
+ "flops_cum": 201123963076608.0
194
+ },
195
+ {
196
+ "step": 2400,
197
+ "loss": 4.1178,
198
+ "lr": 6.183221215612904e-05,
199
+ "iters": 4,
200
+ "steps_per_s": 21.48,
201
+ "flops_cum": 209864682897408.0
202
+ },
203
+ {
204
+ "step": 2500,
205
+ "loss": 4.0966,
206
+ "lr": 4.3933982822017876e-05,
207
+ "iters": 4,
208
+ "steps_per_s": 20.1,
209
+ "flops_cum": 218605402718208.0
210
+ },
211
+ {
212
+ "step": 2600,
213
+ "loss": 4.106,
214
+ "lr": 2.8647450843757897e-05,
215
+ "iters": 4,
216
+ "steps_per_s": 21.5,
217
+ "flops_cum": 227346122539008.0
218
+ },
219
+ {
220
+ "step": 2700,
221
+ "loss": 4.0803,
222
+ "lr": 1.634902137174483e-05,
223
+ "iters": 4,
224
+ "steps_per_s": 21.46,
225
+ "flops_cum": 236086842359808.0
226
+ },
227
+ {
228
+ "step": 2800,
229
+ "loss": 4.0843,
230
+ "lr": 7.34152255572697e-06,
231
+ "iters": 4,
232
+ "steps_per_s": 21.44,
233
+ "flops_cum": 244827562180608.0
234
+ },
235
+ {
236
+ "step": 2900,
237
+ "loss": 4.0506,
238
+ "lr": 1.8467489107293509e-06,
239
+ "iters": 4,
240
+ "steps_per_s": 21.46,
241
+ "flops_cum": 253568282001408.0
242
+ },
243
+ {
244
+ "step": 2999,
245
+ "loss": 4.092,
246
+ "lr": 1.8505504447152708e-10,
247
+ "iters": 4,
248
+ "steps_per_s": 21.7,
249
+ "flops_cum": 262221594624000.0
250
+ }
251
+ ],
252
+ "id_acc": [
253
+ {
254
+ "step": 2500,
255
+ "mean": 0.04187499871477485,
256
+ "per_k": {
257
+ "1": 0.03749999776482582,
258
+ "2": 0.044999998062849045,
259
+ "3": 0.03999999910593033,
260
+ "4": 0.042500000447034836,
261
+ "5": 0.04999999701976776,
262
+ "6": 0.03500000014901161,
263
+ "7": 0.03749999776482582,
264
+ "8": 0.04749999940395355
265
+ }
266
+ },
267
+ {
268
+ "step": 3000,
269
+ "mean": 0.048124998807907104,
270
+ "per_k": {
271
+ "1": 0.03500000014901161,
272
+ "2": 0.03500000014901161,
273
+ "3": 0.044999998062849045,
274
+ "4": 0.04749999940395355,
275
+ "5": 0.05999999865889549,
276
+ "6": 0.044999998062849045,
277
+ "7": 0.05999999865889549,
278
+ "8": 0.057499997317790985
279
+ }
280
+ }
281
+ ],
282
+ "ood": [
283
+ {
284
+ "step": 2500,
285
+ "acc": {
286
+ "16": 0.04749999940395355,
287
+ "32": 0.05249999836087227,
288
+ "64": 0.019999999552965164
289
+ }
290
+ },
291
+ {
292
+ "step": 3000,
293
+ "acc": {
294
+ "16": 0.04999999701976776,
295
+ "32": 0.03500000014901161,
296
+ "64": 0.06499999761581421
297
+ }
298
+ }
299
+ ],
300
+ "iters": [],
301
+ "flops_cum": 262221594624000.0
302
+ }