AbstractPhil commited on
Commit
a399d9a
·
verified ·
1 Parent(s): 8d5c3fe

exp007: pmix8_dense_s5678

Browse files
experiments/exp_007_aleph_routed_attention/runs/pmix8_dense_s5678/checkpoint.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f18886405668698d2f43206dba7d611cc4077bf27b087552b47c3f27ef0fbd69
3
+ size 28088373
experiments/exp_007_aleph_routed_attention/runs/pmix8_dense_s5678/result.json ADDED
@@ -0,0 +1,365 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run": "pmix8_dense_s5678",
3
+ "tag": "pmix8_dense",
4
+ "seed": 5678,
5
+ "dev_final": 0.0066835880279541016,
6
+ "statute_final": "uniform",
7
+ "params": 6922059,
8
+ "minutes": 9.3,
9
+ "bpb": 2.3019221990811647,
10
+ "bpb_eval": NaN,
11
+ "bpb_inbank": NaN,
12
+ "coverage": 0.8013916015625,
13
+ "gate_acc": 0.8504638671875,
14
+ "bpb_bank_cond": 1.9705166756929577,
15
+ "bpb_byte_cond": 2.7855830460374116,
16
+ "mode_spread_deg": 85.92856089117942,
17
+ "mix_entropy": 0.9889900174049001,
18
+ "write_kl": 0.12015839666128159,
19
+ "nn_exact": NaN,
20
+ "ang_err_deg": NaN,
21
+ "budget_deg": NaN,
22
+ "in_budget": NaN,
23
+ "conf_mean": 0.07354755699634552,
24
+ "kernel_invariant": 0.8546377420425415,
25
+ "branching_frac": 1.0,
26
+ "loss": 4.7987284660339355,
27
+ "statute_trajectory": [
28
+ [
29
+ 0,
30
+ {
31
+ "deviation": 0.006669402122497559,
32
+ "pair_fraction": 0.5,
33
+ "statute": "uniform"
34
+ }
35
+ ],
36
+ [
37
+ 250,
38
+ {
39
+ "deviation": 0.0067861080169677734,
40
+ "pair_fraction": 0.53125,
41
+ "statute": "uniform"
42
+ }
43
+ ],
44
+ [
45
+ 500,
46
+ {
47
+ "deviation": 0.006870269775390625,
48
+ "pair_fraction": 0.5,
49
+ "statute": "uniform"
50
+ }
51
+ ],
52
+ [
53
+ 750,
54
+ {
55
+ "deviation": 0.006741762161254883,
56
+ "pair_fraction": 0.5,
57
+ "statute": "uniform"
58
+ }
59
+ ],
60
+ [
61
+ 1000,
62
+ {
63
+ "deviation": 0.006844282150268555,
64
+ "pair_fraction": 0.5,
65
+ "statute": "uniform"
66
+ }
67
+ ],
68
+ [
69
+ 1250,
70
+ {
71
+ "deviation": 0.006862759590148926,
72
+ "pair_fraction": 0.5,
73
+ "statute": "uniform"
74
+ }
75
+ ],
76
+ [
77
+ 1500,
78
+ {
79
+ "deviation": 0.006886124610900879,
80
+ "pair_fraction": 0.5,
81
+ "statute": "uniform"
82
+ }
83
+ ],
84
+ [
85
+ 1750,
86
+ {
87
+ "deviation": 0.006677508354187012,
88
+ "pair_fraction": 0.5,
89
+ "statute": "uniform"
90
+ }
91
+ ],
92
+ [
93
+ 2000,
94
+ {
95
+ "deviation": 0.006614208221435547,
96
+ "pair_fraction": 0.5,
97
+ "statute": "uniform"
98
+ }
99
+ ],
100
+ [
101
+ 2250,
102
+ {
103
+ "deviation": 0.006497025489807129,
104
+ "pair_fraction": 0.5,
105
+ "statute": "uniform"
106
+ }
107
+ ],
108
+ [
109
+ 2500,
110
+ {
111
+ "deviation": 0.006327986717224121,
112
+ "pair_fraction": 0.5,
113
+ "statute": "uniform"
114
+ }
115
+ ],
116
+ [
117
+ 2750,
118
+ {
119
+ "deviation": 0.006353139877319336,
120
+ "pair_fraction": 0.5,
121
+ "statute": "uniform"
122
+ }
123
+ ],
124
+ [
125
+ 3000,
126
+ {
127
+ "deviation": 0.006471037864685059,
128
+ "pair_fraction": 0.5,
129
+ "statute": "uniform"
130
+ }
131
+ ],
132
+ [
133
+ 3250,
134
+ {
135
+ "deviation": 0.006569385528564453,
136
+ "pair_fraction": 0.4375,
137
+ "statute": "uniform"
138
+ }
139
+ ],
140
+ [
141
+ 3500,
142
+ {
143
+ "deviation": 0.006683349609375,
144
+ "pair_fraction": 0.4375,
145
+ "statute": "uniform"
146
+ }
147
+ ],
148
+ [
149
+ 3750,
150
+ {
151
+ "deviation": 0.006607413291931152,
152
+ "pair_fraction": 0.4375,
153
+ "statute": "uniform"
154
+ }
155
+ ],
156
+ [
157
+ 4000,
158
+ {
159
+ "deviation": 0.006688952445983887,
160
+ "pair_fraction": 0.4375,
161
+ "statute": "uniform"
162
+ }
163
+ ],
164
+ [
165
+ 4250,
166
+ {
167
+ "deviation": 0.006772279739379883,
168
+ "pair_fraction": 0.4375,
169
+ "statute": "uniform"
170
+ }
171
+ ],
172
+ [
173
+ 4500,
174
+ {
175
+ "deviation": 0.006843447685241699,
176
+ "pair_fraction": 0.4375,
177
+ "statute": "uniform"
178
+ }
179
+ ],
180
+ [
181
+ 4750,
182
+ {
183
+ "deviation": 0.006940007209777832,
184
+ "pair_fraction": 0.4375,
185
+ "statute": "uniform"
186
+ }
187
+ ],
188
+ [
189
+ 5000,
190
+ {
191
+ "deviation": 0.0070018768310546875,
192
+ "pair_fraction": 0.46875,
193
+ "statute": "uniform"
194
+ }
195
+ ],
196
+ [
197
+ 5250,
198
+ {
199
+ "deviation": 0.007055997848510742,
200
+ "pair_fraction": 0.46875,
201
+ "statute": "uniform"
202
+ }
203
+ ],
204
+ [
205
+ 5500,
206
+ {
207
+ "deviation": 0.007123827934265137,
208
+ "pair_fraction": 0.46875,
209
+ "statute": "uniform"
210
+ }
211
+ ],
212
+ [
213
+ 5750,
214
+ {
215
+ "deviation": 0.00712132453918457,
216
+ "pair_fraction": 0.46875,
217
+ "statute": "uniform"
218
+ }
219
+ ],
220
+ [
221
+ 6000,
222
+ {
223
+ "deviation": 0.00708460807800293,
224
+ "pair_fraction": 0.46875,
225
+ "statute": "uniform"
226
+ }
227
+ ],
228
+ [
229
+ 6250,
230
+ {
231
+ "deviation": 0.007074117660522461,
232
+ "pair_fraction": 0.46875,
233
+ "statute": "uniform"
234
+ }
235
+ ],
236
+ [
237
+ 6500,
238
+ {
239
+ "deviation": 0.007056236267089844,
240
+ "pair_fraction": 0.46875,
241
+ "statute": "uniform"
242
+ }
243
+ ],
244
+ [
245
+ 6750,
246
+ {
247
+ "deviation": 0.007012724876403809,
248
+ "pair_fraction": 0.46875,
249
+ "statute": "uniform"
250
+ }
251
+ ],
252
+ [
253
+ 7000,
254
+ {
255
+ "deviation": 0.006988644599914551,
256
+ "pair_fraction": 0.4375,
257
+ "statute": "uniform"
258
+ }
259
+ ],
260
+ [
261
+ 7250,
262
+ {
263
+ "deviation": 0.006934642791748047,
264
+ "pair_fraction": 0.4375,
265
+ "statute": "uniform"
266
+ }
267
+ ],
268
+ [
269
+ 7500,
270
+ {
271
+ "deviation": 0.00688016414642334,
272
+ "pair_fraction": 0.4375,
273
+ "statute": "uniform"
274
+ }
275
+ ],
276
+ [
277
+ 7750,
278
+ {
279
+ "deviation": 0.006851315498352051,
280
+ "pair_fraction": 0.40625,
281
+ "statute": "uniform"
282
+ }
283
+ ],
284
+ [
285
+ 8000,
286
+ {
287
+ "deviation": 0.006829619407653809,
288
+ "pair_fraction": 0.40625,
289
+ "statute": "uniform"
290
+ }
291
+ ],
292
+ [
293
+ 8250,
294
+ {
295
+ "deviation": 0.00678253173828125,
296
+ "pair_fraction": 0.40625,
297
+ "statute": "uniform"
298
+ }
299
+ ],
300
+ [
301
+ 8500,
302
+ {
303
+ "deviation": 0.006778120994567871,
304
+ "pair_fraction": 0.40625,
305
+ "statute": "uniform"
306
+ }
307
+ ],
308
+ [
309
+ 8750,
310
+ {
311
+ "deviation": 0.006759166717529297,
312
+ "pair_fraction": 0.40625,
313
+ "statute": "uniform"
314
+ }
315
+ ],
316
+ [
317
+ 9000,
318
+ {
319
+ "deviation": 0.006745100021362305,
320
+ "pair_fraction": 0.40625,
321
+ "statute": "uniform"
322
+ }
323
+ ],
324
+ [
325
+ 9250,
326
+ {
327
+ "deviation": 0.006735682487487793,
328
+ "pair_fraction": 0.40625,
329
+ "statute": "uniform"
330
+ }
331
+ ],
332
+ [
333
+ 9500,
334
+ {
335
+ "deviation": 0.0067206621170043945,
336
+ "pair_fraction": 0.40625,
337
+ "statute": "uniform"
338
+ }
339
+ ],
340
+ [
341
+ 9750,
342
+ {
343
+ "deviation": 0.006700277328491211,
344
+ "pair_fraction": 0.40625,
345
+ "statute": "uniform"
346
+ }
347
+ ],
348
+ [
349
+ 10000,
350
+ {
351
+ "deviation": 0.0066835880279541016,
352
+ "pair_fraction": 0.40625,
353
+ "statute": "uniform"
354
+ }
355
+ ]
356
+ ],
357
+ "condition": {
358
+ "tag": "pmix8_dense",
359
+ "head": "hybrid",
360
+ "bank_scorer": "pmix",
361
+ "n_pointers": 8,
362
+ "bank_source": "atlas/bank_16384_tri.pt"
363
+ },
364
+ "steps": 10000
365
+ }
experiments/exp_007_aleph_routed_attention/runs/pmix8_dense_s5678/snapshots.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc4574f3b178c53048be45717575d2ca21cd17178553c3194cc6c552bbff2be7
3
+ size 57350