AbstractPhil commited on
Commit
6c23e2d
·
verified ·
1 Parent(s): ddbb219

exp007: kernel_dense_s5678

Browse files
experiments/exp_007_aleph_routed_attention/runs/kernel_dense_s5678/checkpoint.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6071a928b42be20d862afff67e7b616c30384cc391db3af1a6cb3d26580db90
3
+ size 27410933
experiments/exp_007_aleph_routed_attention/runs/kernel_dense_s5678/result.json ADDED
@@ -0,0 +1,364 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run": "kernel_dense_s5678",
3
+ "tag": "kernel_dense",
4
+ "seed": 5678,
5
+ "dev_final": 0.013119935989379883,
6
+ "statute_final": "uniform",
7
+ "params": 6753090,
8
+ "minutes": 4.8,
9
+ "bpb": 2.8164082991372568,
10
+ "bpb_eval": NaN,
11
+ "bpb_inbank": NaN,
12
+ "coverage": 0.8013916015625,
13
+ "gate_acc": 0.857177734375,
14
+ "bpb_bank_cond": 2.6271058487815573,
15
+ "bpb_byte_cond": 2.747468480263397,
16
+ "mode_spread_deg": NaN,
17
+ "mix_entropy": NaN,
18
+ "write_kl": 0.13203299045562744,
19
+ "nn_exact": NaN,
20
+ "ang_err_deg": NaN,
21
+ "budget_deg": NaN,
22
+ "in_budget": NaN,
23
+ "conf_mean": 0.4771309494972229,
24
+ "kernel_invariant": 0.8547848463058472,
25
+ "branching_frac": 0.94384765625,
26
+ "loss": 5.869759559631348,
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": 0.007070779800415039,
40
+ "pair_fraction": 0.5,
41
+ "statute": "uniform"
42
+ }
43
+ ],
44
+ [
45
+ 500,
46
+ {
47
+ "deviation": 0.007057666778564453,
48
+ "pair_fraction": 0.46875,
49
+ "statute": "uniform"
50
+ }
51
+ ],
52
+ [
53
+ 750,
54
+ {
55
+ "deviation": 0.007502555847167969,
56
+ "pair_fraction": 0.46875,
57
+ "statute": "uniform"
58
+ }
59
+ ],
60
+ [
61
+ 1000,
62
+ {
63
+ "deviation": 0.007939934730529785,
64
+ "pair_fraction": 0.46875,
65
+ "statute": "uniform"
66
+ }
67
+ ],
68
+ [
69
+ 1250,
70
+ {
71
+ "deviation": 0.00851583480834961,
72
+ "pair_fraction": 0.5,
73
+ "statute": "uniform"
74
+ }
75
+ ],
76
+ [
77
+ 1500,
78
+ {
79
+ "deviation": 0.009017586708068848,
80
+ "pair_fraction": 0.4375,
81
+ "statute": "uniform"
82
+ }
83
+ ],
84
+ [
85
+ 1750,
86
+ {
87
+ "deviation": 0.009596824645996094,
88
+ "pair_fraction": 0.40625,
89
+ "statute": "uniform"
90
+ }
91
+ ],
92
+ [
93
+ 2000,
94
+ {
95
+ "deviation": 0.009927988052368164,
96
+ "pair_fraction": 0.4375,
97
+ "statute": "uniform"
98
+ }
99
+ ],
100
+ [
101
+ 2250,
102
+ {
103
+ "deviation": 0.010076642036437988,
104
+ "pair_fraction": 0.46875,
105
+ "statute": "uniform"
106
+ }
107
+ ],
108
+ [
109
+ 2500,
110
+ {
111
+ "deviation": 0.010269641876220703,
112
+ "pair_fraction": 0.46875,
113
+ "statute": "uniform"
114
+ }
115
+ ],
116
+ [
117
+ 2750,
118
+ {
119
+ "deviation": 0.010623812675476074,
120
+ "pair_fraction": 0.40625,
121
+ "statute": "uniform"
122
+ }
123
+ ],
124
+ [
125
+ 3000,
126
+ {
127
+ "deviation": 0.010990142822265625,
128
+ "pair_fraction": 0.40625,
129
+ "statute": "uniform"
130
+ }
131
+ ],
132
+ [
133
+ 3250,
134
+ {
135
+ "deviation": 0.011507391929626465,
136
+ "pair_fraction": 0.40625,
137
+ "statute": "uniform"
138
+ }
139
+ ],
140
+ [
141
+ 3500,
142
+ {
143
+ "deviation": 0.011995911598205566,
144
+ "pair_fraction": 0.40625,
145
+ "statute": "uniform"
146
+ }
147
+ ],
148
+ [
149
+ 3750,
150
+ {
151
+ "deviation": 0.01232147216796875,
152
+ "pair_fraction": 0.4375,
153
+ "statute": "uniform"
154
+ }
155
+ ],
156
+ [
157
+ 4000,
158
+ {
159
+ "deviation": 0.012567877769470215,
160
+ "pair_fraction": 0.46875,
161
+ "statute": "uniform"
162
+ }
163
+ ],
164
+ [
165
+ 4250,
166
+ {
167
+ "deviation": 0.012787699699401855,
168
+ "pair_fraction": 0.5,
169
+ "statute": "uniform"
170
+ }
171
+ ],
172
+ [
173
+ 4500,
174
+ {
175
+ "deviation": 0.012992262840270996,
176
+ "pair_fraction": 0.46875,
177
+ "statute": "uniform"
178
+ }
179
+ ],
180
+ [
181
+ 4750,
182
+ {
183
+ "deviation": 0.013082623481750488,
184
+ "pair_fraction": 0.46875,
185
+ "statute": "uniform"
186
+ }
187
+ ],
188
+ [
189
+ 5000,
190
+ {
191
+ "deviation": 0.013216972351074219,
192
+ "pair_fraction": 0.46875,
193
+ "statute": "uniform"
194
+ }
195
+ ],
196
+ [
197
+ 5250,
198
+ {
199
+ "deviation": 0.013332009315490723,
200
+ "pair_fraction": 0.46875,
201
+ "statute": "uniform"
202
+ }
203
+ ],
204
+ [
205
+ 5500,
206
+ {
207
+ "deviation": 0.013417720794677734,
208
+ "pair_fraction": 0.4375,
209
+ "statute": "uniform"
210
+ }
211
+ ],
212
+ [
213
+ 5750,
214
+ {
215
+ "deviation": 0.013437271118164062,
216
+ "pair_fraction": 0.4375,
217
+ "statute": "uniform"
218
+ }
219
+ ],
220
+ [
221
+ 6000,
222
+ {
223
+ "deviation": 0.013429403305053711,
224
+ "pair_fraction": 0.4375,
225
+ "statute": "uniform"
226
+ }
227
+ ],
228
+ [
229
+ 6250,
230
+ {
231
+ "deviation": 0.01339113712310791,
232
+ "pair_fraction": 0.4375,
233
+ "statute": "uniform"
234
+ }
235
+ ],
236
+ [
237
+ 6500,
238
+ {
239
+ "deviation": 0.0133589506149292,
240
+ "pair_fraction": 0.4375,
241
+ "statute": "uniform"
242
+ }
243
+ ],
244
+ [
245
+ 6750,
246
+ {
247
+ "deviation": 0.013318061828613281,
248
+ "pair_fraction": 0.4375,
249
+ "statute": "uniform"
250
+ }
251
+ ],
252
+ [
253
+ 7000,
254
+ {
255
+ "deviation": 0.013268828392028809,
256
+ "pair_fraction": 0.4375,
257
+ "statute": "uniform"
258
+ }
259
+ ],
260
+ [
261
+ 7250,
262
+ {
263
+ "deviation": 0.013261914253234863,
264
+ "pair_fraction": 0.4375,
265
+ "statute": "uniform"
266
+ }
267
+ ],
268
+ [
269
+ 7500,
270
+ {
271
+ "deviation": 0.01323556900024414,
272
+ "pair_fraction": 0.4375,
273
+ "statute": "uniform"
274
+ }
275
+ ],
276
+ [
277
+ 7750,
278
+ {
279
+ "deviation": 0.013203740119934082,
280
+ "pair_fraction": 0.4375,
281
+ "statute": "uniform"
282
+ }
283
+ ],
284
+ [
285
+ 8000,
286
+ {
287
+ "deviation": 0.013183832168579102,
288
+ "pair_fraction": 0.40625,
289
+ "statute": "uniform"
290
+ }
291
+ ],
292
+ [
293
+ 8250,
294
+ {
295
+ "deviation": 0.013173937797546387,
296
+ "pair_fraction": 0.40625,
297
+ "statute": "uniform"
298
+ }
299
+ ],
300
+ [
301
+ 8500,
302
+ {
303
+ "deviation": 0.013164520263671875,
304
+ "pair_fraction": 0.40625,
305
+ "statute": "uniform"
306
+ }
307
+ ],
308
+ [
309
+ 8750,
310
+ {
311
+ "deviation": 0.013152241706848145,
312
+ "pair_fraction": 0.40625,
313
+ "statute": "uniform"
314
+ }
315
+ ],
316
+ [
317
+ 9000,
318
+ {
319
+ "deviation": 0.013147592544555664,
320
+ "pair_fraction": 0.375,
321
+ "statute": "uniform"
322
+ }
323
+ ],
324
+ [
325
+ 9250,
326
+ {
327
+ "deviation": 0.013133525848388672,
328
+ "pair_fraction": 0.375,
329
+ "statute": "uniform"
330
+ }
331
+ ],
332
+ [
333
+ 9500,
334
+ {
335
+ "deviation": 0.013127803802490234,
336
+ "pair_fraction": 0.375,
337
+ "statute": "uniform"
338
+ }
339
+ ],
340
+ [
341
+ 9750,
342
+ {
343
+ "deviation": 0.013128042221069336,
344
+ "pair_fraction": 0.375,
345
+ "statute": "uniform"
346
+ }
347
+ ],
348
+ [
349
+ 10000,
350
+ {
351
+ "deviation": 0.013119935989379883,
352
+ "pair_fraction": 0.375,
353
+ "statute": "uniform"
354
+ }
355
+ ]
356
+ ],
357
+ "condition": {
358
+ "tag": "kernel_dense",
359
+ "head": "hybrid",
360
+ "bank_scorer": "kernel",
361
+ "bank_source": "atlas/bank_16384_tri.pt"
362
+ },
363
+ "steps": 10000
364
+ }
experiments/exp_007_aleph_routed_attention/runs/kernel_dense_s5678/snapshots.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98863230a8fd252ff533aee5d7062817bb22102552ed7fa8095ba22b4b0d7d97
3
+ size 57350