Polyconvex rotation-frame Jacobian loss / bench_results.json
Mechanism confirmed, baseline not beaten
1{
2 "bench_report": {
3 "bench_version": 1,
4 "track": "deformation_registration",
5 "model": "coordinate_warp",
6 "metric_direction": "lower is better",
7 "n_seeds": 8,
8 "baseline": {
9 "best_config": {
10 "lr": 0.01,
11 "epochs": 8
12 },
13 "sweep": {
14 "0.001": [
15 0.27781713008880615,
16 0.3285594880580902,
17 0.29619064927101135,
18 0.34113916754722595,
19 0.34860163927078247,
20 0.3542182743549347,
21 0.31516358256340027,
22 0.28730571269989014
23 ],
24 "0.003": [
25 0.2060280293226242,
26 0.23130172491073608,
27 0.2125881314277649,
28 0.24440883100032806,
29 0.2516593039035797,
30 0.2537933588027954,
31 0.2329125702381134,
32 0.2088773399591446
33 ],
34 "0.01": [
35 0.1190430298447609,
36 0.13602468371391296,
37 0.14200977981090546,
38 0.1490132212638855,
39 0.14526495337486267,
40 0.18109995126724243,
41 0.1357964277267456,
42 0.14082112908363342
43 ]
44 },
45 "full": {
46 "per_seed": [
47 0.1190430298447609,
48 0.13602468371391296,
49 0.14200977981090546,
50 0.1490132212638855,
51 0.14526495337486267,
52 0.18109995126724243,
53 0.1357964277267456,
54 0.14082112908363342
55 ]
56 }
57 },
58 "idea": {
59 "best_config": {
60 "lr": 0.01,
61 "epochs": 8
62 },
63 "sweep": {
64 "0.001": [
65 0.27786731719970703,
66 0.3285939395427704,
67 0.29622170329093933,
68 0.34119054675102234,
69 0.3486487567424774,
70 0.35429301857948303,
71 0.31522244215011597,
72 0.2873333990573883
73 ],
74 "0.003": [
75 0.2074858844280243,
76 0.2321665734052658,
77 0.21403560042381287,
78 0.24621228873729706,
79 0.2530094385147095,
80 0.25567081570625305,
81 0.23396754264831543,
82 0.21000193059444427
83 ],
84 "0.01": [
85 0.1227923184633255,
86 0.14947034418582916,
87 0.13802805542945862,
88 0.15367528796195984,
89 0.1728336662054062,
90 0.18583709001541138,
91 0.14342737197875977,
92 0.14185279607772827
93 ]
94 },
95 "per_seed": [
96 0.1227923184633255,
97 0.14947034418582916,
98 0.13802805542945862,
99 0.15367528796195984,
100 0.1728336662054062,
101 0.18583709001541138,
102 0.14342737197875977,
103 0.14185279607772827
104 ]
105 },
106 "comparison": {
107 "delta_mean": 0.007355469278991222,
108 "idea_wins": 1,
109 "n_pairs": 8,
110 "per_seed_diffs": [
111 0.0037492886185646057,
112 0.013445660471916199,
113 -0.003981724381446838,
114 0.004662066698074341,
115 0.027568712830543518,
116 0.004737138748168945,
117 0.00763094425201416,
118 0.0010316669940948486
119 ],
120 "p_value": 0.0321,
121 "mde": 0.008003765508655707,
122 "mde_rel_pct": 5.572327803121244,
123 "verdict": "idea worse (significant)",
124 "system_worked": false
125 },
126 "custom_track": {
127 "name": "deformation_registration",
128 "file": "custom_deformation_track.py",
129 "domain": "geometry"
130 },
131 "mechanism_signature": {
132 "prediction": "lifted U/Cof(U)/det penalty improves conditioning and minimum determinant under large deformation",
133 "observed_baseline_mean_condition": 9.625483721494675,
134 "observed_idea_mean_condition": 3.6457439363002777,
135 "observed_baseline_min_det": -0.031383185618324205,
136 "observed_idea_min_det": 0.1291193552315235,
137 "observed_baseline_fold_fraction": 0.05500000016763806,
138 "observed_idea_fold_fraction": 0.0,
139 "confirmed": true
140 }
141 },
142 "math_check": {
143 "cofactor_max_abs": 7.152557373046875e-07,
144 "det_max_abs": 4.76837158203125e-07,
145 "rotation_det_max_abs": 2.384185791015625e-07
146 },
147 "all_sweeps": {
148 "baseline": {
149 "0.001": [
150 {
151 "metric": 0.27781713008880615,
152 "fold_fraction": 0.0,
153 "mean_condition": 1.1320418119430542,
154 "min_det": 0.920944094657898
155 },
156 {
157 "metric": 0.3285594880580902,
158 "fold_fraction": 0.0,
159 "mean_condition": 1.1260122060775757,
160 "min_det": 0.8810373544692993
161 },
162 {
163 "metric": 0.29619064927101135,
164 "fold_fraction": 0.0,
165 "mean_condition": 1.1461265087127686,
166 "min_det": 0.9009076952934265
167 },
168 {
169 "metric": 0.34113916754722595,
170 "fold_fraction": 0.0,
171 "mean_condition": 1.1431663036346436,
172 "min_det": 0.8950434923171997
173 },
174 {
175 "metric": 0.34860163927078247,
176 "fold_fraction": 0.0,
177 "mean_condition": 1.1379283666610718,
178 "min_det": 0.9005656242370605
179 },
180 {
181 "metric": 0.3542182743549347,
182 "fold_fraction": 0.0,
183 "mean_condition": 1.1388452053070068,
184 "min_det": 0.9023783802986145
185 },
186 {
187 "metric": 0.31516358256340027,
188 "fold_fraction": 0.0,
189 "mean_condition": 1.1190083026885986,
190 "min_det": 0.9096402525901794
191 },
192 {
193 "metric": 0.28730571269989014,
194 "fold_fraction": 0.0,
195 "mean_condition": 1.1269221305847168,
196 "min_det": 0.9133870601654053
197 }
198 ],
199 "0.003": [
200 {
201 "metric": 0.2060280293226242,
202 "fold_fraction": 0.0,
203 "mean_condition": 1.6335077285766602,
204 "min_det": 0.493784099817276
205 },
206 {
207 "metric": 0.23130172491073608,
208 "fold_fraction": 0.0,
209 "mean_condition": 1.7948778867721558,
210 "min_det": 0.41981199383735657
211 },
212 {
213 "metric": 0.2125881314277649,
214 "fold_fraction": 0.0,
215 "mean_condition": 1.8645418882369995,
216 "min_det": 0.43604058027267456
217 },
218 {
219 "metric": 0.24440883100032806,
220 "fold_fraction": 0.0,
221 "mean_condition": 1.7859004735946655,
222 "min_det": 0.42782339453697205
223 },
224 {
225 "metric": 0.2516593039035797,
226 "fold_fraction": 0.0,
227 "mean_condition": 1.7938549518585205,
228 "min_det": 0.43748193979263306
229 },
230 {
231 "metric": 0.2537933588027954,
232 "fold_fraction": 0.0,
233 "mean_condition": 1.668728232383728,
234 "min_det": 0.4448341429233551
235 },
236 {
237 "metric": 0.2329125702381134,
238 "fold_fraction": 0.0,
239 "mean_condition": 1.5834689140319824,
240 "min_det": 0.4970932900905609
241 },
242 {
243 "metric": 0.2088773399591446,
244 "fold_fraction": 0.0,
245 "mean_condition": 1.6454168558120728,
246 "min_det": 0.4864053726196289
247 }
248 ],
249 "0.01": [
250 {
251 "metric": 0.1190430298447609,
252 "fold_fraction": 0.0,
253 "mean_condition": 4.642986297607422,
254 "min_det": 0.08073875308036804
255 },
256 {
257 "metric": 0.13602468371391296,
258 "fold_fraction": 0.3449999988079071,
259 "mean_condition": 29.603721618652344,
260 "min_det": -0.3928036689758301
261 },
262 {
263 "metric": 0.14200977981090546,
264 "fold_fraction": 0.0,
265 "mean_condition": 8.519186019897461,
266 "min_det": 0.016476234421133995
267 },
268 {
269 "metric": 0.1490132212638855,
270 "fold_fraction": 0.0,
271 "mean_condition": 4.453977584838867,
272 "min_det": 0.09282688796520233
273 },
274 {
275 "metric": 0.14526495337486267,
276 "fold_fraction": 0.07500000298023224,
277 "mean_condition": 13.297748565673828,
278 "min_det": -0.09539034217596054
279 },
280 {
281 "metric": 0.18109995126724243,
282 "fold_fraction": 0.014999999664723873,
283 "mean_condition": 5.3841352462768555,
284 "min_det": -0.06205438822507858
285 },
286 {
287 "metric": 0.1357964277267456,
288 "fold_fraction": 0.0,
289 "mean_condition": 3.7512576580047607,
290 "min_det": 0.11113739013671875
291 },
292 {
293 "metric": 0.14082112908363342,
294 "fold_fraction": 0.004999999888241291,
295 "mean_condition": 7.350856781005859,
296 "min_det": -0.001996351173147559
297 }
298 ]
299 },
300 "idea": {
301 "0.001": [
302 {
303 "metric": 0.27786731719970703,
304 "fold_fraction": 0.0,
305 "mean_condition": 1.1346538066864014,
306 "min_det": 0.9233546257019043
307 },
308 {
309 "metric": 0.3285939395427704,
310 "fold_fraction": 0.0,
311 "mean_condition": 1.1273115873336792,
312 "min_det": 0.882387638092041
313 },
314 {
315 "metric": 0.29622170329093933,
316 "fold_fraction": 0.0,
317 "mean_condition": 1.1481126546859741,
318 "min_det": 0.9028756022453308
319 },
320 {
321 "metric": 0.34119054675102234,
322 "fold_fraction": 0.0,
323 "mean_condition": 1.145908236503601,
324 "min_det": 0.8978317975997925
325 },
326 {
327 "metric": 0.3486487567424774,
328 "fold_fraction": 0.0,
329 "mean_condition": 1.1401883363723755,
330 "min_det": 0.9029046297073364
331 },
332 {
333 "metric": 0.35429301857948303,
334 "fold_fraction": 0.0,
335 "mean_condition": 1.141953945159912,
336 "min_det": 0.9053830504417419
337 },
338 {
339 "metric": 0.31522244215011597,
340 "fold_fraction": 0.0,
341 "mean_condition": 1.1214745044708252,
342 "min_det": 0.9123570322990417
343 },
344 {
345 "metric": 0.2873333990573883,
346 "fold_fraction": 0.0,
347 "mean_condition": 1.1289077997207642,
348 "min_det": 0.9152964949607849
349 }
350 ],
351 "0.003": [
352 {
353 "metric": 0.2074858844280243,
354 "fold_fraction": 0.0,
355 "mean_condition": 1.7175836563110352,
356 "min_det": 0.5274192094802856
357 },
358 {
359 "metric": 0.2321665734052658,
360 "fold_fraction": 0.0,
361 "mean_condition": 1.833182692527771,
362 "min_det": 0.4331795573234558
363 },
364 {
365 "metric": 0.21403560042381287,
366 "fold_fraction": 0.0,
367 "mean_condition": 1.9415920972824097,
368 "min_det": 0.4649915099143982
369 },
370 {
371 "metric": 0.24621228873729706,
372 "fold_fraction": 0.0,
373 "mean_condition": 1.8535460233688354,
374 "min_det": 0.4599893093109131
375 },
376 {
377 "metric": 0.2530094385147095,
378 "fold_fraction": 0.0,
379 "mean_condition": 1.8582817316055298,
380 "min_det": 0.46327176690101624
381 },
382 {
383 "metric": 0.25567081570625305,
384 "fold_fraction": 0.0,
385 "mean_condition": 1.7491744756698608,
386 "min_det": 0.47469472885131836
387 },
388 {
389 "metric": 0.23396754264831543,
390 "fold_fraction": 0.0,
391 "mean_condition": 1.6311602592468262,
392 "min_det": 0.5201746225357056
393 },
394 {
395 "metric": 0.21000193059444427,
396 "fold_fraction": 0.0,
397 "mean_condition": 1.720042109489441,
398 "min_det": 0.512793242931366
399 }
400 ],
401 "0.01": [
402 {
403 "metric": 0.1227923184633255,
404 "fold_fraction": 0.0,
405 "mean_condition": 3.6080846786499023,
406 "min_det": 0.14010927081108093
407 },
408 {
409 "metric": 0.14947034418582916,
410 "fold_fraction": 0.0,
411 "mean_condition": 4.363121032714844,
412 "min_det": 0.06148712337017059
413 },
414 {
415 "metric": 0.13802805542945862,
416 "fold_fraction": 0.0,
417 "mean_condition": 4.669856071472168,
418 "min_det": 0.09186847507953644
419 },
420 {
421 "metric": 0.15367528796195984,
422 "fold_fraction": 0.0,
423 "mean_condition": 3.651010751724243,
424 "min_det": 0.1293972134590149
425 },
426 {
427 "metric": 0.1728336662054062,
428 "fold_fraction": 0.0,
429 "mean_condition": 3.4115381240844727,
430 "min_det": 0.15038658678531647
431 },
432 {
433 "metric": 0.18583709001541138,
434 "fold_fraction": 0.0,
435 "mean_condition": 2.81828236579895,
436 "min_det": 0.14253130555152893
437 },
438 {
439 "metric": 0.14342737197875977,
440 "fold_fraction": 0.0,
441 "mean_condition": 3.250553846359253,
442 "min_det": 0.1717192828655243
443 },
444 {
445 "metric": 0.14185279607772827,
446 "fold_fraction": 0.0,
447 "mean_condition": 3.3935046195983887,
448 "min_det": 0.14545558393001556
449 }
450 ]
451 }
452 },
453 "device": "cpu"
454}