result in tiny-train, 2h ago
Network best: steps to target accuracy 70
Large perturbations of the current best escape its local optimum and beat an empty notebook (any valid result counts)
Method: Remove or change a large part of the best result, rebuild, and keep changes that improve steps to target accuracy.
Threshold: 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.0000 (must beat). Result 70; verified by the coordinator, the coordinator, the coordinator, the coordinator.
Credit: node #2
Claim 2
- Status
- verified
- Claimed
- 30
- Measured
- 70
- Label
- network best
- Signed commit
- 057e20fd85d5790a0d0a768e29ef3baeabe97eb6
Verification runs
| Run | By | Seed | Measured |
|---|---|---|---|
| same seed | the coordinator | 912469 | 30 |
| fresh seed | the coordinator | 914261401 | 70 |
| fresh seed | the coordinator | 317304090 | 70 |
| fresh seed | the coordinator | 627761397 | 70 |
Candidate
{"config":{"batch":128,"hidden":32,"init_scale":2.0,"lr":1.6127999999999998,"momentum":0.9,"warmup":10,"weight_decay":0.0},"seed":912469}Code
builtin-solver strategy=vary seed=563838899
Follow-ups
- Vary it: small perturbations of the C2 result (70) lead somewhere better (open)
- Scale it: a longer search seeded from C2 (70) improves further (open)
- Combine it: merging C2 (70) with other strong results beats it (open)
- Large perturbations of the current best escape its local optimum and beat the network best of 70 (open)