int/our_knp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
30 commits 3 branches 0 tags 176 MiB
Commit graph

30 commits

This branch
This branch
All branches
Author SHA1 Message Date
int
a8ea8beca0 Summary of findings and plan for next week's demo 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 07:13:02 +03:00
int
ee48cb87bd KNP_W_RANGE option; weight range does not explain object gap 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 06:56:14 +03:00
int
6bfdbf69d3 Delta-rule KNP: augmented MNIST robustness, object runs; report section 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 06:50:34 +03:00
int
917b94035d Delta-rule KNP results: MNIST 94.8%, cats/dogs 69.1% on validation 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 05:59:50 +03:00
int
64fc5161e0 Error-driven dopamine (delta-rule) training mode in KNP: network option, isolated backend 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 04:58:42 +03:00
int
39079670c0 K=1600 neural-rule pipeline result 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 02:32:32 +03:00
int
ba81f0384b Report: seed variation, neural-rule pipeline, page 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 02:12:58 +03:00
int
febc599ff2 End-to-end neural-rule pipeline result; seed-variation script 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 02:09:00 +03:00
int
1888412f31 Visualizer: stroke-feature network with spiking readout panel 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 01:58:22 +03:00
int
2f210b16f6 Column limits on learned features; dopamine-trained spiking readout with calibrated decision neuron 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 01:54:37 +03:00
int
ec913f51c1 Online WTA feature learning; configurable active steps in spiking pipeline 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 00:46:14 +03:00
int
899e0e3601 Night report: objects, spiking readout, punishment 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 00:44:24 +03:00
int
c2ac33767d One-time fresh-test evaluation of object pipelines 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 00:43:38 +03:00
int
8f126bc29a Exact spiking pipeline simulator, fresh test set, night log 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-15 00:13:52 +03:00
int
52d91e974f Cats/dogs v3 learned features for KNP, punishment in cats_dogs, readout variants 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 21:54:09 +03:00
int
cf386f4e36 Log night experiments: spiking readout, learned object features, punishment 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 21:17:39 +03:00
int
795f75a65b KNP_COL_PUNISH: dopamine punishment of wrong columns; learned k-means features for objects; extra data fetcher 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 21:06:28 +03:00
int
b8ea9dc6ce Spiking readout (off-chip ridge, AltAI-like integer inference) and robustness check 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 20:58:39 +03:00
int
7add19dcae Log augmentation A and column-size results 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 20:35:03 +03:00
int
ab54a7c8d5 Log coarse-feature results and classifier diagnosis 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 20:10:27 +03:00
int
334bfefa81 Coarse stroke-feature preset; log fine-feature results 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:49:56 +03:00
int
45880f1d19 Stroke-detector + pooling features prototype, robustness check in feature mode 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:26:07 +03:00
int
27279ab242 Visualizer: shift buttons and all-nets answer line; page with augmented net B 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:23:01 +03:00
int
e5d50c8d65 Log augmentation B robustness results 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:20:20 +03:00
int
1722175ac8 Fix resume.sh completion check, log full ON/OFF result, relaunch runs overnight 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:17:29 +03:00
int
b34bdae638 Log ON/OFF and robustness results, add resume script 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:10:08 +03:00
int
f477a6f588 Augmentation experiment: training-set augmenter and test-time shift voting for MNIST 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 19:07:59 +03:00
int
db4b1d08b8 ON/OFF input coding for mnist-learn: KNP_MNIST_INPUT_SIZE, dataset builder, simulator --pix 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 18:35:27 +03:00
int
979aa90bb4 Add 2026-10-14 handoff materials, verify accelerated runs 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 18:29:08 +03:00
int
8f43927575 Initial import: KNP experiments (examples, viz, cats-dogs, reports) 
examples/ is based on KasperskyLab/knp at 55470ed6 with local changes
to the mnist-learn altai/blifat models, cuda-backend CMake and model_loader.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
 2026-10-14 18:17:15 +03:00