Frozen observer fits: linearised coordinate bank
512 archived A,b coefficient pairs with fitted rank, representative and projector. These are per-case observer fits, not one general pretrained filter. Run python infer.py for the first fit at zero intervention. predict_linearised(fit,u) returns u A^T+b for finite [n,2] interventions.
The diagnostic export alone omitted link labels. CP05 recovered all 512 exact labels from the corresponding original COMPLETE records and verified A,b equality before joining by family/observer/policy. predict_response(fit,u) now applies the recovered linear/tanh link in the original known-reset response coordinate. It is not a raw EEG or arbitrary sensor interface. predict_linearised remains available. Use the separate G04 input-validating fitter to identify a new admissible observer from actual probes. The affine minimum-norm representative is not recovery of unobservable coordinates.
The source row ordering, family, observer and policy are retained. No protected raw observations, private seeds, patient records, new weights or new license are supplied. Public distribution requires author rights approval.
Dependencies and reload check
Install requirements.txt, then run python validate.py to check every saved fit and input rejection. No new fitted data are generated.
Verified packaging revision — 2026-10-20
Preserved model weights reloaded; saved or synthetic example checks passed. No training, held-out evaluation, clinical or hosted verification.
Runtime state: PASS. Rights and release approval are separate. See RELEASE_VERIFICATION.json. No paper or conference submission is included as a new software result.
- Downloads last month
- 12