v4.0.0-beta, patch 1
·
239 commits
to master
since this release
- Kilosort importer reads from NPY files instead of rez.mat (npy-matlab is required for this feature)
- spikesRaw and spikesFilt are written to disk as they are computed, rather than all at once, to save memory (restores v3 behavior)
- removes JRC/saveBinaries due to the above
- New parameter: extractAfterDetect allows you to specify whether you want your features computed after ALL spikes are detected or on a per-chunk, per-file basis.
- Fixes a bug in saveRes which caused fields of inconsistent size to be saved if clusters had been automerged.