Marius Pachitariu

Have you tried to split this unit by amplitude to see if it's really two units? Sometimes a small amplitude unit like that can look bimodal. Also, please send side...

Thanks, can you please show the good cluster view for this entire area like you did for Kilosort4? It's possible that individual cluster exmaples are segmented better by one algorithm...

Looks similar doesn't it? The area in 1800-2200 looks like there are plenty of units for both algorithms and the 2200-2800 is similarly sparse for both. That area 2200-2800 just...

I am not sure if this message means you agree with us, but basically to me it looks like both algorithms perform similarly in terms of how many good units...

This seems to be fixed in Kilosort4 which is live now. Please update and reopen this issue if you still have a problem.

I do the st%NT and histogram that. It shows clearly missing spikes for KS 2.5 and 3, but not 1, 2 and 4, though I haven't tested a lot of...

Sure, but I think the st%NT takes care of these confounds and gets more directly at the source of the problem. In the datasets I looked at, it was a...

Not sure I understand this argument because it is relatively indirect. It can't be a good thing to see the lines at the batch size, then you know for sure...

Ah, thank you so much Adrian, I think this might solve the mystery. I was indeed able to replicate the problem in KS2.5 with non-default batch size, and it's not...

@GaelleChapuis can you please check whether the batch size was changed for the IBL runs? It should be 65,600 in Kilosort 2, 2.5 and 3. You mention 65,536 above.