FEATURE: add optimization tracker

[yaylim]

Addresses issue #62. This PR adds an optional track_progress argument to NCRF.fit() and fit_ncrf() that records the internal optimization state at each iteration, allowing users to inspect how the model evolves during fitting.

[christianbrodbeck]

Why do we need a decorator for this as opposed to a normal argument? I am somewhat worried about the opaqueness of this (a hidden track_progress argument that does not appear in the method signature). This makes it harder to read and understand what's goin on.

[yaylim]

I wanted to write it as a decorator to separate the track_progress out of core fitting logic. I can refactor this to add track_progress=False and tracker=None as explicit arguments directly in the NCRF.fit() and fit_ncrf() signatures, and remove the decorator. Would that make it more understandable?

[christianbrodbeck]

Yes I think that would make it easier to readOn Jun 17, 2026, at 2:25 PM, yaylim ***@***.***> wrote:yaylim left a comment (Eelbrain/neuro-currentRF#63) I wanted to write it as a decorator to separate the track_progress out of core fitting logic. I can refactor this to add track_progress=False and tracker=None as explicit arguments directly in the NCRF.fit() and fit_ncrf() signatures, and remove the decorator. Would that make it more understandable? —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because your review was requested.Message ID: ***@***.***>