[FEATURE] Add a new json field `pruna_version` to smash_config.json

[ParagEkbote]

Is your feature request related to a problem? Please describe.

In order to better diagnose user issues or track regression in algorithms or quantizers, we could add a field like pruna_version which shows the version of pruna used when the model was smashed to smash_config,json . WDYT?

cc: @davidberenstein1957, @johannaSommer

Describe the solution you'd like

Add field of pruna_version to smash_config.json to better track issues in smashed models.

[davidberenstein1957]

#245

[davidberenstein1957]

#247

[davidberenstein1957]

Hi, this would be great. This issue had already been briefly tackled, but should be re-implemented. We would need to provide a warning on the version mismatch by default, and also mention this could cause issues during inference f this is the case, while recommending users to install an older version.

[davidberenstein1957]

I think that would be great! If you want, you can propose an implementations for an approach to tackling this, like outlining the classes you expect to change and the files structure. I am not sure what diffusers, etc, do for versioning, but we can keep that as an outline IMO. @minettekaum and I can then review the proposed solution but before you implement it :)

[github-actions]

This issue has been inactive for 21 days and is now marked as stale. It will be closed in 14 days if there is no further activity.