Andy Ehrenberg

I haven't gotten any videos to show, and yes, I'm on the 0.7 branch running Julia 1.1. Perhaps I can switch to the older version and just run by trained...

I see a few other instances in this repo where the pytorch implementation computes `past_key_values_length` while the flax implementation uses `position_ids` (BART, OPT, etc) - to me, keeping consistency among...

> Mmmm, before merging this PR, there is something wrong going on with the tests: only one of the tests job is actually run (no tests_flax/tests_tf etc...) > > Will...

@sanchit-gandhi - How can I rerun the checks without further commits? The error looks like an account limit overshoot and doesn't seem to do with the two newer commits.

@sanchit-gandhi I think it's ready for another look by you! The torch tests it's failing current seem unrelated to the PR, so rerunning CI may give all passes

> Also sorry! We just modified Whisper quit a bit 😅 @ArthurZucker - Doesn't actually look too bad to catch up with those changes! Can do that soon-ish. I already...

@sanchit-gandhi - I think this is ready for another look - the recent commits (I think) get us to feature parity with the torch version.

@sanchit-gandhi Bump

@sanchit-gandhi @ArthurZucker - Addressed Arthur's comments and cleaned up the timestamp logits processor a bit. Hopefully we're close to getting this merged!

> Very nice @andyehrenberg! Thanks for iterating here - reviewed the new changes and the PR is looking super clean. Last request from me is if we can avoid defining...