Charles Packer

@amitsgh if you're actively working on this you may want to check with @agiletechnologist on Discord since he's also working on adding some logging to a different PR

@davidkundrats what model wrapper + model are you using? I can try and replicate the issue + create a patch. Essentially some model wrappers will do prompt formatting / parsing...

Reopening - the autogen version of this bug will be patched in `0.3.5`. Non-autogen bug should be patched in `0.3.4`.

@sarahwooders issue got auto-closed by #1098, not sure if intended (will reopen can close if so)

> > LGTM (pending tests), thank you! > > Thanks. I don't know what to do (if anything) about the checks that failed. I'm not sure why the remaining tests...

Tests should be OK on main (probs have a bug in our tests @sarahwooders). Thank you for the PR @madgrizzle !!

@OswaldoBornemann can you try using the commit where the experiments were added to the repo [f239041](https://github.com/cpacker/MemGPT/commit/f2390418c4d4902a6ac98f7560fb774ea66855e6)? The MemGPT codebase is under active development and unfortunately we don't have the bandwidth...

@sarahwooders looks like this is the same issue popping up again with tests failing on contrib PRs