Theia Vogel

I'm working on using Glicol to generate music from DNA, and IIRC what I was trying to do here was have the Javascript code that parses the DNA not have...

Thanks for the fix! Don't worry about the formatting CI failure, I can branch off this and fix stuff up.

Hmm, doing some testing with this. Unfortunately it seems to have reduced the silliness at the extremes of some of the vectors, e.g.: ``` ======== old honest @ -2 [INST]...

Awesome, very excited for VLLM support! The control vector gets added **after** the MLP/FC layer. However, I noticed you linked to Mixtral, not sure if that was intentional but FYI...

As far as I can tell, it's some sort of strange conflict between the preinstalled numpy version on colab and the version that repeng wants (as a transitive dependency). For...

is there a specific model you'd be interested in? It's not very hard to add support for models theoretically, so if there's a relatively-popular one you're interested in having supported...

@NanoCode012 the main issue with MoE models is that only a subset of experts are active for a certain forward pass (by design). `repeng` runs a bunch of forward passes...

You can export it for use with llama.cpp with `export_gguf`, but I haven't written a method to read that gguf file back in for HuggingFace use yet. In the meantime,...

Thanks for this--I'll need to take a deeper look at the tests, since I'm not sure why they're passing with the flipped vector (it's possible I reduced the training examples...

make sure you're running the notebook with cwd in the `notebooks` folder, the data folder is `notebooks/data`. alternatively you can just copy the data folder to wherever you need it...