New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Create a variation with mixins of RNN / RWKV style components #365

Open

gkielian opened this issue Jan 19, 2025 · 0 comments

Collaborator

gkielian commented Jan 19, 2025 •

edited

Loading

There is even a fork of nanoGPT which might serve as a great reference:

https://github.com/BlinkDL/nanoRWKV

First we will have to make the attention modular, and allow a a b c style mixtures of attention layer types (with and without weight tying).

Afterwards we should be able to perform a Hymba like approach to mixing language model types.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment