Pinned Loading

1. nanoMoEchat nanoMoEchat Public

   Forked from karpathy/nanochat

   The best ChatGPT that $100 can buy, but an MoE!

   Python 1

2. megablocks-variable megablocks-variable Public

   Forked from databricks/megablocks

   megablocks with variable expert support

   Python

3. nanoMOE nanoMOE Public

   Forked from karpathy/nanoGPT

   Implementing Sparse MoE on Karpathy's nanoGPT.

   Python 3

4. open-concept-steering open-concept-steering Public

   SAEs for all LLMs

   Jupyter Notebook 2 1

5. variable-flex-olmo variable-flex-olmo Public

   Megablocks-based MoE implementation for FlexOlmo models with support for variable-sized experts

   Python