Popular repositories Loading

1. circuit-motifs circuit-motifs Public

   Network motif analysis of LLM attribution graphs — structural fingerprints of neural computation

   Python 1

2. r1-interp r1-interp Public

   Interpretability toolkit for DeepSeek-R1-Distill-Qwen-1.5B: skip transcoders, attribution graphs, and circuit motif analysis

   Python

3. parameter-golf parameter-golf Public

   Forked from openai/parameter-golf

   Train the smallest LM you can that fits in 16MB. Best model wins!

   Python