Open Source Deep Research Alternative to Reason and Search on Private Data. Written in Python.

Simple extension on vLLM to help you speed up reasoning model without training.

Agentic Deep Graph Reasoning Implementation

Lightweight replication study of DeepSeek-R1-Zero. Interesting findings include "No Aha Moment", "Longer CoT ≠ Accuracy", and "Language Mixing in Instruct Models".

Pure RL to post-train base models for social reasoning capabilities. Lightweight replication of DeepSeek-R1-Zero with Social IQa dataset.