Pinned Loading

1. open-r1 open-r1 Public

   A RL framework that supports latest LLMs and VLMs (added new rl features and new verifiers, enhanced profiling)

   Python

2. VLM-reward-hacking-detection VLM-reward-hacking-detection Public

   A framework to train soft tokens and a backbone VLM for detecting reward hacking in target VLMs.

   Python

3. flash-attn-economical-gpu flash-attn-economical-gpu Public

   A set of efficient flash attention implementations based on Triton, beating or reaching comparable performance with Pytorch's cuDNN-backed SDPA on Ampere and Turing GPUs.

   Python

4. recsys-retailrocket recsys-retailrocket Public

   RecSys model training framework on a challenging dataset (finished in 2 days with AI coding)

   Python

5. cs762_project cs762_project Public

   LLM reward hacking detection

   Python

6. FM FM Public

   An efficient implementation of Factorization Machines that supports order=3 features in linear time.

   Python