DQN for Atari Breakout
From scratch implementation of some iconic RL techniques for solving Atari Games
Mathematical Reasoning with GRPO
RL-based post-training method for increasing reasoning capabilities in LLMs
Don't Change My View!
Ideological Bias Auditing in LLMs
Streaminator
Multi-Answer Speculative Decoding for effficent LLM Inference
Schokoban
Monte Carlo Tree Search for Solving Sokoban
C++ Autograd
Basic C++ Autograd Engine written in C++
Torchify
Compiling a json-like file to a torch.nn.Module