RAM is all you need

Anima Machina

PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.

Deep reinforcement learning without experience replay, target networks, or batch updates.

Official implementation of the paper: "ZClip: Adaptive Spike Mitigation for LLM Pre-Training".

Pytorch implementation of preconditioned stochastic gradient descent (Kron and affine preconditioner, low-rank approximation preconditioner and more)

Run Claude Code on OpenAI models

Gymnasium environment for Pokemon Red

PyTorch image models, scripts, pretrained weights -- ResNet, ResNeXT, EfficientNet, EfficientNetV2, NFNet, Vision Transformer, MixNet, MobileNet-V3/V2, RegNet, DPN, CSPNet, and more

Lightweight package that tracks and summarizes code changes using LLMs (Large Language Models)

An implementation of PSGD Kron second-order optimizer for PyTorch

A version of @KellerJordan's Muon that adapts to the scale of the gradients

Simplifying reinforcement learning for complex game environments

Entropy Based Sampling and Parallel CoT Decoding

supporting pytorch FSDP for optimizers

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Schedule-Free Optimization in PyTorch

probe for predicting future hiddenstates on gpt-2 vibes....

NanoGPT (124M) quality in 3.25B tokens

Flux diffusion model implementation using quantized fp8 matmul & remaining layers use faster half precision accumulate, which is ~2x faster on consumer devices.