Pytorch_block_sparse: smaller, faster neural networks

min(DALL·E) is a fast, minimal port of DALL·E Mini to PyTorch - kuprel/min-dalle

I ported Microsoft's TRELLIS.2 (4B parameter image-to-3D model) to run on Apple Silicon via PyTorch MPS. The original requires CUDA with flash_attn, nvdiffrast, and custom sparse convolution kernels: none of which work on Mac. I replaced the CUDA-specific ops with pure-PyTorch alternatives: a gather-scatter sparse 3D convolution, SDPA attention for sparse transformers, and a Python-based mesh extraction replacing CUDA hashmap operations. Total changes are a few hundred lines across 9 files. Generates ~400K vertex meshes from single photos in about 3.5 minutes on M4 Pro (24GB). Not as fast as…

Hello HN, We are pleased to introduce you graphlearn-for-pytorch (https://github.com/alibaba/graphlearn-for-pytorch), an open-source distributed graph neural network library based on PyTorch and compatible with PyG. Our library is designed to make it easy for developers to build and train large-scale graph models in a distributed environment. With graphlearn-for-pytorch, you can leverage GPUs to accelerate graph sampling and utilize UVA to reduce the overheads of feature collection. Following a scalable design, graphlearn-for-pytorch supports training GNN models on…

Fresh and (I think) clean implementations of Faster R-CNN in PyTorch and TensorFlow 2/Keras. I wanted to learn about object detectors and decided to understand and implement a foundational model in the field, Faster R-CNN (elements of which are still used in modern models to this day), using the paper alone. That proved to be more difficult than expected and I had to relent and take a peak at existing implementations to fill in some important gaps. I've documented my struggles and learnings in the README for others to benefit from. I also wanted to solidify my understanding of…

Blazing-fast in-browser neural networks

We're excited to announce that we've open-sourced LeanRL, a lightweight PyTorch reinforcement learning library that provides recipes for fast RL training using torch.compile and CUDA graphs. By leveraging these tools, we've achieved significant speed-ups compared to the original CleanRL implementations - up to 6x faster! Reinforcement learning is notoriously CPU-bound due to the high frequency of small CPU operations. PyTorch's powerful compiler can help alleviate these issues, but comes with its own costs. LeanRL addresses this challenge by providing simple recipes to accelerate your…