PyTorch-native post-training at scale

H-Net: Hierarchical Network with Dynamic Chunking

Code for BLT research paper

Flash Attention in ~100 lines of CUDA (forward pass only)

Training LLMs with QLoRA + FSDP

Codebase for Merging Language Models (ICML 2024)

Robust recipes to align language models with human and AI preferences

TensorRT LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and supports state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. Tensor…