Efficient Triton Kernels for LLM Training

Scalable toolkit for efficient model reinforcement

PyTorch native post-training library

Student version of Assignment 1 for Stanford CS336 - Language Modeling From Scratch

VeOmni: Scaling any Modality Model Training to any Accelerators with PyTorch native Training Framework

Qwen2.5-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.

Verifiers for LLM Reinforcement Learning

Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)

Qwen2.5-Coder is the code version of Qwen2.5, the large language model series developed by Qwen team, Alibaba Cloud.

Evolutionary Scale Modeling (esm): Pretrained language models for proteins

[NeurIPS 2022] Denoising Diffusion Restoration Models -- Official Code Repository

Build-your-own DiffuserCam tutorial