A high-throughput and memory-efficient inference and serving engine for LLMs

EasyR1: An Efficient, Scalable, Multi-Modality RL Training Framework based on veRL

A fast multimodal LLM for real-time voice

Entropy Based Sampling and Parallel CoT Decoding

A framework for efficient model inference with omni-modality models

VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo

Train speculative decoding models effortlessly and port them smoothly to SGLang serving.

Discrete Diffusion Forcing (D2F): dLLMs Can Do Faster-Than-AR Inference

Manages vllm-nccl dependency

Cross-platform transformer training