ZLiao097/AscendExamples
save ascend best examples
save ascend best examples
verl: Volcano Engine Reinforcement Learning for LLMs
Training library for Megatron-based models with bidirectional Hugging Face conversion capability
Ongoing research training transformer models at scale
Provide performance insight capabilities for RL frameworks.
verl data curation
A high-throughput and memory-efficient inference and serving engine for LLMs
Community maintained hardware plugin for vLLM on Ascend