Jiaqi Wang (王家琪)

Jiaqi Wang (王家琪)

🚀 Projects

MiniMax M3.1-Flash project image
MiniMax M3.1-Flash (Contributor: RL scaling, MOPD)
MiniMax's next-generation Flash model, a native multimodal MoE (428B total / 23B active, 1M context) delivering a generational leap in agentic tasks, coding, computer use, and cybersecurity.
[blog] [huggingface] [modelscope]
Trinity-RFT project image
Trinity-RFT (Contributor: on-policy distillation, multi-turn agentic RFT)
A general-purpose, unified framework for reinforcement fine-tuning (RFT) of LLMs. Contributed JSD-based on-policy distillation (#499) and the concatenated multi-turn SFT-then-RFT recipe for ALFWorld (#442).
[paper] [code] [docs] GitHub stars