ChineseHarm-Bench: A Chinese Harmful Content Detection Benchmark

KnowRL: Exploring Knowledgeable Reinforcement Learning for Factuality

Prototype for retrieval-augmented generation (RAG) with vector DBs and open-weight LLMs. Documents early experiments, design ideas, and multimodal model directions.

HcTAG (human-centric Thinking-Augmented Generation) — A prompt augmentation framework for guiding LLMs with layered, human-driven thinking before generation.