paper-buf

• LLM safety
  • Gandalf the red: Adaptive security for LLMs, Jan. 14 2025.
  • AEGIS2.0: A diverse ai safety dataset and risks taxonomy for alignment of LLM guardrails
  • NOTES
• Preference alignment
  • Clone-robust ai alignment
• Others
  • Do transformers really perform bad for graph representation? code

Your eval is more important than the model

• A blogpost that teaches us how to evaluate in a scientific way.

KAG

• “KAG is a logical reasoning and Q&A framework based on the OpenSPG engine and large language models, which is used to build logical reasoning and Q&A solutions for vertical domain knowledge bases. KAG can effectively overcome the ambiguity of traditional RAG vector similarity calculation and the noise problem of GraphRAG introduced by OpenIE. KAG supports logical reasoning and multi-hop fact Q&A, etc., and is significantly better than the current SOTA method.”