I'm SnowCat

Reasearch & Projects

Sincere thanks to all my coauthors for their great work and support. Hope we can make more great things together.

• 

  ACL2026

  Streaming Hallucination Detection in Long Chain-of-Thought Reasoning

  Reasoning Model

  LLM Safety
• 

  ACL Industry Track 2026

  One-Eval: An Agentic System for Automated and Traceable LLM Evaluation

  DCAI

  Benchmark
• 

  CVPR2026

  Reallocating Attention Across Layers to Reduce Multimodal Hallucination

  MLLM

  Inner Safety
• 

  ICASSP2026

  Corrector: an execute-to-correct paradigm for efficient llm secure inference

  LLM

  Security
• 

  LoopLM

  LoopLM: A Self-Iterative Reasoning Framework for Large Language Models

  On Working Yet

Awards & Honors

The moon waxes only to wane; water fills only to overflow.

• Chinese Mathematics Competition

  First Prize, Beijing

  2025

  Beijing, Haidian

• Challenge Cup Innovation & Entrepreneurship Competition

  Third Prize, Beijing

  2025

  Beijing, Haidian

• National College Engineering Robot Competition

  National Excellence Award

  2025

  Beijing, Haidian

• College Engineering Innovation Competition

  First Prize, Beijing — Intelligent Rescue Track

  2025

  Beijing, Haidian

• BUPT Third-Class Scholarship

  2024

  Beijing, Haidian

• National College Network & IT Competition

  Third Prize, National — Track 3

  2024

  Beijing, Haidian

Hobbies & Contributions

Some side projects I built for fun. Most of them are small tools that scratch my own itch.

• One-Eval

  A subtree of DataFlow, focusing on automated evaluation of LLMs.

  2026-01

  Beijing

• Friberg

  An entropy-based script that solves the CS2 Friberg Game in the minimum number of guesses.

  2025-07

  Beijing

What am I doing recently?

I am going to writing some blogs to share my thoughts and ideas on AI safety, alignment, and related topics.