Jiaquan Zhang, Qigan Sun, Chaoning Zhang, Xudong Wang, Zhenzhen Huang, Yitian Zhou, Pengcheng Zheng, Chi-lok Andy Tai, Sung-Ho Bae, Zeyu Ma, Caiyan Qin, Jinyu Guo, Yang Yang, Hengtao Shen
View original ↗Build a library that implements TDA-RC to allow developers to insert logical alignment steps into existing chain-of-thought pipelines. This addresses the common issue of logical drift in multi-step reasoning tasks.
Suggested repo: tda-chain
"Stop your LLM from hallucinating in the middle of a reasoning chain."
Estimated effort: 40h