XingweiT/

SymbReAct-trace-sft

	Enhancing Logical Reasoning in Language Models via Symbolically-Guided Monte Carlo Process Supervision (sft)



【 📦 GitHub repo | 🤗 Paper 】




	
	TL;DR

We propose to enhance LLMs' logical reasoning and generalization by synthesizing symbolic reasoning trajectories using Monte Carlo estimation and integrating them with Direct Preference Optimization and Supervised Fine-Tuning.

@misc{tan2025enhancinglogicalreasoninglanguage, title={Enhancing… See the full description on the dataset page: https://huggingface.co/datasets/XingweiT/SymbReAct-trace-sft.