base_model:
- deepseek-ai/DeepSeek-R1-Distill-Qwen-7B
language:
- en
metrics:
- accuracy
library_name: transformers
pipeline_tag: text-generation
tags:
- Reasoning-while-asking
Reasoning While Asking: Proactive Interactive Reasoning (PIR)
This repository contains the weights for the model presented in the paper Reasoning While Asking: Transforming Reasoning Large Language Models from Passive Solvers to Proactive Inquirers.
- Code: GitHub Repository
Introduction
Reasoning-oriented Large Language Models (LLMs) often remain limited by a blind self-thinking paradigm: performing extensive internal reasoning even when critical information is missing or ambiguous.
PIR (Proactive Interactive Reasoning) is a new reasoning paradigm that transforms LLMs from passive solvers into proactive inquirers that interleave reasoning with clarification. PIR-enabled models detect uncertainty during reasoning and actively ask users for clarification before proceeding, reducing hallucinations and misaligned conclusions.
Demo
Experience the "Reasoning While Asking" capability using the interactive script provided in the official repository:
python run_demo_interactive_generation.py \
--input_file "verl-tool/datasets/mip/gsm8k.json" \
--model_url "http://localhost:1136" \
--model_name Proactive-Interactive-R1-Math-7B \
--output_dir results/
Citation
If you find this work useful, please cite the paper:
@misc{chen2026reasoningaskingtransformingreasoning,
title={Reasoning While Asking: Transforming Reasoning Large Language Models from Passive Solvers to Proactive Inquirers},
author={Xin Chen and Feng Jiang and Yiqian Zhang and Hardy Chen and Shuo Yan and Wenya Xie and Min Yang and Shujian Huang},
year={2026},
eprint={2601.22139},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2601.22139},
}