DISLab/ReFeed-8B

♻️ ReFeed-8B (COLM'25)

Overview

ReFeed rethinks summary refinement by showing that reflective reasoning over multi-dimensional feedback is key to achieving balanced, robust improvements.

Our contributions:

• 🧠 Introduces reflective reasoning for multi-dimensional refinement, enabling models to resolve trade-offs, mitigate order bias, and filter noisy feedback simultaneously
• 🏗️ Releases SumFeed-CoT, a large-scale Long-CoT dataset that distills high-quality reflective reasoning from large reasoning models into lightweight models
• 🚀 Demonstrates strong empirical gains and robustness, achieving consistent improvements across faithfulness, completeness, and conciseness while remaining resilient to feedback quality and ordering

Recommended Prompt

We recommend using the following prompt. For details on the feedback format and the code, please refer to our GitHub repository.

system_prompt = "Your role as an assistant involves thoroughly exploring questions through a systematic long thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Think and Answer. In the Think section, detail your reasoning process using the specified format: <think> {thought with steps separated with '\\n\\n'} </think> Each step should include detailed considerations such as analisying questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Answer section, based on various attempts, explorations, and reflections from the Think section, systematically present the final solution that you deem correct. The solution should remain a logical, accurate, concise expression style and detail necessary step needed to reach the conclusion, formatted as follows: <answer> {final formatted, precise, and clear solution} </answer> Now, try to solve the following question through the above guidelines:"

user_prompt = f"""
Your goal is to deliberate on the provided feedback and propose actionable and specific aggregated feedback based on it.

Instructions:
1. Deliberate on the characteristics an ideal summary should achieve.
2. Assess and choose the validity of the given feedback in improving the summary considering feedback quality criteria:
- Faithfulness: Out-of-article, Entity, Relation, Sentence errors
- Completeness: missing key content
- Conciseness: unnecessary or non-key content
3. Aggregate the valid feedback and revise the summary by incorporating it.
4. Check whether revisions harm other quality dimensions.

Document:
{document}

Summary:
{summary}

Feedback:
{feedback}
"""

Key Results

• 📈 ReFeed achieves the best overall performance across faithfulness, completeness, and conciseness, outperforming all previous refinement methods that optimize a single dimension.
• ⚖️ Reflective reasoning enables balanced improvements, effectively mitigating trade-offs that arise when optimizing multiple dimensions simultaneously.
• 🔀 Strong robustness to feedback order and noise, showing minimal performance variance under shuffled or low-quality feedback settings.
• 🧩 Efficient distillation: a lightweight 8B model matches teacher-level refinement quality while significantly reducing inference cost.

🖇️ Citation

Please consider citation if our paper is useful in your research.

@inproceedings{yun2025refeed,
    title={ReFeed: Multi-dimensional Summarization Refinement with Reflective Reasoning on Feedback},
    author={Taewon Yun and Jihwan Oh and Hyangsuk Min and Yuho Lee and Jihwan Bang and Jason Cai and Hwanjun Song},
    booktitle={Second Conference on Language Modeling},
    year={2025},
    url={https://openreview.net/forum?id=6BGDGKZN7q}
}