AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Toward Effective Tool-Integrated Reasoning via Self-Evolved Preference Learning

Yifei Chen, Guanting Dong, Zhicheng Dou

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning generalization and delta compression

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add pre-aligned DPO after SFT (ablation)Tool-Light (Qwen2.5-7B-Instruct) · 10 reasoning tasks averageaverage score: 56.6 -> 57.4Improved
add self-evolved DPO alignment (2 loops) after pre-aligned DPO (ablation)Tool-Light (Qwen2.5-7B-Instruct) · 10 reasoning tasks averageaverage score: 57.4 -> 58.0Improved
increase self-evolved loops from 2 to 3 (ablation)Tool-Light · 10 reasoning tasks averageperformance: 58.0 -> 56.1Hurt
increase self-evolved loops from 2 to 5 (ablation)Tool-Light · 10 reasoning tasks averageperformance: 58.0 -> 54.1Hurt
use 2 loops instead of 1 loop (ablation)Tool-Light · 10 reasoning tasks averageperformance: 57.9 -> 58.0No clear effect
random positive example selection instead of min-tool-call, low-entropy criterion (ablation)Tool-Light · 10 reasoning tasks averageperformance: 58.0 -> 53.6Hurt
random negative example selection (ablation)Tool-Light · 10 reasoning tasks averageperformance: 58.0 -> 53.9Hurt
mix vanilla and entropy-guided data 1:1 instead of the tuned ratio (ablation)Tool-Light · 10 reasoning tasks averageperformance: 58.0 -> 56.9Hurt
multi-tool-integrated training (Tool-Light/Tool-Star) rather than single-tool trainingQwen2.5-7B-Instruct · math and knowledge tasksaverage score: 50.4 -> 58.0Improved
fine-tuning instead of prompt engineering for tool useQwen2.5-7B-Instruct · 10 tasks averageaverage score: 35.3 -> 56.6Improved
introduce external tools via prompting onlyQwen2.5-7B-Instruct · knowledge-intensive tasksF1: 33.0 -> 35.3Mixed
entropy-guided sampling to learn low-entropy pathsTool-Light · output entropy analysisoutput entropyImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy