AI Paper Atlas

Research map / Large language models / Agents, code and math

RE-TRAC: REcursive TRAjectory Compression for Deep Search Agents

jialiang zhu, Gongrui Zhang, Xiaolong Ma, Lin Xu, Miaosen Zhang, Ruiqi Yang, Song Wang, Kai Qiu, Zhirong Wu, Qi Dai, Ruichun Ma, Bei Liu and 8 more

ICML 2026 · Large language models › Agents, code and math · approach: Long-horizon LLM agents · 0 citations (OpenAlex)

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
use Re-TRAC (8 rounds) instead of single passo4-mini · BrowseComp300accuracy (%): 25.7 -> 46.8Improved
use Re-TRAC (8 rounds) instead of single passo3 · BrowseComp300accuracy (%): 54.9 -> 69.8Improved
use Re-TRAC (8 rounds) instead of single passGPT-5-medium · BrowseComp300accuracy (%): 48.3 -> 66.6Improved
use Re-TRAC (8 rounds) instead of single passDeepSeek-V3.2 · BrowseComp300accuracy (%): 45.3 -> 60.8Improved
use Re-TRAC instead of majority voting at 8GLM-4.7 · BrowseComp300accuracy (%): 41.7 -> 60.7Improved
use Re-TRAC instead of weighted voting at 8GPT-5-medium · BrowseComp300accuracy (%): 64.7 -> 66.6Improved
add Re-TRAC SFT on GLM-4.7 trajectories (ablation)Qwen3-4B-Instruct · BrowseComp300accuracy (%): 2.7 -> 30.0Improved
add Re-TRAC SFT on GLM-4.7 trajectories (ablation)Qwen3-4B-Instruct · GAIAaccuracy (%): 24.4 -> 70.4Improved
add instruction to freely use the summary (ablation)Re-TRAC with o3 · BrowseComp300 round 8accuracy (%): 68.9 -> 71.7Improved
use GLM-4.7 as summarizer instead of the model itself (ablation)RE-TRAC-4B · BrowseComp300accuracy (%): 30.0 -> 38.5Improved
use GLM-4.7 as summarizer instead of the model itself (ablation)RE-TRAC-30B-A3B · BrowseComp300accuracy (%): 53.0 -> 52.4No clear effect
use majority voting instead of single passGLM-4.7 · BrowseComp300accuracy (%): 37.7 -> 41.7Mixed

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