AI Paper Atlas

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

Jupiter: Enhancing LLM Data Analysis Capabilities via Notebook and Inference-Time Value-Guided Search

Shuocheng Li, Yihao Liu, Silin Du, Wenxuan Zeng, Zhe Xu, Mengyu Zhou, Yeye He, Haoyu Dong, Shi Han, Dongmei Zhang

AAAI 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 3 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
finetune on NbQAMistral-7B-Instruct-v0.3 · InfiAgent-DABenchaccuracy by questions (%): 2.33 -> 59.14Improved
finetune on NbQALlama-3.1-8B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 48.25 -> 69.65Improved
finetune on NbQAQwen2.5-7B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 43.97 -> 68.09Improved
finetune on NbQAQwen2.5-14B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 69.65 -> 77.04Improved
add value model to Jupiter search (exploration term removed) (ablation)SFT Qwen2.5-7B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 70.04 -> 77.82Improved
add value model to Jupiter search (exploration term removed) (ablation)SFT Qwen2.5-14B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 79.38 -> 86.38Improved
remove the PUCT exploration term when using value model (ablation)SFT Qwen2.5-14B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 74.71 -> 86.38Improved
remove the PUCT exploration term when using value model (ablation)SFT Qwen2.5-7B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 68.87 -> 77.82Improved
add exploration term without value model (ablation)SFT Qwen2.5-14B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 79.38 -> 75.88Hurt
add exploration term without value model (ablation)SFT Qwen2.5-7B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 70.04 -> 68.87Hurt
use Jupiter value-guided search instead of majority votingSFT Qwen2.5-14B-Instruct · InfiAgent-DABenchaccuracy by questions (%): 83.66 -> 86.38Improved
use value-model-assisted search instead of ReActvanilla Qwen2.5-14B-Instruct · DSBench data modelingtask completion rate (%): 66.22% -> 98.65%Improved

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