AI Paper Atlas

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

DIVE: Scaling Diversity in Agentic Task Synthesis for Generalizable Tool Use

Aili Chen, Chi Zhang, Junteng Liu, Jiangjie Chen, Chengyu Du, Yunji Li, Ming Zhong, Qin Wang, Zhengmao Zhu, Jiayuan Song, Ke Ji, Junxian He and 2 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
scale tool-pool coverage 1 to 4 domains (diversity-only, 12k) rather than quantity 12k to 48k with fixed tools (ablation)SFT Qwen3-8B · OOD benchmarkssuccess rateImproved
scale data quantity 12k to 48k with fixed deep-research tasks and tools (ablation)SFT Qwen3-8B · OOD benchmarkssuccess rateHurt
expand tool pool across domains in addition to toolset variety (ablation)SFT Qwen3-8B (Finance start) · OOD benchmarkssuccess rateImproved
only vary toolsets with fixed tool pool (ablation)SFT Qwen3-8B · OOD benchmarkssuccess rateMixed
add RL after SFT on DIVE dataDIVE-8B · 9 OOD benchmarksaverage success points gain over base: +16.2 (SFT) -> +22.2 (RL)Improved
add RL stage to SFT with more domains (ablation)DIVE-8B · average over benchmarks (24 paths)mean score: 29.1 -> 34.8Improved
RL-SFT gap grows with diversity (ablation)DIVE-8B · average over benchmarksAvgRL - AvgSFT: +4.6 at 1 domain vs. +5.6 at 4 domainsImproved
train on fixed-toolset deep-research specialist data (WebExplorer-8B) vs DIVEQwen3-8B · L3 specialized-tool benchmarkssuccess rate: drops 5.8 points below base modelHurt
train on DIVE (RL) instead of Qwen3-8B baseQwen3-8B · Toolathlonsuccess rate: 0.9 -> 8.3Improved
train on DIVE (RL) instead of EnvScaler-8B (query-first simulated)Qwen3-8B · 9 OOD benchmarksOOD lift: 3.2x larger OOD liftImproved
DIVE (RL) vs DIVE (SFT) on SWE-bench Verified (ablation)Qwen3-8B · SWE-bench Verifiedsuccess rate: 13.2 -> 18.3Improved

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