AI Paper Atlas

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

TaskCraft: Automated Generation of Agentic Tasks

Dingfeng Shi, Jingyi Cao, Qianben Chen, Weichen Sun, Weizhen Li, Hongxuan Lu, Fangchen Dong, Tianrui Qin, King Zhu, Minghao Liu, Yuchen Eleanor Jiang, Jian Yang and 5 more

ICLR 2026 · Large language models › Agents, code and math · approach: Long-horizon LLM agents · 63 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
add tool context (input index and relation) to atomic task generation (ablation)TaskCraft atomic task generation · Atomic task generationpass rate (%): 18.5% -> 43.0%Improved
add tool context to atomic task generation (ablation)TaskCraft atomic task generation · Atomic task generationgeneration time: 119.7s -> 86.7sImproved
add tool context to atomic task generation (ablation)TaskCraft atomic task generation · Atomic task generationaverage tool use: 2.8 -> 2.1Improved
add tool context to atomic task generation (ablation)TaskCraft atomic task generation · Atomic task generationvariance of tool use: 1.2 -> 0.4Improved
add bootstrap few-shot prompt optimization (ablation)atomic task generation · Atomic task generationpass rate (%): 54.9% -> 68.1%Improved
add prompt optimization (ablation)atomic task generation · Atomic task generationtime per document: 29.1s -> 23.5sImproved
add prompt optimization (ablation)depth-wise extension · Depth-wise@6pass rate (%): 41.0% -> 51.2%Improved
add prompt optimization (ablation)depth-wise extension · Depth-wise@6time per task: 31.5s -> 30.2sImproved
add SFT on synthetic trajectoriesQwen2.5-3B-Base base workflow · HotpotQA, Musique, Bamboogle averageaccuracy: 0.034 -> 0.174Improved
add SFT on synthetic trajectoriesQwen2.5-3B-Instruct base workflow · HotpotQA, Musique, Bamboogle averageaccuracy: 0.113 -> 0.173Improved
add SFT on synthetic trajectoriesSearch-R1 Qwen2.5-3B-Base · Bamboogleaccuracy: 0.088 -> 0.280Improved
add SFT on synthetic trajectoriesSearch-R1 Qwen2.5-3B-Base · Musiqueaccuracy: 0.049 -> 0.111Improved

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