AI Paper Atlas

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

AuTAgent: A Reinforcement Learning Framework for Tool-Augmented Audio Reasoning

Siqian Tong, Li Xuan, Yiwei Wang, Baolong Bi, Yujun Cai, Shenghua Liu, Yuchen He, HAO Chengpeng

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
concatenate outputs of all six tools (ablation)Qwen2-Audio-7B-Instruct · MMAU Test-miniaccuracy (%): 50.00 -> 40.90Hurt
use baseline-subtracted differential reward instead of binary reward (ablation)AuTAgent · MMAU Test-miniaccuracy (%): 51.40 -> 54.20Improved
use baseline-subtracted differential reward instead of binary reward (ablation)AuTAgent · MMARaccuracy (%): 39.00 -> 41.30Improved
use differential reward instead of binary reward (ablation)AuTAgent · MMAU Test-miniavg tools invoked: 1.62 -> 1.30Improved
train tool-selection policy with GRPO instead of description-based zero-shot selectionQwen2-Audio-7B-Instruct · MMAU Test-miniaccuracy (%): 51.20 -> 54.20Improved
random tool selection (ablation)Qwen2-Audio-7B-Instruct · MMAU Test-miniaccuracy (%): 50.00 -> 49.10Hurt
random tool selection (ablation)GPT-4o Audio · MMAU Test-miniaccuracy (%): 57.40 -> 53.60Hurt
add trained AuTAgent policy as plugin to GPT-4o AudioGPT-4o Audio · MMAU Test-miniaccuracy (%): 57.40 -> 66.80Improved
plugin policy (trained on other backbone) vs self-trained policy (ablation)AuTAgent · MMAU Test-mini / MMARaccuracy (%): 67.20 -> 66.80 (MMAU, GPT-4o)No clear effect
add tool library to GPT-4o Audio via AuTAgentGPT-4o Audio · MMARaccuracy (%): 55.00 -> 63.00Improved
invoke each single tool T2, T3, T5 individually (ablation)Qwen2-Audio-7B-Instruct · MMAU Test-miniaccuracy (%)Hurt

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