Static SFT and RL training for tool-use agents leads to performance drops under open-world distributional shifts across perception, interaction, reasoning and internalization; perturbation-augmented fine-tuning is proposed as mitigation.
Ruijie Xu, Zengzhi Wang, Run-Ze Fan, and Pengfei Liu
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
New RPS and AGS metrics show within-family distilled LLM agents have 5.9 pp higher tool-use graph similarity than cross-family pairs, with some models exceeding their teachers.
NoisyAgent trains LLM agents with controlled user and tool noise to improve robustness in stochastic environments while also boosting clean-benchmark performance.
citing papers explorer
-
Can Agents Generalize to the Open World? Unveiling the Fragility of Static Training in Tool Use
Static SFT and RL training for tool-use agents leads to performance drops under open-world distributional shifts across perception, interaction, reasoning and internalization; perturbation-augmented fine-tuning is proposed as mitigation.
-
When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors
New RPS and AGS metrics show within-family distilled LLM agents have 5.9 pp higher tool-use graph similarity than cross-family pairs, with some models exceeding their teachers.
-
Learning to Act under Noise: Enhancing Agent Robustness via Noisy Environments
NoisyAgent trains LLM agents with controlled user and tool noise to improve robustness in stochastic environments while also boosting clean-benchmark performance.