WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents

2026-09-29

AI Summary

This study addresses the insufficient post-training performance gains in tool use caused by isolated scaling environments by proposing WEFT, a framework that jointly optimizes all components of agent interaction systems. Methodologically, it introduces a system-wide coupled evolution mechanism, integrating execution-driven self-evolution with MegaMCP state isolation to construct scalable agent architectures. During training, prefix-preserving sampling and atomic-turn credit assignment are employed to enable stable, large-scale post-training. Experimental results demonstrate that WEFT significantly outperforms baselines of comparable scale on benchmarks such as BFCL V4, achieving improvements of up to 12.27 percentage points for 14B-parameter models.

Abstract

Recent efforts to scale tool-use post-training have largely centered on the synthesis of executable environments, which constitute only one component of a broader agentic interaction system comprising the environment, task, agent harness, and evaluator. Scaling environments in isolation, however, does not guarantee commensurate gains in model performance, because reliable learning signals depend on coherent interactions among all components of the agentic interaction system. To address this problem, we introduce WEFT (Whole-system Evolution For Tool-use Post-training), which couples scalable agentic interaction system construction, execution-driven self-evolution, and stable post-training. WEFT scales agentic interaction system construction across environment breadth, task complexity, and interaction diversity. Execution-driven self-evolution iteratively uses execution traces and state evidence to attribute failures and revise the responsible components, with fresh rollouts evaluating the changes and providing evidence for subsequent evolution rounds. For stable post-training at scale, WEFT addresses both optimization and execution reliability: prefix-preserving sampling retains verified progress and atomic-turn credit assignment localizes learning signals, while MegaMCP maintains isolated, recoverable state across concurrent rollouts over shared tool services. Extensive experiments across various models and benchmarks demonstrate the effectiveness of WEFT for tool-use post-training. WEFT-8B and WEFT-14B outperform all evaluated matched-size environment-scaling baselines on BFCL V4, $τ^2$-Bench, and Claw-Eval. In particular, WEFT-14B improves over Agent-World-14B by 6.41, 2.23, and 12.27 percentage points. WEFT-35B-A3B further extends these gains to more challenging long-horizon workflow benchmarks, including Toolathlon-Verified and AutomationBench.

Problem

• tool-use post-training
• agentic interaction system
• environment scaling
• learning signals

Innovation

• Tool-use Post-training
• Whole-system Evolution
• Execution-driven Self-evolution
• Atomic-turn Credit Assignment
• MegaMCP