ArcFlow: Training Flow Policies for Multiple Solver Budgets

Key points

• Flow-matching policies adjust inference cost through the number of solver steps, but reinforcement learning at a single fixed solver budget can hide degradation at cheaper deployment budgets and even erase useful one-step behavior present in a pretrained policy.1
• ArcFlow combines action-aligned supervision, budget exposure, and worst-budget checkpoint selection.1
• Across DM Control, manipulation, and locomotion, ArcFlow attains the highest mean one-step return among five flow-policy baselines on 25 of 30 tasks under each method's reported training settings.1
• These results establish deployment-budget-aware training as an effective route to strong low-step control with one policy serving multiple inference costs.1

Sources (1)

• [1]ArcFlow: Training Flow Policies for Multiple Solver Budgets
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 06:18 PM

  We introduce Action-Ratio-Clip Flow (ArcFlow), a deployment-aware algorithm for training one flow policy across a specified range of solver budgets.

  Flow-matching policies adjust inference cost through the number of solver steps, but reinforcement learning at a single fixed solver budget can hide degradation at cheaper deployment budgets and even erase useful one-step behavior present in a pretrained policy.

Extractive summary: sentences quoted from the sources.