AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Behavioral Mode Discovery for Fine-tuning Multimodal Generative Policies

Alberta Longhini, David Emukpere, Jean-Michel Renders, Seungsu Kim

ICML 2026 · Reinforcement learning › RL from feedback · approach: RL fine-tuning of generative models · 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
add BMD mutual-information regularizationDPPO · Gaussian mixture G1SR_M: 0.58 -> 1.00Improved
add BMD mutual-information regularizationDPPO · Gaussian mixture G2SR_M: 0.42 -> 0.75Improved
add BMD mutual-information regularizationRES · Gaussian mixture G2SR_M: 0.50 -> 1.00Improved
add BMD mutual-information regularizationDSRL · Gaussian mixture G1SR_M: 0.25 -> 0.33Improved
add BMD mutual-information regularizationDSRL · Gaussian mixture G2SR_M: 0.08 -> 0.08No clear effect
add BMD mutual-information regularizationRES · ManiSkill LiftSR_M: 0.50 -> 0.99Improved
add BMD mutual-information regularizationDSRL · ManiSkill LiftSR_M: 0.78 -> 0.88Improved
add BMD mutual-information regularizationDPPO · ManiSkill LiftSR_M: 0.57 -> 0.55No clear effect
add BMD mutual-information regularizationRES · ManiSkill AvoidSR_M: 0.04 -> 0.30Improved
add BMD mutual-information regularizationDSRL · ManiSkill AvoidSR_M: 0.09 -> 0.42Improved
add BMD mutual-information regularizationDSRL · ANYmal locomotionSR_M: 0.25 -> 0.89Improved
add BMD mutual-information regularizationDSRL · Franka KitchenSR_M: 0.04 -> 0.10Improved

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