AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Value Matching: Scalable and Gradient-Free Reward-Guided Flow Adaptation

Cristian Perez Jensen, Luca Schaufelberger, Riccardo De Santi, Kjell Jorner, Andreas Krause

ICLR 2026 · Reinforcement learning › RL from feedback · approach: RL fine-tuning of generative models

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
adapt FlowMol with Value Matching on QED reward (λ=500)FlowMol · FlowMol GEOM-Drugs, QED rewardstable molecules (%): 49.5 -> 67.6Improved
adapt FlowMol with Value Matching on QED rewardFlowMol · FlowMol GEOM-Drugs, QED rewardQED: 0.42 -> 0.49Improved
adapt FlowMol with Value Matching on dipole moment rewardFlowMol · FlowMol GEOM-Drugs, dipole moment rewarddipole moment (Debye): 6.4 -> 7.5Improved
use VM instead of classifier guidance at moderate reward scaleVM · CIFAR-10 aesthetic rewardtraining stability / rewardImproved
use online VM instead of offline classifier guidanceVM · 2D toy environmentKL to tilted distributionImproved
use VM instead of DDPO/DPOK fine-tuningVM · CIFAR-10 (compression, incompression, aesthetic)reward and diversityImproved
use VM instead of CT-PPOVM · CIFAR-10 rewardsrewardNo clear effect
use VM instead of fine-tuning in memoryVM · SD2memory (GB): 250 -> 12Improved
increase value network size from 1.8M to 92.3M parameters (ablation)VM · CIFAR-10 aesthetic reward (λ=100)aesthetic reward: 4.00 -> 3.26No clear effect
use VM guidance instead of SVDD inference-time searchVM · CIFAR-10 aesthetic rewardsampling cost / rewardImproved
add VM value guidance at samplingVM · SD2 samplingtime (s): 122.02 -> 127.44Hurt

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