AI Paper Atlas

Research map / Reinforcement learning / Offline RL

Guided Flow Policy: Learning from High-Value Actions in Offline Reinforcement Learning

Franki NGUIMATSIA TIOFACK, Théotime Le Hellard, Fabian Schramm, Nicolas Perrin-Gilbert, Justin Carpentier

ICLR 2026 · Reinforcement learning › Offline RL · approach: Offline RL sample complexity theory · 6 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 value-aware guidance (GFP) over FQLFQL · OGBench cube-double-noisysuccess rate: 38.2 -> 63.1Improved
add value-aware guidance (GFP) over FQLFQL · OGBench cube-triple-noisysuccess rate: 3.5 -> 24.5Improved
add value-aware guidance (GFP) over FQLFQL · OGBench average (105 tasks)average score: 46.7 -> 53.2Improved
add value-aware guidance (GFP) over ReBRACReBRAC · Minari average (21 tasks)average score: 65.9 -> 74.1Improved
use GFP over ReBRAC on D4RLReBRAC · D4RL average (18 tasks)average score: 64.8 -> 63.0No clear effect
tune discount factor to larger value for ReBRACReBRAC · OGBench antmaze-large-navigatesuccess rate: 81 -> 95.9Improved
use bigger minibatch for FQLFQL · OGBench antmaze-giant-navigatesuccess rate: 9 -> 16.3Improved
add modified Bellman target using VaBC policy (ablation)GFP · OGBench cube-double-playsuccess rate: ~28 -> 47.2Improved
add modified Bellman target using VaBC policy (ablation)GFP · OGBench puzzle-4x4-playsuccess rate: 26.1 -> ~22Hurt
replace soft-max guidance with AWR-style weight (ablation)GFP · OGBench (65 tasks)average score: 38.6 -> 35.9Hurt
replace soft-max guidance with AWR-style weight (ablation)GFP · OGBench cube-triple-noisysuccess rate: 24.5 -> 11.2Hurt
replace soft-max guidance with AWR-style weight (ablation)GFP · OGBench puzzle-3x3-playsuccess rate: 23.1 -> 28.3Improved

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