AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Tilt Matching for Scalable Sampling and Fine-Tuning

Peter Potaptchik, Lee Cheuk Kit, Michael Samuel Albergo

ICML 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Flow matching and flow sampling · 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
fine-tune with ITM (lambda=1) instead of Adjoint Matching (lambda=1)Stable Diffusion 1.5 · ImageRewardImageReward: 0.2170 -> 0.4465Improved
fine-tune with ITM (lambda=1) instead of Adjoint Matching (lambda=1)Stable Diffusion 1.5 · HPSv2HPSv2: 0.2576 -> 0.2659Improved
fine-tune with ITM (lambda=1) instead of base SD 1.5Stable Diffusion 1.5 · ImageRewardImageReward: 0.1873 -> 0.4465Improved
fine-tune with ITM (lambda=1) instead of base SD 1.5Stable Diffusion 1.5 · DreamSimDreamSim (diversity): 0.3849 -> 0.3383Hurt
fine-tune with ITM (lambda=1) instead of base SD 1.5Stable Diffusion 1.5 · CLIPScoreCLIPScore: 0.2746 -> 0.2794Improved
use reward multiplier lambda=100 with Adjoint Matching instead of ITM lambda=1Stable Diffusion 1.5 · ImageRewardImageReward: 0.4465 -> 0.7873Improved
use reward multiplier lambda=100 with Adjoint Matching instead of ITM lambda=1Stable Diffusion 1.5 · CLIPScoreCLIPScore: 0.2794 -> 0.2792No clear effect
use ITM instead of iDEMLJ-13 · LJ-13ESS: 0.231 -> 0.507Improved
use ITM instead of DDSLJ-13 · LJ-13energy W2: 24.61 -> 0.879Improved
use ITM instead of ASBSLJ-55 · LJ-55energy W2: 28.10 -> 26.37Improved
use ITM instead of ASBSLJ-55 · LJ-55Geo W2: 4.00 -> 3.98Improved
use low-variance ITM loss instead of weighted flow matching (ablation)Tilt Matching · LJ-13gradient varianceImproved

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