AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Logit‑KL Flow Matching: Non‑Autoregressive Text Generation via Sampling‑Hybrid Inference

Egor Sevriugov, Nikita Dragunov, Anton Razzhigaev, Andrey Kuznetsov, Ivan Oseledets

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Flow matching distillation · 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
use KL-geodesic (logit-linear) interpolation instead of Fisher-Rao (ablation)KL-Flow (150M) · FineFineWebperplexity (Llama 2): 192 -> 41Improved
use KL-geodesic instead of linear simplex interpolation (ablation)KL-Flow (150M) · FineFineWebperplexity (Llama 2): 1344 -> 41Improved
scale KL-Flow from 150M to 1.5B parametersKL-Flow · FineFineWeb, NFE 1024perplexity (Llama 2): 35.1 -> 32.7Improved
KL-Flow vs SEDD on unconditional generationSEDD · FineFineWeb, NFE 1024perplexity (Llama 2): 47.6 -> 35.1Improved
KL-Flow vs autoregressive GPT-2 on TinyStoriesGPT 2 · TinyStoriesperplexity: 15.4 vs 19.0Hurt
hybrid inference vs sampling inference on Lamini Instruction (ablation)KL-flow (sampling) · Lamini InstructionBLEU (top-5): 7.7 -> 9.5Improved
sampling inference vs hybrid inference on WMT14 De-En (ablation)KL-flow (hybrid) · WMT14 De-EnBLEU (top-5): 23.8 -> 27.0Improved
KL-Flow vs DFM on code infilling (10% masked)DFM · MBPP infilling 10%Pass@1: 11.1 -> 17.4Improved
KL-Flow vs baselines at 90% maskingGPT-2/DFM/SEDD · MBPP infilling 90%Pass@1: 0.1 -> 0.2No clear effect
KL-Flow few-shot refinement with few stepsKL-Flow · FineFineWeb few-shot, 4 stepsentropy: 5.5 -> 3.6Hurt
KL-Flow few-shot refinementDFM/SEDD · FineFineWeb few-shot, 16 stepsperplexity: 321.0 -> 99.8Improved

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