Lune

ACL Findings2026

FLARE: Task-Agnostic Embedding Model Evaluation via Normalizing Flows

Jingzhou Jiang, Yixuan Tang, Yi Yang, Kar Yan Tam

Abstract

Selecting an embedding model for a specific target corpus is difficult when task-specific labels are unavailable. Existing label-free metrics based on kernel estimators or Gaussian mixtures fail in high-dimensional spaces and produce unstable rankings. We propose FLARE (Flow-based Label-free Assessment of Representation Embeddings), which uses normalizing flows to estimate information sufficiency directly from log-likelihoods, avoiding distance-based density estimates. We give finite-sample bounds showing that the estimation error depends on the intrinsic dimension of the data manifold rather than the raw embedding dimension. On 11 datasets and eight embedders, FLARE attains Spearman's ρ up to 0.90 with supervised benchmarks and remains stable for high-dimensional embeddings (d ≥ 3,584), where existing label-free baselines collapse.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

Builds onCited byLeft to right by year; the more cited, the higher.

Builds on9

  1. Understanding Contrastive Representation Learning through Alignment and Uniformity on the HypersphereTongzhou Wang, Phillip IsolaICML 2020 · 2,360 citations
  2. M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge DistillationJianlyu Chen, Shitao Xiao, Peitian Zhang, Kun Luo et al.ACL Findings 2024 · 762 citations
  3. CLUB: A Contrastive Log-ratio Upper Bound of Mutual InformationPengyu Cheng, Weituo Hao, Shuyang Dai, Jiachang Liu et al.ICML 2020 · 512 citations
  4. Scaling Properties of Deep Residual NetworksAlain-Sam Cohen, Rama Cont, Alain Rossier, Renyuan XuICML 2021 · 21 citations
  5. IsoScore: Measuring the Uniformity of Embedding Space UtilizationWilliam Rudman, Nate Gillman, Taylor Rayne, Carsten EickhoffACL Findings 2022 · 15 citations

Related papers

  1. When is an Embedding Model More Promising than Another?Maxime Darrin, Philippe Formont, Ismail Ben Ayed, Jackie CK Cheung et al.NeurIPS 2024 · 10 citations
  2. Intrinsic dimensionality estimation using Normalizing FlowsChristian Horvat, Jean-Pascal PfisterNeurIPS 2022 · 20 citations
  3. Mutual Information Estimation via Normalizing FlowsIvan Butakov, Aleksander Tolmachev, Sofia Malanchuk, Anna Neopryatnaya et al.NeurIPS 2024 · 30 citations
  4. RankMe: Assessing the Downstream Performance of Pretrained Self-Supervised Representations by Their RankQuentin Garrido, Randall Balestriero, Laurent Najman, Yann LeCunICML 2023 · 127 citations
  5. Entropy Estimation via Normalizing FlowZiqiao Ao, Jinglai LiAAAI 2022 · 12 citations
Dusk over the sea between two cliffs drawn in fine vertical lines