Pith.

Machine Learning · arXiv:2603.28301 v3 · submitted 30 Mar 2026

LIBERO-Para: A Diagnostic Benchmark and Metrics for Paraphrase Robustness in VLA Models

Chanyoung Kim, Minwoo Kim, Minseok Kang, Hyunwoo Kim, Dahuin Jung

The paper says

“This degradation is primarily driven by object-level lexical variation: even simple synonym substitutions cause large drops, indicating reliance on surface-level matching rather than semantic grounding.”

from the abstract · picked out as the main claim by a classifier

Pith says

Pith has not reviewed this paper.

Get a Pith Review, $5

Reviews

Reviews by people

No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review

How Pith checks papers · The full audit of this paper

The paper

Abstract

Vision-Language-Action (VLA) models achieve strong performance in robotic manipulation by leveraging pre-trained vision-language backbones. However, in downstream robotic settings, they are typically fine-tuned with limited data, leading to overfitting to specific instruction formulations and leaving robustness to paraphrased instructions underexplored. To study this gap, we introduce LIBERO-Para, a controlled benchmark that independently varies action expressions and object references for fine-grained analysis of linguistic generalization. Across seven VLA configurations (0.6B-7.5B), we observe consistent performance degradation of 22-52 pp under paraphrasing. This degradation is primarily driven by object-level lexical variation: even simple synonym substitutions cause large drops, indicating reliance on surface-level matching rather than semantic grounding. Moreover, 80-96% of failures arise from planning-level trajectory divergence rather than execution errors, showing that paraphrasing disrupts task identification. Binary success rate treats all paraphrases equally, obscuring whether models perform consistently across difficulty levels or rely on easier cases. To address this, we propose PRIDE, a metric that quantifies paraphrase difficulty using semantic and syntactic factors. Our benchmark and corresponding code are available at: https://github.com/cau-hai-lab/LIBERO-Para

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.

abstract compiled · 2026-09-23 · no language-model review

How the main claim is stated: asserted 0.80

What the abstract itself states, as probabilities: claims first 0.09 · states limitation 0.20 · human participants 0.04 · controlled comparison 0.33 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.20 · mentions code or data availability 0.99

The abstract, sentence by sentence

This table labels the abstract only (arXiv listing abstract v2). The number is the probability the classifier gave the label.

SentenceRoleThe authors' own workStates the main conclusionStrength
Vision-Language-Action (VLA) models achieve strong performance in robotic manipulation by leveraging pre-trained vision-language backbones. background 1.00 0.06 0.02 asserted 1.00
However, in downstream robotic settings, they are typically fine-tuned with limited data, leading to overfitting to specific instruction formulations and leaving robustness to paraphrased instructions underexplored. gap 0.95 0.10 0.05 asserted 0.98
To study this gap, we introduce LIBERO-Para, a controlled benchmark that independently varies action expressions and object references for fine-grained analysis of linguistic generalization. contribution 0.96 0.91 0.05 asserted 0.98
Across seven VLA configurations (0.6B-7.5B), we observe consistent performance degradation of 22-52 pp under paraphrasing. result 1.00 0.96 0.31 asserted 1.00
This degradation is primarily driven by object-level lexical variation: even simple synonym substitutions cause large drops, indicating reliance on surface-level matching rather than semantic grounding. interpretation 0.86 0.94 0.39 asserted 0.80
Moreover, 80-96% of failures arise from planning-level trajectory divergence rather than execution errors, showing that paraphrasing disrupts task identification. result 0.79 0.94 0.15 asserted 1.00
Binary success rate treats all paraphrases equally, obscuring whether models perform consistently across difficulty levels or rely on easier cases. gap 0.81 0.23 0.05 asserted 1.00
To address this, we propose PRIDE, a metric that quantifies paraphrase difficulty using semantic and syntactic factors. contribution 0.82 0.90 0.08 asserted 0.99
Our benchmark and corresponding code are available at: https://github.com/cau-hai-lab/LIBERO-Para contribution 0.93 0.35 0.02 asserted 0.90

What kind of paper the classifier reports this is: dataset or benchmark 1.00 · evidence from observation or experiment 0.99

What was examined, and what was not

Sources read

  • arXiv listing abstract v2

Not examined

  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers

Levels run

  • P abstract

Not run

  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

How the labels were made

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 6,928 input tokens.

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON

Cited by 12 Pith papers
How the literature uses it

2 arXiv papers cite this in 6 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

What it cites · 21 sentences without a confident label

Cites as support 3 papers
Cites as background 7 papers
  • Recent work has diversified along several architectural axes: parallel decoding with action chunking, which predicts all actions in a single forward pass for faster inference [ref] ; VLM coupled with a flow-matching action expert, which pairs a billion-scale VLM with a separate action decoder [ref] ; lightweight bridge-based adaptation, which routes VLM representations to a compact policy head via cross-attention [ref] ; and soft-prompted cross-embodiment designs, which encode embodiment-specific knowledge through learnable tokens [ref] .

    2602.12684 · Vision-Language-Action Models · 0.96

  • Recent work has diversified along several architectural axes: parallel decoding with action chunking, which predicts all actions in a single forward pass for faster inference [ref] ; VLM coupled with a flow-matching action expert, which pairs a billion-scale VLM with a separate action decoder [ref] ; lightweight bridge-based adaptation, which routes VLM representations to a compact policy head via cross-attention [ref] ; and soft-prompted cross-embodiment designs, which encode embodiment-specific knowledge through learnable tokens [ref] .

    $\pi_0$: A Vision-Language-Action Flow Model for General Robot Control · Vision-Language-Action Models · 0.96

  • Recent work has diversified along several architectural axes: parallel decoding with action chunking, which predicts all actions in a single forward pass for faster inference [ref] ; VLM coupled with a flow-matching action expert, which pairs a billion-scale VLM with a separate action decoder [ref] ; lightweight bridge-based adaptation, which routes VLM representations to a compact policy head via cross-attention [ref] ; and soft-prompted cross-embodiment designs, which encode embodiment-specific knowledge through learnable tokens [ref] .

    Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success · Vision-Language-Action Models · 0.96

Challenges and discussion

Know something about this paper? Leave a comment.

No comments yet.

Sign in to comment