Saved in:
Bibliographic Details
Main Authors: Bavaresco, Anna, Bernardi, Raffaella, Bertolazzi, Leonardo, Elliott, Desmond, Fernández, Raquel, Gatt, Albert, Ghaleb, Esam, Giulianelli, Mario, Hanna, Michael, Koller, Alexander, Martins, André F. T., Mondorf, Philipp, Neplenbroek, Vera, Pezzelle, Sandro, Plank, Barbara, Schlangen, David, Suglia, Alessandro, Surikuchi, Aditya K, Takmaz, Ece, Testoni, Alberto
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.18403
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918042307395584
author Bavaresco, Anna
Bernardi, Raffaella
Bertolazzi, Leonardo
Elliott, Desmond
Fernández, Raquel
Gatt, Albert
Ghaleb, Esam
Giulianelli, Mario
Hanna, Michael
Koller, Alexander
Martins, André F. T.
Mondorf, Philipp
Neplenbroek, Vera
Pezzelle, Sandro
Plank, Barbara
Schlangen, David
Suglia, Alessandro
Surikuchi, Aditya K
Takmaz, Ece
Testoni, Alberto
author_facet Bavaresco, Anna
Bernardi, Raffaella
Bertolazzi, Leonardo
Elliott, Desmond
Fernández, Raquel
Gatt, Albert
Ghaleb, Esam
Giulianelli, Mario
Hanna, Michael
Koller, Alexander
Martins, André F. T.
Mondorf, Philipp
Neplenbroek, Vera
Pezzelle, Sandro
Plank, Barbara
Schlangen, David
Suglia, Alessandro
Surikuchi, Aditya K
Takmaz, Ece
Testoni, Alberto
contents There is an increasing trend towards evaluating NLP models with LLMs instead of human judgments, raising questions about the validity of these evaluations, as well as their reproducibility in the case of proprietary models. We provide JUDGE-BENCH, an extensible collection of 20 NLP datasets with human annotations covering a broad range of evaluated properties and types of data, and comprehensively evaluate 11 current LLMs, covering both open-weight and proprietary models, for their ability to replicate the annotations. Our evaluations show substantial variance across models and datasets. Models are reliable evaluators on some tasks, but overall display substantial variability depending on the property being evaluated, the expertise level of the human judges, and whether the language is human or model-generated. We conclude that LLMs should be carefully validated against human judgments before being used as evaluators.
format Preprint
id arxiv_https___arxiv_org_abs_2406_18403
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Tasks
Bavaresco, Anna
Bernardi, Raffaella
Bertolazzi, Leonardo
Elliott, Desmond
Fernández, Raquel
Gatt, Albert
Ghaleb, Esam
Giulianelli, Mario
Hanna, Michael
Koller, Alexander
Martins, André F. T.
Mondorf, Philipp
Neplenbroek, Vera
Pezzelle, Sandro
Plank, Barbara
Schlangen, David
Suglia, Alessandro
Surikuchi, Aditya K
Takmaz, Ece
Testoni, Alberto
Computation and Language
There is an increasing trend towards evaluating NLP models with LLMs instead of human judgments, raising questions about the validity of these evaluations, as well as their reproducibility in the case of proprietary models. We provide JUDGE-BENCH, an extensible collection of 20 NLP datasets with human annotations covering a broad range of evaluated properties and types of data, and comprehensively evaluate 11 current LLMs, covering both open-weight and proprietary models, for their ability to replicate the annotations. Our evaluations show substantial variance across models and datasets. Models are reliable evaluators on some tasks, but overall display substantial variability depending on the property being evaluated, the expertise level of the human judges, and whether the language is human or model-generated. We conclude that LLMs should be carefully validated against human judgments before being used as evaluators.
title LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Tasks
topic Computation and Language
url https://arxiv.org/abs/2406.18403