Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Models

https://arxiv.org/abs/2404.18796

View a PDF of the paper titled Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models, by Pat Verga and 8 other authors

View PDF HTML (experimental)

Abstract:As Large Language Models (LLMs) have become more advanced, they have outpaced our abilities to accurately evaluate their quality. Not only is finding data to adequately probe particular model properties difficult, but evaluating the correctness of a model's freeform generation alone is a challenge. To address this, many evaluations now rely on using LLMs themselves as judges to score the quality of outputs from other LLMs. Evaluations most commonly use a single large model like GPT4. While this method has grown in popularity, it is costly, has been shown to introduce intramodel bias, and in this work, we find that very large models are often unnecessary. We propose instead to evaluate models using a Panel of LLm evaluators (PoLL). Across three distinct judge settings and spanning six different datasets, we find that using a PoLL composed of a larger number of smaller models outperforms a single large judge, exhibits less intra-model bias due to its composition of disjoint model families, and does so while being over seven times less expensive.

Submission history

From: Pat Verga [view email]
[v1] Mon, 29 Apr 2024 15:33:23 UTC (7,795 KB)
[v2] Wed, 1 May 2024 15:37:11 UTC (7,795 KB)

{
  "by": "Jimmc414",
  "descendants": 4,
  "id": 40215100,
  "kids": [
    40216933,
    40216434,
    40216381,
    40217318
  ],
  "score": 47,
  "time": 1714504816,
  "title": "Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Models",
  "type": "story",
  "url": "https://arxiv.org/abs/2404.18796"
}

{
  "author": "Pat Verga",
  "date": "2024-04-29T12:00:00.000Z",
  "description": "As Large Language Models (LLMs) have become more advanced, they have outpaced our abilities to accurately evaluate their quality. Not only is finding data to adequately probe particular model properties difficult, but evaluating the correctness of a model’s freeform generation alone is a challenge. To address this, many evaluations now rely on using LLMs themselves as judges to score the quality of outputs from other LLMs. Evaluations most commonly use a single large model like GPT4. While this method has grown in popularity, it is costly, has been shown to introduce intramodel bias, and in this work, we find that very large models are often unnecessary. We propose instead to evaluate models using a Panel of LLm evaluators (PoLL). Across three distinct judge settings and spanning six different datasets, we find that using a PoLL composed of a larger number of smaller models outperforms a single large judge, exhibits less intra-model bias due to its composition of disjoint model families, and does so while being over seven times less expensive.",
  "image": "https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png",
  "logo": "https://logo.clearbit.com/arxiv.org",
  "publisher": "arXiv.org",
  "title": "Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models",
  "url": "https://arxiv.org/abs/2404.18796v2"
}

{
  "url": "https://arxiv.org/abs/2404.18796",
  "title": "Replacing Judges with Juries: Evaluating LLM Generations with a...",
  "description": "As Large Language Models (LLMs) have become more advanced, they have outpaced our abilities to accurately evaluate their quality. Not only is finding data to adequately probe particular model...",
  "links": [
    "https://arxiv.org/abs/2404.18796v2",
    "https://arxiv.org/abs/2404.18796"
  ],
  "image": "https://static.arxiv.org/icons/twitter/arxiv-logo-twitter-square.png",
  "content": "<div>\n                <p>View a PDF of the paper titled Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models, by Pat Verga and 8 other authors</p>\n    <p><a target=\"_blank\" href=\"https://arxiv.org/pdf/2404.18796\">View PDF</a>\n    <a target=\"_blank\" href=\"https://arxiv.org/html/2404.18796v2\">HTML (experimental)</a></p><blockquote>\n            <span>Abstract:</span>As Large Language Models (LLMs) have become more advanced, they have outpaced our abilities to accurately evaluate their quality. Not only is finding data to adequately probe particular model properties difficult, but evaluating the correctness of a model's freeform generation alone is a challenge. To address this, many evaluations now rely on using LLMs themselves as judges to score the quality of outputs from other LLMs. Evaluations most commonly use a single large model like GPT4. While this method has grown in popularity, it is costly, has been shown to introduce intramodel bias, and in this work, we find that very large models are often unnecessary. We propose instead to evaluate models using a Panel of LLm evaluators (PoLL). Across three distinct judge settings and spanning six different datasets, we find that using a PoLL composed of a larger number of smaller models outperforms a single large judge, exhibits less intra-model bias due to its composition of disjoint model families, and does so while being over seven times less expensive.\n    </blockquote>\n  </div><div>\n      <h2>Submission history</h2><p> From: Pat Verga [<a target=\"_blank\" href=\"https://arxiv.org/show-email/0e344dbe/2404.18796\">view email</a>]      <br />            <strong><a target=\"_blank\" href=\"https://arxiv.org/abs/2404.18796v1\">[v1]</a></strong>\n        Mon, 29 Apr 2024 15:33:23 UTC (7,795 KB)<br />\n    <strong>[v2]</strong>\n        Wed, 1 May 2024 15:37:11 UTC (7,795 KB)<br />\n</p></div>",
  "author": "",
  "favicon": "https://arxiv.org/static/browse/0.3.4/images/icons/favicon-16x16.png",
  "source": "arxiv.org",
  "published": "",
  "ttr": 46,
  "type": "website"
}