Correlating and Predicting Human Evaluations of Language Models from Natural Language Processing Benchmarks

• URL: http://arxiv.org/abs/2502.18339v1
• Date: Mon, 24 Feb 2025 01:01:02 GMT
• Title: Correlating and Predicting Human Evaluations of Language Models from Natural Language Processing Benchmarks
• Authors: Rylan Schaeffer, Punit Singh Koura, Binh Tang, Ranjan Subramanian, Aaditya K Singh, Todor Mihaylov, Prajjwal Bhargava, Lovish Madaan, Niladri S. Chatterji, Vedanuj Goswami, Sergey Edunov, Dieuwke Hupkes, Sanmi Koyejo, Sharan Narang,
• Abstract summary: We compare four Chat Llama 2 models against extensive human preferences on more than 11k single-turn and 2k multi-turn dialogues from over 2k human annotators.<n>Most NLP benchmarks strongly correlate with human evaluations, suggesting that cheaper, automated metrics can serve as surprisingly reliable predictors of human preferences.
• Score: 47.40240774236047
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The explosion of high-performing conversational language models (LMs) has spurred a shift from classic natural language processing (NLP) benchmarks to expensive, time-consuming and noisy human evaluations - yet the relationship between these two evaluation strategies remains hazy. In this paper, we conduct a large-scale study of four Chat Llama 2 models, comparing their performance on 160 standard NLP benchmarks (e.g., MMLU, ARC, BIG-Bench Hard) against extensive human preferences on more than 11k single-turn and 2k multi-turn dialogues from over 2k human annotators. Our findings are striking: most NLP benchmarks strongly correlate with human evaluations, suggesting that cheaper, automated metrics can serve as surprisingly reliable predictors of human preferences. Three human evaluations, such as adversarial dishonesty and safety, are anticorrelated with NLP benchmarks, while two are uncorrelated. Moreover, through overparameterized linear regressions, we show that NLP scores can accurately predict human evaluations across different model scales, offering a path to reduce costly human annotation without sacrificing rigor. Overall, our results affirm the continued value of classic benchmarks and illuminate how to harness them to anticipate real-world user satisfaction - pointing to how NLP benchmarks can be leveraged to meet evaluation needs of our new era of conversational AI.

Related papers

• HypoEval: Hypothesis-Guided Evaluation for Natural Language Generation [25.193026443079987]
  HypoEval is a Hypothesis-guided Evaluation framework for large language models (LLMs) With only 30 human evaluations, HypoEval achieves state-of-the-art performance in alignment with both human rankings (Spearman correlation) and human scores (Pearson correlation) We conduct systematic studies to assess the robustness of HypoEval, highlighting its effectiveness as a reliable and interpretable automated evaluation framework.
  arXiv  Detail & Related papers  (2025-04-09T18:00:01Z)
• ContrastScore: Towards Higher Quality, Less Biased, More Efficient Evaluation Metrics with Contrastive Evaluation [38.49478115331574]
  We introduce ContrastScore, a contrastive evaluation metric designed to enable higher-quality, less biased, and more efficient assessment of generated text. Experimental results show that ContrastScore consistently achieves stronger correlation with human judgments than both single-model and ensemble-based baselines.
  arXiv  Detail & Related papers  (2025-04-02T20:11:45Z)
• Accelerating Unbiased LLM Evaluation via Synthetic Feedback [17.597195550638343]
  We propose a statistically principled framework that integrates human and synthetic feedback to reduce reliance on human annotations.<n>Our experiments demonstrate a reduction in human annotations by up to 12.2% with an off-the-shelf synthetic evaluator and up to 24.8% with a finetuned variant.
  arXiv  Detail & Related papers  (2025-02-14T21:27:09Z)
• A LLM-Based Ranking Method for the Evaluation of Automatic Counter-Narrative Generation [14.064465097974836]
  This paper proposes a novel approach to evaluate Counter Narrative (CN) generation using a Large Language Model (LLM) as an evaluator. We show that traditional automatic metrics correlate poorly with human judgements and fail to capture the nuanced relationship between generated CNs and human perception.
  arXiv  Detail & Related papers  (2024-06-21T15:11:33Z)
• ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate [57.71597869337909]
  We build a multi-agent referee team called ChatEval to autonomously discuss and evaluate the quality of generated responses from different models. Our analysis shows that ChatEval transcends mere textual scoring, offering a human-mimicking evaluation process for reliable assessments.
  arXiv  Detail & Related papers  (2023-08-14T15:13:04Z)
• INSTRUCTSCORE: Explainable Text Generation Evaluation with Finegrained Feedback [80.57617091714448]
  We present InstructScore, an explainable evaluation metric for text generation. We fine-tune a text evaluation metric based on LLaMA, producing a score for generated text and a human readable diagnostic report.
  arXiv  Detail & Related papers  (2023-05-23T17:27:22Z)
• FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation [176.56131810249602]
  evaluating the factuality of long-form text generated by large language models (LMs) is non-trivial. We introduce FACTSCORE, a new evaluation that breaks a generation into a series of atomic facts and computes the percentage of atomic facts supported by a reliable knowledge source.
  arXiv  Detail & Related papers  (2023-05-23T17:06:00Z)
• G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment [64.01972723692587]
  We present G-Eval, a framework of using large language models with chain-of-thoughts (CoT) and a form-filling paradigm to assess the quality of NLG outputs. We show that G-Eval with GPT-4 as the backbone model achieves a Spearman correlation of 0.514 with human on summarization task, outperforming all previous methods by a large margin.
  arXiv  Detail & Related papers  (2023-03-29T12:46:54Z)
• Revisiting the Gold Standard: Grounding Summarization Evaluation with Robust Human Evaluation [136.16507050034755]
  Existing human evaluation studies for summarization either exhibit a low inter-annotator agreement or have insufficient scale. We propose a modified summarization salience protocol, Atomic Content Units (ACUs), which is based on fine-grained semantic units. We curate the Robust Summarization Evaluation (RoSE) benchmark, a large human evaluation dataset consisting of 22,000 summary-level annotations over 28 top-performing systems.
  arXiv  Detail & Related papers  (2022-12-15T17:26:05Z)
• Dynamic Human Evaluation for Relative Model Comparisons [8.843915018287476]
  We present a dynamic approach to measure the required number of human annotations when evaluating generated outputs in relative comparison settings. We propose an agent-based framework of human evaluation to assess multiple labelling strategies and methods to decide the better model in a simulation and a crowdsourcing case study.
  arXiv  Detail & Related papers  (2021-12-15T11:32:13Z)
• Towards Automatic Evaluation of Dialog Systems: A Model-Free Off-Policy Evaluation Approach [84.02388020258141]
  We propose a new framework named ENIGMA for estimating human evaluation scores based on off-policy evaluation in reinforcement learning. ENIGMA only requires a handful of pre-collected experience data, and therefore does not involve human interaction with the target policy during the evaluation. Our experiments show that ENIGMA significantly outperforms existing methods in terms of correlation with human evaluation scores.
  arXiv  Detail & Related papers  (2021-02-20T03:29:20Z)
• Learning to Compare for Better Training and Evaluation of Open Domain Natural Language Generation Models [23.62054164511058]
  We propose to evaluate natural language generation models by learning to compare a pair of generated sentences by fine-tuning BERT. While able to be trained in a fully self-supervised fashion, our model can be further fine-tuned with a little amount of human preference annotation.
  arXiv  Detail & Related papers  (2020-02-12T15:52:21Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.