Foundational Autoraters: Taming Large Language Models for Better Automatic Evaluation

TL;DR

The paper introduces FLAMe, a family of large autorater models trained on diverse human judgments, significantly improving automatic evaluation of LLM outputs across multiple benchmarks and reducing bias compared to proprietary models.

Contribution

We develop FLAMe, a new large-scale autorater model trained on over 5 million human judgments, enhancing generalization and performance in automatic LLM evaluation tasks.

Findings

FLAMe outperforms proprietary models like GPT-4 on many benchmarks.

FLAMe achieves 87.8% accuracy on RewardBench with a 24B parameter model.

Tail-patch fine-tuning reduces training data needs by 25x with competitive results.

Abstract

As large language models (LLMs) advance, it becomes more challenging to reliably evaluate their output due to the high costs of human evaluation. To make progress towards better LLM autoraters, we introduce FLAMe, a family of Foundational Large Autorater Models. FLAMe is trained on our large and diverse collection of 100+ quality assessment tasks comprising 5M+ human judgments, curated and standardized using publicly released human evaluations from previous research. FLAMe significantly improves generalization to a wide variety of held-out tasks, outperforming LLMs trained on proprietary data like GPT-4 and Claude-3 on many tasks. We show that FLAMe can also serve as a powerful starting point for further downstream fine-tuning, using reward modeling evaluation as a case study (FLAMe-RM). Notably, on RewardBench, our FLAMe-RM-24B model (with an accuracy of 87.8%) is the top-performing generative model trained exclusively on permissively licensed data, outperforming both GPT-4-0125 (85.9%) and GPT-4o (84.7%). Additionally, we explore a more computationally efficient approach using a novel tail-patch fine-tuning strategy to optimize our FLAMe multitask mixture for reward modeling evaluation (FLAMe-Opt-RM), offering competitive RewardBench performance while requiring approximately 25x less training datapoints. Overall, our FLAMe variants outperform all popular proprietary LLM-as-a-Judge models we consider across 8 out of 12 autorater evaluation benchmarks, encompassing 53 quality assessment tasks, including RewardBench and LLM-AggreFact. Finally, our analysis reveals that FLAMe is significantly less biased than these LLM-as-a-Judge models on the CoBBLEr autorater bias benchmark, while effectively identifying high-quality responses for code generation.