How to Evaluate Reward Models for RLHF
- URL: http://arxiv.org/abs/2410.14872v2
- Date: Tue, 22 Oct 2024 22:18:14 GMT
- Title: How to Evaluate Reward Models for RLHF
- Authors: Evan Frick, Tianle Li, Connor Chen, Wei-Lin Chiang, Anastasios N. Angelopoulos, Jiantao Jiao, Banghua Zhu, Joseph E. Gonzalez, Ion Stoica,
- Abstract summary: We introduce a new benchmark for reward models that quantifies their ability to produce strong language models through RLHF (Reinforcement Learning from Human Feedback)
We build a predictive model of downstream LLM performance by evaluating the reward model on proxy tasks.
We launch an end-to-end RLHF experiment on a large-scale crowdsourced human preference platform to view real reward model downstream performance as ground truth.
- Score: 51.31240621943791
- License:
- Abstract: We introduce a new benchmark for reward models that quantifies their ability to produce strong language models through RLHF (Reinforcement Learning from Human Feedback). The gold-standard approach is to run a full RLHF training pipeline and directly probe downstream LLM performance. However, this process is prohibitively expensive. To address this, we build a predictive model of downstream LLM performance by evaluating the reward model on proxy tasks. These proxy tasks consist of a large-scale human preference and a verifiable correctness preference dataset, in which we measure 12 metrics across 12 domains. To investigate which reward model metrics are most correlated to gold-standard RLHF outcomes, we launch an end-to-end RLHF experiment on a large-scale crowdsourced human preference platform to view real reward model downstream performance as ground truth. Ultimately, we compile our data and findings into Preference Proxy Evaluations (PPE), the first reward model benchmark explicitly linked to post-RLHF real-world human preference performance, which we open-source for public use and further development. Our code and evaluations can be found at https://github.com/lmarena/PPE .
Related papers
- Bootstrapping Language Models with DPO Implicit Rewards [45.68366127605774]
Direct preference optimization (DPO) has greatly simplified the process from past work in reinforcement learning from human feedback.
In this work, we make a novel observation that this implicit reward model can by itself be used in a bootstrapping fashion to further align the LLM.
Our approach, named self-alignment with DPO ImpliCit rEwards (DICE), shows great improvements in alignment and achieves superior performance.
arXiv Detail & Related papers (2024-06-14T06:57:18Z) - RLHF Workflow: From Reward Modeling to Online RLHF [79.83927049253924]
We present the workflow of Online Iterative Reinforcement Learning from Human Feedback (RLHF) in this technical report.
RLHF is widely reported to outperform its offline counterpart by a large margin in the recent large language model (LLM) literature.
We show that supervised fine-tuning (SFT) and iterative RLHF can obtain state-of-the-art performance with fully open-source datasets.
arXiv Detail & Related papers (2024-05-13T15:50:39Z) - Weak-to-Strong Extrapolation Expedites Alignment [135.12769233630362]
We propose a method called ExPO to boost models' alignment with human preference.
We demonstrate that ExPO consistently improves off-the-shelf DPO/RLHF models.
We shed light on the essence of ExPO amplifying the reward signal learned during alignment training.
arXiv Detail & Related papers (2024-04-25T17:39:50Z) - RewardBench: Evaluating Reward Models for Language Modeling [100.28366840977966]
We present RewardBench, a benchmark dataset and code-base for evaluation of reward models.
The dataset is a collection of prompt-chosen-rejected trios spanning chat, reasoning, and safety.
On the RewardBench leaderboard, we evaluate reward models trained with a variety of methods.
arXiv Detail & Related papers (2024-03-20T17:49:54Z) - PRDP: Proximal Reward Difference Prediction for Large-Scale Reward Finetuning of Diffusion Models [13.313186665410486]
Reward finetuning has emerged as a promising approach to aligning foundation models with downstream objectives.
Existing reward finetuning methods are limited by their instability in large-scale prompt datasets.
We propose Proximal Reward Difference Prediction (PRDP) to enable stable black-box reward finetuning for diffusion models.
arXiv Detail & Related papers (2024-02-13T18:58:16Z) - Improving Reinforcement Learning from Human Feedback with Efficient Reward Model Ensemble [67.4269821365504]
Reinforcement Learning from Human Feedback (RLHF) is a widely adopted approach for aligning large language models with human values.
However, RLHF relies on a reward model that is trained with a limited amount of human preference data.
We contribute a reward ensemble method that allows the reward model to make more accurate predictions.
arXiv Detail & Related papers (2024-01-30T00:17:37Z) - The Alignment Ceiling: Objective Mismatch in Reinforcement Learning from
Human Feedback [5.037876196534672]
Reinforcement learning from human feedback (RLHF) has emerged as a powerful technique to make large language models (LLMs) more capable in complex settings.
In this paper, we illustrate the causes of this issue, reviewing relevant literature from model-based reinforcement learning, and argue for solutions.
arXiv Detail & Related papers (2023-10-31T21:52:41Z) - Direct Preference Optimization: Your Language Model is Secretly a Reward Model [119.65409513119963]
We introduce a new parameterization of the reward model in RLHF that enables extraction of the corresponding optimal policy in closed form.
The resulting algorithm, which we call Direct Preference Optimization (DPO), is stable, performant, and computationally lightweight.
Our experiments show that DPO can fine-tune LMs to align with human preferences as well as or better than existing methods.
arXiv Detail & Related papers (2023-05-29T17:57:46Z)
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.