Clear Preferences Leave Traces: Reference Model-Guided Sampling for Preference Learning

• URL: http://arxiv.org/abs/2501.15109v1
• Date: Sat, 25 Jan 2025 07:21:50 GMT
• Title: Clear Preferences Leave Traces: Reference Model-Guided Sampling for Preference Learning
• Authors: Nirav Diwan, Tolga Ergen, Dongsub Shim, Honglak Lee,
• Abstract summary: Direct Preference Optimization (DPO) has emerged as a de-facto approach for aligning language models with human preferences. Recent work has shown DPO's effectiveness relies on training data quality. We discover that reference model probability space naturally detects high-quality training samples.
• Score: 59.11519451499754
• License:
• Abstract: Direct Preference Optimization (DPO) has emerged as a de-facto approach for aligning language models with human preferences. Recent work has shown DPO's effectiveness relies on training data quality. In particular, clear quality differences between preferred and rejected responses enhance learning performance. Current methods for identifying and obtaining such high-quality samples demand additional resources or external models. We discover that reference model probability space naturally detects high-quality training samples. Using this insight, we present a sampling strategy that achieves consistent improvements (+0.1 to +0.4) on MT-Bench while using less than half (30-50%) of the training data. We observe substantial improvements (+0.4 to +0.98) for technical tasks (coding, math, and reasoning) across multiple models and hyperparameter settings.

Related papers

• Optimizing LLMs with Direct Preferences: A Data Efficiency Perspective [4.548047308860141]
  This study investigates the impact of different type of preference data on model performance. It aims to reduce their dependency on extensive amounts of preference data, which is expensive to collect.
  arXiv  Detail & Related papers  (2024-10-22T00:11:41Z)
• When Does Visual Prompting Outperform Linear Probing for Vision-Language Models? A Likelihood Perspective [57.05315507519704]
  We propose a log-likelihood ratio (LLR) approach to analyze the comparative benefits of visual prompting and linear probing. Our measure attains up to a 100-fold reduction in run time compared to full training, while achieving prediction accuracies up to 91%.
  arXiv  Detail & Related papers  (2024-09-03T12:03:45Z)
• Unpacking DPO and PPO: Disentangling Best Practices for Learning from Preference Feedback [110.16220825629749]
  Learning from preference feedback has emerged as an essential step for improving the generation quality and performance of modern language models. In this work, we identify four core aspects of preference-based learning: preference data, learning algorithm, reward model, and policy training prompts. Our findings indicate that all aspects are important for performance, with better preference data leading to the largest improvements.
  arXiv  Detail & Related papers  (2024-06-13T16:17:21Z)
• Uncertainty Aware Learning for Language Model Alignment [97.36361196793929]
  We propose uncertainty-aware learning (UAL) to improve the model alignment of different task scenarios. We implement UAL in a simple fashion -- adaptively setting the label smoothing value of training according to the uncertainty of individual samples. Experiments on widely used benchmarks demonstrate that our UAL significantly and consistently outperforms standard supervised fine-tuning.
  arXiv  Detail & Related papers  (2024-06-07T11:37:45Z)
• Towards An Online Incremental Approach to Predict Students Performance [0.8287206589886879]
  We propose a memory-based online incremental learning approach for updating an online classifier. Our approach achieves a notable improvement in model accuracy, with an enhancement of nearly 10% compared to the current state-of-the-art.
  arXiv  Detail & Related papers  (2024-05-03T17:13:26Z)
• How to Train Data-Efficient LLMs [56.41105687693619]
  We study data-efficient approaches for pre-training language models (LLMs) We find that Ask-LLM and Density sampling are the best methods in their respective categories. In our comparison of 19 samplers, involving hundreds of evaluation tasks and pre-training runs, we find that Ask-LLM and Density are the best methods in their respective categories.
  arXiv  Detail & Related papers  (2024-02-15T02:27:57Z)
• Revisiting Sample Size Determination in Natural Language Understanding [18.637079595450366]
  Knowing exactly how many data points need to be labeled to achieve a certain model performance is a beneficial step towards reducing the overall budgets for annotation. We derived a simple yet effective approach to predict the maximum achievable model performance based on small amount of training samples.
  arXiv  Detail & Related papers  (2023-07-01T16:08:52Z)
• ZeroGen$^+$: Self-Guided High-Quality Data Generation in Efficient Zero-Shot Learning [97.2907428983142]
  ZeroGen attempts to purely use PLM to generate data and train a tiny model without relying on task-specific annotation. We propose a noise-robust bi-level re-weighting framework which is able to learn the per-sample weights measuring the data quality without requiring any gold data.
  arXiv  Detail & Related papers  (2022-05-25T11:38:48Z)

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.