Bayesian Reward Models for LLM Alignment

To ensure that large language model (LLM) responses are helpful and non-toxic, we usually fine-tune a reward model on human preference data. We then select policy responses with high rewards (best-of-n sampling) or further optimize the policy to produce responses with high rewards (reinforcement learning from human feedback). However, this process is vulnerable to reward overoptimization or hacking, in which the responses selected have high rewards due to errors in the reward model rather than a genuine preference. This is especially problematic as the prompt or response diverges from the training data. It should be possible to mitigate these issues by training a Bayesian reward model, which signals higher uncertainty further from the training data distribution. Therefore, we trained Bayesian reward models using Laplace-LoRA (Yang et al., 2024) and found that the resulting uncertainty estimates can successfully mitigate reward overoptimization in best-of-n sampling.

Bayesian low-rank adaptation for large language models

Parameter-efficient fine-tuning (PEFT) has emerged as a new paradigm for cost-efficient fine-tuning of large language models (LLMs), with low-rank adaptation (LoRA) being a widely adopted choice. However, fine-tuned LLMs often become overconfident especially when fine-tuned on small datasets. Bayesian methods, with their inherent ability to estimate uncertainty, serve as potent tools to mitigate overconfidence and enhance calibration. In this work, we introduce Laplace-LoRA, a straightforward yet effective Bayesian method, which applies the Laplace approximation to the LoRA parameters and, considerably boosts the calibration of fine-tuned LLMs.

MONGOOSE: Path-wise Smooth Bayesian Optimisation via Meta-learning

In Bayesian optimisation, we often seek to minimise the black-box objective functions that arise in real-world physical systems. A primary contributor to the cost of evaluating such black-box objective functions is often the effort required to prepare the system for measurement. We consider a common scenario where preparation costs grow as the distance between successive evaluations increases. In this setting, smooth optimisation trajectories are preferred and the jumpy paths produced by the standard myopic (i.e.\ one-step-optimal) Bayesian optimisation methods are sub-optimal. Our algorithm, MONGOOSE, uses a meta-learnt parametric policy to generate smooth optimisation trajectories, achieving performance gains over existing methods when optimising functions with large movement costs.

Deep kernel processes

We define deep kernel processes in which positive definite Gram matrices are progressively transformed by nonlinear kernel functions and by sampling from (inverse) Wishart distributions. Remarkably, we find that deep Gaussian processes (DGPs), Bayesian neural networks (BNNs), infinite BNNs, and infinite BNNs with bottlenecks can all be written as deep kernel processes. For DGPs the equivalence arises because the Gram matrix formed by the inner product of features is Wishart distributed, and as we show, standard isotropic kernels can be written entirely in terms of this Gram matrix -- we do not need knowledge of the underlying features. We define a tractable deep kernel process, the deep inverse Wishart process, and give a doubly-stochastic inducing-point variational inference scheme that operates on the Gram matrices, not on the features, as in DGPs. We show that the deep inverse Wishart process gives superior performance to DGPs and infinite BNNs on standard fully-connected baselines.