Probabilistic Inference in Language Models via Twisted Sequential Monte Carlo

• URL: http://arxiv.org/abs/2404.17546v1
• Date: Fri, 26 Apr 2024 17:18:32 GMT
• Title: Probabilistic Inference in Language Models via Twisted Sequential Monte Carlo
• Authors: Stephen Zhao, Rob Brekelmans, Alireza Makhzani, Roger Grosse,
• Abstract summary: We leverage the rich toolkit of Sequential Monte Carlo (SMC) for probabilistic inference problems. We use learned twist functions to estimate the expected future value of the potential at each timestep. We present methods for evaluating the accuracy of language model inference techniques.
• Score: 7.182174507225034
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Numerous capability and safety techniques of Large Language Models (LLMs), including RLHF, automated red-teaming, prompt engineering, and infilling, can be cast as sampling from an unnormalized target distribution defined by a given reward or potential function over the full sequence. In this work, we leverage the rich toolkit of Sequential Monte Carlo (SMC) for these probabilistic inference problems. In particular, we use learned twist functions to estimate the expected future value of the potential at each timestep, which enables us to focus inference-time computation on promising partial sequences. We propose a novel contrastive method for learning the twist functions, and establish connections with the rich literature of soft reinforcement learning. As a complementary application of our twisted SMC framework, we present methods for evaluating the accuracy of language model inference techniques using novel bidirectional SMC bounds on the log partition function. These bounds can be used to estimate the KL divergence between the inference and target distributions in both directions. We apply our inference evaluation techniques to show that twisted SMC is effective for sampling undesirable outputs from a pretrained model (a useful component of harmlessness training and automated red-teaming), generating reviews with varied sentiment, and performing infilling tasks.

Related papers

• Amortizing intractable inference in large language models [56.92471123778389]
  We use amortized Bayesian inference to sample from intractable posterior distributions. We empirically demonstrate that this distribution-matching paradigm of LLM fine-tuning can serve as an effective alternative to maximum-likelihood training. As an important application, we interpret chain-of-thought reasoning as a latent variable modeling problem.
  arXiv  Detail & Related papers  (2023-10-06T16:36:08Z)
• Observation-Guided Diffusion Probabilistic Models [41.749374023639156]
  We propose a novel diffusion-based image generation method called the observation-guided diffusion probabilistic model (OGDM) Our approach reestablishes the training objective by integrating the guidance of the observation process with the Markov chain. We demonstrate the effectiveness of our training algorithm using diverse inference techniques on strong diffusion model baselines.
  arXiv  Detail & Related papers  (2023-10-06T06:29:06Z)
• Practical Probabilistic Model-based Deep Reinforcement Learning by Integrating Dropout Uncertainty and Trajectory Sampling [7.179313063022576]
  This paper addresses the prediction stability, prediction accuracy and control capability of the current probabilistic model-based reinforcement learning (MBRL) built on neural networks. A novel approach dropout-based probabilistic ensembles with trajectory sampling (DPETS) is proposed.
  arXiv  Detail & Related papers  (2023-09-20T06:39:19Z)
• Value-Distributional Model-Based Reinforcement Learning [59.758009422067]
  Quantifying uncertainty about a policy's long-term performance is important to solve sequential decision-making tasks. We study the problem from a model-based Bayesian reinforcement learning perspective. We propose Epistemic Quantile-Regression (EQR), a model-based algorithm that learns a value distribution function.
  arXiv  Detail & Related papers  (2023-08-12T14:59:19Z)
• ProbVLM: Probabilistic Adapter for Frozen Vision-Language Models [69.50316788263433]
  We propose ProbVLM, a probabilistic adapter that estimates probability distributions for the embeddings of pre-trained vision-language models. We quantify the calibration of embedding uncertainties in retrieval tasks and show that ProbVLM outperforms other methods. We present a novel technique for visualizing the embedding distributions using a large-scale pre-trained latent diffusion model.
  arXiv  Detail & Related papers  (2023-07-01T18:16:06Z)
• Cheap and Deterministic Inference for Deep State-Space Models of Interacting Dynamical Systems [38.23826389188657]
  We present a deep state-space model which employs graph neural networks in order to model the underlying interacting dynamical system. The predictive distribution is multimodal and has the form of a Gaussian mixture model, where the moments of the Gaussian components can be computed via deterministic moment matching rules. Our moment matching scheme can be exploited for sample-free inference, leading to more efficient and stable training compared to Monte Carlo alternatives.
  arXiv  Detail & Related papers  (2023-05-02T20:30:23Z)
• Low-variance estimation in the Plackett-Luce model via quasi-Monte Carlo sampling [58.14878401145309]
  We develop a novel approach to producing more sample-efficient estimators of expectations in the PL model. We illustrate our findings both theoretically and empirically using real-world recommendation data from Amazon Music and the Yahoo learning-to-rank challenge.
  arXiv  Detail & Related papers  (2022-05-12T11:15:47Z)
• Distributionally Robust Models with Parametric Likelihood Ratios [123.05074253513935]
  Three simple ideas allow us to train models with DRO using a broader class of parametric likelihood ratios. We find that models trained with the resulting parametric adversaries are consistently more robust to subpopulation shifts when compared to other DRO approaches.
  arXiv  Detail & Related papers  (2022-04-13T12:43:12Z)
• Learning Operators with Coupled Attention [9.715465024071333]
  We propose a novel operator learning method, LOCA, motivated from the recent success of the attention mechanism. In our architecture the input functions are mapped to a finite set of features which are then averaged with attention weights that depend on the output query locations. By coupling these attention weights together with an integral transform, LOCA is able to explicitly learn correlations in the target output functions.
  arXiv  Detail & Related papers  (2022-01-04T08:22:03Z)
• Nonparametric Conditional Density Estimation In A Deep Learning Framework For Short-Term Forecasting [0.34410212782758043]
  Many machine learning techniques give a single-point prediction of the conditional distribution of the target variable. We propose a technique that simultaneously estimates the entire conditional distribution and flexibly allows for machine learning techniques to be incorporated.
  arXiv  Detail & Related papers  (2020-08-17T22:31:19Z)
• Learned Factor Graphs for Inference from Stationary Time Sequences [107.63351413549992]
  We propose a framework that combines model-based algorithms and data-driven ML tools for stationary time sequences. neural networks are developed to separately learn specific components of a factor graph describing the distribution of the time sequence. We present an inference algorithm based on learned stationary factor graphs, which learns to implement the sum-product scheme from labeled data.
  arXiv  Detail & Related papers  (2020-06-05T07:06:19Z)

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.