paper-with-me

Papers

On Masked Pre-training and the Marginal Likelihood

2023-09-21 · NeurIPS 2023 11

Masked pre-training removes random input dimensions and learns a model that can predict the missing values. Empirical results indicate that this intuitive form of self-supervised learning yields models that generalize very well to new domains. A theoretical understanding is, however, lacking. This paper shows that masked pre-training with a suitable cumulative scoring function corresponds to maximizing the model's marginal likelihood, which is de facto the Bayesian model selection measure of generalization. Beyond shedding light on the success of masked pre-training, this insight also suggests that Bayesian models can be trained with appropriately designed self-supervision. Empirically, we confirm the developed theory and explore the main learning principles of masked pre-training in large language models.

📄 PDF Abstract BibTeX

Code (1)

pmorenoz/mpt-lml 공식 구현 pytorch

Similar Papers 제목 키워드 기반

Marginal Post Processing of Bayesian Inference Products with Normalizing Flows and Kernel Density Estimators

2022-05-25 · Harry T. J. Bevins, William J. Handley, Pablo Lemos, Peter H. Sims 외

Bayesian analysis has become an indispensable tool across many different cosmological fields including the study of gravitational waves, the Cosmic Microwave Background and the 21-cm signal from the Cosmic Dawn among oth…

Bayesian InferenceExperimental Design

Discrete Tilt Matching

2026-04-20 · Yuyuan Chen, Shiyi Wang, Peter Potaptchik, Jaeyeon Kim 외 arxiv

Masked diffusion large language models (dLLMs) are a promising alternative to autoregressive generation. While reinforcement learning (RL) methods have recently been adapted to dLLM fine-tuning, their objectives typicall…

Reinforcement Learning

Scalable Marginal Likelihood Estimation for Model Selection in Deep Learning

2021-04-11 · Alexander Immer, Matthias Bauer, Vincent Fortuin, Gunnar Rätsch 외

Marginal-likelihood based model-selection, even though promising, is rarely used in deep learning due to estimation difficulties. Instead, most approaches rely on validation data, which may not be readily available. In t…

Deep Learningimage-classificationImage ClassificationModel Selection+2

Last Layer Marginal Likelihood for Invariance Learning

2021-06-14 · Pola Schwöbel, Martin Jørgensen, Sebastian W. Ober, Mark van der Wilk

Data augmentation is often used to incorporate inductive biases into models. Traditionally, these are hand-crafted and tuned with cross validation. The Bayesian paradigm for model selection provides a path towards end-to…

Data AugmentationGaussian ProcessesModel Selection

On the marginal likelihood and cross-validation

2019-05-21 · Edwin Fong, Chris Holmes

In Bayesian statistics, the marginal likelihood, also known as the evidence, is used to evaluate model fit as it quantifies the joint probability of the data under the prior. In contrast, non-Bayesian models are typicall…

scoring rule