paper-with-me

Papers

Partition Generative Modeling: Masked Modeling Without Masks

2025-05-24 · Justin Deschenaux, Lan Tran, Caglar Gulcehre

We introduce ``Partition Generative Models'' (PGMs), a novel approach to masked generative modeling (MGMs), particularly effective for masked diffusion language modeling (MDLMs). PGM divides tokens into two distinct groups and employs sparse attention patterns to prevent cross-group information exchange. Hence, the model is trained to predict tokens in one group based solely on information from the other group. This partitioning strategy eliminates the need for MASK tokens entirely. While traditional MGMs inefficiently process MASK tokens during generation, PGMs achieve greater computational efficiency by operating exclusively on unmasked tokens. Our experiments on OpenWebText with a context length of 1024 tokens demonstrate that PGMs deliver at least 5x improvements in both latency and throughput compared to MDLM when using the same number of sampling steps, while generating samples with better generative perplexity than MDLM. Finally, we show that PGMs can be distilled with Self-Distillation Through Time (SDTT), a method originally devised for MDLM, in order to achieve further inference gains.

📄 PDF Abstract BibTeX arXiv:2505.18883

Code (1)

kuleshov-group/mdlm 공식 구현 pytorch

Tasks

Computational EfficiencyLanguage ModelingLanguage Modelling

Methods 이 논문이 사용한 방법론

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Attention 설명 없음
Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…
PGM A regularization criterion that, differently from dropout and its variants, is deterministic rather than random. It grounds on the…

Similar Papers 제목 키워드 기반

Designing BERT for Convolutional Networks: Sparse and Hierarchical Masked Modeling

2023-01-09 · Keyu Tian, Yi Jiang, Qishuai Diao, Chen Lin 외

We identify and overcome two key obstacles in extending the success of BERT-style pre-training, or the masked image modeling, to convolutional networks (convnets): (i) convolution operation cannot handle irregular, rando…

2D Object DetectionContrastive LearningDecoderImage Classification+5

RECTOR: Masked Region-Channel-Temporal Modeling for Affective and Cognitive Representation Learning

2026-06-13 · Jinhan Liu, Mahsa Shoaran arxiv

Affective and cognitive disorders manifest as distributed, time-varying brain network dynamics across regions, channels, and time, challenging robust representation learning from EEG/sEEG for clinical diagnosis. We propo…

EEG Emotion RecognitionRepresentation Learning

VL-BEiT: Generative Vision-Language Pretraining

2022-06-02 · Hangbo Bao, Wenhui Wang, Li Dong, Furu Wei

We introduce a vision-language foundation model called VL-BEiT, which is a bidirectional multimodal Transformer learned by generative pretraining. Our minimalist solution conducts masked prediction on both monomodal and …

image-classificationImage ClassificationImage-text RetrievalLanguage Modeling+9

Masked Image Modeling Boosting Semi-Supervised Semantic Segmentation

2024-11-13 · Yangyang Li, Xuanting Hao, Ronghua Shang, Licheng Jiao

In view of the fact that semi- and self-supervised learning share a fundamental principle, effectively modeling knowledge from unlabeled data, various semi-supervised semantic segmentation methods have integrated represe…

Self-Supervised LearningSemantic SegmentationSemi-Supervised Semantic Segmentation

Depth-Wise Representation Development Under Blockwise Self-Supervised Learning for Video Vision Transformers

2026-01-14 · Jonas Römer, Timo Dickscheid arxiv

End-to-end backpropagation couples all layers through a global error signal, enabling coordinated learning but requiring long-range credit assignment. Motivated by recent progress in blockwise self-supervised learning (B…

Self-Supervised Learning