[en] Attention-based models such as Transformers and recurrent models like state space models (SSMs) have emerged as successful methods for autoregressive sequence modeling. Although both enable parallel training, none enable parallel generation due to their autoregressiveness. We propose the variational SSM (VSSM), a variational autoencoder (VAE) where both the encoder and decoder are SSMs. Since sampling the latent variables and decoding them with the SSM can be parallelized, both training and generation can be conducted in parallel. Moreover, the decoder recurrence allows generation to be resumed without reprocessing the whole sequence. Finally, we propose the autoregressive VSSM that can be conditioned on a partial realization of the sequence, as is common in language generation tasks. Interestingly, the autoregressive VSSM still enables parallel generation. We highlight on toy problems (MNIST, CIFAR) the empirical gains in speed-up and show that it competes with traditional models in terms of generation quality (Transformer, Mamba SSM).
Disciplines :
Computer science
Author, co-author :
Lambrechts, Gaspard ✱; Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
Claes, Yann ✱; Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
Geurts, Pierre ; Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
Ernst, Damien ; Université de Liège - ULiège > Montefiore Institute of Electrical Engineering and Computer Science
✱ These authors have contributed equally to this work.
Language :
English
Title :
Parallelizing Autoregressive Generation with Variational State Space Models
Publication date :
July 2024
Event name :
ICML Workshop on Next Generation of Sequence Modeling Architectures