arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.08415cs.LGstat.ML
收藏

基于变分状态空间模型的自回归生成并行化方法

Parallelizing Autoregressive Generation with Variational State Space Models

Montefiore Institute, University of Liège(列日大学蒙泰菲奥雷研究所)

Gaspard Lambrechts, Yann Claes, Pierre Geurts, Damien Ernst

AI总结:

针对自回归序列模型无法并行生成的问题,提出编码器和解码器均为状态空间模型的变分自编码器VSSM,实现训练与生成全并行,且自回归变体仍支持并行生成,在速度和生成质量上表现优异。

摘要:

基于注意力的模型(如Transformer)和循环类模型(如状态空间模型SSM)已成为自回归序列建模的有效方法。尽管二者都支持并行训练,但由于自回归特性,均无法实现并行生成。本文提出变分状态空间模型(VSSM),这是一种编码器和解码器均采用SSM的变分自编码器(VAE)。由于潜在变量采样及利用SSM解码的过程可并行化,训练和生成均可并行执行。此外,解码器的循环特性支持生成过程恢复,无需重新处理完整序列。最后,本文提出可基于序列部分观测条件生成的自回归VSSM(适用于语言生成等常见任务),值得注意的是,该自回归VSSM仍支持并行生成。我们在玩具问题(MNIST、CIFAR)上验证了其加速效果,并表明其在生成质量上可与传统模型(Transformer、Mamba SSM)相媲美。

英文摘要:

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).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑