Gendistiller: Distilling Pre-trained Language Models Based On An Autoregressive Generative Model

Abstract

Pre-trained speech language models such as HuBERT and WavLM leverage unlabeled speech data for self-supervised learning and offer powerful representations for numerous downstream tasks. Despite the success of these models, their high requirements for memory and computing resource hinder their application on resource restricted devices. Therefore, this paper introduces GenDistiller, a novel knowledge distillation framework which generates the hidden representations of the pre-trained teacher model directly by a much smaller student network. The proposed method takes the previous hidden layer as history and implements a layer-by-layer prediction of the teacher model autoregressively. Experiments on SUPERB reveal the advantage of GenDistiller over the baseline distilling method without an autoregressive framework, with 33% fewer parameters, similar time consumption and better performance on most of the SUPERB tasks. Ultimately, the proposed GenDistiller reduces the size of WavLM by 82%.

Stats

• citations1
• S2 citations—
• github stars0
• HF likes0
• heat score2.26
• arxiv keygao2024gendistiller

Related papers

• Gendistiller: Distilling Pre-trained Language Models Based On Generative Models (2023)0.00
• Distilhubert: Speech Representation Learning By Layer-wise Distillation Of Hidden-unit BERT (2021)15.06
• Star: Distilling Speech Temporal Relation For Lightweight Speech Self-supervised Learning Models (2023)5.24
• Fithubert: Going Thinner And Deeper For Knowledge Distillation Of Speech Self-supervised Learning (2022)10.97
• Application Of Knowledge Distillation To Multi-task Speech Representation Learning (2022)2.26
• SKILL: Similarity-aware Knowledge Distillation For Speech Self-supervised Learning (2024)3.58
• An Efficient End-to-end Approach To Noise Invariant Speech Features Via Multi-task Learning (2024)0.00
• Recycle-and-distill: Universal Compression Strategy For Transformer-based Speech SSL Models With Attention Map Reusing And Masking Distillation (2023)5.84