Align-to-Distill: Trainable Attention Alignment for Knowledge Distillation in Neural Machine Translation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917621408989184 |
|---|---|
| author | Jin, Heegon Son, Seonil Park, Jemin Kim, Youngseok Noh, Hyungjong Lee, Yeonsoo |
| author_facet | Jin, Heegon Son, Seonil Park, Jemin Kim, Youngseok Noh, Hyungjong Lee, Yeonsoo |
| contents | The advent of scalable deep models and large datasets has improved the performance of Neural Machine Translation. Knowledge Distillation (KD) enhances efficiency by transferring knowledge from a teacher model to a more compact student model. However, KD approaches to Transformer architecture often rely on heuristics, particularly when deciding which teacher layers to distill from. In this paper, we introduce the 'Align-to-Distill' (A2D) strategy, designed to address the feature mapping problem by adaptively aligning student attention heads with their teacher counterparts during training. The Attention Alignment Module in A2D performs a dense head-by-head comparison between student and teacher attention heads across layers, turning the combinatorial mapping heuristics into a learning problem. Our experiments show the efficacy of A2D, demonstrating gains of up to +3.61 and +0.63 BLEU points for WMT-2022 De->Dsb and WMT-2014 En->De, respectively, compared to Transformer baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_01479 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Align-to-Distill: Trainable Attention Alignment for Knowledge Distillation in Neural Machine Translation Jin, Heegon Son, Seonil Park, Jemin Kim, Youngseok Noh, Hyungjong Lee, Yeonsoo Computation and Language Artificial Intelligence 68T50 I.2.7 The advent of scalable deep models and large datasets has improved the performance of Neural Machine Translation. Knowledge Distillation (KD) enhances efficiency by transferring knowledge from a teacher model to a more compact student model. However, KD approaches to Transformer architecture often rely on heuristics, particularly when deciding which teacher layers to distill from. In this paper, we introduce the 'Align-to-Distill' (A2D) strategy, designed to address the feature mapping problem by adaptively aligning student attention heads with their teacher counterparts during training. The Attention Alignment Module in A2D performs a dense head-by-head comparison between student and teacher attention heads across layers, turning the combinatorial mapping heuristics into a learning problem. Our experiments show the efficacy of A2D, demonstrating gains of up to +3.61 and +0.63 BLEU points for WMT-2022 De->Dsb and WMT-2014 En->De, respectively, compared to Transformer baselines. |
| title | Align-to-Distill: Trainable Attention Alignment for Knowledge Distillation in Neural Machine Translation |
| topic | Computation and Language Artificial Intelligence 68T50 I.2.7 |
| url | https://arxiv.org/abs/2403.01479 |