Analyzing Feed-Forward Blocks in Transformers through the Lens of Attention Maps
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2023
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917639548305408 |
|---|---|
| author | Kobayashi, Goro Kuribayashi, Tatsuki Yokoi, Sho Inui, Kentaro |
| author_facet | Kobayashi, Goro Kuribayashi, Tatsuki Yokoi, Sho Inui, Kentaro |
| contents | Transformers are ubiquitous in wide tasks. Interpreting their internals is a pivotal goal. Nevertheless, their particular components, feed-forward (FF) blocks, have typically been less analyzed despite their substantial parameter amounts. We analyze the input contextualization effects of FF blocks by rendering them in the attention maps as a human-friendly visualization scheme. Our experiments with both masked- and causal-language models reveal that FF networks modify the input contextualization to emphasize specific types of linguistic compositions. In addition, FF and its surrounding components tend to cancel out each other's effects, suggesting potential redundancy in the processing of the Transformer layer. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2302_00456 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Analyzing Feed-Forward Blocks in Transformers through the Lens of Attention Maps Kobayashi, Goro Kuribayashi, Tatsuki Yokoi, Sho Inui, Kentaro Computation and Language Transformers are ubiquitous in wide tasks. Interpreting their internals is a pivotal goal. Nevertheless, their particular components, feed-forward (FF) blocks, have typically been less analyzed despite their substantial parameter amounts. We analyze the input contextualization effects of FF blocks by rendering them in the attention maps as a human-friendly visualization scheme. Our experiments with both masked- and causal-language models reveal that FF networks modify the input contextualization to emphasize specific types of linguistic compositions. In addition, FF and its surrounding components tend to cancel out each other's effects, suggesting potential redundancy in the processing of the Transformer layer. |
| title | Analyzing Feed-Forward Blocks in Transformers through the Lens of Attention Maps |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2302.00456 |