What Makes Two Language Models Think Alike?
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913402381664256 |
|---|---|
| author | Salle, Jeanne Jalouzot, Louis Lan, Nur Chemla, Emmanuel Lakretz, Yair |
| author_facet | Salle, Jeanne Jalouzot, Louis Lan, Nur Chemla, Emmanuel Lakretz, Yair |
| contents | Do architectural differences significantly affect the way models represent and process language? We propose a new approach, based on metric-learning encoding models (MLEMs), as a first step to answer this question. The approach provides a feature-based comparison of how any two layers of any two models represent linguistic information. We apply the method to BERT, GPT-2 and Mamba. Unlike previous methods, MLEMs offer a transparent comparison, by identifying the specific linguistic features responsible for similarities and differences. More generally, the method uses formal, symbolic descriptions of a domain, and use these to compare neural representations. As such, the approach can straightforwardly be extended to other domains, such as speech and vision, and to other neural systems, including human brains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_12620 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | What Makes Two Language Models Think Alike? Salle, Jeanne Jalouzot, Louis Lan, Nur Chemla, Emmanuel Lakretz, Yair Computation and Language Do architectural differences significantly affect the way models represent and process language? We propose a new approach, based on metric-learning encoding models (MLEMs), as a first step to answer this question. The approach provides a feature-based comparison of how any two layers of any two models represent linguistic information. We apply the method to BERT, GPT-2 and Mamba. Unlike previous methods, MLEMs offer a transparent comparison, by identifying the specific linguistic features responsible for similarities and differences. More generally, the method uses formal, symbolic descriptions of a domain, and use these to compare neural representations. As such, the approach can straightforwardly be extended to other domains, such as speech and vision, and to other neural systems, including human brains. |
| title | What Makes Two Language Models Think Alike? |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2406.12620 |