Enhancing Latent Computation in Transformers with Latent Tokens
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866913845775171584 |
|---|---|
| author | Sun, Yuchang Chen, Yanxi Li, Yaliang Ding, Bolin |
| author_facet | Sun, Yuchang Chen, Yanxi Li, Yaliang Ding, Bolin |
| contents | Augmenting large language models (LLMs) with auxiliary tokens has emerged as a promising strategy for enhancing model performance. In this work, we introduce a lightweight method termed latent tokens; these are dummy tokens that may be non-interpretable in natural language but steer the autoregressive decoding process of a Transformer-based LLM via the attention mechanism. The proposed latent tokens can be seamlessly integrated with a pre-trained Transformer, trained in a parameter-efficient manner, and applied flexibly at inference time, while adding minimal complexity overhead to the existing infrastructure of standard Transformers. We propose several hypotheses about the underlying mechanisms of latent tokens and design synthetic tasks accordingly to verify them. Numerical results confirm that the proposed method noticeably outperforms the baselines, particularly in the out-of-distribution generalization scenarios, highlighting its potential in improving the adaptability of LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_12629 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Enhancing Latent Computation in Transformers with Latent Tokens Sun, Yuchang Chen, Yanxi Li, Yaliang Ding, Bolin Machine Learning Computation and Language Augmenting large language models (LLMs) with auxiliary tokens has emerged as a promising strategy for enhancing model performance. In this work, we introduce a lightweight method termed latent tokens; these are dummy tokens that may be non-interpretable in natural language but steer the autoregressive decoding process of a Transformer-based LLM via the attention mechanism. The proposed latent tokens can be seamlessly integrated with a pre-trained Transformer, trained in a parameter-efficient manner, and applied flexibly at inference time, while adding minimal complexity overhead to the existing infrastructure of standard Transformers. We propose several hypotheses about the underlying mechanisms of latent tokens and design synthetic tasks accordingly to verify them. Numerical results confirm that the proposed method noticeably outperforms the baselines, particularly in the out-of-distribution generalization scenarios, highlighting its potential in improving the adaptability of LLMs. |
| title | Enhancing Latent Computation in Transformers with Latent Tokens |
| topic | Machine Learning Computation and Language |
| url | https://arxiv.org/abs/2505.12629 |