Abstract
We present Hidden-State Optimization (HSO), a gradient-based method for improving the performance of transformer language models at inference time. Similar to dynamic evaluation (Krause et al., 2018), HSO computes the gradient of the log-probability the language model assigns to an evaluation text, but uses it to update the cached hidden states rather than the model parameters. We test HSO with pretrained Transformer-XL and GPT-2 language models, finding improvement on the WikiText-103 and PG-19 datasets in terms of perplexity, especially when evaluating a model outside of its training distribution. We also demonstrate downstream applicability by showing gains in the recently developed prompt-based few-shot evaluation setting, again with no extra parameters or training data.- Anthology ID:
- 2021.findings-emnlp.346
- Volume:
- Findings of the Association for Computational Linguistics: EMNLP 2021
- Month:
- November
- Year:
- 2021
- Address:
- Punta Cana, Dominican Republic
- Editors:
- Marie-Francine Moens, Xuanjing Huang, Lucia Specia, Scott Wen-tau Yih
- Venue:
- Findings
- SIG:
- SIGDAT
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 4099–4105
- Language:
- URL:
- https://aclanthology.org/2021.findings-emnlp.346
- DOI:
- 10.18653/v1/2021.findings-emnlp.346
- Cite (ACL):
- Davis Yoshida and Kevin Gimpel. 2021. Reconsidering the Past: Optimizing Hidden States in Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 4099–4105, Punta Cana, Dominican Republic. Association for Computational Linguistics.
- Cite (Informal):
- Reconsidering the Past: Optimizing Hidden States in Language Models (Yoshida & Gimpel, Findings 2021)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-3/2021.findings-emnlp.346.pdf
- Data
- AG News, PG-19, SST, SST-2, WikiText-103