Learn What Is Possible, Then Choose What Is Best: Disentangling One-To-Many Relations in Language Through Text-based Games

Benjamin Towle, Ke Zhou


Abstract
Language models pre-trained on large self-supervised corpora, followed by task-specific fine-tuning has become the dominant paradigm in NLP. These pre-training datasets often have a one-to-many structure—e.g. in dialogue there are many valid responses for a given context. However, only some of these responses will be desirable in our downstream task. This raises the question of how we should train the model such that it can emulate the desirable behaviours, but not the undesirable ones. Current approaches train in a one-to-one setup—only a single target response is given for a single dialogue context—leading to models only learning to predict the average response, while ignoring the full range of possible responses. Using text-based games as a testbed, our approach, PASA, uses discrete latent variables to capture the range of different behaviours represented in our larger pre-training dataset. We then use knowledge distillation to distil the posterior probability distribution into a student model. This probability distribution is far richer than learning from only the hard targets of the dataset, and thus allows the student model to benefit from the richer range of actions the teacher model has learned. Results show up to 49% empirical improvement over the previous state-of-the-art model on the Jericho Walkthroughs dataset.
Anthology ID:
2022.findings-emnlp.364
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2022
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4955–4965
Language:
URL:
https://aclanthology.org/2022.findings-emnlp.364
DOI:
10.18653/v1/2022.findings-emnlp.364
Bibkey:
Cite (ACL):
Benjamin Towle and Ke Zhou. 2022. Learn What Is Possible, Then Choose What Is Best: Disentangling One-To-Many Relations in Language Through Text-based Games. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 4955–4965, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Learn What Is Possible, Then Choose What Is Best: Disentangling One-To-Many Relations in Language Through Text-based Games (Towle & Zhou, Findings 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.findings-emnlp.364.pdf
Video:
 https://aclanthology.org/2022.findings-emnlp.364.mp4