We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CL

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Lifelong Pretraining: Continually Adapting Language Models to Emerging Corpora

Abstract: Pretrained language models (PTLMs) are typically learned over a large, static corpus and further fine-tuned for various downstream tasks. However, when deployed in the real world, a PTLM-based model must deal with data from a new domain that deviates from what the PTLM was initially trained on, or newly emerged data that contains out-of-distribution information. In this paper, we study a lifelong language model pretraining challenge where a PTLM is continually updated so as to adapt to emerging data. Over a domain-incremental research paper stream and a chronologically ordered tweet stream, we incrementally pretrain a PTLM with different continual learning algorithms, and keep track of the downstream task performance (after fine-tuning) to analyze its ability of acquiring new knowledge and preserving learned knowledge. Our experiments show continual learning algorithms improve knowledge preservation, with logit distillation being the most effective approach. We further show that continual pretraining improves generalization when training and testing data of downstream tasks are drawn from different time steps, but do not improve when they are from the same time steps. We believe our problem formulation, methods, and analysis will inspire future studies towards continual pretraining of language models.
Comments: 8 pages
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2110.08534 [cs.CL]
  (or arXiv:2110.08534v1 [cs.CL] for this version)

Submission history

From: Xisen Jin [view email]
[v1] Sat, 16 Oct 2021 09:59:33 GMT (298kb,D)

Link back to: arXiv, form interface, contact.