Orthogonal Language and Task Adapters in Zero-Shot Cross-Lingual Transfer
This work offers an incremental improvement for researchers working on efficient cross-lingual transfer in natural language processing.
This paper proposes orthogonal language and task adapters (orthoadapters) for zero-shot cross-lingual transfer, aiming to encode complementary language- and task-specific information. Experiments across three tasks (POS-tagging, NER, NLI) and 10 languages show their usefulness, particularly for the NLI task, though optimal configurations vary by task and language.
Adapter modules, additional trainable parameters that enable efficient fine-tuning of pretrained transformers, have recently been used for language specialization of multilingual transformers, improving downstream zero-shot cross-lingual transfer. In this work, we propose orthogonal language and task adapters (dubbed orthoadapters) for cross-lingual transfer. They are trained to encode language- and task-specific information that is complementary (i.e., orthogonal) to the knowledge already stored in the pretrained transformer's parameters. Our zero-shot cross-lingual transfer experiments, involving three tasks (POS-tagging, NER, NLI) and a set of 10 diverse languages, 1) point to the usefulness of orthoadapters in cross-lingual transfer, especially for the most complex NLI task, but also 2) indicate that the optimal adapter configuration highly depends on the task and the target language. We hope that our work will motivate a wider investigation of usefulness of orthogonality constraints in language- and task-specific fine-tuning of pretrained transformers.