styfeng/DataAug4NLP

Organized by 15+ NLP task categories (text classification, translation, QA, sequence tagging, parsing, dialogue, multimodal, and more), this curated repository systematically maps augmentation techniques to specific problem domains with linked implementations and benchmark datasets. Grounded in an ACL 2021 survey paper, it combines peer-reviewed research with a community contribution model via pull requests, enabling practitioners to identify task-appropriate augmentation strategies from backtranslation and synonym replacement to recent LLM-based approaches like GPT3Mix and automated augmentation methods.

831 stars. No commits in the last 6 months.