GLUECoS is an evaluation benchmark for code-switched NLP. The current version of the benchmark has eleven datasets, spanning six tasks and two language pairs (English-Hindi and English-Spanish). The tasks included in the benchmark are :

• Language Identification (LID)
• POS Tagging (POS)
• Named Entity Recognition (NER)
• Sentiment Analysis (SA)
• Question Answering (QA)
• Natural Language Inference (NLI)

GLUECoS is a continuous effort and we hope to make this an ever-growing test-bed for code-mixed language understanding, and plan to include more tasks and diverse language pairs in future versions.

Scripts for data preprocessing and evaluation for the benchmark can be found on our Github page which can be used to download all the datasets, process them and evaluate models on the benchmark.

To be included in the leaderboard, please refer to the README in the above link for submission instructions.