The main feature of SQuAD v1.1 is its capability. In this version, the answer to every question is a segment of text (a "span") from the corresponding reading passage. Key Features of SQuAD v1.1

: Models are primarily evaluated based on two metrics: Exact Match (EM) and F1 Score .

: The official evaluation script evaluate-v1.1.py is used to calculate the performance of your model's predictions.

: Every answer is directly taken from the provided Wikipedia article context.

: The data is provided in a standard JSON format, typically split into train-v1.1.json and dev-v1.1.json . Downloads and Resources