FABRIC is a benchmark for evaluating how reliably large language models provide financial advice for Indian markets across six languages.

from datasets import load_dataset

dataset = load_dataset("agenticclass/fabric")

q = dataset["train"][0] print(q["question_en"]) # English version print(q["question_hi"]) # Hindi version print(q["answer"]) # Ground truth answer… See the full description on the dataset page: https://huggingface.co/datasets/agenticclass/fabric.