albert-base-v2-squad

This model is a fine-tuned version of albert-base-v2 on the SQuAD 1.1 and adversarial_qa datasets. It achieves the following results on the SQuAD 1.1 evaluation set:

• Exact Match(EM): 84.68
• F1: 91.40

Inference API

You can test the model directly using the Hugging Face Inference API:

from transformers import pipeline

# Load the pipeline
qa_pipeline = pipeline("question-answering", model="xichenn/albert-base-v2-squad")

# Run inference
result = qa_pipeline(question="What is the capital of France?", context="France is a country in Europe. Its capital is Paris.")

print(result)