Implementation:Neuml Txtai Extractive QA

Knowledge Sources	Neuml_Txtai
Domains	Machine Learning, NLP, Question Answering, Transformers
Last Updated	2026-02-10 01:00 GMT

Overview

Concrete tool for running extractive question answering over question-context pairs provided by txtai.

Description

Questions extends HFPipeline and wraps the Hugging Face question-answering pipeline to perform extractive QA. For each question-context pair, the pipeline extracts the best answer span from the context. Answers with a confidence score below 0.05 are filtered out and returned as None. The pipeline handles null or empty questions/contexts gracefully by returning None for those entries.

Usage

Use Questions when you need to extract specific answers from text passages given a set of questions. This is the standard extractive QA approach where answers are spans of text directly from the provided context, not generated text.

Code Reference

Source Location

• Repository: Neuml_Txtai
• File: src/python/txtai/pipeline/text/questions.py

Signature

class Questions(HFPipeline):
    def __init__(self, path=None, quantize=False, gpu=True, model=None, **kwargs)
    def __call__(self, questions, contexts, workers=0)

Import

from txtai.pipeline.text.questions import Questions

I/O Contract

Inputs

Name	Type	Required	Description
questions	list	Yes	List of question strings. Entries can be None or empty.
contexts	list	Yes	List of context strings corresponding to each question. Must be the same length as questions.
workers	int	No	Number of concurrent workers for data processing. Defaults to 0.

Outputs

Name	Type	Description
answers	list	List of answer strings extracted from contexts. Returns None for entries where the question or context is empty, or where the confidence score is below 0.05.

Usage Examples

from txtai.pipeline.text.questions import Questions

# Create a question answering pipeline
qa = Questions("distilbert-base-cased-distilled-squad", gpu=True)

# Answer questions from contexts
questions = [
    "What is the capital of France?",
    "Who founded Microsoft?"
]
contexts = [
    "Paris is the capital and largest city of France.",
    "Microsoft was founded by Bill Gates and Paul Allen in 1975."
]

answers = qa(questions, contexts)
# Returns: ["Paris", "Bill Gates and Paul Allen"]

# Handles empty inputs gracefully
answers = qa(["What is AI?", None], ["AI is artificial intelligence", "Some context"])
# Returns: ["artificial intelligence", None]

Related Pages

• Environment:Neuml_Txtai_Python_Core_Dependencies