AIbase
EN
Home
Distilbert Qasports
D

Distilbert Qasports

Developed by laurafcamargos
A question-answering model fine-tuned on distilbert-base-uncased-distilled-squad, achieving an F1 score of 81.33 on the evaluation set
Question Answering System
Transformers
Open Source License:Apache-2.0#Question Answering System#High F1 Score#Sports Domain
Downloads 18
Release Time : 2/8/2025

Model Overview

This model is a lightweight question-answering system based on DistilBERT, specifically optimized for QA tasks, capable of quickly and accurately answering questions based on given context.

Model Features

Efficient QA
Based on the lightweight DistilBERT architecture, it improves inference speed while maintaining high accuracy.
Excellent Performance
Achieves an F1 score of 81.33 and an exact match rate of 76.87 on the evaluation set.
Optimized Training
Uses techniques like linear learning rate scheduling and mixed-precision training to optimize the training process.

Model Capabilities

Context-based question answering
Text comprehension
Answer extraction

Use Cases

Customer Support
Product FAQ Responses
Automatically answers common user questions about product features.
Accuracy rate of 76.87%
Education
Reading Comprehension Assistance
Helps students quickly find answers to questions from given texts.
F1 score of 81.33
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase