Open-source Question - Answering Model question-answering-roberta-base-s-v2 - Precise Inference of Answers and Confidence Scores

Question Answering Roberta Base S V2

Developed by consciousAI

A RoBERTa-based question answering model specialized in inferring answer text, span, and confidence scores given a question and context.

Question Answering System

Transformers

Open Source License:Apache-2.0 #High-precision Q&A #Context Understanding #SQUAD Fine-tuning

Downloads 1,832

Release Time : 11/21/2022

Model Overview

This model adopts an encoder-only architecture, based on deepset/roberta-base-squad2, equipped with a question-answering language model head, and fine-tuned on the SQUADx dataset, suitable for question-answering tasks.

Model Features

High-performance Q&A

Fine-tuned on the SQUADx dataset, achieving an exact match score of 84.83 and an F1 score of 91.80.

Multi-framework Support

Supports seamless collaboration among Jax, PyTorch, and TensorFlow, the three mainstream deep learning libraries.

Sub-block Processing

Capable of handling longer text blocks by dividing them into sub-blocks with a maximum context length of 384 tokens and a stride of 128 tokens.

Model Capabilities

Question Answering System

Text Understanding

Answer Span Inference

Use Cases

Education

Automated Q&A System

Used in educational settings for automated Q&A systems, helping students quickly obtain answers to questions.

High exact match and F1 scores, providing accurate answers.

Customer Support

Intelligent Customer Service

Used in customer support systems to automatically answer common customer questions.

Quick response times, improving customer satisfaction.

🚀 Question Answering

This model is designed for the Q&A task. Given a question and context, it attempts to infer the answer text, answer span, and confidence score. It's an encoder-only model (deepset/roberta-base-squad2) with a QuestionAnswering LM Head, fine-tuned on the SQUADx dataset, achieving performance scores of exact_match: 84.83 and f1: 91.80.

🚀 Quick Start

Live Demo: Question Answering Encoders vs Generative

Please follow this link for Encoder based Question Answering V1 Please follow this link for Generative Question Answering

💻 Usage Examples

Basic Usage

from transformers import pipeline

model_checkpoint = "consciousAI/question-answering-roberta-base-s-v2"

context = """
🤗 Transformers is backed by the three most popular deep learning libraries — Jax, PyTorch and TensorFlow — with a seamless integration
between them. It's straightforward to train your models with one before loading them for inference with the other.
"""
question = "Which deep learning libraries back 🤗 Transformers?"

question_answerer = pipeline("question-answering", model=model_checkpoint)
question_answerer(question=question, context=context)

📚 Documentation

Training and evaluation data

SQUAD Split

Training procedure

Preprocessing

SQUAD Data longer chunks were sub-chunked with input context max-length 384 tokens and stride as 128 tokens.
Target answers readjusted for sub-chunks, sub-chunks with no-answers or partial answers were set to target answer span as (0,0)

Metrics

Adjusted accordingly to handle sub-chunking.
n best = 20
skip answers with length zero or higher than max answer length (30)

Training hyperparameters

Custom Training Loop: The following hyperparameters were used during training:

learning_rate: 2e-5
train_batch_size: 32
eval_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 2

Training results

{'exact_match': 84.83443708609272, 'f1': 91.79987545811638}

Framework versions

Transformers 4.23.0.dev0
Pytorch 1.12.1+cu113
Datasets 2.5.2
Tokenizers 0.13.0

📄 License

This project is licensed under the Apache-2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご