T

Tapas Large Finetuned Sqa

Developed by google
This model is the large version of TAPAS, fine-tuned for sequential question answering (SQA) tasks, suitable for table-related question answering scenarios.
Downloads 71
Release Time : 3/2/2022

Model Overview

TAPAS is a BERT-like Transformer model specifically designed for question answering tasks involving tabular data and related text. The model was pre-trained on English Wikipedia tables and fine-tuned on the SQA dataset.

Model Features

Table Question Answering Capability
A QA model specifically designed for tabular data, capable of understanding table structures and content.
Sequential Question Answering Support
Supports sequential question answering in conversational scenarios, capable of handling interrelated series of questions.
Two-stage Training
First pre-trained with masked language modeling, then undergoes intermediate pre-training to enhance numerical reasoning capabilities.

Model Capabilities

Table Understanding
Sequential Question Answering
Numerical Reasoning
Text-Table Association Analysis

Use Cases

Business Intelligence
Financial Statement Analysis
Automatically answers various questions about financial statement data
Achieved 72.89% accuracy on the SQA dataset
Customer Service
Product Information Query
Answers customer inquiries based on product specification tables
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase