bert-base-uncased-sparse-70-unstructured Open Source Model - High Sparsity Version Supporting Fine-tuning for Downstream Tasks

Bert Base Uncased Sparse 70 Unstructured

Developed by Intel

This is a pruned version of the BERT base model with 70% sparsity, suitable for fine-tuning on downstream tasks.

Downloads 28

Release Time : 3/2/2022

Model Overview

This model is a pruned version of the BERT base model with built-in sparsity, suitable for natural language processing tasks.

High Sparsity

The model has been pruned to achieve 70% sparsity, reducing computational resource requirements.

Sparsity Preservation

Built-in sparsity allows zero-value weights to remain unchanged during fine-tuning via masking.

BERT-based Architecture

Based on the BERT base model, inheriting its powerful natural language processing capabilities.

Text classification

Named entity recognition

Question answering

Text similarity calculation

Natural Language Processing

Sentiment Analysis

Analyze the sentiment polarity of text.

Text Classification

Classify text into predefined categories.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base