AIbase
EN
Home
Vit Base Patch32 Clip 224.metaclip 2pt5b
V

Vit Base Patch32 Clip 224.metaclip 2pt5b

Developed by timm
A vision Transformer model trained on the MetaCLIP-2.5B dataset, compatible with both open_clip and timm frameworks
Image Classification
Safetensors
#Zero-shot image classification#Multi-framework compatibility#Large-scale pre-training
Downloads 5,571
Release Time : 10/23/2024

Model Overview

This is a dual-framework compatible vision Transformer model primarily designed for zero-shot image classification tasks, supporting usage under both open_clip and timm frameworks.

Model Features

Dual-framework compatibility
Supports both open_clip and timm frameworks, offering more flexible usage options
Large-scale pre-training
Trained on the large-scale MetaCLIP-2.5B dataset, possessing powerful visual representation capabilities
Fast inference
Utilizes 32x32 patch size and quickgelu activation function, balancing accuracy and speed

Model Capabilities

Zero-shot image classification
Image feature extraction
Cross-modal representation learning

Use Cases

Computer vision
Zero-shot image classification
Classify images of new categories without requiring specific category training data
Image retrieval
Retrieve relevant images based on text queries
Multimodal applications
Image-text matching
Determine whether images and text descriptions match
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