clip加载text encoder

最新推荐文章于 2025-05-14 08:49:12 发布

Wanderer X

最新推荐文章于 2025-05-14 08:49:12 发布

阅读量637

点赞数 1

分类专栏：代码文章标签：深度学习人工智能

本文链接：https://blog.csdn.net/wandererXX/article/details/142823894

版权

代码专栏收录该内容

17 篇文章

订阅专栏

import os
import clip
import torch

# Load the model
device = "cuda" if torch.cuda.is_available() else "cpu"
model, preprocess = clip.load('ViT-B/32', device)
text_inputs = clip.tokenize('hello').to(device)
with torch.no_grad():
    text_features = model.encode_text(text_inputs)
text_features /= text_features.norm(dim=-1, keepdim=True)