TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@1	69.2	# 4
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@5	89.9	# 5
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@10	96.1	# 5
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@1	62.2	# 8
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@5	86.6	# 8
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@10	94.9	# 6
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@1	64.0	# 7
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@5	89.2	# 7
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@10	94.4	# 8
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@1	64.7	# 6
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@5	89.6	# 6
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@10	94.6	# 7
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@1	48.1	# 12
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@5	81.3	# 10
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@10	90.5	# 10
Zero-shot Text-to-Image Retrieval	COCO-CN	Chinese CLIP ViT-H/14	Recall@1	69.2	# 3
Zero-shot Text-to-Image Retrieval	COCO-CN	Chinese CLIP ViT-H/14	Recall@5	89.9	# 3
Zero-shot Text-to-Image Retrieval	COCO-CN	Chinese CLIP ViT-H/14	Recall@10	96.1	# 3
Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@1	77.0	# 5
Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@5	97.1	# 1
Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@10	99.0	# 3
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@1	78.9	# 4
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@5	96.3	# 5
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@10	99.0	# 3
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@1	80.1	# 2
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@5	96.7	# 3
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@10	99.2	# 1
Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@1	81.5	# 1
Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@5	96.9	# 2
Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@10	99.1	# 2
Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@1	66.8	# 9
Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@5	91.1	# 9
Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@10	97.0	# 8
Zero-shot Text Retrieval	Flickr30k-CN	WukongViT-L/14	R@1	76.1	# 7
Zero-shot Text Retrieval	Flickr30k-CN	WukongViT-L/14	R@5	94.8	# 7
Zero-shot Text Retrieval	Flickr30k-CN	WukongViT-L/14	R@10	97.5	# 7
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@1	68.0	# 8
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@5	89.7	# 8
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@10	94.4	# 8
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@1	69.0	# 7
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@5	90.7	# 6
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@10	95.4	# 6
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@1	71.2	# 5
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@5	91.4	# 5
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@10	95.5	# 4
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14@336px	R@1	83.3	# 3
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14@336px	R@5	97.2	# 4
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14@336px	R@10	98.5	# 5
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@1	84.4	# 3
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@5	97.1	# 3
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@10	98.7	# 1
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@1	82.7	# 6
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@5	96.7	# 5
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@10	98.6	# 2
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@1	79.1	# 7
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@5	94.8	# 7
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@10	97.4	# 5
Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@1	66.7	# 11
Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@5	89.4	# 11
Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@10	94.1	# 11
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@1	48.8	# 12
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@5	76.0	# 12
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@10	84.6	# 12
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@1	62.7	# 9
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@5	86.9	# 9
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@10	92.8	# 9
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-H/14	R@1	81.6	# 4
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-H/14	R@5	97.5	# 2
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-H/14	R@10	98.8	# 3
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@1	83.8	# 5
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@5	96.9	# 4
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@10	98.6	# 2
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14	R@1	80.2	# 5
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14	R@5	96.6	# 6
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14	R@10	98.2	# 6
Zero-shot Text Retrieval	Flickr30k-CN	R2D2ViT-L/14	R@1	77.6	# 6
Zero-shot Text Retrieval	Flickr30k-CN	R2D2ViT-L/14	R@5	96.7	# 5
Zero-shot Text Retrieval	Flickr30k-CN	R2D2ViT-L/14	R@10	98.9	# 2
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@1	52.1	# 4
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@5	76.7	# 4
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@10	84.4	# 4
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	Mean Recall	71.1	# 4
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@1	63.0	# 1
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@5	84.1	# 1
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@10	89.2	# 1
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	Mean Recall	78.8	# 1
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@1	58.4	# 5
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@5	83.6	# 4
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@10	90.0	# 4
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	Mean Recall	77.4	# 5
Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@1	48.6	# 7
Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@5	75.1	# 7
Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@10	84.0	# 7
Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	Mean Recall	69.2	# 7
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@1	42.6	# 7
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@5	68.6	# 7
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@10	77.9	# 7
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	Mean Recall	63.0	# 7
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@1	59.0	# 2
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@5	81.4	# 2
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@10	87.8	# 2
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	Mean Recall	76.1	# 2
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@1	68.9	# 1
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@5	88.7	# 1
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@10	93.1	# 1
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	Mean Recall	83.6	# 1
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@1	65.3	# 2
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@5	86.7	# 2
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@10	92.1	# 2
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	Mean Recall	81.3	# 2
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@1	63.3	# 3
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@5	85.6	# 3
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@10	91.3	# 3
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	Mean Recall	80.1	# 3
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@1	56.3	# 3
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@5	79.8	# 3
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@10	86.2	# 3
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	Mean Recall	74.1	# 3

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/image-retrieval-on-coco-cn)](https://paperswithcode.com/sota/image-retrieval-on-coco-cn?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/zero-shot-image-retrieval-on-muge-retrieval)](https://paperswithcode.com/sota/zero-shot-image-retrieval-on-muge-retrieval?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/image-retrieval-on-muge-retrieval)](https://paperswithcode.com/sota/image-retrieval-on-muge-retrieval?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/zero-shot-text-to-image-retrieval-on-coco-cn)](https://paperswithcode.com/sota/zero-shot-text-to-image-retrieval-on-coco-cn?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/zero-shot-text-retrieval-on-flickr30k-cn)](https://paperswithcode.com/sota/zero-shot-text-retrieval-on-flickr30k-cn?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/image-retrieval-on-flickr30k-cn)](https://paperswithcode.com/sota/image-retrieval-on-flickr30k-cn?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/zero-shot-image-retrieval-on-coco-cn)](https://paperswithcode.com/sota/zero-shot-image-retrieval-on-coco-cn?p=chinese-clip-contrastive-vision-language)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/chinese-clip-contrastive-vision-language/zero-shot-image-retrieval-on-flickr30k-cn)](https://paperswithcode.com/sota/zero-shot-image-retrieval-on-flickr30k-cn?p=chinese-clip-contrastive-vision-language)`

Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese

2 Nov 2022 · An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, Chang Zhou ·

The tremendous success of CLIP (Radford et al., 2021) has promoted the research and application of contrastive learning for vision-language pretraining. In this work, we construct a large-scale dataset of image-text pairs in Chinese, where most data are retrieved from publicly available datasets, and we pretrain Chinese CLIP models on the new dataset. We develop 5 Chinese CLIP models of multiple sizes, spanning from 77 to 958 million parameters. Furthermore, we propose a two-stage pretraining method, where the model is first trained with the image encoder frozen and then trained with all parameters being optimized, to achieve enhanced model performance. Our comprehensive experiments demonstrate that Chinese CLIP can achieve the state-of-the-art performance on MUGE, Flickr30K-CN, and COCO-CN in the setups of zero-shot learning and finetuning, and it is able to achieve competitive performance in zero-shot image classification based on the evaluation on the ELEVATER benchmark (Li et al., 2022). We have released our codes, models, and demos in https://github.com/OFA-Sys/Chinese-CLIP

PDF Abstract

Code

Add Remove Mark official

ofa-sys/chinese-clip official

3,640

Tasks

Add Remove

Contrastive Learning

Image Classification

Image Retrieval

Representation Learning

Retrieval

Zero-Shot Image Classification

Zero-shot Image Retrieval

Zero-Shot Learning

Zero-shot Text Retrieval

Zero-shot Text-to-Image Retrieval

Datasets

Introduced in the Paper:

ImageNet_CN

Used in the Paper:

CIFAR-10

ImageNet

CIFAR-100

Visual Question Answering

Oxford 102 Flower

Caltech-101

EuroSAT

FGVC-Aircraft ELEVATER

COCO-CN

Flickr30k-CNA

Results from the Paper

Edit

Ranked #1 on Zero-shot Image Retrieval on MUGE Retrieval

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@1	69.2	# 4	Compare
			R@5	89.9	# 5	Compare
			R@10	96.1	# 5	Compare
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@1	62.2	# 8	Compare
			R@5	86.6	# 8	Compare
			R@10	94.9	# 6	Compare
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@1	64.0	# 7	Compare
			R@5	89.2	# 7	Compare
			R@10	94.4	# 8	Compare
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@1	64.7	# 6	Compare
			R@5	89.6	# 6	Compare
			R@10	94.6	# 7	Compare
Zero-shot Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@1	48.1	# 12	Compare
			R@5	81.3	# 10	Compare
			R@10	90.5	# 10	Compare
Zero-shot Text-to-Image Retrieval	COCO-CN	Chinese CLIP ViT-H/14	Recall@1	69.2	# 3	Compare
			Recall@5	89.9	# 3	Compare
			Recall@10	96.1	# 3	Compare
Image Retrieval	COCO-CN	CN-CLIP (ViT-B/16)	R@1	77.0	# 5	Compare
			R@5	97.1	# 1	Compare
			R@10	99.0	# 3	Compare
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14)	R@1	78.9	# 4	Compare
			R@5	96.3	# 5	Compare
			R@10	99.0	# 3	Compare
Image Retrieval	COCO-CN	CN-CLIP (ViT-L/14@336px)	R@1	80.1	# 2	Compare
			R@5	96.7	# 3	Compare
			R@10	99.2	# 1	Compare
Image Retrieval	COCO-CN	CN-CLIP (ViT-H/14)	R@1	81.5	# 1	Compare
			R@5	96.9	# 2	Compare
			R@10	99.1	# 2	Compare
Image Retrieval	COCO-CN	CN-CLIP (RN50)	R@1	66.8	# 9	Compare
			R@5	91.1	# 9	Compare
			R@10	97.0	# 8	Compare
Zero-shot Text Retrieval	Flickr30k-CN	WukongViT-L/14	R@1	76.1	# 7	Compare
			R@5	94.8	# 7	Compare
			R@10	97.5	# 7	Compare
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@1	68.0	# 8	Compare
			R@5	89.7	# 8	Compare
			R@10	94.4	# 8	Compare
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@1	69.0	# 7	Compare
			R@5	90.7	# 6	Compare
			R@10	95.4	# 6	Compare
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@1	71.2	# 5	Compare
			R@5	91.4	# 5	Compare
			R@10	95.5	# 4	Compare
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14@336px	R@1	83.3	# 3	Compare
			R@5	97.2	# 4	Compare
			R@10	98.5	# 5	Compare
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14@336px)	R@1	84.4	# 3	Compare
			R@5	97.1	# 3	Compare
			R@10	98.7	# 1	Compare
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-L/14)	R@1	82.7	# 6	Compare
			R@5	96.7	# 5	Compare
			R@10	98.6	# 2	Compare
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@1	79.1	# 7	Compare
			R@5	94.8	# 7	Compare
			R@10	97.4	# 5	Compare
Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@1	66.7	# 11	Compare
			R@5	89.4	# 11	Compare
			R@10	94.1	# 11	Compare
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (RN50)	R@1	48.8	# 12	Compare
			R@5	76.0	# 12	Compare
			R@10	84.6	# 12	Compare
Zero-shot Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-B/16)	R@1	62.7	# 9	Compare
			R@5	86.9	# 9	Compare
			R@10	92.8	# 9	Compare
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-H/14	R@1	81.6	# 4	Compare
			R@5	97.5	# 2	Compare
			R@10	98.8	# 3	Compare
Image Retrieval	Flickr30k-CN	CN-CLIP (ViT-H/14)	R@1	83.8	# 5	Compare
			R@5	96.9	# 4	Compare
			R@10	98.6	# 2	Compare
Zero-shot Text Retrieval	Flickr30k-CN	CN-CLIPViT-L/14	R@1	80.2	# 5	Compare
			R@5	96.6	# 6	Compare
			R@10	98.2	# 6	Compare
Zero-shot Text Retrieval	Flickr30k-CN	R2D2ViT-L/14	R@1	77.6	# 6	Compare
			R@5	96.7	# 5	Compare
			R@10	98.9	# 2	Compare
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@1	52.1	# 4	Compare
			R@5	76.7	# 4	Compare
			R@10	84.4	# 4	Compare
			Mean Recall	71.1	# 4	Compare
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@1	63.0	# 1	Compare
			R@5	84.1	# 1	Compare
			R@10	89.2	# 1	Compare
			Mean Recall	78.8	# 1	Compare
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-B/16)	R@1	58.4	# 5	Compare
			R@5	83.6	# 4	Compare
			R@10	90.0	# 4	Compare
			Mean Recall	77.4	# 5	Compare
Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@1	48.6	# 7	Compare
			R@5	75.1	# 7	Compare
			R@10	84.0	# 7	Compare
			Mean Recall	69.2	# 7	Compare
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (RN50)	R@1	42.6	# 7	Compare
			R@5	68.6	# 7	Compare
			R@10	77.9	# 7	Compare
			Mean Recall	63.0	# 7	Compare
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@1	59.0	# 2	Compare
			R@5	81.4	# 2	Compare
			R@10	87.8	# 2	Compare
			Mean Recall	76.1	# 2	Compare
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-H/14)	R@1	68.9	# 1	Compare
			R@5	88.7	# 1	Compare
			R@10	93.1	# 1	Compare
			Mean Recall	83.6	# 1	Compare
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14@336px)	R@1	65.3	# 2	Compare
			R@5	86.7	# 2	Compare
			R@10	92.1	# 2	Compare
			Mean Recall	81.3	# 2	Compare
Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@1	63.3	# 3	Compare
			R@5	85.6	# 3	Compare
			R@10	91.3	# 3	Compare
			Mean Recall	80.1	# 3	Compare
Zero-shot Image Retrieval	MUGE Retrieval	CN-CLIP (ViT-L/14)	R@1	56.3	# 3	Compare
			R@5	79.8	# 3	Compare
			R@10	86.2	# 3	Compare
			Mean Recall	74.1	# 3	Compare

Methods

Add Remove

CLIP • Contrastive Learning

Edit Social Preview

Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit

Methods Edit Add Remove

Code

Add Remove Mark official

Tasks

Add Remove

Datasets

Results from the Paper

Edit

Methods

Add Remove