Closed XixinYang closed 10 months ago
As a text-image alignment foundation mode, CLIP is not appropriate to place in examples
. We can put it under mindone
As a text-image alignment foundation mode, CLIP is not appropriate to place in
examples
. We can put it undermindone
Take back by words. Let MindCV be its root.
TODO: add an API check list in a susquent PR.
TODO:Compare with the original CLIP paper on the zero-shot and linear-probe ACC (imagenet)
TODO: add an API check list in a susquent PR.
done
Thank you for your contribution to the MindCV repo. Before submitting this PR, please make sure:
Motivation
Add mindspore-version CLIP which is totally rewritten based on openAI-CLIP.
Test Plan
These code have been tested on the first example in ./README.md with both ViT and ResNet ckpts, which has used all the main APIS, and the result differences are as below: ViT: ResNet: After discussion, all the diffences are considered within an acceptable range.
Related Issues and PRs
(Is this PR part of a group of changes? Link the other relevant PRs and Issues here. Use https://help.github.com/en/articles/closing-issues-using-keywords for help on GitHub syntax)