⚡⚡ CLIP - Retraining topics

[QIN2DIM]

If you find that our previously released model does not handle the current binary challenge image_binary_challenge well, i.e., the PROMPT remains the same, but the dataset features have drifted significantly.

🔀🔀 At this point you can submit an issue to alert me to the release of a new model (it only takes me a few minutes to retrain the model, but perhaps due to different time zones, I may sometimes see it half a day after you submit the issue)🥹

• [Re-train] Please click on the largest animal. #881
• [Challenge] furniture #859
• [Challenge] [re-train] Please click each image containing an item of office equipment #862

⚡⚡ [PyPI v0.9.2-post3] At this point you can use the CLIP model for self-supervised classification. The same process operates for new challenges.

see more details;

• feat(onnx): ViT zero-shot tasks #858
• example