azshue / TPT

Test-time Prompt Tuning (TPT) for zero-shot generalization in vision-language models (NeurIPS 2022))
https://azshue.github.io/TPT/
MIT License
136 stars 16 forks source link

Reproducing papers results - CoOp and CoCoop weights #3

Closed taesiri closed 1 year ago

taesiri commented 1 year ago

Hello,

Thank you for your excellent work and paper.

I want to reproduce the paper's results related to both ResNet and ViT-B/16 models. (Table 1 in the paper). Could you please highlight which weight files you used for CoOp and CoCoOp ?

azshue commented 1 year ago

Hello,

Thank you for your interest in our work!

For CoOp, we use the pre-trained weights released by its authors. The official pre-trained CoCoOp is not publicly available, so we trained our own CoCoOp model using the hyper-parameter setting provided in the original paper. However, we do not plan to release our reproduced CoCoOp. You could contact the author of CoCoOp or train a model yourself using its official implementation.

Best, Manli