Closed na-fi closed 3 weeks ago
Thanks for the intriguing work!
I have a question about the t-SNE experiment. For generating the embeddings, did you fine-tune on a downstream task, like classification, first? Or did you simply use the encoder with the pre-trained MAE weights?
Thank you
Thank you for getting interested in our paper.
We used the pre-trained encoder weights and generated the embeddings. (didn't fine-tune on a downstream task)
Thank you :)
Thanks for the intriguing work!
I have a question about the t-SNE experiment. For generating the embeddings, did you fine-tune on a downstream task, like classification, first? Or did you simply use the encoder with the pre-trained MAE weights?
Thank you