-
Hello, I'm just wondering how do you implement the memory buffer in SimCLRV2?
Did you follow exactly the same implementation with MoCo ? (i.e. use the output of the target encoder as the positive emb…
-
![MoCov1](https://github.com/youngtboy/Awesome-Self-Supervised-Vision-Pretrain/assets/66102178/0db281de-9c7c-4292-8e52-2fe125fb4afa)
![MoCov2](https://github.com/youngtboy/Awesome-Self-Supervised-Vis…
-
I would like to know, to compare result of SimCLR with result of supervised (baseline), I must initialised weights of ResNet50 with None or Imagenet
`base = tf.keras.applications.ResNet50(weights=No…
-
I am interested in downloading a pre-trained simCLR model with the projection head, to retrieve the latent features z, upon which the contrastive loss was applied.
Is this layer + pre-trained weight…
-
For the SSL pretrained and supervised trained ImageNet models, what was the sk_ratio? Also, was the se_ratio = 0? What does the se_ratio represent? Thanks.
-
Hi, your work is so interesting! However, I'm not very clear about the inverse transformations in your paper and codes. I noticed that in your ICCV paper, Eq.3, the dim of 'zi' is mx2, may I ask why t…
-
Hi what is the use of no_aug_train_10p set ?
When i tried to refine with imagenet, it eexpected a no_aug train set. What is its purpose?
-
Dear Ting,
Thanks for adding the TF2 SavedModels. Looks like the `trainable_variables_list' for the SavedModels are empty, and I cannot finetune the full network from these SavedModels. Is there…
-
1、In the configuration file gta5_stm.yml mean: [0.0, 0.0, 0.0] std: [1., 1., 1.]. In gta5_maps_pla.yml and gta5_st.yml, it is mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225] Why are the normal…
-
This question was asked in #85, and the answer at the time was "wait for the TF2 implementation". Are there any guidelines on how one might run finetuning on multiple GPUs with the TF2 implementation?…