You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jul 24, 2024. It is now read-only.
I am a beginner who is just studying SSL.
As far as I know, there are two training sessions in SSL.
1)One is the training the encoder in pretrain session, 2)the other is layer training in linear evaluation.
In the figure 9 of this paper, there is a result with different batch size and epochs.
In which training did you record the batch size and epoch?
If it is the latter, what is the pretrained model used? ResNet-50 trained with ImageNet 1000epochs?
The text was updated successfully, but these errors were encountered:
On Mon, Sep 12, 2022 at 7:28 PM Joonsun Auh ***@***.***> wrote:
I am a beginner who is just studying SSL.
As far as I know, there are two training sessions in SSL.
1)One is the training the encoder in pretrain session, 2)the other is
layer training in linear evaluation.
In the figure 9 of this paper, there is a result with different batch size
and epochs.
In which training did you record the batch size and epoch?
If it is the latter, what is the pretrained model used? ResNet-50 trained
with ImageNet 1000epochs?
—
Reply to this email directly, view it on GitHub
<#206>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAKERUJM5YJK6I4S6UGJDWTV57RENANCNFSM6AAAAAAQK773BI>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
Then how many times last layer has been trained in step linear evaluation?
When I operate the linear evaluation, last layer always has a training step.
I can't find the number of epochs in linear evaluation.
Thank you.
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
I am a beginner who is just studying SSL.
As far as I know, there are two training sessions in SSL.
1)One is the training the encoder in pretrain session, 2)the other is layer training in linear evaluation.
In the figure 9 of this paper, there is a result with different batch size and epochs.
In which training did you record the batch size and epoch?
If it is the latter, what is the pretrained model used? ResNet-50 trained with ImageNet 1000epochs?
The text was updated successfully, but these errors were encountered: