search

yoctta / multiple-attention

The code of multi-attention deepfake detection
241 stars 54 forks source link

About the tricks during the training #17

Closed 2686873626 closed 2 years ago

2686873626 commented 2 years ago

After the pretraining of backbone, I froze it and then train the multi-attentional without AGDA loss. But the accuracy is still no more than 97%. Is there any other trick used during the training of attention layer.

yoctta commented 2 years ago

Then you should finetune the whole network with AGDA for some epochs.

From: 2686873626 Sent: Tuesday, February 22, 2022 12:11 PM To: yoctta/multiple-attention Cc: Subscribed Subject: [yoctta/multiple-attention] About the tricks during the training (Issue #17)

After the pretraining of backbone, I froze it and then train the multi-attentional without AGDA loss. But the accuracy is still no more than 97%. Is there any other trick used during the training of attention layer. — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: @.***>

2686873626 commented 2 years ago

Then you should finetune the whole network with AGDA for some epochs. From: 2686873626 Sent: Tuesday, February 22, 2022 12:11 PM To: yoctta/multiple-attention Cc: Subscribed Subject: [yoctta/multiple-attention] About the tricks during the training (Issue #17) After the pretraining of backbone, I froze it and then train the multi-attentional without AGDA loss. But the accuracy is still no more than 97%. Is there any other trick used during the training of attention layer. — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: @.***>

Thank you very much

damengdameng commented 2 years ago

Then you should finetune the whole network with AGDA for some epochs. From: 2686873626 Sent: Tuesday, February 22, 2022 12:11 PM To: yoctta/multiple-attention Cc: Subscribed Subject: [yoctta/multiple-attention] About the tricks during the training (Issue #17) After the pretraining of backbone, I froze it and then train the multi-attentional without AGDA loss. But the accuracy is still no more than 97%. Is there any other trick used during the training of attention layer. — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: @.***>

Thank you very much

hi, can you share the environment details? Thank you!

2686873626 commented 2 years ago

Then you should finetune the whole network with AGDA for some epochs. From: 2686873626 Sent: Tuesday, February 22, 2022 12:11 PM To: yoctta/multiple-attention Cc: Subscribed Subject: [yoctta/multiple-attention] About the tricks during the training (Issue #17) After the pretraining of backbone, I froze it and then train the multi-attentional without AGDA loss. But the accuracy is still no more than 97%. Is there any other trick used during the training of attention layer. — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: @.***>

Thank you very much

hi, can you share the environment details? Thank you!

I've already sent it to your email

xuyingzhongguo commented 1 year ago

Then you should finetune the whole network with AGDA for some epochs. From: 2686873626 Sent: Tuesday, February 22, 2022 12:11 PM To: yoctta/multiple-attention Cc: Subscribed Subject: [yoctta/multiple-attention] About the tricks during the training (Issue #17) After the pretraining of backbone, I froze it and then train the multi-attentional without AGDA loss. But the accuracy is still no more than 97%. Is there any other trick used during the training of attention layer. — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: @.***>

Thank you very much

hi, can you share the environment details? Thank you!

I've already sent it to your email

Hi, do you mind sharing it with me please? Thank you!