ZhuoYulang / IF-MMIN

MIT License
25 stars 3 forks source link

Questions about IFIM module encoder inputs #2

Open liulian233 opened 9 months ago

liulian233 commented 9 months ago

Hello, you wrote in your article that the input of the next encoder in the IFIM module is the output of the previous encoder, but in the code you provided, in the ResidualAE module it says x_in = x x_out = shared latents = [] for i in range(self.nblocks): encoder = getattr(self, 'encoder' + str(i)) decoder = getattr(self, 'decoder_' + str(i)) x_in = x_in + shared # x_out latent = encoder(x_in) x_out = decoder(latent) Should these two lines "x_in = x" "x_in = x_in + shared" be "x_out = x" "x_in = x_out + shared"?

ZhuoYulang commented 9 months ago

To whom it may concern: Thank you for your question. It's my pleasure that you read my code carefully. I agree with you that "x_in = x_out + shared". Actually,this is my question that upload a incorrect file to github and cause your confusion. I will update this file and send this to you. But "x_out = x" is not right. "x_out" is the output of every AE layers. It should be from decoder of AEs. So in my code, "x_out = decoder(latent)". And you can see it at the new file.

------------------ 原始邮件 ------------------ 发件人: "ZhuoYulang/IF-MMIN" @.>; 发送时间: 2023年12月6日(星期三) 下午2:38 @.>; @.***>; 主题: [ZhuoYulang/IF-MMIN] Questions about IFIM module encoder inputs (Issue #2)

Hello, you wrote in your article that the input of the next encoder in the IFIM module is the output of the previous encoder, but in the code you provided, in the ResidualAE module it says x_in = x x_out = shared latents = [] for i in range(self.nblocks): encoder = getattr(self, 'encoder' + str(i)) decoder = getattr(self, 'decoder_' + str(i)) x_in = x_in + shared # x_out latent = encoder(x_in) x_out = decoder(latent) Should these two lines "x_in = x" "x_in = x_in + shared" be "x_out = x" "x_in = x_out + shared"?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>

liulian233 commented 9 months ago

Thank you very much for the code!