Justin-Tan
commented
5 years ago Hey,
It’s from section 4.1 of the paper - concatenation of the representation with noise drawn from a fixed prior to form some latent vector Z.
Cheers, Justin
On Tue, 30 Jul 2019 at 8:13 pm, shellhuang1227 notifications@github.com wrote:
Hello, while reading the paper, it proposed "When encoding the channels of w^ to a bit-stream, we use an arithmetic encoder where frequencies are stored for each channel separately and then encode them in a static (non-adaptive) manner (i.e. context model and adaptive arithmetic encoding)." But in your code, the quantized value w^ is concatenated with noise v to form the latent vector z, and then the generator G tries to generate an image x^ = G(z). I did not find the step: encoding the channels of w^ to a bit-stream in context model and adaptive arithmetic encoding. Did you not conside this step？ Thank you very much.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Justin-Tan/generative-compression/issues/32?email_source=notifications&email_token=AGRNY6H66G66JDIPUK3EO23QCAH3LA5CNFSM4IH2WNI2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HCIIAIA, or mute the thread https://github.com/notifications/unsubscribe-auth/AGRNY6C6R2XJEIHD3PHGUCLQCAH3LANCNFSM4IH2WNIQ .
shellhuang1227
Hello, while reading the paper, it proposed "When encoding the channels of w^ to a bit-stream, we use an arithmetic encoder where frequencies are stored for each channel separately and then encode them in a static (non-adaptive) manner (i.e. context model and adaptive arithmetic encoding)." But in your code, the quantized value w^ is concatenated with noise v to form the latent vector z, and then the generator G tries to generate an image x^ = G(z). I did not find the step: encoding the channels of w^ to a bit-stream in context model and adaptive arithmetic encoding. Did you not conside this step？ Thank you very much.