Closed rafaelorozco closed 1 year ago
Patch coverage: 96.00
% and project coverage change: +0.05
:tada:
Comparison is base (
671e9c1
) 88.47% compared to head (7fb07d9
) 88.52%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
How about something like this previous commit @mloubout ? you make the summarized net like this:
sum_net = ResNet(n_cond, 16, 3; norm=nothing) # make sure it doesnt have any weird normalizati8ons
# Network and input
flow = NetworkConditionalGlow(n_in, n_cond, n_hidden, L, K; ndims=length(N))
G = SummarizedNet(flow, sum_net)
And the main operations are the same as before
Y, ZCond = G.forward(X,Cond)
X_ = G.inverse(Y,ZCond)
G.backward(Y, Y, ZCond; Y_save = Cond)
First try at adding a summary network to conditional glow.
I could use some extra eyes on how to implement this cleanly. Although it gets the job done and works with arbitrary flux layers, it feel a little bit clunky.