Update.

diff --git a/tiny_vae.py b/tiny_vae.py
index 577f717..d33cc4b 100755 (executable)
--- a/tiny_vae.py
+++ b/tiny_vae.py
@@ -24,9 +24,13 @@ device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
 ######################################################################
 
-parser = argparse.ArgumentParser(description="Tiny LeNet-like auto-encoder.")
+parser = argparse.ArgumentParser(
+    description="Very simple implementation of a VAE for teaching."
+)
+
+parser.add_argument("--nb_epochs", type=int, default=25)
 
-parser.add_argument("--nb_epochs", type=int, default=100)
+parser.add_argument("--learning_rate", type=float, default=1e-3)
 
 parser.add_argument("--batch_size", type=int, default=100)
 
@@ -40,6 +44,11 @@ parser.add_argument("--nb_channels", type=int, default=128)
 
 parser.add_argument("--no_dkl", action="store_true")
 
+# With that option, do not follow the setup of the original VAE paper
+# of forcing the variance of X|Z to 1 during training and to 0 for
+# sampling, but optimize and use the variance.
+parser.add_argument("--no_hacks", action="store_true")
+
 args = parser.parse_args()
 
 log_file = open(args.log_filename, "w")
@@ -135,6 +144,8 @@ class ImageGivenLatentNet(nn.Module):
     def forward(self, z):
         output = self.model(z.view(z.size(0), -1, 1, 1))
         mu, log_var = output[:, 0:1], output[:, 1:2]
+        if not args.no_hacks:
+            log_var[...] = 0
         return mu, log_var
 
 
@@ -160,7 +171,7 @@ model_p_X_given_z = ImageGivenLatentNet(
 
 optimizer = optim.Adam(
     itertools.chain(model_p_X_given_z.parameters(), model_q_Z_given_x.parameters()),
-    lr=4e-4,
+    lr=args.learning_rate,
 )
 
 model_p_X_given_z.to(device)
@@ -227,14 +238,20 @@ save_image(x, "input.png")
 mean_q_Z_given_x, log_var_q_Z_given_x = model_q_Z_given_x(x)
 z = sample_gaussian(mean_q_Z_given_x, log_var_q_Z_given_x)
 mean_p_X_given_z, log_var_p_X_given_z = model_p_X_given_z(z)
-x = sample_gaussian(mean_p_X_given_z, log_var_p_X_given_z)
+if args.no_hacks:
+    x = sample_gaussian(mean_p_X_given_z, log_var_p_X_given_z)
+else:
+    x = mean_p_X_given_z
 save_image(x, "output.png")
 
 # Generate a bunch of images
 
 z = sample_gaussian(mean_p_Z.expand(x.size(0), -1), log_var_p_Z.expand(x.size(0), -1))
 mean_p_X_given_z, log_var_p_X_given_z = model_p_X_given_z(z)
-x = sample_gaussian(mean_p_X_given_z, log_var_p_X_given_z)
+if args.no_hacks:
+    x = sample_gaussian(mean_p_X_given_z, log_var_p_X_given_z)
+else:
+    x = mean_p_X_given_z
 save_image(x, "synth.png")
 
 ######################################################################