Skip to content

Open-Sora: Democratizing Efficient Video Production for All

Python

debug

Shen-Chenhui30353e1351ff
diff --git a/opensora/models/vae/vae_3d_v2.py b/opensora/models/vae/vae_3d_v2.py
index 36319f7..3e176ce 100644
--- a/opensora/models/vae/vae_3d_v2.py
+++ b/opensora/models/vae/vae_3d_v2.py
@@ -814,6 +814,7 @@ class VAE_3D_V2(nn.Module):
kl_loss = torch.mean(kl_loss) / kl_loss.shape[0]
# TODO: DOUBLE add more sophisticated discrminator loss
+ gen_loss = self.zero
if self.adversarial_loss_weight is not None and self.adversarial_loss_weight > 0:
if video_contains_first_frame:
# video_len = video.shape[2]
@@ -829,6 +830,7 @@ class VAE_3D_V2(nn.Module):
# perceptual loss
# SCH: NOTE: if mse can pick single frame, if use sum of errors, need to calc for all frames!
+ perceptual_loss = self.zero
if self.perceptual_loss_weight is not None and self.perceptual_loss_weight > 0:
frame_indices = torch.randn((batch, frames)).topk(1, dim = -1).indices
@@ -847,8 +849,7 @@ class VAE_3D_V2(nn.Module):
recon_vgg_feats = self.vgg(recon_vgg_input)
perceptual_loss = F.mse_loss(input_vgg_feats, recon_vgg_feats)
# perceptual_loss = self.lpips(input_vgg_input.contiguous(), recon_vgg_input.contiguous())
- else:
- perceptual_loss = self.zero
+
total_loss = recon_loss \