Skip to content

Open-Sora: Democratizing Efficient Video Production for All

Python

config

Shen-Chenhui0da15473d06a
diff --git a/configs/vae_magvit_v2/train/pipeline_16x128x128.py b/configs/vae_magvit_v2/train/pipeline_16x128x128.py
index 2d4f1a9..46ec996 100644
--- a/configs/vae_magvit_v2/train/pipeline_16x128x128.py
+++ b/configs/vae_magvit_v2/train/pipeline_16x128x128.py
@@ -42,11 +42,11 @@ model = dict(
discriminator = dict(
type="DISCRIMINATOR_3D",
- image_size = image_size,
+ image_size = (16, 16), # NOTE: here image size is different
num_frames = num_frames,
in_channels = 4,
filters = 128,
- use_pretrained=False, # NOTE: set to False only if we want to fresh train using a different discriminator!
+ use_pretrained=True, # NOTE: set to False only if we want to fresh train using a different discriminator!
# channel_multipliers = (2,4,4,4,4), # (2,4,4,4) for 64x64 resolution
channel_multipliers= (2,4,4) # since on intermediate layer, 16 x 16 x 16 dimension z
)