n0p/mlx-video
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

- Refactor video generation script

Browse Source 
- Introduced argparse for parameter handling, streamlined model loading, and enhanced denoising functions.
- Updated VAE weight sanitization for compatibility and improved activation function handling in text projection.
- Added support for saving individual frames and refined output video generation process.
This commit is contained in:
Prince Canuma
2026-01-12 14:04:53 +01:00
parent d1ca36a315
commit 7114b023bd
6 changed files with 270 additions and 304 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
mlx_video/models/ltx/text_projection.py
View File

@@ -16,7 +16,7 @@ class PixArtAlphaTextProjection(nn.Module):
out_features = out_features or hidden_size
self.linear1 = nn.Linear(in_features, hidden_size, bias=bias)
self.act = nn.GELU(approx="precise")
self.act = nn.GELU(approx="tanh") # Must match PyTorch's approximate="tanh"
self.linear2 = nn.Linear(hidden_size, out_features, bias=bias)
def __call__(self, x: mx.array) -> mx.array: