Basic Hunyuan Video model support.
This commit is contained in:
@@ -194,6 +194,7 @@ def make_time_attn(
|
||||
attn_kwargs=None,
|
||||
alpha: float = 0,
|
||||
merge_strategy: str = "learned",
|
||||
conv_op=ops.Conv2d,
|
||||
):
|
||||
return partialclass(
|
||||
AttnVideoBlock, in_channels, alpha=alpha, merge_strategy=merge_strategy
|
||||
|
||||
Reference in New Issue
Block a user