Open ErwannMillon opened 2 months ago
for the batch option you need to add
def __init__(self):
self.unfold_batch = True
but for the attention masking I need to do some testing with animations
lol oops typo, tested with the unfold_batch = True but ctrl-z'ed before commit.
was working for weight scheduling + regional conditioning, did not test attn masking
hm, doesn't seem to work with regional conditioning when trying to use weight scheduling on one ipa while the other has constant weight. Will check it out
This PR creates a new node,
IPAdapterFromParamsBatch
, which inherits from IPAdapterBatch.It was previously impossible to use weight schedules with the
IPAdapterFromParams
node asunfold_batch
was not enabled.