Is this docstring of `encoder_hidden_states` argument in `Transformer2DModel` correct?

[offchan42]

diffusers/src/diffusers/models/transformer_2d.py

Lines 228 to 230 in e607a58

	encoder_hidden_states ( `torch.LongTensor` of shape `(batch size, encoder_hidden_states dim)`, *optional*):
	Conditional embeddings for cross attention layer. If not given, cross-attention defaults to
	self-attention.

The shape is (batch_size, seq_length, embedding_dim), isn't it? Also it's supposed to be float tensor. Or maybe it supports both long and float?

[patrickvonplaten]

You're right it should be a float tensor and the shape should be (batch_size, seq_length, embedding_dim) - mind opening a PR here? :-)

[offchan42]

Sure. Done it.