Function half

models/transformer/wan/modules/attention.py:59–60 · view source on GitHub ↗

(x)

Source from the content-addressed store, hash-verified

57	b, lq, lk, out_dtype = q.size(0), q.size(1), k.size(1), q.dtype
58
59	def half(x):
60	return x if x.dtype in half_dtypes else x.to(dtype)
61
62	# preprocess query
63	if q_lens is None:

flash_attentionFunction · 0.70

no outgoing calls

no test coverage detected