wangbo/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-04-11 19:12:34 +08:00
Code Issues Actions 10 Packages Projects Releases Wiki Activity

fix: prevent autocast crash with fp8 weights in WAN model addcmul ops

Browse Source 
PyTorch 2.8's autocast `prioritize` function doesn't handle FP8
ScalarTypes, causing sporadic "Unexpected floating ScalarType in
at::autocast::prioritize" RuntimeError when fp8_e4m3fn_fast weights
are used with the WAN 2.1 model. This wraps all torch.addcmul calls
in the WAN attention blocks with autocast-disabled context when
autocast is active, matching the existing pattern used in
sub_quadratic_attention.py and other ComfyUI modules.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Deep Mehta 2026-04-09 17:31:08 -07:00
parent 359559c913
commit 8e4bc0edac
1 changed files with 18 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
comfy/ldm/wan/model.py
View File

@ -174,6 +174,15 @@ def repeat_e(e, x):
return torch.repeat_interleave(e, repeats + 1, dim=1)[:, :x.size(1)] return torch.repeat_interleave(e, repeats + 1, dim=1)[:, :x.size(1)]
def _addcmul(x, y, z):
"""torch.addcmul wrapper that disables autocast to avoid
'Unexpected floating ScalarType in at::autocast::prioritize' with fp8 weights."""
if torch.is_autocast_enabled():
with torch.autocast(device_type=x.device.type, enabled=False):
return torch.addcmul(x, y, z)
return torch.addcmul(x, y, z)
class WanAttentionBlock(nn.Module): class WanAttentionBlock(nn.Module):
def __init__(self, def __init__(self,
@ -242,10 +251,10 @@ class WanAttentionBlock(nn.Module):
# self-attention # self-attention
x = x.contiguous() # otherwise implicit in LayerNorm x = x.contiguous() # otherwise implicit in LayerNorm
y = self.self_attn( y = self.self_attn(
torch.addcmul(repeat_e(e[0], x), self.norm1(x), 1 + repeat_e(e[1], x)), _addcmul(repeat_e(e[0], x), self.norm1(x), 1 + repeat_e(e[1], x)),
freqs, transformer_options=transformer_options) freqs, transformer_options=transformer_options)
x = torch.addcmul(x, y, repeat_e(e[2], x)) x = _addcmul(x, y, repeat_e(e[2], x))
del y del y
# cross-attention & ffn # cross-attention & ffn
@ -255,8 +264,8 @@ class WanAttentionBlock(nn.Module):
for p in patches["attn2_patch"]: for p in patches["attn2_patch"]:
x = p({"x": x, "transformer_options": transformer_options}) x = p({"x": x, "transformer_options": transformer_options})
y = self.ffn(torch.addcmul(repeat_e(e[3], x), self.norm2(x), 1 + repeat_e(e[4], x))) y = self.ffn(_addcmul(repeat_e(e[3], x), self.norm2(x), 1 + repeat_e(e[4], x)))
x = torch.addcmul(x, y, repeat_e(e[5], x)) x = _addcmul(x, y, repeat_e(e[5], x))
return x return x
@ -371,7 +380,7 @@ class Head(nn.Module):
else: else:
e = (comfy.model_management.cast_to(self.modulation, dtype=x.dtype, device=x.device).unsqueeze(0) + e.unsqueeze(2)).unbind(2) e = (comfy.model_management.cast_to(self.modulation, dtype=x.dtype, device=x.device).unsqueeze(0) + e.unsqueeze(2)).unbind(2)
x = (self.head(torch.addcmul(repeat_e(e[0], x), self.norm(x), 1 + repeat_e(e[1], x)))) x = (self.head(_addcmul(repeat_e(e[0], x), self.norm(x), 1 + repeat_e(e[1], x))))
return x return x
@ -1453,17 +1462,17 @@ class WanAttentionBlockAudio(WanAttentionBlock):
# self-attention # self-attention
y = self.self_attn( y = self.self_attn(
torch.addcmul(repeat_e(e[0], x), self.norm1(x), 1 + repeat_e(e[1], x)), _addcmul(repeat_e(e[0], x), self.norm1(x), 1 + repeat_e(e[1], x)),
freqs, transformer_options=transformer_options) freqs, transformer_options=transformer_options)
x = torch.addcmul(x, y, repeat_e(e[2], x)) x = _addcmul(x, y, repeat_e(e[2], x))
# cross-attention & ffn # cross-attention & ffn
x = x + self.cross_attn(self.norm3(x), context, context_img_len=context_img_len, transformer_options=transformer_options) x = x + self.cross_attn(self.norm3(x), context, context_img_len=context_img_len, transformer_options=transformer_options)
if audio is not None: if audio is not None:
x = self.audio_cross_attn_wrapper(x, audio, transformer_options=transformer_options) x = self.audio_cross_attn_wrapper(x, audio, transformer_options=transformer_options)
y = self.ffn(torch.addcmul(repeat_e(e[3], x), self.norm2(x), 1 + repeat_e(e[4], x))) y = self.ffn(_addcmul(repeat_e(e[3], x), self.norm2(x), 1 + repeat_e(e[4], x)))
x = torch.addcmul(x, y, repeat_e(e[5], x)) x = _addcmul(x, y, repeat_e(e[5], x))
return x return x
class DummyAdapterLayer(nn.Module): class DummyAdapterLayer(nn.Module):