openvinotoolkit/stable-diffusion-webui
1
1
Fork 0
mirror of https://github.com/openvinotoolkit/stable-diffusion-webui.git synced 2024-12-14 22:53:25 +03:00
Code Issues Packages Projects Releases Wiki Activity

Add hypernetwork support to split cross attention v1

Browse Source 
* Add hypernetwork support to split_cross_attention_forward_v1
* Fix device check in esrgan_model.py to use devices.device_esrgan instead of shared.device
This commit is contained in:
brkirch 2022-10-08 01:47:02 -04:00 committed by AUTOMATIC1111
parent a958f9b3fd
commit f2055cb1d4
2 changed files with 15 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
modules/esrgan_model.py
View File

@ -111,7 +111,7 @@ class UpscalerESRGAN(Upscaler):
print("Unable to load %s from %s" % (self.model_path, filename)) print("Unable to load %s from %s" % (self.model_path, filename))
return None return None
pretrained_net = torch.load(filename, map_location='cpu' if shared.device.type == 'mps' else None) pretrained_net = torch.load(filename, map_location='cpu' if devices.device_esrgan.type == 'mps' else None)
crt_model = arch.RRDBNet(3, 3, 64, 23, gc=32) crt_model = arch.RRDBNet(3, 3, 64, 23, gc=32)
pretrained_net = fix_model_layers(crt_model, pretrained_net) pretrained_net = fix_model_layers(crt_model, pretrained_net)

18
modules/sd_hijack_optimizations.py
View File

@ -12,13 +12,22 @@ from modules import shared
def split_cross_attention_forward_v1(self, x, context=None, mask=None): def split_cross_attention_forward_v1(self, x, context=None, mask=None):
h = self.heads h = self.heads
q = self.to_q(x) q_in = self.to_q(x)
context = default(context, x) context = default(context, x)
k = self.to_k(context)
v = self.to_v(context) hypernetwork = shared.selected_hypernetwork()
hypernetwork_layers = (hypernetwork.layers if hypernetwork is not None else {}).get(context.shape[2], None)
if hypernetwork_layers is not None:
k_in = self.to_k(hypernetwork_layers[0](context))
v_in = self.to_v(hypernetwork_layers[1](context))
else:
k_in = self.to_k(context)
v_in = self.to_v(context)
del context, x del context, x
q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v)) q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q_in, k_in, v_in))
del q_in, k_in, v_in
r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device) r1 = torch.zeros(q.shape[0], q.shape[1], v.shape[2], device=q.device)
for i in range(0, q.shape[0], 2): for i in range(0, q.shape[0], 2):
@ -31,6 +40,7 @@ def split_cross_attention_forward_v1(self, x, context=None, mask=None):
r1[i:end] = einsum('b i j, b j d -> b i d', s2, v[i:end]) r1[i:end] = einsum('b i j, b j d -> b i d', s2, v[i:end])
del s2 del s2
del q, k, v
r2 = rearrange(r1, '(b h) n d -> b n (h d)', h=h) r2 = rearrange(r1, '(b h) n d -> b n (h d)', h=h)
del r1 del r1