Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

cleanup logs

Browse Source
This commit is contained in:
Damian at mba 2022-10-24 11:16:43 +02:00
parent 63902f3d34
commit 0564397ee6
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
ldm/models/diffusion/shared_invokeai_diffusion.py
View File

@ -82,18 +82,18 @@ class InvokeAIDiffuserComponent:
# flip because sigmas[0] is for the fully denoised image # flip because sigmas[0] is for the fully denoised image
# percent_through must be <1 # percent_through must be <1
percent_through = 1.0 - float(sigma_index.item() + 1) / float(self.model.sigmas.shape[0]) percent_through = 1.0 - float(sigma_index.item() + 1) / float(self.model.sigmas.shape[0])
print('estimated percent_through', percent_through, 'from sigma', sigma.item()) #print('estimated percent_through', percent_through, 'from sigma', sigma.item())
cross_attention_control_types_to_do = CrossAttentionControl.get_active_cross_attention_control_types_for_step(self.cross_attention_control_context, percent_through) cross_attention_control_types_to_do = CrossAttentionControl.get_active_cross_attention_control_types_for_step(self.cross_attention_control_context, percent_through)
if len(cross_attention_control_types_to_do)==0: if len(cross_attention_control_types_to_do)==0:
print('not doing cross attention control') #print('not doing cross attention control')
# faster batched path # faster batched path
x_twice = torch.cat([x]*2) x_twice = torch.cat([x]*2)
sigma_twice = torch.cat([sigma]*2) sigma_twice = torch.cat([sigma]*2)
both_conditionings = torch.cat([unconditioning, conditioning]) both_conditionings = torch.cat([unconditioning, conditioning])
unconditioned_next_x, conditioned_next_x = self.model_forward_callback(x_twice, sigma_twice, both_conditionings).chunk(2) unconditioned_next_x, conditioned_next_x = self.model_forward_callback(x_twice, sigma_twice, both_conditionings).chunk(2)
else: else:
print('pct', percent_through, ': doing cross attention control on', cross_attention_control_types_to_do) #print('pct', percent_through, ': doing cross attention control on', cross_attention_control_types_to_do)
# slower non-batched path (20% slower on mac MPS) # slower non-batched path (20% slower on mac MPS)
# We are only interested in using attention maps for conditioned_next_x, but batching them with generation of # We are only interested in using attention maps for conditioned_next_x, but batching them with generation of
# unconditioned_next_x causes attention maps to *also* be saved for the unconditioned_next_x. # unconditioned_next_x causes attention maps to *also* be saved for the unconditioned_next_x.