change callback_fn

cfg_sample.py

@@ -48,6 +48,12 @@ def resize_and_center_crop(image, size):
     image = image.resize((int(fac * image.size[0]), int(fac * image.size[1])), Image.LANCZOS)
     return TF.center_crop(image, size[::-1])
 
+def callback_fn(info):
+    if info['i'] % 50==0:
+        out = info['pred'].add(1).div(2)
+        save_image(out, f"interm_output_{info['i']:05d}.png")
+        if IS_NOTEBOOK:
+            display.display(display.Image(f"interm_output_{info['i']:05d}.png",height=300))
 
 def main():
     p = argparse.ArgumentParser(description=__doc__,
@@ -132,13 +138,6 @@ def main():
 
     torch.manual_seed(args.seed)
 
-    def callback_fn(pred, i):
-        if i % 50==0 or i==args.steps:
-            out = pred.add(1).div(2)
-            save_image(out, f"interm_output_{i:05d}.png")
-            if IS_NOTEBOOK:
-                display.display(display.Image(f"interm_output_{i:05d}.png",height=300))
-
     def cfg_model_fn(x, t):
         n = x.shape[0]
         n_conds = len(target_embeds)
@@ -235,13 +234,6 @@ def run_diffusion_cfg(prompts,images=None,steps=1000,init=None,model="cc12m_1_cf
 
     torch.manual_seed(args.seed)
 
-    def callback_fn(pred, i):
-        if i % display_freq==0 or i==args.steps:
-            out = pred.add(1).div(2)
-            save_image(out, f"interm_output_{i:05d}.png")
-            if IS_NOTEBOOK:
-                display.display(display.Image(f"interm_output_{i:05d}.png",height=300))
-
     def cfg_model_fn(x, t):
         n = x.shape[0]
         n_conds = len(target_embeds)

clip_sample.py

@@ -77,6 +77,12 @@ def resize_and_center_crop(image, size):
     image = image.resize((int(fac * image.size[0]), int(fac * image.size[1])), Image.LANCZOS)
     return TF.center_crop(image, size[::-1])
 
+def callback_fn(info):
+    if info['i'] % 50==0:
+        out = info['pred'].add(1).div(2)
+        save_image(out, f"interm_output_{info['i']:05d}.png")
+        if IS_NOTEBOOK:
+            display.display(display.Image(f"interm_output_{info['i']:05d}.png",height=300))
 
 def main():
     p = argparse.ArgumentParser(description=__doc__,
@@ -176,13 +182,6 @@ def main():
 
     torch.manual_seed(args.seed)
 
-    def callback_fn(pred, i):
-        if i % 50==0 or i==args.steps:
-            out = pred.add(1).div(2)
-            save_image(out, f"interm_output_{i:05d}.png")
-            if IS_NOTEBOOK:
-                display.display(display.Image(f"interm_output_{i:05d}.png",height=300))
-
     def cond_fn(x, t, pred, clip_embed):
         clip_in = normalize(make_cutouts((pred + 1) / 2))
         image_embeds = clip_model.encode_image(clip_in).view([args.cutn, x.shape[0], -1])
@@ -295,13 +294,6 @@ def run_diffusion(prompts,images=None,steps=1000,init=None,model="yfcc_2",size=[
 
     torch.manual_seed(args.seed)
 
-    def callback_fn(pred, i):
-        if i % display_freq==0 or i==args.steps:
-            out = pred.add(1).div(2)
-            save_image(out, f"interm_output_{i:05d}.png")
-            if IS_NOTEBOOK:
-                display.display(display.Image(f"interm_output_{i:05d}.png",height=300))
-
     def cond_fn(x, t, pred, clip_embed):
         clip_in = normalize(make_cutouts((pred + 1) / 2))
         image_embeds = clip_model.encode_image(clip_in).view([args.cutn, x.shape[0], -1])

diffusion/sampling.py

@@ -44,8 +44,6 @@ def sample(model, x, steps, eta, extra_args, callback=None):
             if eta:
                 x += torch.randn_like(x) * ddim_sigma
 
-        if callback_fn:
-            callback_fn(pred,i)
 
     # If we are on the last timestep, output the denoised image
     return pred
@@ -101,8 +99,6 @@ def cond_sample(model, x, steps, eta, extra_args, cond_fn, callback=None):
             if eta:
                 x += torch.randn_like(x) * ddim_sigma
 
-        if callback_fn:
-            callback_fn(pred,i)
 
     # If we are on the last timestep, output the denoised image
     return pred