commit

app.py

@@ -38,7 +38,10 @@ from refiners.foundationals.latent_diffusion.stable_diffusion_1.multi_upscaler i
     MultiUpscaler,
     UpscalerCheckpoints,
 )
+from datetime import datetime
 
+def log(msg):
+    print(f'{datetime.now().time()} {msg}')
 
 Tile = tuple[int, int, Image.Image]
 Tiles = list[tuple[int, int, list[Tile]]]
@@ -500,7 +503,7 @@ function custom(){
 # torch pipes
 
 image_pipe = FluxPipeline.from_pretrained(base, torch_dtype=torch.bfloat16).to(device)
-image_pipe.enable_model_cpu_offload()
+#image_pipe.enable_model_cpu_offload()
 
 # functionality
 
@@ -509,7 +512,7 @@ def upscaler(
     input_image: Image.Image,
     prompt: str = "masterpiece, best quality, highres",
     negative_prompt: str = "worst quality, low quality, normal quality",
-    seed: int = 42,
+    seed: int = int(str(random.random()).split(".")[1]),
     upscale_factor: int = 8,
     controlnet_scale: float = 0.6,
     controlnet_decay: float = 1.0,
@@ -520,10 +523,15 @@ def upscaler(
     num_inference_steps: int = 18,
     solver: str = "DDIM",
 ) -> Image.Image:
+
+    log(f'CALL upscaler')
+
     manual_seed(seed)
 
     solver_type: type[Solver] = getattr(solvers, solver)
 
+    log(f'DBG upscaler 1')
+
     enhanced_image = enhancer.upscale(
         image=input_image,
         prompt=prompt,
@@ -539,6 +547,8 @@ def upscaler(
         solver_type=solver_type,
     )
 
+    log(f'RET upscaler')
+
     return enhanced_image
 
 @spaces.GPU(duration=180)
@@ -547,12 +557,15 @@ def summarize_text(
     pegasus_tokenizer = PegasusTokenizerFast.from_pretrained("google/pegasus-xsum"),
     pegasus_model = PegasusForConditionalGeneration.from_pretrained("google/pegasus-xsum")
 ):
-    return pegasus_tokenizer.decode( pegasus_model.generate(
+    log(f'CALL summarize_text')
+    summary = pegasus_tokenizer.decode( pegasus_model.generate(
         pegasus_tokenizer(text,return_tensors="pt").input_ids,
         max_length=max_length,
         num_beams=num_beams,
         early_stopping=early_stopping
     )[0], skip_special_tokens=True)
+    log(f'RET summarize_text with summary as {summary}')
+    return summary
 
 def generate_random_string(length):
     characters = str(ascii_letters + digits)
@@ -560,7 +573,8 @@ def generate_random_string(length):
 
 @spaces.GPU(duration=180)
 def pipe_generate(p1,p2):
-    return image_pipe(
+    log(f'CALL pipe_generate')
+    img = image_pipe(
             prompt=p1,
             negative_prompt=p2,
             height=height,
@@ -571,9 +585,13 @@ def pipe_generate(p1,p2):
             max_sequence_length=seq,
             generator=torch.Generator(device).manual_seed(int(str(random.random()).split(".")[1]))
     ).images[0]
+    log(f'RET pipe_generate')
+    return img
 
 def handle_generate(artist,song,genre,lyrics):
 
+    log(f'CALL handle_generate')
+
     pos_artist = re.sub("([ \t\n]){1,}", " ", artist).strip()
     pos_song = re.sub("([ \t\n]){1,}", " ", song).strip()
     pos_song = ' '.join(word[0].upper() + word[1:] for word in pos_song.split())