Update app.py

app.py

@@ -8,6 +8,10 @@ import random
 import numpy as np
 import os
 import spaces
+import huggingface_hub
+import time
+
+huggingface_hub.constants.HF_HUB_DOWNLOAD_TIMEOUT = 60
 
 try:
     import basicsr
@@ -30,16 +34,24 @@ pipe = FluxImg2ImgPipeline.from_pretrained(
 ).to(device)
 pipe.enable_vae_tiling()  # To help with memory for large images
 
-# Initialize Florence model with float32 to avoid dtype mismatch
-florence_model = AutoModelForCausalLM.from_pretrained(
-    'microsoft/Florence-2-large', 
-    trust_remote_code=True,
-    torch_dtype=torch.float32
-).to(device).eval()
-florence_processor = AutoProcessor.from_pretrained(
-    'microsoft/Florence-2-large', 
-    trust_remote_code=True
-)
+# Initialize Florence model with float32 to avoid dtype mismatch, with retry
+for attempt in range(5):
+    try:
+        florence_model = AutoModelForCausalLM.from_pretrained(
+            'microsoft/Florence-2-large', 
+            trust_remote_code=True,
+            torch_dtype=torch.float32
+        ).to(device).eval()
+        florence_processor = AutoProcessor.from_pretrained(
+            'microsoft/Florence-2-large', 
+            trust_remote_code=True
+        )
+        break
+    except Exception as e:
+        print(f"Attempt {attempt+1} to load Florence-2 failed: {e}")
+        time.sleep(10)
+else:
+    raise RuntimeError("Failed to load Florence-2 after multiple attempts")
 
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 2048
@@ -113,7 +125,8 @@ def tiled_flux_img2img(image, prompt, strength, num_inference_steps, guidance_sc
                     mask = Image.new('L', (tile_right - tile_left, effective_overlap))
                     for i in range(mask.width):
                         for j in range(mask.height):
-                            mask.putpixel((i, j), int(255 * (j / effective_overlap)))
+                            divisor = effective_overlap - 1 if effective_overlap > 1 else 1
+                            mask.putpixel((i, j), int(255 * (j / divisor)))
                     blend_region = Image.composite(
                         generated_tile.crop((0, 0, mask.width, mask.height)),
                         result.crop((tile_left, tile_top, tile_right, tile_top + mask.height)),
@@ -131,7 +144,8 @@ def tiled_flux_img2img(image, prompt, strength, num_inference_steps, guidance_sc
                     mask_h = Image.new('L', (effective_overlap_h, tile_bottom - tile_top))
                     for i in range(mask_h.width):
                         for j in range(mask_h.height):
-                            mask_h.putpixel((i, j), int(255 * (i / effective_overlap_h)))
+                            divisor_h = effective_overlap_h - 1 if effective_overlap_h > 1 else 1
+                            mask_h.putpixel((i, j), int(255 * (i / divisor_h)))
                     blend_region_h = Image.composite(
                         generated_tile.crop((0, 0, mask_h.width, mask_h.height)),
                         result.crop((tile_left, tile_top, tile_left + mask_h.width, tile_bottom)),