Update app.py

app.py

@@ -4,20 +4,31 @@ import torch
 from PIL import Image
 from transformers import AutoProcessor, AutoModelForCausalLM
 
-# Attempt to install flash-attn
+# Upgrade transformers to the latest version
+try:
+    subprocess.run('pip install --upgrade transformers', check=True, shell=True)
+    print("Successfully upgraded transformers.")
+except subprocess.CalledProcessError as e:
+    print(f"Error upgrading transformers: {e}")
+    print("Continuing with the current version, but this may cause issues.")
+
+# Attempt to install flash-attn (optional, for performance)
 try:
     subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, check=True, shell=True)
+    print("Successfully installed flash-attn.")
 except subprocess.CalledProcessError as e:
     print(f"Error installing flash-attn: {e}")
     print("Continuing without flash-attn.")
 
 # Determine the device to use
 device = "cuda" if torch.cuda.is_available() else "cpu"
+print(f"Using device: {device}")
 
 # Load the base model and processor
 try:
     vision_language_model_base = AutoModelForCausalLM.from_pretrained('microsoft/Florence-2-base', trust_remote_code=True).to(device).eval()
     vision_language_processor_base = AutoProcessor.from_pretrained('microsoft/Florence-2-base', trust_remote_code=True)
+    print("Base model and processor loaded successfully.")
 except Exception as e:
     print(f"Error loading base model: {e}")
     vision_language_model_base = None
@@ -27,6 +38,7 @@ except Exception as e:
 try:
     vision_language_model_large = AutoModelForCausalLM.from_pretrained('microsoft/Florence-2-large', trust_remote_code=True).to(device).eval()
     vision_language_processor_large = AutoProcessor.from_pretrained('microsoft/Florence-2-large', trust_remote_code=True)
+    print("Large model and processor loaded successfully.")
 except Exception as e:
     print(f"Error loading large model: {e}")
     vision_language_model_large = None