Use MPS device (via Pytorch for Apple silicon) when available instead of cpu

app.py

@@ -30,8 +30,14 @@ def fixed_get_imports(filename: str | os.PathLike) -> list[str]:
 @spaces.GPU
 def get_device_type():
     import torch
-    return "cuda" if torch.cuda.is_available() else "cpu"
-
+    if torch.cuda.is_available():  
+        return "cuda"
+    else: 
+        if (torch.backends.mps.is_available() and torch.backends.mps.is_built()):
+            return "mps"
+        else:
+            return "cpu"
+        
 model_id = 'microsoft/Florence-2-base-ft'
 
 import subprocess
@@ -40,11 +46,13 @@ if (device == "cuda"):
     subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
     model = AutoModelForCausalLM.from_pretrained("microsoft/Florence-2-base-ft", trust_remote_code=True)
     processor = AutoProcessor.from_pretrained("microsoft/Florence-2-base-ft", trust_remote_code=True)
+    model.to(device)
 else:
     #https://huggingface.co/microsoft/Florence-2-base-ft/discussions/4
     with patch("transformers.dynamic_module_utils.get_imports", fixed_get_imports):
         model = AutoModelForCausalLM.from_pretrained("microsoft/Florence-2-base-ft", trust_remote_code=True)
         processor = AutoProcessor.from_pretrained("microsoft/Florence-2-base-ft", trust_remote_code=True)
+        model.to(device)
 
 DESCRIPTION = "# [Florence-2 base-ft Demo with CPU inference support](https://huggingface.co/microsoft/Florence-2-base-ft)"