update

app.py

@@ -34,7 +34,7 @@ magma_qa_prompt = "<image>\n{} Answer the question briefly."
 magma_model_id = "microsoft/Magma-8B"
 magam_model = AutoModelForCausalLM.from_pretrained(magma_model_id, trust_remote_code=True, torch_dtype=dtype)
 magma_processor = AutoProcessor.from_pretrained(magma_model_id, trust_remote_code=True)
-magam_model.to("cuda")
+# magam_model.to("cuda")
 
 # Download the entire repository
 # snapshot_download(repo_id=repo_id, local_dir=local_dir)
@@ -92,7 +92,7 @@ def get_som_response(instruction, image_som):
     inputs = magma_processor(images=[image_som], texts=prompt, return_tensors="pt")
     inputs['pixel_values'] = inputs['pixel_values'].unsqueeze(0)
     inputs['image_sizes'] = inputs['image_sizes'].unsqueeze(0)
-    inputs = inputs.to("cuda").to(dtype)
+    inputs = inputs.to(dtype)
 
     magam_model.generation_config.pad_token_id = magma_processor.tokenizer.pad_token_id
     with torch.inference_mode():
@@ -129,7 +129,7 @@ def get_qa_response(instruction, image):
     inputs = magma_processor(images=[image], texts=prompt, return_tensors="pt")
     inputs['pixel_values'] = inputs['pixel_values'].unsqueeze(0)
     inputs['image_sizes'] = inputs['image_sizes'].unsqueeze(0)
-    inputs = inputs.to("cuda")
+    inputs = inputs.to(dtype)
 
     magam_model.generation_config.pad_token_id = magma_processor.tokenizer.pad_token_id
     with torch.inference_mode():