Update app.py

app.py

@@ -0,0 +1,67 @@
+from transformers import AutoProcessor, AutoModelForCausalLM, AutoConfig
+import gradio as gr
+import torch
+
+# # Load the processor and model
+processor = AutoProcessor.from_pretrained("microsoft/git-base")
+# config = AutoConfig.from_pretrained("./adapter_config.json")
+# # model = AutoModelForCausalLM.from_pretrained("microsoft/git-base")
+
+# model_path = "./adapter_model.safetensors"
+# model = AutoModelForCausalLM.from_pretrained(model_path)
+
+from transformers import AutoModelForCausalLM
+from peft import PeftModel
+
+#Base model on your local filesystem
+base_model_dir = "microsoft/git-base"
+base_model = AutoModelForCausalLM.from_pretrained(base_model_dir)
+
+#Adaptor directory on your local filesystem
+adaptor_dir = "./"
+merged_model = PeftModel.from_pretrained(base_model,adaptor_dir)
+
+merged_model = merged_model.merge_and_unload()
+merged_model.save_pretrained("./Merged-Model/")
+
+model = merged_model
+
+
+def predict(image):
+    try:
+        # Prepare the image using the processor
+        inputs = processor(images=image, return_tensors="pt")
+
+        # Move inputs to the appropriate device
+        device = "cuda" if torch.cuda.is_available() else "cpu"
+        inputs = {key: value.to(device) for key, value in inputs.items()}
+        model.to(device)
+
+        # Generate the caption
+        outputs = model.generate(**inputs)
+
+        # Decode the generated caption
+        caption = processor.batch_decode(outputs, skip_special_tokens=True)[0]
+
+        return caption
+
+    except Exception as e:
+        print("Error during prediction:", str(e))
+        return "Error: " + str(e)
+
+# https://www.gradio.app/guides
+with gr.Blocks() as demo:
+    image = gr.Image(type="pil")
+    predict_btn = gr.Button("Predict", variant="primary")
+    output = gr.Label(label="Generated Caption")
+
+    inputs = [image]
+    outputs = [output]
+
+    predict_btn.click(predict, inputs=inputs, outputs=outputs)
+
+if __name__ == "__main__":
+    demo.launch()  # Local machine only
+    # demo.launch(server_name="0.0.0.0")  # LAN access to local machine
+    # demo.launch(share=True)  # Public access to local machine
+