Load the model using Hugging Face's transformers pipeline

app.py

@@ -1,13 +1,15 @@
 import gradio as gr
+from transformers import pipeline
 
-# Load the model outside the main function to ensure it is only loaded once
-model = gr.Interface.load("microsoft/Phi-3-mini-128k-instruct")
+def main():
+    # Load the model using Hugging Face's transformers pipeline
+    model = pipeline("text-generation", model="microsoft/Phi-3-mini-128k-instruct")
 
-def chat_with_model(prompt):
-    # Directly use the model's prediction function
-    return model(prompt)
+    def generate_text(input_text):
+        # Generate text using the model
+        result = model(input_text, max_length=50)
+        return result[0]['generated_text']
 
-def main():
     with gr.Blocks() as blocks:
         with gr.Tab("Model Info"):
             gr.Markdown("""
@@ -22,11 +24,10 @@ def main():
                 The information provided should not be seen as a modification of the licensing under which the model is released.
                 [More information](https://huggingface.co/microsoft/Phi-3-mini-128k-instruct)
             """)
-        with gr.Tab("Chat"):
-            with gr.Row():
-                input_text = gr.Textbox(placeholder="Type here to talk to the model")
-                output_text = gr.Textbox(label="Model Response")
-            input_text.change(fn=chat_with_model, inputs=input_text, outputs=output_text)
+        with gr.Tab("Generate Text"):
+            input_text = gr.Textbox(placeholder="Type here to generate text")
+            output_text = gr.Textbox(label="Generated Text")
+            input_text.submit(fn=generate_text, inputs=input_text, outputs=output_text)
 
         blocks.launch()