Update app.py

app.py

@@ -237,7 +237,7 @@ def compare_models(
     Args:
         prompt (str): The input prompt for text generation.
         temperature (float): Sampling temperature.
-        top_p (float): Top-p sampling probability.
+        top_p (float): Sampling top-p.
         min_new_tokens (int): Minimum number of new tokens to generate.
         max_new_tokens (int): Maximum number of new tokens to generate.
 
@@ -385,7 +385,7 @@ def chat_rag(
         user_input (str): The user's chat input.
         history (list[list[str]]): The chat history.
         temperature (float): Sampling temperature.
-        top_p (float): Top-p sampling probability.
+        top_p (float): Sampling top-p.
         min_new_tokens (int): Minimum number of new tokens to generate.
         max_new_tokens (int): Maximum number of new tokens to generate.
 
@@ -429,27 +429,37 @@ with gr.Blocks() as demo:
     gr.Markdown("# QLoRA Fine-tuning & RAG-based Chat Demo using Custom R1 Model")
     gr.Markdown("---")
 
-    gr.TabbedInterface(
+    with gr.TabbedInterface(
         [
             gr.Interface(
                 fn=finetune_small_subset,
                 inputs=None,
                 outputs=gr.Textbox(label="Fine-tuning Status", interactive=False),
                 title="⚙️ Fine-tuning (Optional)",
-                description="This section allows you to fine-tune the custom R1 model on a small subset of the ServiceNow dataset. This step is optional but can potentially improve the model's performance on ServiceNow-related tasks. **Note:** This process may take up to 5 minutes."
+                description="""
+                    ### Optional Fine-tuning
+                    This section allows you to fine-tune the custom R1 model on a small subset of the ServiceNow dataset.
+                    This step is **optional** but can potentially improve the model's performance on ServiceNow-related tasks.
+
+                    **Note:** This process may take up to 5 minutes. Click the button below to start fine-tuning.
+                """
             ),
             gr.Interface(
                 fn=predict,
                 inputs=[
                     gr.Textbox(lines=3, label="Input Prompt", placeholder="Enter your prompt here..."),
-                    gr.Slider(0.0, 1.5, step=0.1, value=0.7, label="Temperature (Creativity)"),
-                    gr.Slider(0.0, 1.0, step=0.05, value=0.9, label="Top-p (Sampling Nucleus)"),
-                    gr.Slider(1, 2500, value=50, step=10, label="Min New Tokens"),
-                    gr.Slider(1, 2500, value=200, step=50, label="Max New Tokens")
+                    gr.Slider(0.0, 1.5, step=0.1, value=0.7, label="Temperature (Creativity)", info="Adjust the randomness of the output. Higher values mean more creative but potentially less coherent text."),
+                    gr.Slider(0.0, 1.0, step=0.05, value=0.9, label="Top-p (Sampling Nucleus)", info="Controls the sampling pool. Lower values make the output more focused."),
+                    gr.Slider(1, 2500, value=50, step=10, label="Min New Tokens", info="Minimum number of tokens to generate."),
+                    gr.Slider(1, 2500, value=200, step=50, label="Max New Tokens", info="Maximum number of tokens to generate.")
                 ],
                 outputs=gr.Textbox(label="Custom R1 Output", lines=8, interactive=False),
                 title="✍️ Direct Generation",
-                description="Enter a prompt to generate text directly using the custom R1 model. This is standard text generation without retrieval augmentation."
+                description="""
+                    ### Direct Text Generation
+                    Enter a prompt to generate text directly using the custom R1 model.
+                    This is standard text generation without retrieval augmentation.
+                """
             ),
             gr.Interface(
                 fn=compare_models,
@@ -465,16 +475,24 @@ with gr.Blocks() as demo:
                     gr.Textbox(label="Official R1 Output", lines=6, interactive=False)
                 ],
                 title="🆚 Model Comparison",
-                description="Enter a prompt to compare the text generation of your fine-tuned custom R1 model with the official DeepSeek-R1-Distill-Llama-8B model."
+                description="""
+                    ### Model Output Comparison
+                    Enter a prompt to compare the text generation of your fine-tuned custom R1 model with the official DeepSeek-R1-Distill-Llama-8B model.
+                    This allows you to see the differences in output between the two models.
+                """
             ),
             gr.ChatInterface(
                 fn=chat_rag,
                 chatbot=gr.Chatbot(label="RAG Chatbot"),
                 textbox=gr.Textbox(placeholder="Ask a question to the RAG Chatbot...", lines=2, show_label=False),
                 title="💬 RAG Chat",
-                description="Chat with the custom R1 model, enhanced with retrieval-augmented memory. The model retrieves relevant info for informed responses."
+                description="""
+                    ### RAG-Enhanced Chat with Custom R1
+                    Chat with the custom R1 model, enhanced with retrieval-augmented generation (RAG).
+                    The model retrieves relevant information to provide more informed and context-aware responses.
+                """
             )
         ]
-    )
+    ).render() # Added .render() here for potential future theme application
 
 demo.launch()