Update app.py

app.py

@@ -34,7 +34,7 @@ def respond(
     - top_p: top-p (nucleus) sampling
     - frequency_penalty: penalize repeated tokens in the output
     - seed: a fixed seed for reproducibility; -1 will mean 'random'
-    - custom_model: the user-provided custom model name (if any)
+    - custom_model: the final model name in use, which may be set by selecting from the Featured Models radio or by typing a custom model
     """
 
     print(f"Received message: {message}")
@@ -42,7 +42,7 @@ def respond(
     print(f"System message: {system_message}")
     print(f"Max tokens: {max_tokens}, Temperature: {temperature}, Top-P: {top_p}")
     print(f"Frequency Penalty: {frequency_penalty}, Seed: {seed}")
-    print(f"Custom model: {custom_model}")
+    print(f"Selected model (custom_model): {custom_model}")
 
     # Convert seed to None if -1 (meaning random)
     if seed == -1:
@@ -65,7 +65,7 @@ def respond(
     # Append the latest user message
     messages.append({"role": "user", "content": message})
 
-    # Determine which model to use: either custom_model or a default
+    # If user provided a model, use that; otherwise, fall back to a default
     model_to_use = custom_model.strip() if custom_model.strip() != "" else "meta-llama/Llama-3.3-70B-Instruct"
     print(f"Model selected for inference: {model_to_use}")
 
@@ -75,7 +75,7 @@ def respond(
 
     # Make the streaming request to the HF Inference API via openai-like client
     for message_chunk in client.chat.completions.create(
-        model=model_to_use,              # Use either the user-provided custom model or default
+        model=model_to_use,              # Use either the user-provided or default model
         max_tokens=max_tokens,
         stream=True,                     # Stream the response
         temperature=temperature,
@@ -93,104 +93,137 @@ def respond(
 
     print("Completed response generation.")
 
+# -------------------------
+# GRADIO UI CONFIGURATION
+# -------------------------
+
 # Create a Chatbot component with a specified height
 chatbot = gr.Chatbot(height=600)
 print("Chatbot interface created.")
 
-# Create the Gradio ChatInterface
-# We add two new sliders for Frequency Penalty, Seed, and now a new "Custom Model" text box.
+# We'll create text boxes & sliders for system prompt, tokens, etc.
+system_message_box = gr.Textbox(value="", label="System message")
+
+max_tokens_slider = gr.Slider(
+    minimum=1,
+    maximum=4096,
+    value=512,
+    step=1,
+    label="Max new tokens"
+)
+temperature_slider = gr.Slider(
+    minimum=0.1,
+    maximum=4.0,
+    value=0.7,
+    step=0.1,
+    label="Temperature"
+)
+top_p_slider = gr.Slider(
+    minimum=0.1,
+    maximum=1.0,
+    value=0.95,
+    step=0.05,
+    label="Top-P"
+)
+frequency_penalty_slider = gr.Slider(
+    minimum=-2.0,
+    maximum=2.0,
+    value=0.0,
+    step=0.1,
+    label="Frequency Penalty"
+)
+seed_slider = gr.Slider(
+    minimum=-1,
+    maximum=65535,
+    value=-1,
+    step=1,
+    label="Seed (-1 for random)"
+)
+
+# The custom_model_box is what the respond function sees as "custom_model"
+custom_model_box = gr.Textbox(
+    value="",
+    label="Custom Model",
+    info="(Optional) Provide a custom Hugging Face model path. Overrides any selected featured model."
+)
+
+# Define a function that, when a user selects a model from the radio, populates `custom_model_box`
+def set_custom_model_from_radio(selected):
+    """
+    This function will get triggered whenever someone picks a model from the 'Featured Models' radio.
+    We will update the Custom Model text box with that selection automatically.
+    """
+    return selected
+
+# The main ChatInterface object
 demo = gr.ChatInterface(
     fn=respond,
+    # For ChatInterface, we can pass additional inputs in order to feed them into the "respond" function
     additional_inputs=[
-        gr.Textbox(value="", label="System message"),
-        gr.Slider(
-            minimum=1,
-            maximum=4096,
-            value=512,
-            step=1,
-            label="Max new tokens"
-        ),
-        gr.Slider(
-            minimum=0.1,
-            maximum=4.0,
-            value=0.7,
-            step=0.1,
-            label="Temperature"
-        ),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-P"
-        ),
-        gr.Slider(
-            minimum=-2.0,
-            maximum=2.0,
-            value=0.0,
-            step=0.1,
-            label="Frequency Penalty"
-        ),
-        gr.Slider(
-            minimum=-1,
-            maximum=65535,
-            value=-1,
-            step=1,
-            label="Seed (-1 for random)"
-        ),
-        gr.Textbox(
-            value="",
-            label="Custom Model",
-            info="(Optional) Provide a custom Hugging Face model path. This will override the default model if not empty."
-        ),
+        system_message_box,
+        max_tokens_slider,
+        temperature_slider,
+        top_p_slider,
+        frequency_penalty_slider,
+        seed_slider,
+        custom_model_box
     ],
     fill_height=True,
     chatbot=chatbot,
     theme="Nymbo/Nymbo_Theme",
 )
-print("Gradio interface initialized.")
 
-# --------------------------------------------------------
-# NEW FEATURE: "Featured Models" Accordion with Filtering
-# Adapted from Serverless-ImgGen-Hub's approach
-# --------------------------------------------------------
+# -----------
+# ADDING THE "FEATURED MODELS" ACCORDION
+# -----------
 with demo:
     with gr.Accordion("Featured Models", open=False):
-        # Textbox to search/filter models
-        model_search = gr.Textbox(
+        model_search_box = gr.Textbox(
             label="Filter Models",
             placeholder="Search for a featured model...",
             lines=1
         )
-        
-        # For demonstration purposes, here is a sample list of possible text-generation models
+
+        # Sample list of popular text models
         models_list = [
             "meta-llama/Llama-3.3-70B-Instruct",
-            "meta-llama/Llama-3.1-8B-Instruct",
-            "microsoft/Phi-3.5-mini-instruct",
-            "mistralai/Mistral-7B-Instruct-v0.3",
+            "bigscience/bloomz-7b1",
+            "OpenAssistant/oasst-sft-1-pythia-12b",
+            "OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5",
             "tiiuae/falcon-7b-instruct",
-            "Qwen/Qwen2.5-72B-Instruct",
+            "OpenAI/gpt-3.5-turbo",
+            "OpenAI/gpt-4-32k",
+            "meta-llama/Llama-2-13B-chat-hf",
+            "meta-llama/Llama-2-70B-chat-hf",
         ]
-        
-        # Radio buttons to display and select from the featured models
-        # This won't directly override the "Custom Model" field, but you can copy it from here
-        featured_model = gr.Radio(
+
+        featured_model_radio = gr.Radio(
             label="Select a model below",
             choices=models_list,
             value="meta-llama/Llama-3.3-70B-Instruct",
             interactive=True
         )
 
-        # Filtering function to update model list based on search input
+        # Filter function for the radio
         def filter_models(search_term):
-            # Filter the list by checking if the search term is in each model name
             filtered = [m for m in models_list if search_term.lower() in m.lower()]
             return gr.update(choices=filtered)
 
-        # When the user types in the search box, we update the featured_model radio choices
-        model_search.change(filter_models, inputs=model_search, outputs=featured_model)
+        # Whenever we type in the search box, update the radio with the filtered list
+        model_search_box.change(
+            fn=filter_models,
+            inputs=model_search_box,
+            outputs=featured_model_radio
+        )
 
+        # Whenever we select a featured model, populate the 'Custom Model' textbox
+        featured_model_radio.change(
+            fn=set_custom_model_from_radio,
+            inputs=featured_model_radio,
+            outputs=custom_model_box
+        )
+
+print("Gradio interface initialized.")
 
 if __name__ == "__main__":
     print("Launching the demo application.")