Update app.py

app.py

@@ -22,193 +22,152 @@ def respond(
     top_p,
     frequency_penalty,
     seed,
-    model
+    model_selection,
+    custom_model
 ):
     """
-    This function handles the chatbot response. It takes in:
-    - message: the user's new message
-    - history: the list of previous messages, each as a tuple (user_msg, assistant_msg)
-    - system_message: the system prompt
-    - max_tokens: the maximum number of tokens to generate in the response
-    - temperature: sampling temperature
-    - top_p: top-p (nucleus) sampling
-    - frequency_penalty: penalize repeated tokens in the output
-    - seed: a fixed seed for reproducibility; -1 will mean 'random'
-    - model: the selected model for text generation
+    This function handles the chatbot response.
     """
+    selected_model = custom_model if custom_model.strip() != "" else model_selection
+    print(f"Selected model: {selected_model}")
 
-    print(f"Received message: {message}")
-    print(f"History: {history}")
-    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}, Model: {model}")
-
-    # Convert seed to None if -1 (meaning random)
     if seed == -1:
         seed = None
 
-    # Construct the messages array required by the API
     messages = [{"role": "system", "content": system_message}]
-
-    # Add conversation history to the context
     for val in history:
-        user_part = val[0]
-        assistant_part = val[1]
-        if user_part:
-            messages.append({"role": "user", "content": user_part})
-            print(f"Added user message to context: {user_part}")
-        if assistant_part:
-            messages.append({"role": "assistant", "content": assistant_part})
-            print(f"Added assistant message to context: {assistant_part}")
-
-    # Append the latest user message
+        if val[0]:
+            messages.append({"role": "user", "content": val[0]})
+        if val[1]:
+            messages.append({"role": "assistant", "content": val[1]})
     messages.append({"role": "user", "content": message})
 
-    # Start with an empty string to build the response as tokens stream in
     response = ""
-    print("Sending request to OpenAI API.")
-
-    # Make the streaming request to the HF Inference API via openai-like client
     for message_chunk in client.chat.completions.create(
-        model=model,  # Use the selected model
+        model=selected_model,
         max_tokens=max_tokens,
-        stream=True,  # Stream the response
+        stream=True,
         temperature=temperature,
         top_p=top_p,
         frequency_penalty=frequency_penalty,
         seed=seed,
         messages=messages,
     ):
-        # Extract the token text from the response chunk
         token_text = message_chunk.choices[0].delta.content
-        print(f"Received token: {token_text}")
         response += token_text
         yield response
 
-    print("Completed response generation.")
-
 # Create a Chatbot component with a specified height
 chatbot = gr.Chatbot(height=600)
-print("Chatbot interface created.")
 
-# List of featured models (placeholder models for now)
+# Define placeholder models
 featured_models = [
     "meta-llama/Llama-3.3-70B-Instruct",
-    "gpt-3.5-turbo",
-    "gpt-4",
-    "mistralai/Mistral-7B-Instruct-v0.1",
-    "tiiuae/falcon-40b-instruct"
+    "gpt2",
+    "bert-base-uncased",
+    "facebook/bart-base",
+    "google/flan-t5-base"
 ]
 
-# Function to filter models based on search input
-def filter_models(search_term):
-    filtered_models = [m for m in featured_models if search_term.lower() in m.lower()]
-    return gr.update(choices=filtered_models)
-
 # Create the Gradio ChatInterface
-demo = gr.ChatInterface(
-    respond,
-    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.Radio(label="Select a model below", value="meta-llama/Llama-3.3-70B-Instruct", choices=featured_models, interactive=True, elem_id="model-radio")
-    ],
-    fill_height=True,
-    chatbot=chatbot,
-    theme="Nymbo/Nymbo_Theme",
-)
+with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
+    gr.Markdown("# Serverless Text Generation Hub")
 
-# Add a "Custom Model" text box and "Featured Models" accordion
-with demo:
-    with gr.Tab("Model Settings"):
+    with gr.Tab("Basic Settings"):
         with gr.Row():
             with gr.Column():
-                # Textbox for custom model input
-                custom_model = gr.Textbox(label="Custom Model", info="Hugging Face model path (optional)", placeholder="username/model-name")
-                # Accordion for selecting featured models
+                # Textbox for system message
+                system_message = gr.Textbox(value="", label="System message")
+        with gr.Row():
+            with gr.Column():
+                # Model selection
                 with gr.Accordion("Featured Models", open=True):
-                    # Textbox for searching models
-                    model_search = gr.Textbox(label="Filter Models", placeholder="Search for a featured model...", lines=1, elem_id="model-search-input")
-                    # Radio buttons to select the desired model
-                    model_radio = gr.Radio(label="Select a model below", value="meta-llama/Llama-3.3-70B-Instruct", choices=featured_models, interactive=True, elem_id="model-radio")
-                    # Update model list when search box is used
-                    model_search.change(filter_models, inputs=model_search, outputs=model_radio)
-
-    # Add an "Information" tab with accordions
-    with gr.Tab("Information"):
+                    model_search = gr.Textbox(label="Filter Models", placeholder="Search for a featured model...")
+                    model = gr.Radio(label="Select a model", choices=featured_models, value="meta-llama/Llama-3.3-70B-Instruct")
+
+                    def filter_models(search_term):
+                        filtered_models = [m for m in featured_models if search_term.lower() in m.lower()]
+                        return gr.update(choices=filtered_models)
+
+                    model_search.change(filter_models, inputs=model_search, outputs=model)
+        with gr.Row():
+            with gr.Column():
+                # Custom model input
+                custom_model = gr.Textbox(label="Custom Model", placeholder="Enter a custom model name")
+
+    with gr.Tab("Advanced Settings"):
+        with gr.Row():
+            max_tokens = gr.Slider(minimum=1, maximum=4096, value=512, step=1, label="Max new tokens")
+            temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
         with gr.Row():
-            # Accordion for "Featured Models" with a table
-            with gr.Accordion("Featured Models (WiP)", open=False):
-                gr.HTML(
-                    """
-                <p><a href="https://huggingface.co/models?inference=warm&pipeline_tag=text-generation&sort=trending">See all available models</a></p>
+            top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-P")
+            frequency_penalty = gr.Slider(minimum=-2.0, maximum=2.0, value=0.0, step=0.1, label="Frequency Penalty")
+        with gr.Row():
+            seed = gr.Slider(minimum=-1, maximum=65535, value=-1, step=1, label="Seed (-1 for random)")
+
+    with gr.Tab("Information"):
+        with gr.Accordion("Featured Models", open=False):
+            gr.Markdown(
+                """
                 <table style="width:100%; text-align:center; margin:auto;">
                     <tr>
                         <th>Model Name</th>
-                        <th>Typical Use Case</th>
-                        <th>Notes</th>
+                        <th>Description</th>
                     </tr>
                     <tr>
                         <td>meta-llama/Llama-3.3-70B-Instruct</td>
-                        <td>General-purpose instruction following</td>
-                        <td>High-quality, large-scale model</td>
+                        <td>Highly capable Llama model</td>
                     </tr>
                     <tr>
-                        <td>gpt-3.5-turbo</td>
-                        <td>Chat and general text generation</td>
-                        <td>Fast and efficient</td>
+                        <td>gpt2</td>
+                        <td>Generative Pre-trained Transformer 2</td>
                     </tr>
                     <tr>
-                        <td>gpt-4</td>
-                        <td>Advanced text generation</td>
-                        <td>State-of-the-art performance</td>
-                    </tr>
-                    <tr>
-                        <td>mistralai/Mistral-7B-Instruct-v0.1</td>
-                        <td>Instruction following</td>
-                        <td>Lightweight and efficient</td>
-                    </tr>
-                    <tr>
-                        <td>tiiuae/falcon-40b-instruct</td>
-                        <td>Instruction following</td>
-                        <td>High-quality, large-scale model</td>
+                        <td>bert-base-uncased</td>
+                        <td>Bidirectional Encoder Representations from Transformers</td>
                     </tr>
                 </table>
                 """
-                )
-
-            # Accordion for "Parameters Overview" with markdown
-            with gr.Accordion("Parameters Overview", open=False):
-                gr.Markdown(
+            )
+        with gr.Accordion("Parameters Overview", open=False):
+            gr.Markdown(
                 """
                 ## System Message
-                ###### This is the initial prompt that sets the behavior of the model. It can be used to define the tone, style, or role of the assistant.
+                ###### Sets the behavior and tone of the assistant.
 
-                ## Max Tokens
-                ###### This controls the maximum length of the generated response. Higher values allow for longer responses but may take more time to generate.
+                ## Max New Tokens
+                ###### Determines the maximum length of the response.
 
                 ## Temperature
-                ###### This controls the randomness of the output. Lower values make the model more deterministic, while higher values make it more creative.
+                ###### Controls the randomness of the output. Lower values make the output more deterministic.
 
                 ## Top-P
-                ###### This controls the diversity of the output by limiting the model to the most likely tokens. Lower values make the output more focused, while higher values allow for more diversity.
+                ###### Used for nucleus sampling. Higher values include more tokens in consideration.
 
                 ## Frequency Penalty
-                ###### This penalizes repeated tokens in the output. Higher values discourage repetition, while lower values allow for more repetitive outputs.
+                ###### Penalizes the model for repeating the same tokens.
 
                 ## Seed
-                ###### This sets a fixed seed for reproducibility. A value of -1 means the seed is random.
-
-                ## Model
-                ###### This selects the model used for text generation. You can choose from featured models or specify a custom model.
+                ###### Ensures reproducibility of results.
                 """
-                )
-
-print("Gradio interface initialized.")
+            )
+
+    # Chat interface
+    demo = gr.ChatInterface(
+        respond,
+        additional_inputs=[
+            system_message,
+            max_tokens,
+            temperature,
+            top_p,
+            frequency_penalty,
+            seed,
+            model,
+            custom_model
+        ],
+        chatbot=chatbot,
+        theme="Nymbo/Nymbo_Theme"
+    )
 
 if __name__ == "__main__":
     print("Launching the demo application.")