Update app.py

app.py

@@ -22,8 +22,7 @@ def respond(
     top_p,
     frequency_penalty,
     seed,
-    model,
-    custom_model
+    model
 ):
     """
     This function handles the chatbot response. It takes in:
@@ -35,8 +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'
-    - model: the selected model from the featured list
-    - custom_model: a custom model specified by the user
+    - model: the selected model for text generation
     """
 
     print(f"Received message: {message}")
@@ -44,15 +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"Model: {model}, Custom Model: {custom_model}")
-
-    # Determine the model to use
-    if custom_model.strip() != "":
-        selected_model = custom_model.strip()
-    else:
-        selected_model = model
-
-    print(f"Selected model for inference: {selected_model}")
+    print(f"Model: {model}")
 
     # Convert seed to None if -1 (meaning random)
     if seed == -1:
@@ -77,25 +67,24 @@ def respond(
 
     # Start with an empty string to build the response as tokens stream in
     response = ""
-    print(f"Sending request to OpenAI API using model: {selected_model}.")
+    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=selected_model,
+        model=model,   # Use the selected model
         max_tokens=max_tokens,
         stream=True,  # Stream the response
         temperature=temperature,
         top_p=top_p,
-        frequency_penalty=frequency_penalty,
-        seed=seed,
+        frequency_penalty=frequency_penalty,  # <-- NEW
+        seed=seed,                             # <-- NEW
         messages=messages,
     ):
         # Extract the token text from the response chunk
         token_text = message_chunk.choices[0].delta.content
-        if token_text is not None:
-            print(f"Received token: {token_text}")
-            response += token_text
-            yield response
+        print(f"Received token: {token_text}")
+        response += token_text
+        yield response
 
     print("Completed response generation.")
 
@@ -103,108 +92,175 @@ def respond(
 chatbot = gr.Chatbot(height=600)
 print("Chatbot interface created.")
 
-# Define featured models
-featured_models_list = [
-    "meta-llama/Llama-3.3-70B-Instruct",
-    "mistralai/Mistral-7B-v0.1",
-    "google/gemma-7b",
-]
-
 # Create the Gradio ChatInterface
-with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
-    with gr.Tab("Chat"):
+# We add two new sliders for Frequency Penalty and Seed
+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,  # Arbitrary upper limit for demonstration
+            value=-1,
+            step=1,
+            label="Seed (-1 for random)"
+        ),
+        gr.Textbox(label="Custom Model", info="Model Hugging Face path (optional)", placeholder="meta-llama/Llama-3.3-70B-Instruct"),
+    ],
+    fill_height=True,
+    chatbot=chatbot,
+    theme="Nymbo/Nymbo_Theme",
+)
+print("Gradio interface initialized.")
+
+# Custom CSS to hide the footer in the interface
+css = """
+* {}
+footer {visibility: hidden !important;}
+"""
+
+print("Initializing Gradio interface...")  # Debug log
+
+# Define the Gradio interface
+with gr.Blocks(theme='Nymbo/Nymbo_Theme_5') as textgen:
+    # Tab for basic settings
+    with gr.Tab("Basic Settings"):
+        with gr.Row():
+            with gr.Column(elem_id="prompt-container"):
+                with gr.Row():
+                    # Textbox for user to input the prompt
+                    text_prompt = gr.Textbox(label="Prompt", placeholder="Enter a prompt here", lines=3, elem_id="prompt-text-input")
+                with gr.Row():
+                    # Textbox for custom model input
+                    custom_model = gr.Textbox(label="Custom Model", info="Model Hugging Face path (optional)", placeholder="meta-llama/Llama-3.3-70B-Instruct")
+                with gr.Row():
+                    # Accordion for selecting the model
+                    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")
+                        models_list = (
+                            "meta-llama/Llama-3.3-70B-Instruct",
+                            "meta-llama/Llama-3.3-30B-Instruct",
+                            "meta-llama/Llama-3.3-13B-Instruct",
+                            "meta-llama/Llama-3.3-7B-Instruct",
+                        )
+
+                        # Radio buttons to select the desired model
+                        model = gr.Radio(label="Select a model below", value="meta-llama/Llama-3.3-70B-Instruct", choices=models_list, interactive=True, elem_id="model-radio")
+
+                        # Filtering models based on search input
+                        def filter_models(search_term):
+                            filtered_models = [m for m in models_list if search_term.lower() in m.lower()]
+                            return gr.update(choices=filtered_models)
+
+                        # Update model list when search box is used
+                        model_search.change(filter_models, inputs=model_search, outputs=model)
+
+    # Tab for advanced settings
+    with gr.Tab("Advanced Settings"):
         with gr.Row():
-            with gr.Column():
-                # Chat interface
-                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.Dropdown(label="Featured Models", choices=featured_models_list, value="meta-llama/Llama-3.3-70B-Instruct", interactive=True),
-                        gr.Textbox(value="", label="Custom Model (Optional)"),
-                    ],
-                    fill_height=True,
-                    chatbot=chatbot,
-                )
-            with gr.Column():
-                # Featured models accordion
-                with gr.Accordion("Featured Models", open=True):
-                    model_search = gr.Textbox(label="Filter Models", placeholder="Search for a featured model...", lines=1)
-                    model_radio = gr.Radio(label="Select a model below", choices=featured_models_list, value="meta-llama/Llama-3.3-70B-Instruct", interactive=True)
-
-                    def filter_models(search_term):
-                        filtered_models = [m for m in featured_models_list if search_term.lower() in m.lower()]
-                        return gr.update(choices=filtered_models)
-
-                    model_search.change(filter_models, inputs=model_search, outputs=model_radio)
-
-                # Custom model textbox
-                custom_model_textbox = gr.Textbox(label="Custom Model", placeholder="Enter a custom model path here (optional)", lines=1)
+            # Slider for setting the maximum number of new tokens
+            max_tokens = gr.Slider(label="Max new tokens", value=512, minimum=1, maximum=4096, step=1)
+        with gr.Row():
+            # Slider for adjusting the temperature
+            temperature = gr.Slider(label="Temperature", value=0.7, minimum=0.1, maximum=4.0, step=0.1)
+        with gr.Row():
+            # Slider for adjusting the top-p (nucleus) sampling
+            top_p = gr.Slider(label="Top-P", value=0.95, minimum=0.1, maximum=1.0, step=0.05)
+        with gr.Row():
+            # Slider for adjusting the frequency penalty
+            frequency_penalty = gr.Slider(label="Frequency Penalty", value=0.0, minimum=-2.0, maximum=2.0, step=0.1)
+        with gr.Row():
+            # Slider for setting the seed for reproducibility
+            seed = gr.Slider(label="Seed", value=-1, minimum=-1, maximum=65535, step=1)
 
+    # Tab to provide information to the user
     with gr.Tab("Information"):
-        with gr.Accordion("Featured Models", open=False):
+        with gr.Row():
+            # Display a sample prompt for guidance
+            gr.Textbox(label="Sample prompt", value="{prompt} | ultra detail, ultra elaboration, ultra quality, perfect.")
+
+        # Accordion displaying featured models
+        with gr.Accordion("Featured Models (WiP)", open=False):
             gr.HTML(
                 """
-                <p><a href="https://huggingface.co/models?pipeline_tag=text-generation&sort=trending">See all available models</a></p>
-                <table style="width:100%; text-align:center; margin:auto;">
-                    <tr>
-                        <th>Model Name</th>
-                        <th>Notes</th>
-                    </tr>
-                    <tr>
-                        <td>meta-llama/Llama-3.3-70B-Instruct</td>
-                        <td>Powerful large language model.</td>
-                    </tr>
-                    <tr>
-                        <td>mistralai/Mistral-7B-v0.1</td>
-                        <td>A smaller, efficient model.</td>
-                    </tr>
-                    <tr>
-                        <td>google/gemma-7b</td>
-                        <td>Google's language model.</td>
-                    </tr>
-                </table>
-                """
+            <p><a href="https://huggingface.co/models?inference=warm&pipeline_tag=text-generation&sort=trending">See all available models</a></p>
+            <table style="width:100%; text-align:center; margin:auto;">
+                <tr>
+                    <th>Model Name</th>
+                    <th>Typography</th>
+                    <th>Notes</th>
+                </tr>
+                <tr>
+                    <td>meta-llama/Llama-3.3-70B-Instruct</td>
+                    <td>✅</td>
+                    <td></td>
+                </tr>
+                <tr>
+                    <td>meta-llama/Llama-3.3-30B-Instruct</td>
+                    <td>✅</td>
+                    <td></td>
+                </tr>
+                <tr>
+                    <td>meta-llama/Llama-3.3-13B-Instruct</td>
+                    <td>✅</td>
+                    <td></td>
+                </tr>
+                <tr>
+                    <td>meta-llama/Llama-3.3-7B-Instruct</td>
+                    <td>✅</td>
+                    <td></td>
+                </tr>
+            </table>
+            """
             )
 
+        # Accordion providing an overview of advanced settings
         with gr.Accordion("Parameters Overview", open=False):
             gr.Markdown(
-                """
-                ## Parameters Overview
-
-                ### System Message
-                The system message is an initial instruction or context that you provide to the chatbot. It sets the stage for the conversation and can be used to guide the chatbot's behavior or persona.
-
-                ### Max New Tokens
-                This parameter limits the length of the chatbot's response. It specifies the maximum number of tokens (words or subwords) that the chatbot can generate in a single response.
+            """
+            ## System Message
+            ###### This box is for setting the system prompt, which guides the AI's behavior and context.
 
-                ### Temperature
-                Temperature controls the randomness of the chatbot's responses. A higher temperature (e.g., 1.0) makes the output more random and creative, while a lower temperature (e.g., 0.2) makes the output more focused and deterministic.
+            ## Max New Tokens
+            ###### This slider allows you to specify the maximum number of tokens (words or parts of words) the AI will generate in response to your prompt. The default value is 512.
 
-                ### Top-P
-                Top-P, also known as nucleus sampling, is another way to control the randomness of the responses. It sets a threshold for the cumulative probability of the most likely tokens. The chatbot will only consider tokens whose cumulative probability is below this threshold.
+            ## Temperature
+            ###### Temperature controls the randomness of the AI's output. A higher temperature makes the output more random and creative, while a lower temperature makes it more predictable and focused.
 
-                ### Frequency Penalty
-                This parameter discourages the chatbot from repeating the same tokens or phrases too often. A higher value (e.g., 1.0) penalizes repetition more strongly, while a lower value (e.g., 0.0) has no penalty.
+            ## Top-P (Nucleus Sampling)
+            ###### Top-P sampling is a technique that selects the smallest set of top tokens whose cumulative probability exceeds a threshold (p). This helps in generating more coherent and relevant responses.
 
-                ### Seed
-                The seed is a number that initializes the random number generator used by the chatbot. If you set a specific seed, you will get the same response every time you run the chatbot with the same parameters. If you set the seed to -1, a random seed will be used, resulting in different responses each time.
+            ## Frequency Penalty
+            ###### This parameter penalizes repeated tokens in the output, encouraging the AI to generate more diverse responses. A higher value means more penalty for repetition.
 
-                ### Featured Models
-                You can select a featured model from the dropdown list. These models have been pre-selected for their performance and capabilities.
+            ## Seed
+            ###### The seed is a value that ensures reproducibility. If you set a specific seed, the AI will generate the same output for the same input. Setting it to -1 means the seed will be random.
 
-                ### Custom Model
-                If you have a specific model that you want to use, you can enter its path in the Custom Model textbox. This allows you to use models that are not included in the featured list.
-                """
+            ### Remember, these settings are all about giving you control over the text generation process. Feel free to experiment and see what each one does. And if you're ever in doubt, the default settings are a great place to start. Happy creating!
+            """
             )
 
-print("Gradio interface initialized.")
+    # Row containing the 'Run' button to trigger the text generation
+    with gr.Row():
+        text_button = gr.Button("Run", variant='primary', elem_id="gen-button")
+    # Row for displaying the generated text output
+    with gr.Row():
+        text_output = gr.Textbox(label="Text Output", elem_id="text-output")
+
+    # Set up button click event to call the respond function
+    text_button.click(respond, inputs=[text_prompt, chatbot, gr.Textbox(value="", label="System message"), max_tokens, temperature, top_p, frequency_penalty, seed, model], outputs=text_output)
 
-if __name__ == "__main__":
-    print("Launching the demo application.")
-    demo.launch()
+print("Launching Gradio interface...")  # Debug log
+# Launch the Gradio interface without showing the API or sharing externally
+textgen.launch(show_api=False, share=False)