Update app.py

app.py

@@ -1,228 +1,96 @@
 import gradio as gr
-from openai import OpenAI
-import os
-
-ACCESS_TOKEN = os.getenv("HF_TOKEN")
-print("Access token loaded.")
-
-client = OpenAI(
-    base_url="https://api-inference.huggingface.co/v1/",
-    api_key=ACCESS_TOKEN,
-)
-print("OpenAI client initialized.")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    frequency_penalty,
-    seed,
-    custom_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}")
-    print(f"Selected model (custom_model): {custom_model}")
-
-    # Convert seed to None if -1 (meaning random)
-    if seed == -1:
-        seed = None
-
-    messages = [{"role": "system", "content": system_message}]
-    print("Initial messages array constructed.")
-
-    # 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
-    messages.append({"role": "user", "content": message})
-    print("Latest user message appended.")
-
-    # If user provided a model, use that; otherwise, fall back to a default model
-    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}")
-
-    # Start with an empty string to build the response as tokens stream in
-    response = ""
-    print("Sending request to OpenAI API.")
-
-    for message_chunk in client.chat.completions.create(
-        model=model_to_use,
+from huggingface_hub import InferenceClient
+import json
+
+# Список доступных моделей
+models_list = [
+    "google/gemma-3-27b-it",
+    "meta-llama/Llama-3.3-70B-Instruct",
+    "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B",
+    "Qwen/Qwen2.5-72B-Instruct"
+]
+
+def add_message(role, content, messages):
+    messages.append({"role": role, "content": content})
+    return messages, len(messages), str(messages)
+
+def clear_messages(messages):
+    return [], 0, "[]"
+
+def start_conversation(model, messages, max_tokens, temperature, top_p):
+    client = InferenceClient(model)
+    response = client.chat_completion(
+        messages,
         max_tokens=max_tokens,
-        stream=True,
+        stream=False,
         temperature=temperature,
         top_p=top_p,
-        frequency_penalty=frequency_penalty,
-        seed=seed,
-        messages=messages,
-    ):
-        token_text = message_chunk.choices[0].delta.content
-        print(f"Received token: {token_text}")
-        response += token_text
-        yield response
-
-    print("Completed response generation.")
-
-# GRADIO UI
-
-chatbot = gr.Chatbot(height=600, show_copy_button=True, placeholder="Select a model and begin chatting", likeable=True, layout="panel")
-print("Chatbot interface created.")
-
-system_message_box = gr.Textbox(value="", placeholder="You are a helpful assistant.", label="System Prompt")
-
-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.",
-    placeholder="meta-llama/Llama-3.3-70B-Instruct"
-)
-
-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.
-    """
-    print(f"Featured model selected: {selected}")
-    return selected
-
-demo = gr.ChatInterface(
-    fn=respond,
-    additional_inputs=[
-        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("ChatInterface object created.")
+    )
+    return response.choices[0].message.content
+
+def show_messages(messages):
+    return str(messages)
+
+def get_messages_api(messages):
+    return json.dumps(messages, indent=4)
+
+demo = gr.Blocks()
 
 with demo:
-    with gr.Accordion("Model Selection", open=False):
-        model_search_box = gr.Textbox(
-            label="Filter Models",
-            placeholder="Search for a featured model...",
-            lines=1
-        )
-        print("Model search box created.")
-
-        models_list = [
-            "meta-llama/Llama-3.3-70B-Instruct",
-            "meta-llama/Llama-3.1-70B-Instruct",
-            "meta-llama/Llama-3.0-70B-Instruct",
-            "meta-llama/Llama-3.2-3B-Instruct",
-            "meta-llama/Llama-3.2-1B-Instruct",
-            "meta-llama/Llama-3.1-8B-Instruct",
-            "NousResearch/Hermes-3-Llama-3.1-8B",
-            "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
-            "mistralai/Mistral-Nemo-Instruct-2407",
-            "mistralai/Mixtral-8x7B-Instruct-v0.1",
-            "mistralai/Mistral-7B-Instruct-v0.3",
-            "mistralai/Mistral-7B-Instruct-v0.2",
-            "Qwen/Qwen2.5-72B-Instruct",
-            "Qwen/Qwen2.5-3B-Instruct",
-            "Qwen/Qwen2.5-0.5B-Instruct",
-            "Qwen/QwQ-32B-Preview",
-            "Qwen/Qwen2.5-Coder-32B-Instruct",
-            "microsoft/Phi-3.5-mini-instruct",
-            "microsoft/Phi-3-mini-128k-instruct",
-            "microsoft/Phi-3-mini-4k-instruct",
-            "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B",
-            "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",
-            "HuggingFaceH4/zephyr-7b-beta",
-            "HuggingFaceTB/SmolLM2-360M-Instruct",
-            "tiiuae/falcon-7b-instruct",
-            "01-ai/Yi-1.5-34B-Chat",
-        ]
-        print("Models list initialized.")
-
-        featured_model_radio = gr.Radio(
-            label="Select a model below",
-            choices=models_list,
-            value="meta-llama/Llama-3.3-70B-Instruct",
-            interactive=True
-        )
-        print("Featured models radio button created.")
-
-        def filter_models(search_term):
-            print(f"Filtering models with search term: {search_term}")
-            filtered = [m for m in models_list if search_term.lower() in m.lower()]
-            print(f"Filtered models: {filtered}")
-            return gr.update(choices=filtered)
-
-        model_search_box.change(
-            fn=filter_models,
-            inputs=model_search_box,
-            outputs=featured_model_radio
-        )
-        print("Model search box change event linked.")
-
-        featured_model_radio.change(
-            fn=set_custom_model_from_radio,
-            inputs=featured_model_radio,
-            outputs=custom_model_box
-        )
-        print("Featured model radio button change event linked.")
-
-print("Gradio interface initialized.")
+    gr.Markdown("# Chat Interface")
+    role_input = gr.Textbox(label="Role")
+    content_input = gr.Textbox(label="Content")
+    messages_state = gr.State(value=[])
+    messages_output = gr.Textbox(label="Messages", value="[]")
+    count_output = gr.Number(label="Count", value=0)
+    response_output = gr.Textbox(label="Response")
+    messages_api_output = gr.Textbox(label="Messages API")
+
+    add_button = gr.Button("Add")
+    clear_button = gr.Button("Clear")
+    start_button = gr.Button("Start")
+    show_button = gr.Button("Show messages")
+    get_api_button = gr.Button("Get messages API")
+
+    model_input = gr.Radio(
+        label="Select a model",
+        choices=models_list,
+        value=models_list[0],
+    )
+
+    max_tokens_slider = gr.Slider(minimum=1, maximum=2048, 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 (nucleus sampling)")
+
+    add_button.click(
+        add_message,
+        inputs=[role_input, content_input, messages_state],
+        outputs=[messages_state, count_output, messages_output],
+    )
+
+    clear_button.click(
+        clear_messages,
+        inputs=[messages_state],
+        outputs=[messages_state, count_output, messages_output],
+    )
+
+    start_button.click(
+        start_conversation,
+        inputs=[model_input, messages_state, max_tokens_slider, temperature_slider, top_p_slider],
+        outputs=[response_output],
+    )
+
+    show_button.click(
+        show_messages,
+        inputs=[messages_state],
+        outputs=[messages_output],
+    )
+
+    get_api_button.click(
+        get_messages_api,
+        inputs=[messages_state],
+        outputs=[messages_api_output],
+    )
 
 if __name__ == "__main__":
-    print("Launching the demo application.")
     demo.launch()