Falcon-H1 is a new SoTA hybrid model by TII in Abu Dhabi. It is open source and available on Hugging Face. This demo is powered by OpenInnovationAI. Try out our chat interface.
Today: {today_date}
""" def stream_chat( message: str, history: list, model_name: str, temperature: float = 0.1, max_new_tokens: int = 1024, top_p: float = 1.0, top_k: int = 20, presence_penalty: float = 1.2, ): """Chat function that streams responses from the selected model""" cfg = MODEL_CONFIGS[model_name] api_key = os.getenv(cfg["api_key_env"]) base_url = os.getenv(cfg.get("base_url_env", ""), None) if not api_key: yield f"❌ Env-var `{cfg['api_key_env']}` not set." return if cfg.get("base_url_env") and not base_url: yield f"❌ Env-var `{cfg['base_url_env']}` not set." return client = openai.OpenAI(api_key=api_key, base_url=base_url) msgs = [{"role": "system", "content": SYSTEM_PROMPT},] for u, a in history: msgs += [{"role": "user", "content": u}, {"role": "assistant", "content": a}] msgs.append({"role": "user", "content": message}) try: stream = client.chat.completions.create( model=cfg["model_id"], messages=msgs, temperature=temperature, top_p=top_p, max_tokens=max_new_tokens, presence_penalty=presence_penalty, stream=True, ) partial = "" for chunk in stream: if (delta := chunk.choices[0].delta).content: partial += delta.content yield partial except Exception as e: yield f"❌ Error: {str(e)}" # Create the Gradio interface with gr.Blocks(css=CSS, theme="soft") as demo: # Header section gr.HTML(TITLE) gr.HTML(SUB_TITLE.format(today_date=today.strftime('%B %d, %Y'))) gr.DuplicateButton(value="Duplicate Space", elem_classes="duplicate-button") # Create model selection with descriptions model_options = list(MODEL_CONFIGS.keys()) with gr.Row(): model_dropdown = gr.Dropdown( choices=model_options, value=model_options[0], label="Select Falcon-H1 Model", info="Choose which model checkpoint to use" ) # Create chatbot chatbot = gr.Chatbot(height=600, latex_delimiters=[ {"left": "$$", "right": "$$", "display": True}, # For display mode math {"left": "$", "right": "$", "display": False}, # For inline math {"left": "\\(", "right": "\\)", "display": False}, # Common inline delimiters {"left": "\\[", "right": "\\]", "display": True} # Common display delimiters ] ) # Message input area with a cleaner layout with gr.Row(): with gr.Column(scale=0.85): msg = gr.Textbox( scale=1, show_label=False, placeholder="Enter text and press enter", container=False ) with gr.Column(scale=0.15, min_width=0): submit_btn = gr.Button("Submit", variant="primary") # Parameters in accordion similar to second code with gr.Accordion("⚙️ Parameters", open=False, elem_classes="accordion"): max_new_tokens = gr.Slider(minimum=64, maximum=4096*8, value=1024, step=64, label="Max new tokens", info="Maximum length of generated response") top_p = gr.Slider(minimum=0, maximum=1, value=1.0, step=0.05, label="top_p", info="1.0 means no filtering") top_k = gr.Slider(minimum=1, maximum=20, step=1, value=20, label="top_k") presence_penalty = gr.Slider(minimum=0, maximum=2, value=1.2, step=0.1, label="Presence penalty", info="Penalizes repetition") # Examples section gr.Examples( examples=[ ["Hello there, can you suggest a few places to visit in UAE?"], ["What is UAE known for?"], ], inputs=msg, ) # Chat handler function def user(user_message, history): return "", history + [[user_message, None]] def bot(history, model_choice, max_tokens, top_p_val, top_k_val, penalty): temp = 0.1 user_message = history[-1][0] history[-1][1] = "" for character in stream_chat( user_message, history[:-1], model_choice, temp, max_tokens, top_p_val, top_k_val, penalty ): history[-1][1] = character yield history # Set up event handlers msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then( bot, [chatbot, model_dropdown, max_new_tokens, top_p, top_k, presence_penalty], chatbot ) submit_btn.click(user, [msg, chatbot], [msg, chatbot], queue=False).then( bot, [chatbot, model_dropdown, max_new_tokens, top_p, top_k, presence_penalty], chatbot ) if __name__ == "__main__": demo.launch()