import gradio as gr from llama_cpp import Llama # Initialize the model model = Llama( model_path="Fgot_Official_3B.Q4_K_M.gguf", # Replace with your model path n_ctx=4096, # Context window n_threads=2 # Number of CPU threads to use ) def format_response(text): # Обрабатываем переносы строк для HTML return text.replace('\n', '
') def respond( message, history: list[tuple[str, str]], system_message, max_tokens, temperature, top_p, ): # Format the conversation history messages = [{"role": "system", "content": system_message}] for user_msg, assistant_msg in history: if user_msg: messages.append({"role": "user", "content": user_msg}) if assistant_msg: messages.append({"role": "assistant", "content": assistant_msg}) messages.append({"role": "user", "content": message}) # Generate response response = "" stream = model.create_chat_completion( messages=messages, max_tokens=max_tokens, stream=True, temperature=temperature, top_p=top_p ) for chunk in stream: # Extract content from the chunk if 'choices' in chunk and len(chunk['choices']) > 0: if 'text' in chunk['choices'][0]: content = chunk['choices'][0]['text'] elif 'delta' in chunk['choices'][0] and 'content' in chunk['choices'][0]['delta']: content = chunk['choices'][0]['delta']['content'] else: continue if content is not None: response += content yield format_response(response) # Create the Gradio interface with HTML formatting demo = gr.ChatInterface( respond, additional_inputs=[ gr.Textbox(value="(Пиши посты)", label="System prompt"), gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Максимум новых токенов"), gr.Slider(minimum=0.1, maximum=2.0, value=0.42, step=0.01, label="Температура (рандомность)"), gr.Slider( minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (работает почти как температура, 0.95 = 95%)", ), ], ) if __name__ == "__main__": demo.launch()