I am wondering if there is the possibility of using a stream for Marino notebooks in the ui.chat, something along the lines of: from ollama import chat model = 'llama3.2' def simple_stream(mess, config): stream = chat(messages=[{'role':'user', 'content':mess}], model=model, stream=True) for chunk in stream: yield chunk['message']['content'] mo.ui.chat(simple_stream, prompts=['']) As it is, ui.chat does not really know how to handle the generator. Maybe there is a simple hack that I should consider? Many thanks for your help! Continue reading...