Jupyter AI

Description
The profile is based on the official Jupyter profile Datascience. Therefore, common packages for Python, Julia, and R are already pre-installed. In addition, the profile contains Jupyter AI, a chat-based AI assistant that supports the development of Jupyter Notebooks.
After the profile has been started, a new chat icon appears in the left sidebar. The chatbot can be used there and the underlying LLM can be configured.


Configuration
When using the AI assistant for the first time, you must specify the LLMs that will process your requests. The configuration is saved across all profiles so that the models ideally only need to be configured once.
When selecting a model, you will receive a list of numerous providers. You cannot simply select and use the models. Almost all providers ask for an API key so that the use of the models can be assigned and billed accordingly. As there is currently no state service that provides LLMs, all universities and users must ensure that they have access to LLMs.

 

If you only want to "try out" the AI assistant briefly, you can select one of the "bwJupyter :: [model]" models. The models are provided free of charge by the GWDG as part of its AI services. No separate API key is required for these LLMs. A list of all available LLMs and their capabilities can be found in the GWDG documentation. Please note that the use of the GWDG models is limited. If you encounter the error "429 Rate limit reached", the quota has already been used. You can try again after one hour.

 

In the following, we will show you how to integrate different providers. If you have access to LLMs but do not know how to integrate the models with bwJupyter, please send us a brief email to support∂bwjupyter.de.

 

  • Some universities run OpenWebUI. OpenWebUI provides a web interface similar to ChatGPT. Administrators can configure several LLMs, which can then be addressed via the chat. To be able to use the models via OpenWebUI in bwJupyter, you need an API key and the model ID. The API key can be generated via "Settings >> Account >> API key". The model ID is displayed when you select a model by dragging the cursor over it.
    The model ID is displayed in brackets after the model.

    The information for the "OpenAI (general interface) :: *" model can then be entered as follows:
    The API key must be entered under "API Keys".