jupyter-ai

jupyter-ai copied to clipboard

Add test handlers, option to choose mock language model for testing

1

### Problem Currently to test something involving interaction with language model or to test chat UI, real API key or locally hosted language model is required. This is not optimal...

andrii-i

## Description In the chat settings panel, when "SageMaker endpoint :: *" is selected as the language model, the default region name appears with "Region name (required)" superimposed on it....

JasonWeill

Make the default chat window memory size configurable

2

### Problem Chat conversation memory window of 2 is a bit small especially as people are used to long memories from their experiences with ChatGPT. ### Proposed Solution * Make...

michaelchia

Unable to use ChatVertexAI as provider due to default chat's ChatPromptTemplate

5

### Problem * I am unable to use ChatVertexAI as a Provider for chat as the final message ChatPromptTemplate is an AIMessage. https://github.com/jupyterlab/jupyter-ai/blob/0f619061430be6320bf9b87c63a1b939597998f9/packages/jupyter-ai/jupyter_ai/chat_handlers/default.py#L35-L52 * Langchain's ChatVertexAI requires that the last...

michaelchia

Generate command can create sections with no code cells (no retry logic)

3

## Description The `/generate` command in the chat UI can generate a notebook with no code cells, or with empty code cells where code was expected. You can see this...

JasonWeill

Add meaningful errors in the chat interface

1

## Summary If an error occurs during a chat session for any reason, an error related to the indexing is always returned to the user, along with the exception stack...

3coins

### Problem When I send a prompt to a SageMaker endpoint `jumpstart-dft-hf-translation-t5-small`, which expects plain text, not JSON, as input, I get an error (see below). This is because Jupyter...

JasonWeill

### Problem Each LLM provider and model has limits on the input size of prompts. When a user asks questions about indexed docs with `/ask`, the chunk size used to...

3coins

Folders can be learned with /learn multiple times

4

## Summary If a directory has been already indexed with a previous `/learn` command, and is submitted for learning again, this proceeds without any warning. However, when the `/ask` command...

3coins

Update prompts for locally installed models

6

### Problem Current prompts in Jupyter AI work well with remote providers, but are not optimized for locally installed models provided by GPT4All. See the discussion on #190 to some...

3coins