v1.13.0

1.13.0

This release uses the new LangChain partner packages for Anthropic and OpenAI. This allows Jupyter AI to support additional Anthropic and OpenAI models that are not available from the langchain package alone.

Existing users upgrading to this version will have to take additional action, outlined below.

• If you are using Anthropic models: please install the langchain-anthropic package.
• If you are using OpenAI models: please install the langchain-openai package.
• Otherwise, no action is needed.

(Full Changelog)

Enhancements made

• Improve support for custom providers #713 (@dlqqq)
• Update Anthropic providers to use langchain_anthropic partner package #700 (@dlqqq)
• Use AZURE_OPENAI_API_KEY for Azure OpenAI provider #691 (@aroffe99)
• [1.x] Add Claude-3-Haiku (#696) #697 (@srdas)
• /export added #658 (@apurvakhatri)

Bugs fixed

• Fix rendering of model IDs with a colon in their name #704 (@dlqqq)
• Update Anthropic providers to use langchain_anthropic partner package #700 (@dlqqq)
• Use new langchain-openai partner package #653 (@startakovsky)

Contributors to this release

(GitHub contributors page for this release)

@dlqqq | @lumberbot-app | @meeseeksmachine | @srdas | @welcome

v2.12.0

2.12.0

(Full Changelog)

Enhancements made

• Add Anthropic Claude 3 models to providers #672 (@srdas)
• Add support for Gemini #666 (@giswqs)
• %ai version added #665 (@apurvakhatri)
• Together.ai provider added #654 (@MahdiDavari)

Bugs fixed

• Fix selecting models with a colon in their ID #682 (@dlqqq)
• Use regex in TeX replace function to catch repeating symbol occurrences #675 (@andrii-i)
• Resolves chat panel initialization error #660 (@abbott)
• fix bug: check before using the variables #656 (@ya0guang)

Documentation improvements

• Update nodejs installation #679 (@giswqs)

Contributors to this release

(GitHub contributors page for this release)

@abbott | @andrii-i | @apurvakhatri | @dlqqq | @giswqs | @lumberbot-app | @MahdiDavari | @srdas | @welcome | @ya0guang

v1.12.0

1.12.0

(Full Changelog)

Enhancements made

• Add support for Gemini #666 (@giswqs)
• %ai version added #665 (@apurvakhatri)
• Together.ai provider added #654 (@MahdiDavari)

Bugs fixed

• Fix selecting models with a colon in their ID #682 (@dlqqq)
• Use regex in TeX replace function to catch repeating symbol occurrences #675 (@andrii-i)

Contributors to this release

(GitHub contributors page for this release)

@dlqqq | @giswqs | @lumberbot-app | @meeseeksmachine | @welcome

v2.11.0

2.11.0

This release notably includes a significant UI improvement for the chat side panel. The chat UI now uses the native JupyterLab frontend to render Markdown, code blocks, and TeX markup instead of a third party package. Thank you to @andrii-i for building this feature!

(Full Changelog)

Enhancements made

• Fix cookiecutter template #637 (@dlqqq)
• Add OpenAI text-embedding-3-small, -large models #628 (@JasonWeill)
• Add new OpenAI models #625 (@EduardDurech)
• Use @jupyterlab/rendermime for in-chat markdown rendering #564 (@andrii-i)

Bugs fixed

• Unifies parameters to instantiate llm while incorporating model params #632 (@JasonWeill)

Documentation improvements

• Add nodejs=20 to the contributing docs #645 (@jtpio)
• Update docs to mention langchain_community.llms #642 (@jtpio)
• Fix cookiecutter template #637 (@dlqqq)
• Fix conda-forge typo in readme #626 (@droumis)

Contributors to this release

(GitHub contributors page for this release)

@andrii-i | @dlqqq | @droumis | @EduardDurech | @JasonWeill | @jtpio | @krassowski | @lalanikarim | @lumberbot-app | @welcome | @Wzixiao

v1.11.0

1.11.0

(Full Changelog)

Enhancements made

• Fix cookiecutter template #637 (@dlqqq)
• Add new OpenAI models #625 (@EduardDurech)
• [1.x] Use @jupyterlab/rendermime for in-chat markdown rendering #631 (@andrii-i)
• Add OpenAI text-embedding-3-small, -large models #628 (@JasonWeill)

Bugs fixed

• Unifies parameters to instantiate llm while incorporating model params #632 (@JasonWeill)

Documentation improvements

• Update docs to mention langchain_community.llms #642 (@jtpio)
• Fix cookiecutter template #637 (@dlqqq)
• Fix conda-forge typo in readme #626 (@droumis)

Contributors to this release

(GitHub contributors page for this release)

@andrii-i | @dlqqq | @JasonWeill | @krassowski | @lalanikarim | @lumberbot-app | @meeseeksmachine | @welcome | @Wzixiao

v2.10.0

2.10.0

This is the first public release of Jupyter AI inline completion, initially developed by @krassowski. Inline completion requires jupyterlab==4.1.0 to work, so make sure you have that installed if you want to try it out! 🎉

(Full Changelog)

Enhancements made

• Bump @jupyterlab/completer resolution to ^4.1.0 #621 (@dlqqq)
• Restyles model names in Markdown to avoid wrapping model names #606 (@JasonWeill)
• Expose templates for customisation in providers #581 (@krassowski)
• Add nvidia provider #579 (@stevie-35)
• Reflect theme changes without a refresh #575 (@garsonbyte)
• Setting default model providers #421 (@aws-khatria)

Bugs fixed

• Allow usage without NVIDIA partner package #622 (@dlqqq)
• fix to conda install instructions in readme #610 (@Tom-A-Lynch)
• Uses invoke() to call custom chains. Handles dict output format. #600 (@JasonWeill)
• Removes deprecated models, adds updated models for openai #596 (@JasonWeill)
• Upgrades cohere dependency, model list #594 (@JasonWeill)

Documentation improvements

• Mentions conda install instructions in docs #611 (@JasonWeill)
• Add Kaggle to supported platforms #577 (@adriens)

Other merged PRs

• Correction to default models configuration. #605 (@3coins)

Contributors to this release

(GitHub contributors page for this release)

@3coins | @adriens | @aws-khatria | @dlqqq | @garsonbyte | @JasonWeill | @krassowski | @lumberbot-app | @stevie-35 | @Tom-A-Lynch | @welcome

v1.10.0

1.10.0

(Full Changelog)

Enhancements made

• Restyles model names in Markdown to avoid wrapping model names #606 (@JasonWeill)
• Add nvidia provider #579 (@stevie-35)
• Setting default model providers #421 (@aws-khatria)
• Expose templates for customisation in providers #581 (@krassowski)
• Backport PR #575: Reflect theme changes without a refresh #599 (@dlqqq)

Bugs fixed

• Allow usage without NVIDIA partner package #622 (@dlqqq)
• Backport PR #610: fix to conda install instructions in readme #612 (@JasonWeill)
• Uses invoke() to call custom chains. Handles dict output format. #600 (@JasonWeill)
• Removes deprecated models, adds updated models for openai #596 (@JasonWeill)
• Upgrades cohere dependency, model list #594 (@JasonWeill)

Documentation improvements

• Mentions conda install instructions in docs #611 (@JasonWeill)
• Backport PR #610: fix to conda install instructions in readme #612 (@JasonWeill)
• Correction to default models configuration. #605 (@3coins)

Contributors to this release

(GitHub contributors page for this release)

@3coins | @dlqqq | @JasonWeill | @jtpio | @krassowski | @lumberbot-app | @meeseeksmachine | @stevie-35 | @welcome

v2.10.0-beta.1

2.10.0-beta.1

(Full Changelog)

Bugs fixed

• Fix streaming, add minimal tests #592 (@krassowski)

Contributors to this release

(GitHub contributors page for this release)

@krassowski

v2.10.0-beta.0

2.10.0-beta.0

(Full Changelog)

Enhancements made

• Inline completion support #582 (@krassowski)

Contributors to this release

(GitHub contributors page for this release)

@dlqqq | @jtpio | @krassowski

v2.9.1

2.9.1

(Full Changelog)

Enhancements made

• LangChain v0.1.0 #572 (@dlqqq)

Bugs fixed

• Update Cohere model IDs #584 (@dlqqq)

Contributors to this release

(GitHub contributors page for this release)

@dlqqq | @JasonWeill | @jtpio | @welcome