shroominic/funcchain
⛓️ build cognitive systems, pythonic
13 Releases
Latest: 1y ago
⛓️v0.3.6 - langchain v0.3 support and typing fixesv0.3.6Latest
shroominic·1y ago·November 19, 2024📋 Changes
- langchain v0.3 support
- typing fixes
⛓️v0.3.4 - azure, groq and fixesv0.3.4
shroominic·2y ago·June 15, 2024📋 Changes
- support for azure openai
- support for groq json mode
- multiple small bug fixes
- added new openai model identifiers (gpt-4o and turbo)
- simplifying config files
- update deps
⛓️ v0.3.3 - fix jinja format instructionsv0.3.3
shroominic·2y ago·April 20, 2024📋 Changes
- fix format instructions when using jinja
⛓️ v0.3.2 - claude3 supportv0.3.2
shroominic·2y ago·March 10, 2024⛓️v0.3.0 - funcchain dependenciesv0.3.0
shroominic·2y ago·February 15, 2024Introducing a dependency injection system like you know from FastAPI!
⛓️ v0.2.3v0.2.3
shroominic·2y ago·February 15, 2024📋 Changes
- fix .env loading with settings
- fix input dict type
- fix prefix for model selection
- fix support for new openai models
- fix default settings.llm handling
- improved README shields
⛓️ v0.2.2v0.2.2
shroominic·2y ago·February 12, 2024📋 Changes
- support for primitive dict types
- improved local model info
- fix in output type selector
- fix .env loading
- fix langchain deprecation warning
⛓️v0.2.1 - fix fix + small improvementsv0.2.1
shroominic·2y ago·February 11, 2024📋 Changes
- fix primitive type parsing when using openai
- Fix typo strem_to -> stream_to by @ahuang11 in https://github.com/shroominic/funcchain/pull/17
- fix README.md by @shroominic in https://github.com/shroominic/funcchain/pull/25
- fix pop model_name by @ahuang11 in https://github.com/shroominic/funcchain/pull/24
- add `ollama_base_url` setting by @jquesnelle in https://github.com/shroominic/funcchain/pull/21
- add system prompt to chain by @ahuang11 in https://github.com/shroominic/funcchain/pull/20
- improve chat handlers
✨ New Contributors
- @ahuang11 made their first contribution in https://github.com/shroominic/funcchain/pull/17
- @jquesnelle made their first contribution in https://github.com/shroominic/funcchain/pull/21
- Full Changelog: https://github.com/shroominic/funcchain/compare/v0.2.0...v0.2.1
⛓️v0.2.0 - new features + docsv0.2.0
shroominic·2y ago·February 2, 2024NEW DOCUMENTATION: https://shroominic.github.io/funcchain/ IMPROVEMENTS: 🔗 langchain lcel compatibility (runnable decorator) 🔍 primitive type parsers 🔄 dynamic chat routers 📄 full jinja2 templating 🚦 enums, literal output types 🦙 llamacpp + ollama support 📡 openai json mode streaming 🔁 auto retry for pydantic validations 👁️ better vision support ✍️ create custom parsing types 🏗️ codebase restructuring ⬆️ updated deps **Full Changelog**: https://github.com/shroominic/funcchain/compare/v0.1.10...v0.2.0
⛓️ v0.2.0-alpha.1v0.2.0-alpha.1Pre-release
shroominic·2y ago·December 15, 2023📋 Changes
- settings namespace changes to lowercase (breaking)
- easy local model support
- function input arguments ignore everything not string
- more tests
- auto gguf model download from huggingface
- dependency split into
- core `pip install funcchain`
- local `pip install "funcchain[local]"`
- + 3 more
⛓️ v0.1.10 - Universal Model Loaderv0.1.10
shroominic·2y ago·December 12, 2023📋 Changes
- universal model loader
- SettingOverrides
- Auto GGUF Download
- select optional deps
- improve examples
- other small improvements
⬆️ v0.1.9 - Local Model LlamaCpp Supportv0.1.9
shroominic·2y ago·December 5, 2023⛓️ v0.1.8 - pydantic v2 + morev0.1.8
shroominic·2y ago·November 26, 2023📋 Changes
- all BaseModels are not Pydantic V2
- pydantic field validators now supported with retry when parsing fails
- retry with context of the error