You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A library that provides high level pythonic abstractions like typed functions dynamically compiling grammars for return pydantic structs (funcchain) that would also allow you to drop down to customize generation (outlines )either within or around these functions would be amazing. High level mypy Typed boundaries around outlines programs.
Analogy would be Pytorch:triton::funcchain/equivalent:outlines
Aside from the funcchain feature, for my use case I'd love to see:
Custom unpacking of pydantic structs: Looping/ programmatically accessing fields into prompts
Customizing generation of pydantic output structs
Mixing and matching regular python types and pydantic inputs and outputs
Our baseline MVP will be using OpenAI models initially. For this to be computationally feasible, I think we'd need function calling, which seems to be planned?
This discussion was converted from issue #528 on January 18, 2024 22:17.
Heading
Bold
Italic
Quote
Code
Link
Numbered list
Unordered list
Task list
Attach files
Mention
Reference
Menu
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Opening per @remic33 in https://twitter.com/lateinteraction/status/1736804963760976092?t=5Pypu6f6AWPljWiWUbWM8g&s=19
I really like outlines. I also really like https://github.com/shroominic/funcchain
A library that provides high level pythonic abstractions like typed functions dynamically compiling grammars for return pydantic structs (funcchain) that would also allow you to drop down to customize generation (outlines )either within or around these functions would be amazing. High level mypy Typed boundaries around outlines programs.
Analogy would be Pytorch:triton::funcchain/equivalent:outlines
Aside from the funcchain feature, for my use case I'd love to see:
Our baseline MVP will be using OpenAI models initially. For this to be computationally feasible, I think we'd need function calling, which seems to be planned?
Related to: #508
Beta Was this translation helpful? Give feedback.
All reactions