Check out my latest article: Constrained and Provable LLM Code Generation linkedin.com/pulse/constrai… via LinkedIn

There's a new bill, SB-1047 "Safe and Secure Innovation for Frontier Artificial Intelligence Models Act". I think it could do a great deal of harm to startups, American innovation, open source, and safety. So I've written a response to the authors: 🧵 answer.ai/posts/2024-04-…

F# example to serve ONNX vision models (e.g. phi-3 vision) for online and batch modes. Supports multiple parallel request if GPU hardware allows. github.com/fwaris/OnnxVis…

Check out my latest article: Phi-3 Vision is a Surprisingly Useful Gem linkedin.com/pulse/phi-3-vi… via LinkedIn

Need of the hour: Parallel but self-throttled LLM API calls to stay below allowed capacity. #fsharp #asyncseq extensions github.com/fwaris/OnnxVis…

New look for F# FsOpenAI rag/chat app. Test link below (need to use own OpenAI key) fsopenaiserver1.azurewebsites.net Now with Radzen Blazor components Source: (github.com)

Revisiting Logic Puzzle with "o1" linkedin.com/pulse/revisiti… via LinkedIn

OpenAI's o1 seems like a pretty capable F# code generator. See this post for more. linkedin.com/posts/activity…

Experimental 'chat with code evaluator' feature added to FsOpenAI to generate and evaluate F# code interactively. …zdthndaey.eastus-01.azurewebsites.net (need own OpenAI key) Evaluation sandbox utilizes typed AST to only allow types from set namespaces.

Pretty impressive performance by o3 #OpenAI12Days. Still, o3 can consume $1K+ worth of electricity on complex tasks (@ megawatts). Human brains are in @ 25 watts range, so machines have some way to go, yet. However, with Googles quantum leap, the gap may be closing fast.

F# (windows) service and GUI client (Avalonia, FuncUI) for transcription and diarization. linkedin.com/posts/activity…

I'm pleased to announce OxCaml! OxCaml is Jane Street's branch of OCaml. We've given it a new name and a snazzy logo, and done a bunch of work to make it easy for people to try.

Automated Prompt Tuning with FsGepa (F#) linkedin.com/pulse/automate… via LinkedIn

In the next lab from my Tiny Systems course, you learn how type inference in ML-like languages works and write your own Hindley-Milner inference algorithm! 🎞️ Watch the videos: youtube.com/watch?v=FUFQ4j… ⌨️ Get the #fsharp code: github.com/tpetricek/tiny…

Biology doesn’t compute the way our current AI systems assume. Neurons aren’t static units. They operate across multiple scales, electrical, mechanical, structural, and quantum-timing, all interacting in real time. Recent neuroscience points to: • fractal dendritic integration

OpenAI Codex (preview) is the best code gen model I have used thus for F# code gen. (Sonnet 4.5 is not at the same level). Did a flawless job of creating Fabulous-Maui bindings for a new Maui controls lib. See: github.com/fabulous-dev/F…

GenAI meets old school Symbolic AI linkedin.com/posts/activity…