Very interesting. How does this approach work for complex agentic workflows where the LLM is expected to orchestrate across multiple tools (such as when using MCP)? Or is this mainly for simple cases like the ones presented in the blog post?
re: different tools (apis vs mcps). in my mind, there should be no real difference at what kind of tools is called at this moment since I model this as a softmax over a label set of tools.
that said, an idea I want to investigate is whether tools can live in a learned embedding space, where selection isn’t a softmax over discrete labels but a nearest-neighbor or attention mechanism over continuous vectors.
this is the intuition I'm developing as we speak and in some of my other comments on this thread (see differentiable state machine comment).
The work described appears as if it would handle a complex set of multiple tools just fine, but you do train the controller on a specific tool set, so you would presumably need to train (or at least something like "fine tune") a controller for each toolset you wanted to use.
for sure, there's a way here where I think we ought to be able to learn multiple tool calls and prompts together with real world data. investigating that next.
"Professor Vitaly Melnikov, 77, who had headed the Department of Rocket and Space Systems at RSC Energia, Moscow's leading spacecraft manufacturer, had been battling a sudden and grave illness before his death. The source of his poisoning was inedible mushrooms, Moskovsky Komsomolets, a Moscow newspaper reported."
chessprogramming.org is a treasure trove of knowledge on building chess engines. With its help, I wrote one in C++ a few years ago that got quite good (2100+ rating on FICS but that's nowhere close to the likes of Stockfish). In fact, writing a reasonably strong chess engine is straightforward (and incredibly fun) but at the top end of strength, there's immense depth, and after a point making improvements gets increasingly resource intensive (tuning params, running experiments to verify strength gain all takes a lot of compute).
Chess programming is also extremely addictive. On forums like talkchess.com, you see folks hanging out who have been doing it for decades (most of them are also super helpful to newbies).
> running experiments to verify strength gain all takes a lot of compute
Fun fact, in my master thesis I proved that quantum computers can verify this using quadratically fewer iterations than on a classical computer. That is, if it takes a classical computer c*n iterations to say with 99% certainty that agent A is stronger than agent B, a quantum computer can do it in d*sqrt(n) iterations, where c and d are agent-independent constants (obviously n is not agent-independent as two closely matched agents are harder to distinguish than a steamroll).
The number of qubits needed put this into the far future of quantum computing, but it's neat nonetheless.
Just completed part 1 (used Python instead of Java to spice it up a bit) and it was a great experience! Looking forward to working through part 2, but it will be a while before I can get around to it. I hope more books follow this engaging writing pattern (when it makes sense). Kudos to Bob Nystrom for this great book.
And yet there are hardly any software written in Haskell or Ocaml that are widely used or have any notable positive impact on the modern digital world, compared to those written in languages like Go, C++ or Python.
edit: ps: Big fan of OCaml, but have since moved on to Go and Python for getting things done in the real world.
OCaml is mostly used in speciality applications. I have a few friends that work on flight control software for satellites and their stuff is all OCaml which is apparently not that uncommon in their field.
That said none of them knew OCaml before joining said company and thus none were hired because of it... so that does somewhat invalidate the "right" people argument but I guess if you are doing literal rocket science I think that is pretty self-selecting for smart folk lol.
> And yet there are hardly any software written in Haskell or Ocaml that are widely used or have any notable positive impact on the modern digital world,
It's used in supply chain management at target.
IIRC at Starbucks too.
Oh, and ever use Facebook chat? Haskell is running over your messages to filter out spam.
