This article has a bunch of collaterals - first of all, Linfa, the ML framework I am planning to work in 2020: https://crates.io/crates/linfa/all …
-
-
Prikaži ovu nit
-
Its Python bindings, currently available for Linux (macOs and Windows coming as soon as I figure out a good CI):https://github.com/LukeMathWalker/linfa-python …
Prikaži ovu nit -
The benchmarking code, with examples of how to wire up a basic gRPC service in Rust using Tonic (awesome work
@lucio_d_franco):https://github.com/LukeMathWalker/clustering-benchmarks …Prikaži ovu nit
Kraj razgovora
Novi razgovor -
-
-
Awesome article! Would it be reasonable to try to serve scikit-learn predictions from a Rust server? Or would that not work? Seriously considering picking up some Rust!
-
Yes! You could either load the Python model directly and call it from Rust or implement the inference side of scikit models in Rust and deserialise Pickle files to the Rust version. The second option gives you best results, but you are tied to scikit internals.
- Još 3 druga odgovora
Novi razgovor -
-
-
It was great to work with you
@algo_luca ! I will look at Linfa’s roadmap and see if I can help. Also I hope your next blog post will use vega-lite-3 rust crate for the plots :D -
I hope it as well!
Kraj razgovora
Novi razgovor -
-
-
Very nice work! As a side note, for KMeans it's also that the current scikit-learn implementation is a bit suboptimal and will be significantly improved once https://github.com/scikit-learn/scikit-learn/pull/11950 … is merged.
-
That's quite an interesting PR - thanks for linking it :)
Kraj razgovora
Novi razgovor -
-
-
I'm waiting for the time when Rust replaces Python in the so-called frontend too. Number crunching performance isn't even the primary reason for that. It's ease of writing correct code. It's *extremely* hard to write correct code in an untyped language.
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.