Luca Palmieri  

@algo_luca

Mathematician by training, ML by trade, Rustacean by night. Currently at . He/him. Opinions are my own.

Vrijeme pridruživanja: srpanj 2017.

Medijski sadržaj

  1. 2. velj
    Odgovor korisniku/ci

    Stealing some of youra for my February iteration on the habit tracker.

  2. 2. pro 2019.

    Finally finished the write-up of our ML experiments from - a 25x speedup is way more than we were expecting! Beyond the raw numbers, a long read on Python, , the scientific computing space and its outlook in the next few years.

    Prikaži ovu nit
  3. 24. stu 2019.

    Small spoiler from an upcoming blog post about for .

  4. 10. stu 2019.

    Ndarray workshop at , day 2! Working on getting our K-means working end to end!

  5. 9. stu 2019.

    A full house for the Ndarray workshop at !

  6. 8. stu 2019.

    Getting a solution branch ready - really appreciating `include!` macro expansion from : getting all IDE features on those files is super useful.

  7. 13. lis 2019.
    Odgovor korisnicima

    You bring those perspectives back in your non-FP code unconsciously, so it's definitely worth doing it. I should spend some quality time with Haskell as well, just too much going on already to pick a project where I can actually try it out. Even Reddit is pushing me...

  8. 22. ruj 2019.

    My current look.

  9. 7. ruj 2019.

    Cargo is magical: compiling projects easily transforms my laptop into a radiator. Thankfully winter is coming.

  10. 27. kol 2019.
    Odgovor korisnicima

    And the future is coming from this method

      pub fn log(
      &self,
      r: AsyncRecord,
  ) -> impl Future<Output = Result<PutRecordOutput, RusotoError<PutRecordError>>> {
      // [...] Irrilevant stuff
      self.client.put_record(firehose_record).compat()
  }
  11. 27. kol 2019.
    Odgovor korisniku/ci

    Mmmh, what do you mean? This is a stripped down version of what I am trying right now:

    fn spawn_thread(self, drain: FirehoseDrain) -> (thread::JoinHandle<()>, Sender<AsyncMsg>) {
    let (tx, rx) = crossbeam_channel::bounded(self.chan_size);
    let mut builder = thread::Builder::new();
    let join = builder
        .spawn(move || async_std::task::block_on(consume_log_channel(rx, drain)))
        .unwrap();

    (join, tx)
}

async fn consume_log_channel<C>(rx: Receiver<AsyncMsg>, drain: FirehoseDr...
  12. 16. srp 2019.

    Not too bad I'd say!

  13. 16. lip 2019.
    Odgovor korisniku/ci
  14. 22. svi 2019.

    Keeping this in mind when building the story for is going to be key for success.

  15. 11. svi 2019.

    The system itself is probably nowhere near the complexity of what is running, but I can relate to what says in this talk, from a couple of years ago: [/2]

    Prikaži ovu nit
  16. 20. ožu 2019.
    Odgovor korisniku/ci
  17. 16. ruj 2018.

    The video is here:

    Prikaži ovu nit
  18. 2. ruj 2018.

    I am quite overwhelmed: I wasn't expecting this reaction to the launch of my blog series on Reinforcement Learning! It gives me lot of motivation to keep pushing forward!

    Prikaži ovu nit

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.

    Možda bi vam se svidjelo i ovo:

    ·