Eric Jang   

@ericjang11

Research scientist at Robotics @ Google. Formerly at Pixar Animation Studios and Two Sigma Investments. Brown '16. I love science. All opinions are my own.

Vrijeme pridruživanja: siječanj 2014.

Medijski sadržaj

  1. having a good ergo setup is a good long-term investment. The office furniture folks at kindly sent me one of their ErgoChairs () to review. Pros: nice breathable back, all the adjustable options of a pro office chair.

    Prikaži ovu nit
  2. I propose a grand challenge for dextrous robotic manipulation: cook eggs in all the ways shown in this video

    Prikaži ovu nit
  3. Odgovor korisnicima
  4. A single bad semantic triple can lead to an entire ontological web collapsing. Reminds me of this quote from Alien, Covenant: "When one note is off, it eventually destroys the whole symphony, David."

    Prikaži ovu nit
  5. That design choice alone might be fatal. HN user wrnr provides a brilliant example of the surprising complexity and uncertainty underlying the simplest of Semantic triples

    Prikaži ovu nit
  6. Edit: Thanks to Stephan Hoyer, who pointed out that my benchmark only ran on 1 TPU core, whereas a Cloud TPU has 8 available. Updated post, chart and code to reflect that using the whole TPU does speed things up. My implementation could do a better job of parallelizing though.

    Prikaži ovu nit
  7. Anything that can be implemented in JAX *will* be implemented in JAX. Here's a differentiable path tracer (and a tutorial!) Blog Post: Code:

    Prikaži ovu nit
  8. FB Reality team should have compressed the DeepFovea presentation video with DeepFovea for maximum wow factor. Seriously though, this is awesome video compression work!

  9. Cool single-actuator hand mechanism I saw at

  10. Self-training with Noisy Student: A semi-supervised approach by Google/CMU that outperforms Facebook's "weakly labeled 3.5B Instagram" method on ImageNet.

  11. Prikaži ovu nit
  12. Blog post on a "financial vulnerability" in 's margin lending system.

  13. Rendering & sim with spatial sparsity spends 1% time during the actual relevant computation and 99% doing data structure overhead (hash lookup, cache misses, branching). The Taichi Programming Language is an elegant (and fast!) solution! [SIGGRAPH 2019]

    Prikaži ovu nit
  14. Odgovor korisniku/ci

    's "self-supervised / predictive learning talk"

  15. JuliaRobotics: Making robots walk with Julia | Robin Deits via

  16. Move aside, Safe For Work Sasuke

  17. Antony Starr was BORN to play Homelander

  18. Prikaži ovu nit
  19. If you've ever wondered what "bits-per-pixel" actually means, why logistic distributions are awesome, how to improve training stability of normalizing flows and more... Check out this blog post on training likelihood models

    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:

    ·