Josh Tobin

@josh_tobin_

Former research scientist , AI PhD , investor , consultant .

San Francisco, CA
Vrijeme pridruživanja: srpanj 2011.

Tweetovi

Blokirali ste korisnika/cu @josh_tobin_

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @josh_tobin_

  1. Prikvačeni tweet
    12. stu 2019.

    1/ Excited to share our NeurIPS paper, Geometry-Aware Neural Rendering! . With , , and the rest of the robotics team

    Prikaži ovu nit
    Poništi
  2. proslijedio/la je Tweet
    29. sij

    Couldn’t be more excited about what we are building! Yet, even more excited about team, all brilliant and exceptional at working across team boundaries, engineering, research, business, we are all having fun and getting it done together!

    Poništi
  3. proslijedio/la je Tweet
    25. sij

    [VIDEO] for my guest lecture at is now available. "Lessons learned from building practical DL systems"

    Prikaži ovu nit
    Poništi
  4. proslijedio/la je Tweet
    16. sij

    To train bug-free deep learning models, we need to treat building them as an iterative process. In this post, I'd like to share a mental recipe to troubleshoot deep neural networks, as codified by at

    Prikaži ovu nit
    Poništi
  5. proslijedio/la je Tweet
    10. sij

    Workera recently published a report on AI career pathways. It doesn't mention hardware. I also don't see the difference b/w SWE-ML & ML Engineer. But it highlights some important distinctions. I also like talk on the structure of AI teams

    Poništi
  6. proslijedio/la je Tweet
    5. sij

    Data science has never been as much about ML as it has about cleaning, shaping, and moving data from place to place. In this post, I shared 5 data management concepts to ensure data quality for your models, as a courtesy of at

    Poništi
  7. proslijedio/la je Tweet
    16. pro 2019.

    Are We Making Real Progress in Simulated Environments? Yes — if the simulator is optimized. It turns out correlation between simulation and reality can be quantified and leveraged to identify/combat ways agents cheat in simulation.

    Prikaži ovu nit
    Poništi
  8. proslijedio/la je Tweet
    24. pro 2019.

    Every day, hundreds of Cruise engineers interact with data collected on the road and in simulation by our AVs. In this presentation, our AV Tools team dives into two of the open-source tools they built to visualize & explore this data: Webviz & Worldview.

    Poništi
  9. proslijedio/la je Tweet
    19. pro 2019.

    The craft of running Machine Learning teams effectively is not yet matured. In this blog post, I'd like to share key insights on how to do this productively, courtesy of at

    Poništi
  10. 12. pro 2019.

    Slides from my talk at NeurIPS yesterday. Geometry-Aware Neural Rendering: improving neural rendering by incorporating network primitives inspired by classical 3D computer vision

    Poništi
  11. 10. pro 2019.

    Ah, NeurIPS, our one chance a year to catch up with all of our friends from the Bay Area we’re too busy to see in real life

    Poništi
  12. 10. pro 2019.

    On my way to Vancouver for NeurIPS! Hope to see many of you there. DM me if you want to meet up, and come see my talk about “Geometry-Aware Neural Rendering” on Wednesday at 4:25 in west ballrooms A+B.

    Poništi
  13. proslijedio/la je Tweet
    4. pro 2019.

    The portion of ML code in a real-world ML system is a lot smaller than the infrastructure needed for its support. This blog post shares 7 questions that you should ask to handle ML technical debt, courtesy of at

    Prikaži ovu nit
    Poništi
  14. proslijedio/la je Tweet
    25. stu 2019.

    85% of machine learning projects ultimately fail to deliver on their intended promises to business. In this blog post, I would like to share the 5 steps that you can use to set up your projects up for success, courtesy of at

    Poništi
  15. 24. stu 2019.

    What are the best resources on avoiding or reducing bias in ML systems?

    Poništi
  16. proslijedio/la je Tweet
    20. stu 2019.

    Only course that stacks up well against mettle of free Deep learning course by is Full Stack Deep Learning by , , . In-fact FSDL is extremely well suited complement to 's course. 👍

    Poništi
  17. 20. stu 2019.

    Honest question: why do people write on medium? Isn’t the whole point of blogging to get your point of view out there? Paywalling your content limits its reach

    Poništi
  18. 19. stu 2019.

    The guide mostly focuses on domain randomization and its extensions, but I touch on simulator design, system ID, and domain adaptation as well. Hope you enjoy!

    Prikaži ovu nit
    Poništi
  19. 19. stu 2019.

    I presented it earlier today in 's Advanced Robotics class ()

    Prikaži ovu nit
    Poništi
  20. 19. stu 2019.

    I wrote a guide on transferring neural networks from simulation to the real world:

    Prikaži ovu nit
    Poništi
  21. proslijedio/la je Tweet

    This weekend 120+ practitioners joined us from 🇯🇵/🇫🇷/🇺🇾/🇧🇷/🇲🇽/🇬🇧 and around the for a weekend of nonstop lectures, hands-on deployment-level labs, and illuminating lectures by guest speakers Franziska Bell.

    Prikaži ovu nit
    Poništi

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:

    ·