Jeff Deanযাচাইকৃত অ্যাকাউন্ট

@JeffDean

Senior Fellow, Google AI. Google Brain team lead and co-founder. Co-designer/implementor of software systems like , MapReduce, BigTable, Spanner, ..

যোগদান করেছেন সেপ্টেম্বর ২০১৭

মিডিয়া

  1. ১৪ ঘন্টা আগে

    This illustrates one aspect discussed, a clever way of exploring many possible layouts of photos to find one that looks good: "Finding the optimum photo grid (ie the one with the lowest combined set of bad rows) is as simple as calculating the shortest path through the graph."

    এই থ্রেডটি দেখান
  2. ১৯ জুন

    Google Cloud TPUs now offer preemptible pricing at ~70% off the reserved instance pricing. This means, for example, that you can train a ResNet-50 model for ~$7.50 instead of $25, or a Transformer neural translation model for ~$13 instead of $41. See:

    এই থ্রেডটি দেখান
  3. ৯ মে

    If you want to learn more about why we're so excited about TPUs and how they make machine learning research and applications so much more efficient, I highly recommend watching Zak Stone's excellent talk about TPUs from today's Google I/O session on TPUs:

  4. ১৭ এপ্রিল

    We just posted new DAWNBench results for ImageNet classification training time and cost using Google Cloud TPUs+AmoebaNet (architecture learned via evolutionary search). You can train a model to 93% top-5 accuracy in <7.5 hours for <$50. Results:

    এই থ্রেডটি দেখান
  5. ১১ এপ্রিল

    I enjoyed chatting with at the TensorFlow Developer Summit (although I'll now chide myself for not sitting up straight).

  6. ৩০ মার্চ
    -কে উত্তর দিচ্ছেন

    I'll share pictures from my own experience from a few years ago, which was not quite as amazing as this video (ours was on top of the vehicle and nearby vehicles, but an amazing experience nonetheless) you might like the video in the parent tweet

  7. ২৭ মার্চ

    Stanford's DAWNBench is a new benchmark suite measuring a variety of deep learning training and inference tasks. Google Cloud TPU results are now up in the Imagenet category and are #1 in both ImageNet Training Time and ImageNet Training Cost.

    এই থ্রেডটি দেখান
  8. ১২ ফেব

    1/ TPUs are now in Beta for people that want access to high speed accelerators for training machine learning models. Details in blog post at:

    এই থ্রেডটি দেখান
  9. ২০ জানু
    -কে উত্তর দিচ্ছেন

    Agreed. The underlying neural architecture search models are actually better than those by world class ml/computer vision researchers, though. Black dots vs. red dots in this picture (Figure 5 from )

  10. ১৬ ডিসেম্বর, ২০১৭

    This article on Feature Visualization by Chris Olah (), Alexander Mordvintsev (), and Ludwig Schubert () is excellent for understanding what's going inside computer vision machine learning models.

  11. ১৪ ডিসেম্বর, ২০১৭

    Congrats to Google Brain team member Chris Shallue and his collaborator Andrew Vandenburg for their work on using machine learning to find lots of new exoplanets! Blog: Paper: AMA:

  12. ৯ ডিসেম্বর, ২০১৭

    enjoyed my cheetah photos from 2012 when I chatted with her at the workshop dinner. Others might like them as well..

লোড হতে বেশ কিছুক্ষণ সময় নিচ্ছে।

টুইটার তার ক্ষমতার বাইরে চলে গেছে বা কোনো সাময়িক সমস্যার সম্মুখীন হয়েছে আবার চেষ্টা করুন বা আরও তথ্যের জন্য টুইটারের স্থিতি দেখুন।

    আপনিও পছন্দ করতে পারেন

    ·