Machine Learning Tokyo

@__MLT__

Award-winning nonprofit dedicated to democratizing Machine Learning. Open Source, Science & Education. Tweets by & 🤖🧠

Tokyo-to, Japan
Vrijeme pridruživanja: kolovoz 2018.

Tweetovi

Blokirali ste korisnika/cu @__MLT__

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

  1. Prikvačeni tweet
    27. ruj 2019.

    Some impressions of our MLT NPO Launch event at Mistletoe of Tokyo! Thank you for capturing these special moments! ✨ Find more pics here

    Poništi
  2. proslijedio/la je Tweet

    No natural graph? No problem! In episode 3 of Neural Structured Learning, Software Engineer Arjun Gopalan goes over how graphs can be synthesized from raw input data and used to train neural networks. 📈Watch now →

    Neural Structured Learning thumbnail
    Poništi
  3. prije 7 sati
    Poništi
  4. proslijedio/la je Tweet
    4. velj

    "How to do machine learning efficiently". There's so much to love about this wonderful article.

    Poništi
  5. proslijedio/la je Tweet
    3. velj

    "One of the things I really like about this article is how it integrates work from the fields of artificial intelligence, psychology, neuroscience, and evolutionary theory." editor , picks Reinforcement Learning, Fast and Slow as her review of 2019

    Poništi
  6. proslijedio/la je Tweet
    3. velj

    Our ML Math Reading Sessions with , and many more are so. much. fun. and the discussions are incredibly valuable. I collected a few resources that cover fundamental concepts – feel free to add to the list and send a PR.

    Poništi
  7. proslijedio/la je Tweet
    4. velj

    Thank you to for providing the Tinker Edge T board for the community to explore and experiment with! 🙌 it's capable of performing four tera-operations per second (TOPS) using 0.5 watts per unit of computation!! 🤯

    Poništi
  8. proslijedio/la je Tweet
    3. velj

    MIT's new CS class teaches you things that all the other classes don't teach you, like... 🖥️Shell tools and scripting 🖥️Vim 🖥️Data wrangling 🖥️Command-line environment 🖥️Version control Watch all 11 lectures for free here:

    Poništi
  9. proslijedio/la je Tweet
    3. velj

    ELIZA could do the same, or better, in a tiny fraction of the cost. Over 50 years ago! Yet an example of how is being oversold. is not the solution to all your problems! Proper engineering is: thinks first about requirements and design options!

    Poništi
  10. proslijedio/la je Tweet
    31. sij

    On Tuesday, in my class, we have learnt that all a neural net does is stretching / contracting the space fabric. For example this 3-layer net (1 hidden layer of 100 positive neurons) gets its 5D logits (2D projections) linearly separable by the classifier hyperplanes (lines).

    Poništi
  11. proslijedio/la je Tweet
    4. velj

    The biggest reasons why PyTorch became so popular on 1. Amazing model zoo. Very imp to start with a pretrained model in any comp based on NNs. 2. Compared to TF 1.x, writing models in it felt like writing C++ vs Python

    Poništi
  12. proslijedio/la je Tweet

    Is there a technical podcast (or audiobook or anything else to *listen* to) that teaches ML, data science, or statistics without visual aid?

    Prikaži ovu nit
    Poništi
  13. 4. velj

    You don't need a PhD, you don't even need to finish high school, as long as you pass the "hardcore coding test" and demonstrate a deep understanding of AI and its deployment. Did we get this right ?

    Poništi
  14. proslijedio/la je Tweet
    3. velj

    Our NN is initially in Python for rapid iteration, then converted to C++/C/raw metal driver code for speed (important!). Also, tons of C++/C engineers needed for vehicle control & entire rest of car. Educational background is irrelevant, but all must pass hardcore coding test.

    Prikaži ovu nit
    Poništi
  15. proslijedio/la je Tweet
    3. velj

    Really excited to publish this Perspective today, open access! A huge project since 2001.

    Poništi
  16. proslijedio/la je Tweet
    4. velj
    Poništi
  17. proslijedio/la je Tweet
    3. velj

    Added ImageNet validation results for 164 pretrained models on several datasets, incl ImageNet-A, ImageNetV2, and Imagenet-Sketch. No surprise, models with exposure to more data do quite well. Without extra, EfficientNets are holding their own.

    Prikaži ovu nit
    Poništi
  18. proslijedio/la je Tweet
    3. velj

    yep. most of the people dont care about what degree you have as long as you understand stuff and can implement it.

    Poništi
  19. proslijedio/la je Tweet
    2. velj

    Wow: Google's "Meena" chatbot was trained on a full TPUv3 pod (2048 TPU cores) for **30 full days** - That's more than $1,400,000 of compute time to train this chatbot model. (! 100+ petaflops of sustained compute !)

    Poništi
  20. proslijedio/la je Tweet
    3. velj

    Diagrams of neural networks back when people still called them “artificial neural networks” looked more like biological neural networks.

    Poništi
  21. 2. velj
    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:

    ·