Adam King

@AdamDanielKing

Neural networks. My latest experiment:

Toronto, Canada
Vrijeme pridruživanja: siječanj 2017.

Tweetovi

Blokirali ste korisnika/cu @AdamDanielKing

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

  1. Prikvačeni tweet
    5. stu 2019.

    The "too dangerous to release" GPT-2 text generator is finally fully released! I've put it up on for those who like to live dangerously. A sample:

    Poništi
  2. proslijedio/la je Tweet
    12. sij

    The funny thing about AIDungeon is due to all the GPT-2 pre-training on swaths of random web text, you can create an adventure game about literally anything. Here's one about installing OpenCV from source:

    Poništi
  3. 11. stu 2019.

    Changing up the examples on and adding a personal favorite of . What else should I add?

    Poništi
  4. 25. lis 2019.

    'When it comes to the quality of [Google's] search results, ... “this is the single biggest ... most positive change we’ve had in the last five years and perhaps one of the biggest since the beginning.”'

    Poništi
  5. proslijedio/la je Tweet
    19. ruj 2019.

    We've fine-tuned GPT-2 using human feedback for tasks such as summarizing articles, matching the preferences of human labelers (if not always our own). We're hoping this brings safety methods closer to machines learning values by talking with humans.

    Poništi
  6. 20. kol 2019.

    OpenAI just released a bigger, smarter version of GPT-2! Try it out:

    Prikaži ovu nit
    Poništi
  7. proslijedio/la je Tweet
    17. srp 2019.

    Amazing!! Deep Learning-based NLP techniques are going to revolutionize the way we write software. Here's Deep TabNine, a GPT-2 model trained on around 2 million files from GitHub. Details at

    Prikaži ovu nit
    Poništi
  8. proslijedio/la je Tweet
    13. lip 2019.

    The beta version is now available to everyone as a web service via AI Playground A short illustration video is available in May everybody have fun with the app. ,

    Poništi
  9. proslijedio/la je Tweet
    13. lip 2019.

    At NAACL last week we built a new side project, Write With Transformer. It lets you trigger GPT-2 completions multiple times, in a Google Doc-like interface. 🦄 It's like having a unicorn friend that completes your thoughts 🦄 cc Try it:

    Prikaži ovu nit
    Poništi
  10. 25. svi 2019.

    What's next on 's list of neural net-generated lists? Let's ask a neural net.

    Poništi
  11. proslijedio/la je Tweet
    24. svi 2019.

    now that science has a Very Large Telescope, an Extremely Large Telescope, and an Overwhelmingly Large Telescope, how will we name future telescopes? neural net to the rescue

    1. Very Large Telescope
2. Extremely Large Telescope
3. Overwhelmingly Large Telescope
4. World-class Wide Field Telescope
5. World-class Extra Large Telescope
6. Most Extreme Very Large Telescope
7. Most Extreme Very Large Telescope with a Giant Antenna

4. Extremely Huge Telescope
5. All-around Bigger Telescope
6. Biggest Telescope in the Universe

4. Amazingly Large Telescope
5. Star Spangled Monster
    1. Very Large Telescope
2. Extremely Large Telescope
3. Overwhelmingly Large Telescope
4. Large Telescope With No Exoplanets
5. Not Any One Is Allowed
6. Small Telescope
7. Small Telescope With Most Exoplanets
8. Small Telescope But Not Allowed
9. No Exoplanets
10. No Exoplanets
11. Large Telescope With Few Exoplanets
12. Large Telescope With Several Exoplanets
13. Large Telescope With Very Little Exoplanets
    Prikaži ovu nit
    Poništi
  12. proslijedio/la je Tweet
    22. svi 2019.

    Another great paper from Samsung AI lab! @egorzakharovdl et al. animate heads using only few shots of target person (or even 1 shot). Keypoints, adaptive instance norms and GANs, no 3D face modelling at all. ▶️ 📝

    Poništi
  13. proslijedio/la je Tweet
    20. svi 2019.

    The birth of Fruit Genie is still my favorite performance (by ). More behind-the-scenes at .

    Poništi
  14. 13. svi 2019.

    About to go on Dublin's FM104 station and talk with about my site, .

    Poništi
  15. 9. svi 2019.

    Thanks! Glad you like it. And many thanks to the GPT-2 team for creating this model.

    Poništi
  16. 8. svi 2019.

    Thanks to everyone who's been trying out and sharing . I'm blown away. The site is generating about 400,000 characters of text per minute on 8 GPUs. That's about one 44-million-word Encyclopedia Britannica every 10 hours.

    Poništi
  17. 7. svi 2019.

    Yesterday I launched Talk to Transformer, a site where you can try out 's new text-generating language model on your own custom text. One example is below.

    Poništi
  18. proslijedio/la je Tweet
    22. tra 2019.

    Final OpenAI Five stats: - Competitive: 7,215 wins, 42 losses (99.4% winrate, 15,019 total players) - Cooperative: 35,466 games (18,689 total players) gg all!

    Poništi
  19. proslijedio/la je Tweet
    13. tra 2019.

    OpenAI Five makes history by winning a best-of-three versus , the Dota 2 world champions!!!! Huge congrats to the team and to OG for an extremely well-played match.

    Prikaži ovu nit
    Poništi
  20. 11. tra 2019.

    Photos from Crude Sketches: NVIDIA's GauGAN Explained Visually New blog post explaining how NVIDIA's app converts simple blobs of color into photorealistic landscapes. And generative adversarial networks in general.

    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:

    ·