shawwn

See <a href="https://github.com/shawwn/tpudiepie#ml-community">https://github.com/shawwn/tpudiepie#ml-community</a> and&nbsp;<a href="https://news.ycombinator.com/item?id=23479257">https://news.ycombinator.com/item?id=23479257</a> and&nbsp;<a href="https://www.tensorfork.com/tpus">tensorfork.com/tpus</a> for examples of the kinds of stuff we make. I also post most of my results to twitter: <a href="https://twitter.com/theshawwn">https://twitter.com/theshawwn </a>Tensorfork's research results are carefully documented at&nbsp;<a href="https://github.com/tensorfork/tensorfork/issues">https://github.com/tensorfork/tensorfork/issues</a> mostly thanks to Gwern. I'll be fleshing this out soon. In the meantime, I super appreciate any contributions you're able to make. Here's a brief list of highlights: - A Newsweek article <a href="https://www.newsweek.com/openai-text-generator-gpt-2-video-game-walkthrough-most-tedious-1488334">https://www.newsweek.com/openai-text-generator-gpt-2-video-g...</a> - GPT-2 chess <a href="https://www.theregister.com/2020/01/10/gpt2_chess/">https://www.theregister.com/2020/01/10/gpt2_chess/</a> - ... which DeepMind referenced: <a href="https://twitter.com/theshawwn/status/1226916484938530819">https://twitter.com/theshawwn/status/1226916484938530819</a> - GPT-2 music <a href="https://soundcloud.com/theshawwn/sets/ai-generated-videogame-music">https://soundcloud.com/theshawwn/sets/ai-generated-videogame...</a> - Swarm training (WIP) <a href="https://www.docdroid.net/faDq8Bu/swarm-training-v01a.pdf">https://www.docdroid.net/faDq8Bu/swarm-training-v01a.pdf </a>

Free

Waifu

+ Ask me questions + My everlasting gratitude

GPT-2 Small

Your name (or pseudonym) will appear on the Wall of Thanks on shawwn.com for as long as you maintain this pledge. + All above rewards

GPT-2 Medium

Early access to my research, including the option of downloading models that are currently being trained (which will eventually be made public) + All above rewards

GPT-2 Large

Your name in the footnotes of all shawwn.com essays, tutorials, and open-source software while you maintain this pledge. Your contribution will directly fund an n1-standard-2 instance, which we use to train BigGAN, StyleGAN, GPT-2 1.5B, and other models. + All above rewards

Publisher

Your contribution will directly impact the accessibility of machine learning research by funding day-to-day operational costs, e.g. the VMs we use to train models. + All above rewards

Recent posts by shawwn