Neural Machine Translation Without Tokenization
[ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://arxiv.org/abs/2105.13626)
I think using characters directly as tokens, without a tokenizer like SentencePiece, like described in this paper will make sense for Argos Translate 2.0. This combined with [seq2seq sentence boundary detection](https://www.youtube.com/watch?v=TyFRbg7rsuE&list=PLe6dpCSdH0zSXUJiZhWfwYzoIWmvCzdhH) would allow translation using only CTranslate2.
Skribilo: A Document Programming Framework
Skribilo: A Document Programming Framework - https://www.nongnu.org/skribilo/
DeepMind says reinforcement learning is ‘enough’ to reach general AI
Scientists at U.K.-based AI lab DeepMind argue true artificial intelligence will emerge from sticking to the principle of reward maximization.
How Hackers Used Slack to Break into EA Games
A representative for the hackers explained to Motherboard how the group stole a wealth of data from the game publishing giant.
Old Textbooks Galore https://hackaday.com/2021/06/04/old-textbooks-galore/
Original tweet : https://twitter.com/hackaday/status/1400996022135013380
Saving P̶r̶i̶v̶a̶t̶e̶ #SciHub : Good people on #Reddit launch « A Rescue Mission for Sci-Hub and #OpenScience » 🔎 https://old.reddit.com/r/DataHoarder/comments/nc27fv/rescue_mission_for_scihub_and_open_science_we_are/
Architect, Computational Designer, Transhumanist, and Free Software Enthusiast.
Fosstodon is an English speaking Mastodon instance that is open to anyone who is interested in technology; particularly free & open source software.