Category: Uncategorized
-
Token, Tokenizer (ko)
Token이란 …one can talk about tokens, individual occurrences of something, and types, the different things present. …English in general is not a stationary ergodic process. But we can nevertheless model it with various stochastic approximations. Tokenizer란 Tokenizer의 종류 WordPiece Byte Pair Encoding (BPE) Unigram LM …The unigram language model makes…
-
Recent articles from Google Research (2018-2019)
Autoencoders 2019 Improving Interpolation in Autoencoders (ICLR 2019 / David Berthelot, Colin Raffel, Aurko Roy, and Ian Goodfellow) NY Stacked Capsule Autoencoders (NeuroIPS 2019 / Adam Roman Kosiorek, Sara Sabour, Yee Whye Teh, and Geoffrey Hinton) NY Unsupervised speech representation learning using WaveNet autoencoders (IEEE TASLP 2019 / Jan Chorowski,…
Frodo's 