Twitter Beğeni Hilesi

Huge congratulations to Xiang Lisa Li on winning #emnlp2019 Best Paper! (From 2800+ submissions.) And on her delightful talk: video online. Lisa did this work with me as a junior. First time any *ACL best paper award has gone to an undergrad project?? cs.jhu.edu/~jason/papers/…

Xiang Lisa Li And congrats to my close colleagues Percy Liang (at #microsoft #semanticmachines) and Philipp Koehn (at Johns Hopkins JHU CLSP). They + co-authors won Best Paper runner-up aclweb.org/anthology/D19-… and Best Resource Paper aclweb.org/anthology/D19-… .

Interested in diglossia, multilingual models, and dialectal adaptation? Please consider participating in our dialectal speech translation shared task at IWSLT iwslt.org/2022/dialect

Fun visualization and artwork by Google's deep neural networks! googleresearch.blogspot.jp/2015/06/incept…

Very interesting blog post about Deep Nets (and limitations thereof) from the Bayesian perspective by Yarin mlg.eng.cam.ac.uk/yarin/blog_3d8…

Slides from the various speakers at the ACL Semantics for Statistical Machine Translation workshop are now available! hlt.suda.edu.cn/workshop/s2mt/…

Worth checking out new Deep Learning library for Python "CGT replicates Theano's API but has very short compile time" joschu.github.io

How do neural encoder-decoders help reranking in machine translation? Nice analysis by Graham Neubig et al.: phontron.com/paper/neubig15…

ACL-IJCNLP2015 talks are online, thanks to efforts of the local organizers and speakers: techtalks.tv/acl-ijcnlp-201…

Recurrent nets reply your email for you. On Google blog: googleresearch.blogspot.se/2015/11/comput…

#NLP question: Are there any efforts on translating the English Web Treebank (LDC2012T13), e.g. into Russian, Spanish, Chinese?

Deep learning free course on Udacity - Built by Google udacity.com/course/deep-le…

#emnlp2016 accepted papers are up: emnlp2016.net/accepted-paper…

“50 things I learned at NIPS 2016” by Andreas Stuhlmüller blog.ought.com/nips-2016-875b…

Our guest post on the ACL2017 PC blog "Reject before review - we're sorry for it!" chairs-blog.acl2017.org/2017/01/10/rej…

"DyNet: The Dynamic Neural Network Toolkit," Neubig et al. (lots of folks from all over, inc. (((ل()(ل() 'yoav))))👾 and Naomi Saphra): arxiv.org/abs/1701.03980

I wasn't really happy with existing LSTM computation graph visualizations. Here is my attempt.

Google's TPU paper is out. 8-bit matmul processor, large on-die memory (28MB). Excited to finally read the details.. drive.google.com/file/d/0Bx4haf…

Large vocabs in neural MT? Try binary error correcting codes! Small models, 12x faster on CPU, easy on GPU #ACL2017 arxiv.org/abs/1704.06918