Tweetovi
- Tweetovi, trenutna stranica.
- Tweetovi i odgovori
Blokirali ste korisnika/cu @dnanhkhoa
Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @dnanhkhoa
-
Khoa Duong proslijedio/la je Tweet
The slides of all talks from the NLP session at
@DeepIndaba are now online, featuring talks from@RichardSocher@sgouws@alienelf@vukosi@HermanKamper@negamuhia & me https://sites.google.com/view/nlpdlindaba/2019 …pic.twitter.com/yOW7hcxOth
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
@kobai19@yoavgo@srush_nlp@dcpage3@philipmlong 1/ The blog post https://myrtle.ai/how-to-train-your-resnet-7-batch-norm/ … is a nice demonstration of some phenomena which, luckily, we have a deep literature on. So allow me to relay some of my current knowledge. Here's a graphical summary again.pic.twitter.com/O8o15Q4caz
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
There is a trend for huge Transformers. We went the other way: decreasing the size!
Introducing DistilBERT: a smaller, faster, cheaper, lighter BERT trained w/ distillation!
95% of BERT's GLUE perf w/ 66M parameters.
: https://medium.com/huggingface/distilbert-8cf3380435b5 …
: https://github.com/huggingface/pytorch-transformers …pic.twitter.com/TraVr39J6v
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
New blog post and ICLR paper on the *Universal Transformer*. Generalizing the Transformer to give it a recurrent inductive bias and make it computationally universal. Work of Amsterdam's Mostafa Dehghani (
@m__dehghani) while at Google Brain.https://twitter.com/Thom_Wolf/status/1125670551882801152 …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
Interesting developments happened in 2018/2019 for natural language generation decoding algorithms: here's a thread with some papers & code So, the two most common decoders for language generation used to be greedy-decoding (GD) and beam-search (BS). [1/9]
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
We used architecture search to find a better architecture for object detection. Results: Better and faster architectures than Mask-RCNN, FPN and SSD architectures. Architecture also looks unexpected and pretty funky. Link: https://arxiv.org/abs/1904.07392 pic.twitter.com/00KKNlnybv
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
NEWSROOM https://summari.es/ -- a corpus of 1.3M (1,321,995) article-summary pairs for automated summarization. It's big, it's diverse, and it's an open challenge. Oh, and we are pretty excited about it! Joint work with Max Grusky and
@informor#NLProc#naacl2018Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
I'm really loving this article on "Rules of Machine Learning" by Martin Zinkevich http://martin.zinkevich.org/rules_of_ml/rules_of_ml.pdf … ... some favorites:pic.twitter.com/UgQp2ypMdq
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi -
Khoa Duong proslijedio/la je Tweet
Happy to announce our QANet models, #1 on
@stanfordnlp question answering dataset (SQuAD). 3 ideas: deep & fast arch (130+ layers), data augmentation, transfer learning. Joint work /w@AdamsYu@dmdohan@oahziur, Quoc Le, et al. See our#ICLR2018 paper https://openreview.net/pdf?id=B14TlG-RW …pic.twitter.com/wSdKHp4nCt
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
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.