Twitter | Pretraživanje | |
Quoc Le
Principal Scientist, Google Brain Team.
150
Tweetovi
78
Pratim
15.032
Osobe koje vas prate
Tweetovi
Quoc Le proslijedio/la je tweet
Daniel Adiwardana 28. sij
Enabling people to converse with chatbots about anything has been a passion of a lifetime for me, and I'm sure of others as well. So I'm very thankful to be able to finally share our results with you all. Hopefully, this will help inform efforts in the area. (1/4)
Reply Retweet Označi sa "sviđa mi se"
Quoc Le proslijedio/la je tweet
Connor Shorten 29. sij
This video explains 's amazing new Meena chatbot! An Evolved Transformer with 2.6B parameters on 341 GB / 40B words of conversation data to achieves remarkable chatbot performance! "Horses go to Hayvard!"
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 30. sij
Odgovor korisniku/ci @xpearhead @lmthang @Blonkhart
I had another conversation with Meena just now. It's not as funny and I don't understand the first answer. But the replies to the next two questions are quite funny.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 29. sij
Odgovor korisniku/ci @lintool
We cited another Riloff paper but thanks for the reference to Riloff 1996. We will look into it and cite in the next revision.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 29. sij
Odgovor korisniku/ci @dnouri @xpearhead i 2 ostali
"steer" is a type of cows:
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 29. sij
Odgovor korisniku/ci @xpearhead @lmthang @Blonkhart
My favorite conversation is below. The Hayvard pun was funny but I totally missed the steer joke at the end until it was pointed out today by
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 28. sij
Odgovor korisniku/ci @xpearhead @lmthang
You can find some sample conversations with the bot here:
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 28. sij
New paper: Towards a Human-like Open-Domain Chatbot. Key takeaways: 1. "Perplexity is all a chatbot needs" ;) 2. We're getting closer to a high-quality chatbot that can chat about anything Paper: Blog:
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 19. pro
Odgovor korisniku/ci @hardmaru
We have a few data points that suggest such improvements are meaningful: 1. Better ImageNet models transfer better to other datasets: 2. Better accuracy on ImageNet gives vast improvements in out-of-distro generalization:
Reply Retweet Označi sa "sviđa mi se"
Quoc Le proslijedio/la je tweet
Connor Shorten 3. pro
This video explains AdvProp from ! This technique leverages Adversarial Examples for ImageNet classification by using separate Batch Normalization layers for clean and adversarial mini-batches.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le proslijedio/la je tweet
Jeff Dean 26. stu
Some nice case studies about how 's AutoML products can help tackle real-world problems in visual inspection across a number of different manufacturing domains, being used by companies like Global Foundries and Siemens.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 26. stu
Odgovor korisniku/ci @AwokeKnowing
Yes, you're probably right. We see similar results with other data augmentation methods: we need bigger models to learn from the extra augmented data.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 26. stu
Odgovor korisniku/ci @wightmanr
Pretrained checkpoints in Pytorch: h/t to
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 26. stu
Odgovor korisniku/ci @quocleix
AdvProp improves accuracy for a wide range of image models, from small to large. But the improvement seems bigger when the model is larger.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 25. stu
Odgovor korisniku/ci @quocleix
As a data augmentation method, adversarial examples are more general than other image processing techniques. So I expect AdvProp to be useful everywhere (language, structured data etc.), not just image recognition.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 25. stu
Odgovor korisniku/ci @quocleix
Many of us tried to use adversarial examples as data augmentation and observed a drop in accuracy. And it seems that simply using two BatchNorms overcomes this mysterious drop in accuracy.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 25. stu
AdvProp: One weird trick to use adversarial examples to reduce overfitting. Key idea is to use two BatchNorms, one for normal examples and another one for adversarial examples. Significant gains on ImageNet and other test sets.
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 22. stu
Odgovor korisniku/ci @nir_benz @ogrisel i 2 ostali
You have a good point. Please take a look at my additional tweet above for latency on CPU and GPU (Figure 4 in the paper).
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 22. stu
Odgovor korisniku/ci @tanmingxing @ruomingpang
And latency on CPU and GPU:
Reply Retweet Označi sa "sviđa mi se"
Quoc Le 22. stu
Odgovor korisniku/ci @quocleix
Architecture of EfficientDet
Reply Retweet Označi sa "sviđa mi se"