I'm thrilled to announce the public release of #SpeechBrain!
#SpeechBrain is an #opensource toolkit designed to make research and development of #speech technologies faster. It is flexible, modular, easy-to-use, well documented.
https://t.co/a1wqxLucgw
#DeepLearning#AI
« Oui Poutine est violent, mais il a lutté contre les oligarques »: que tous ceux qui ont utilisé cet argument une fois lisent cela à défaut de voir le docu. Ce régime est sans doute la plus grande kleptocratie mondiale. Et Poutine est son voleur en chef. https://t.co/4ZQZIVJKPh
Vaccination contre le coronavirus en France : à ce rythme-là, il faudrait attendre... l’an 3181 pour vacciner 60% de la population et avoir l’immunité collective.
Bonne année à tous ! 🙃
#Quotidien
Navalny appelle l’un des agents chargés de son assassinat en se faisant passer pour un chef du FSB et enregistre ses aveux. Sans doute la 1ere fois que la victime d’un crime politique mène l’enquête et ridiculise ainsi le régime qui voulut le tuer. Génie. https://t.co/lHMFX8reNI
Great progress in speech recognition: wav2vec 2.0 pre-training + self-training with just 10 minutes of labeled data rivals the best published systems trained on 960 hours of labeled data from just a year ago.
Paper: https://t.co/niBzDiei1j
Models: https://t.co/frCK1GJMZQ
#NeurIPS2020 Authors! Wondering what a broader impact statement is and how to write one? We realize this may be a challenge, especially with no precedents to follow from past years. Looking on the bright side, you'll be a part of history! Here are some resources and tips. 1/5
Matlab: on commence à compter par 1
C/C++: on commence à compter par zéro
Ecoliers Français: hold my beer. On commence par 6 et on termine par zéro mais on dit "terminal".
Set of illustrated Machine Learning cheatsheets covering the content of Stanford's CS 229 class:
Deep Learning: https://t.co/MMCu0tsN2V
Supervised Learning: https://t.co/GwVyoPktcB
Unsupervised Learning: https://t.co/PlhLdKt8N1
Tips and tricks: https://t.co/z8IAP4yU1M
Today we're open sourcing Horizon, an end-to-end applied reinforcement learning platform built on @PyTorch 1.0. Horizon uses RL to optimize systems in large-scale production environments and we're excited to make it accessible to anyone using RL at scale. https://t.co/X6zTmvrI7O
Best GAN samples ever yet? Very impressive ICLR submission! BigGAN improves Inception Scores by >100.
Paper: https://t.co/DEVUtFaPd3
Lots more samples: https://t.co/FSUqs2L7Xm
PyTorch 0.4.0 is out! https://t.co/LwDzyfwLQb lots of welcome additions: Variables/Tensor merge, more numpy-likeness (dtypes, *_like, pro indexing...), much easier to write CPU/GPU agnostic code, gradient checkpointing for memory-efficient backprop, reduce=False, distributions 👏
State representation learning for Deep Reinforcement Learning taken to the next level. Self-supervision for RL was done before, but this is going beyond. I like the "interactive" paper, you can play with the latent representation of the VAE for VizDoom. Great work. https://t.co/Sd3Txxoi0s