Pierre-Antoine retweeted
Le franchouille "YMCA" (signé par nos compatriotes Henri Belolo et Jacques Morali) d'office pour célébrer Trump.
Hymne gay par excellence devenu le parangon du MAGA.
On est bien, Tintin.
#VillagePeople
Pierre-Antoine retweeted
can @huggingface quickly start a Twitter clone for AI/ML folks 🤪
Best. Blog. Ever.
Training large models demands a lot of GPU memory and a long training time. With several training parallelism strategies and a variety of memory saving designs, it is possible to train very large neural networks across many GPUs.
lilianweng.github.io/lil-log…
Pierre-Antoine retweeted
Contrastive learning aims to learn representation such that similar samples stay close, while dissimilar ones are far apart. It can be applied to supervised / unsupervised data and has been shown to achieve good results on various tasks.
📚 A long read: lilianweng.github.io/lil-log…
Pierre-Antoine retweeted
Video shows how high you would be able to jump on each planet. This includes (!) Pluto where our spaceman friend can jump really high. Source: buff.ly/2PHwXVW
Pierre-Antoine retweeted
An empirical study on the regularization effects of Batch Normalization. We find that preventing explosive growth at the final layer at initialization and during training can recover a large part of BatchNorm’s regularization boost. openreview.net/pdf?id=d-XzF8…
Automatic Mixed Precision (AMP) is now a native feature of PyTorch. pytorch.org/blog/acceleratin…
Pierre-Antoine retweeted
I have five open PhD positions in deep ml @unige_en
fleuret.org/francois/hiring.…
Please apply or circulate!
Pierre-Antoine retweeted
On commencera la journée avec @trypag Co-founder de @radiostic et Data Scientist qui animera le talk :
"Toute la vérité sur l'utilisation du #DeepLearning en imagerie médicale"
Pour ne pas manquer ça... rendez-vous à 13h ici 👉 lnkd.in/dGq6MxT
#DataScience #Data
Data Echoing : why not reuse the last batch while waiting for the next one ! we all know pre-processing can be cpu heavy and lead to long cpu/gpu transfer time. Data echoing is a smart trick to train faster ! ai.googleblog.com/2020/05/sp…
TorchServe Starter Pack (tutorials and doc) :
aws.amazon.com/blogs/machine…
medium.com/@freidankm_39840/…
pytorch.org/serve/
/ @PyTorch @awscloud
Pierre-Antoine retweeted
Deux événements dans la communauté @PyTorch cette semaine :
- [08/04] PyTorch Community Update live.globalai.community/#tal…
- [09/04] Deep Learning at Scale with Azure
databricks.com/p/webinar/dee…
It's still time for you to pick the best tool
We're standardizing OpenAI's deep learning framework on PyTorch to increase our research productivity at scale on GPUs (and have just released a PyTorch version of Spinning Up in Deep RL): openai.com/blog/openai-pytor…
1
Pigeons reviewing breast cancer images. For real. journals.plos.org/plosone/ar…
Pierre-Antoine retweeted
We've hand-picked our 💙10 favorite #Python libraries💙 of the year. Check out the complete list including #Metaflow by @NetflixResearch, #FastAPI by @tiangolo or #Detectron2 by @facebookai
tryolabs.com/blog/2019/12/10…
Pierre-Antoine retweeted
Scikit-learn 0.22 is out! New website, new plotting API, permutation variable importances, support for missing values in GBRT, KNN Imputer, Decision Tree pruning and much more.
Highlights:
scikit-learn.org/stable/auto…
Full changelog: scikit-learn.org/stable/what…
We just released the final 0.22, and you can start using it: `pip install -U scikit-learn` ✨
Check out scikit-learn.org/stable/auto… for a short list of highlights.
Object detection model card by google. Seems pretty solid on unexpected content 😋 Try your own modelcards.withgoogle.com/ob…
1
Get your free copy of Deep Learning with PyTorch ! Limited offer pytorch.org/deep-learning-wi…

















