In 2020, my life was turned upside down -- I was encountering failure after failure and in deep depression. But that also turned out to be the most enlightening period. In October I was invited to give a talk at Google (thanks to @orf_bnw). It's out today: https://t.co/hYFLnH7JHV pic.twitter.com/gIYdnsMRgI
— Rosanne Liu (@savvyRL) October 7, 2021
Inside a Jupyter notebook:
— Aurélien Geron (@aureliengeron) October 7, 2021
BAD:
!pip install somelib # may install in the wrong environment!
or
!conda install somelib # same problem
GOOD:
import sys
!{sys.executable} -m pip install somelib
or
!conda install --yes --prefix {sys.prefix} somelib
A Dot Product Attention Free Transformer
— AK (@ak92501) October 6, 2021
abs: https://t.co/d19I4MHPRb
an efficient variant of Transformers (Vaswani et al., 2017) that eliminates the query-key dot product in self attention pic.twitter.com/LKrO2uzlpk
Large NLP models are expensive and opaque, but maybe it doesn't have to be that way. This exciting work with Omar Khattab and @ChrisGPotts uses retrieval to set SotA results in hard NLP tasks at low cost. Our Baleen paper will be a spotlight at NeurIPS. https://t.co/vGp0YD52CU
— Matei Zaharia (@matei_zaharia) October 6, 2021
What if you could predict with your own BERT models in 1ms on GPU and 3ms on CPU, in a Docker container that runs anywhere? Please meet Infinity by @huggingface Please watch https://t.co/uCrCcWPIy0 and sign up for the trial at https://t.co/cdh5b2RowR #MachineLearning #NLP #MLOps
— Julien Simon (@julsimon) October 4, 2021
Do Self-Supervised and Supervised Methods Learn
— AK (@ak92501) October 4, 2021
Similar Visual Representations?
pdf: https://t.co/AxQ3hj4Qdn
abs: https://t.co/474VQXVxTA pic.twitter.com/F1l05AIHdy
threadpoolctl 3.0.0 is out: this is a Python utility to introspect & control the number of threads used by OpenMP parallel loops and BLAS linear algebra kernels, typically used in computational libraries (eg NumPy, SciPy and machine learning libraries).https://t.co/TqSb2LXRaJ
— Olivier Grisel (@ogrisel) October 1, 2021
We’re using KaoKore, a machine learning-friendly dataset, to decipher cursive and illustrations in historical Japanese art. Learn how machine learning can be used for humanities research and contribute to cultural preservation. https://t.co/nYSvuehWoR
— Boo-gle 👻 (@Google) September 30, 2021
Age of people at every job...
— Nathan Yau (@flowingdata) September 30, 2021
- umpires and fast food workers tend youngest
- school bus drivers and tailors tend oldesthttps://t.co/GbhWtFsmVH pic.twitter.com/E5Ck01uf7r
This is the wrong way to look at it. Facebook employs people who consider themselves researchers but who are actually PR fronts. https://t.co/LGzNvz5dqj
— Cathy O'Neil (@mathbabedotorg) September 30, 2021
The problem with p-hacking is not the “hacking,” it’s the “p” (or, Fisher is just fine on this one) https://t.co/Rc0DoYeMmj
— Andrew Gelman et al. (@StatModeling) September 30, 2021
DeepRobust is a PyTorch adversarial library for attack and defense methods on images and graphs. https://t.co/DygJTpo0sF #deeplearning #machinelearning #ml #ai #neuralnetworks #datascience #PyTorch
— PyTorch Best Practices (@PyTorchPractice) September 30, 2021
