Contestability: when building an algorithmic system, include mechanisms for questioning & disagreeing with results as part of the system
— Rachel Thomas (@math_rachel) November 22, 2020
rather than having disagreement or attempts to shape the decision making process as an external add onhttps://t.co/AkXdtT1u81 pic.twitter.com/4e0P46pCRo
A series of courses on “Tiny ML” taught by Vijay Janapa Reddi, @lmoroney @petewarden on @edXOnline
— hardmaru (@hardmaru) November 22, 2020
Details: https://t.co/mjvBstjhCWhttps://t.co/nMbHX3xUP3
Seems like a bunch of folks found this useful! So, here are a few more pytest flags that you might also like:
— jacobian (@jacobian) November 20, 2020
--ff → "fail fast" - stop on the first failure
--nf → run only new test cases since your last run
-k something → only run tests whose name contains 'something'
Quick pytest tip: ran some tests, got a failure, not sure what's up?
— jacobian (@jacobian) November 20, 2020
Try this: `pytest --lf --pdb`
What this means:
--lf → "run the last failing test(s)"
--pdb → "and drop into pdb when the test fails"
(see https://t.co/6IHg5Euq55 if you need help w/pdb)
I posted my #SustaiNLP2020 slides on "More is Less? Non-parametric Language Models and Efficiency" (https://t.co/N9wynDRcHB), where I overview and survey non-parametric models in NLP, and introduce our new work on sparse prototypes for text generation (https://t.co/BhNTr9zkM2)! https://t.co/cELvzdMDe7 pic.twitter.com/AMz0GaUEWZ
— Graham Neubig (@gneubig) November 20, 2020
The powerful mT5 model including all pre-trained models is now part of Transformers.
— Hugging Face (@huggingface) November 20, 2020
mT5 was trained on 101 languages and yields SOTA on many multilingual tasks.
Thanks, @lintingxue, Noah Constant, @ada_rob, @colinraffel for releasing the weights!
➡️ https://t.co/sWdM7VNeLw. pic.twitter.com/mOj9hEkpCG
Some Intuition on Neural Tangent Kernel
— Ferenc Huszár🇪🇺 (@fhuszar) November 20, 2020
new post with a (basic) colab notebook you can play withhttps://t.co/kNP2Ltpff2
⚠️ SO YOU WANT TO BE A BAYESIAN⚠️ :
— Chelsea Parlett-Pelleriti (@ChelseaParlett) November 20, 2020
(since I compiled a quick list of Bayesian resources today, I figured I should share; these are just my opinion!) https://t.co/9incv2VOfs
Check out my new paper "FROST: Faster and more Robust One-shot Semi-supervised Training" at https://t.co/lhfEfeCcKB. FROST is cool and it is the most practical method for one-shot semi-supervised training.
— Leslie Smith (@lnsmith613) November 19, 2020
This animation of the NBA's scoring pattern over the years reminded me of the growth pattern of Australian cities before COVID: Lots of growth in the outer suburbs and in the city centre but a "missing middle". Source: https://t.co/UXANoO0ahA pic.twitter.com/X2MSPSKvwi
— Simon Kuestenmacher (@simongerman600) November 19, 2020
These tutorial slides on "High Perf NLP" are really impressive. Every slide is current to the minute. Amazing set of diagrams.https://t.co/o0o4SY6chR
— Sasha Rush (@srush_nlp) November 19, 2020
(@gabriel_ilharco @Tim_Dettmers @IuliaTurc @kentonctlee Felipe Ferreira Cesar Ilharco) pic.twitter.com/TdfVx9lBsU
It’s often repeated in the media – or here on Twitter – that life expectancy has increased ‘only’ because child mortality has fallen. But it is not true.
— Max Roser (@MaxCRoser) November 19, 2020
My post: It’s not just about child mortality, life expectancy improved at all ages.https://t.co/iX9GkVUuof
