@FinansTor@OAlexanderDK A quick look at the data from Nov1st to Dec13th shows that the ratio of (monday_positives)/(sunday_positives) looks fairly normal.
Dec13th was 1.43, which is less than both Dec8th (1.47) and Nov15th (1.47). The mean is over the period is 1.35.
A sketch to explain how a new variant may appear milder even with no change in underlying virulence. This can occur because, when calculating the fraction of cases that are severe, the denominator now includes many re-infections that had previously been averted. A thread. 1/8
@jon_barron@ericmjl It's not uncommon to use a magic sharpness parameter in softplus, i.e. log(e^(k·x)+1) / k, so no need to feel ashamed ;)
Squareplus is ~2.5-3.5x faster than softplus in Numpy. Nice!
I made a Python library for reading images from multiple webcams in a synchronized fashion with minimum latency.
Check it out! https://t.co/0jlJ4o69AI
#ComputerVision#Python#MachineLearning