ResNet-50 Encoder-based modelling for Schrödinger's Zhirinovsky Paradox
Image: Zhirinovsky and some unknown man.
In the name of recent events, we aim to compute the probability of whether Zhirinovsky- otherwise known as Squealer is alive or dead.
We use ImageNet-pretrained ResNet-50 encoder to compute probs over 341th class using randomly sampled images of Zhirinovsky.
As long as Russian media propaganda tries to change the outcome every day, we add up some normal noise to the input to simulate the superposition of Zhirinovsky.
As an author, I wish the real-world probability is less than 0.45.
Glory to Ukraine! 🇺🇦
Install the requirements:
python3 -m pip install -r requirements.txt
Run the evaluation with 1 trial:
python3 eval.py
Or run the evaluation with N trials, utilizing gpu.
python3 eval.py --trials N --device cuda
Example output:
5 Trial Runs have been finished.
Congratulations! Theoretically he turns out to be DEAD with the probability of 0.83.