M2M Day 373: Can it go all the way?
This post is part of Month to Master, a 12-month accelerated learning project. For October, my goal is to defeat world champion Magnus Carlsen at a game of chess.
I got to Germany today at around 6pm, so most of my day was spent attempting to sleep on an airplane.
However, while in the air, one of my machine learning models continued to train.
Yesterday, at 10am, when the model started running, it could correctly identify chess positions as “good” or “bad” with 68% accuracy:
As of a few minutes ago, the model is now at 75% accuracy:
While there is still a long way to go, this is the most promising training session I’ve run to date: With each set of 100 iterations, the model’s accuracy is getting better and better with no sign of tapering off.
The progress is gradual, but steady.
This model might just go all the way…
I’m heading to bed now, but am eager to see its progress when I wake up.
Read the next post. Read the previous post.
