Note

TODO

• review 0-intro

• review 1-tensor

• review 2-model

• review 3-data

• review 4-autograd-loss-optimizer

• review 5-train

• Find a better linear model for IRIS

• In 5-train, explain that the train loss is decreasing but validation is random

• Copy the latest output to 5-train

• 10_transfer_learning and fine-tuning