parameter	value	explain
lr0	0.01	Initial learning rate (initial value of 0.01).
lrf	0.01	Final learning rate (initial learning rate multiplied by this value). The final learning rate is 0.01 * 0.01 = 0.0001.
momentum	0.937	Momentum of the SGD optimizer. Momentum is used to accelerate convergence and reduce oscillations.
weight_decay	0.0005	The weights of the optimizer are attenuated to prevent overfitting with a value of 5 × 10−4.