Skip to main content
European Transport Research Review

An Open Access Journal

Table 2 Hyper-parameters of the RL algorithm

From: Optimization of a physical internet based supply chain using reinforcement learning

Hyper-parameter name

Value

Batch size

32

Discounting factor

0.99

Optimizer

Adam

Learning-rate

5e-4

Synchronization freq. (ε)

1e-2 (16)

Experience replay mem. size

1000

Policy for exploration

Boltzmann-sampl. (τ=1.0) eq. 15

Training length

10000

Back to article page