Homa Maleki The weights from the main network are replicated to the target network every N step. Using both of these networks improves learning stability and allows the algorithm to learn more successfully. Every 100 steps, the main network weights replace the target network weights in our implementation.

Dear Homa Maleki,

I suggest you seelinks on the topic.

https://www.nature.com/articles/s41598-019-47154-z

https://ieeexplore.ieee.org/document/9287318

Article Double Deep Q-Learning and Faster R-CNN-Based Autonomous Veh...

Article An Intelligent Anti-Jamming Scheme for Cognitive Radio Based...

Article Double Deep Q-Network with a Dual-Agent for Traffic Signal Control

Best regards

Qamar Ul Islam Thank you for your answer. could you please explain a little bit more? why do you say every 100 steps?

Homa Maleki Yes, you can refer https://towardsdatascience.com/deep-q-learning-tutorial-mindqn-2a4c855abffc