Open a2108870 opened 1 year ago
You can find the RL Parameters named _filter_usagepenalty in config. If you set it to 1 and each time the agent reuse a filter, a penalty is subtracted from the reward. So the model will not make decisions you discribed.
Dear Dr. Hu,
I hope this message finds you well. I am writing to inquire about your reinforcement learning model designed for retouching images. Specifically, I am interested in understanding whether the model is capable of selecting duplicate retouching operations for a single image. For instance, I would like to know if the model can make decisions that result in the following sequence of retouching operations: gamma, contrast, exposure, contrast, with the contrast operation being repeated twice within this sequence.
Thank you for your time and consideration. I look forward to your response.