How to train the agents by using HER （hindsight experience replay，HER）?

[beiketaoerge]

In stable-baseline3,we can use the code model = DDPG("MultiInputPolicy", env,replay_buffer_class=HerReplayBuffer, verbose=1, tensorboard_log=logs_dir,batch_size=256,device="cuda") to achieve HER.If we use the code memories = RandomMemory(memory_size=15000, device=device, replacement=False)in skrl, can we achieve HER? Thank you for your answer.

[Toni-SM]

Hi @beiketaoerge

Sorry for late response.
Unfortunately, skrl doesn't implement HER for the time being.
It is in the roadmap but I have not found the time for it 🙈