search

shariqiqbal2810 / MAAC

Code for "Actor-Attention-Critic for Multi-Agent Reinforcement Learning" ICML 2019
MIT License
676 stars 173 forks source link

About query, key and value input embedding #25

Closed TianQi-777 closed 3 years ago

TianQi-777 commented 4 years ago

In the code: the input of sel_ext(query) is state_encodings the input of k_ext(key) is state_action_encodings the input of v_ext(value) is state_action_encodings In the paper, the input of key and query should be state_action_encodings.

I think the correct input should be the input of sel_ext(query) is state_action_encodings(change) the input of k_ext(key) is state_action_encodings the input of v_ext(value) is state_encodings(change)

Could you explain why this is done in the code?

shariqiqbal2810 commented 3 years ago

We output an action-value for each possible action, rather than feeding a specific action as input. This is explained in the section of the paper entitled "Multi-Agent Advantage Function"