Can you explain the main differences between DT and the submitted paper? DT is getting all the attention (ha! pun!)... maybe it is faster EDIT: ah. Here is a

Can you explain the main differences between DT and the submitted paper? DT is getting all the attention (ha! pun!)... maybe it is faster EDIT: ah. Here is a nice quote: Decision Transformer [5] bypasses the need for computing cumulative rewards through dynamic programming, but rather generates future actions conditioning on the desired returns, past states and actions