0.0.21

multiple actions is ready for q-learning!

0.0.20

almost there

0.0.19

move single action q head into own module

0.0.18

final refactor before venturing out into multiple actions

0.0.17

backup and bring code to only single actions, also fix adaptive layer…

0.0.16

fix non-nstep

0.0.15

oops

0.0.14

allow for min reward and monte carlo return to be set when instantiat…

0.0.12

allow for one to customize the min reward for the conservative reg lo…

0.0.11

enable flash attention for robotics transformer, to remove memory bot…