-
-
-
submission-v3na
normalize adv by rollouts
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
submission-v1.8a
ds2x-modelx2-fs2-2nd-layer
normalize adv by rollouts
recalculate advantages
gamma 9993
fs2 no rnorm
fs2 clip25
no fs best rew
best model deque100
gray framestack 2
layernorm
rnorm fix
no framestack
reward norm
coeff bugfix
ray rollouts
custom torch
torch baseline
lr 5e-4
ds2x-modelx2-fs2-2nd-layer