H H _sA
brain.Event:2R.
,tensorboard.summary.writer.event_file_writeral" x= E)sA*
objective/kl cw$' F =E)sA*
objective/kl_coefL> S' F E)sA*
objective/entropyBz/ m] P E)sA*"
ppo/mean_non_score_reward D% 6 E)sA*
ppo/mean_scoresEF=} x$ B+M E)sA*
ppo/std_scores=>su- AMJ( pJ 6F)sA*
ppo/policy/entropye@;=) 7_ 'F)sA*
ppo/policy/approxkl%>!) 7_ QF)sA*
ppo/policy/policykl=ȥ) 7_ gF)sA*
ppo/policy/clipfrac$>0 _ l F)sA*#
!
ppo/policy/advantages_meanffh3(& sO F)sA*
ppo/returns/mean?6% 6 F)sA*
ppo/returns/var>wj# wC F)sA*
ppo/val/vpredB@ΐ# wC F)sA*
ppo/val/error@0;& sO F)sA*
ppo/val/clipfracff%?t^" x= >F)sA*
ppo/val/mean6X@! { iF)sA*
ppo/val/varx@$w$+ K F)sA*
ppo/val/var_explained9G ' F F)sA*
ppo/learning_rate
l7=+ K F)sA*
time/ppo/forward_pass|>m}. W F)sA*!
time/ppo/compute_rewards Ғ<$1 F)sA*$
"
time/ppo/compute_advantages ;Y?, E F)sA*
time/ppo/optimize_stepp? I() 7_ F)sA*
time/ppo/calc_stats\h>Equ$ B+M 'F)sA*
time/ppo/total8? $ % 6 8F)sA*
env/reward_meanEF=~o$ B+M GF)sA*
env/reward_std=>
b" x= ɝ;sA*
objective/kl ?(' F uɝ;sA*
objective/kl_coef?L>SGd' F ɝ;sA*
objective/entropy B0#e/ m] P ɝ;sA*"
ppo/mean_non_score_reward-] ~% 6 ɝ;sA*
ppo/mean_scores=)~$ B+M ɝ;sA*
ppo/std_scores=5.- l( pJ ϝ;sA*
ppo/policy/entropyla@e## wC ֝;sA*
ppo/val/vpred%@
ũ# wC ם;sA*
ppo/val/error@Ŕ& sO ڝ;sA*
ppo/val/clipfrac> *" x= ڝ;sA*
ppo/val/meanb4@! { ڝ;sA*
ppo/val/varP?t+ K ڝ;sA*
ppo/val/var_explained)\q9' F ;sA*
ppo/learning_rate
l7!
+ K ;sA*
time/ppo/forward_pass@z>d. W ";sA*!
time/ppo/compute_rewards :1 3;sA*$
"
time/ppo/compute_advantages ;舾, E C;sA*
time/ppo/optimize_step02?P9) 7_ T;sA*
time/ppo/calc_statsh>_ס$ B+M ;sA*
time/ppo/total_? Y% 6 ܝ;sA*
env/reward_mean=($ B+M dܝ;sA*
env/reward_std=2E