<s>
Proximal	B-Algorithm
Policy	I-Algorithm
Optimization	I-Algorithm
(	O
PPO	O
)	O
is	O
a	O
family	O
of	O
model-free	B-Algorithm
reinforcement	O
learning	O
algorithms	O
developed	O
at	O
OpenAI	O
in	O
2017	O
.	O
</s>
<s>
PPO	O
algorithms	O
have	O
some	O
of	O
the	O
benefits	O
of	O
trust	B-Algorithm
region	I-Algorithm
policy	O
optimization	O
(	O
TRPO	O
)	O
algorithms	O
,	O
but	O
they	O
are	O
simpler	O
to	O
implement	O
,	O
more	O
general	O
,	O
and	O
have	O
better	O
sample	O
complexity	O
.	O
</s>
