Detailed Framework
Overview

・Multiprocessing flow (Worker-Memory-Trainer)

・Multiprocessing flow (Worker-Trainer)

Play flow

・1 Episode loop

Multiplay flow

Class diagram
RL

Env

Run

Space(srl.base.spaces)
Class |
Type |
SpaceType |
DiscreteSpace |
int |
DISCRETE |
ArrayDiscreteSpace |
list[int] |
DISCRETE |
ContinuousSpace |
float |
CONTINUOUS |
ArrayContinuousSpace |
list[float] |
CONTINUOUS |
NpArraySpace |
NDArray[int] or NDArray[float] # 1D |
DISCRETE or CONTINUOUS |
BoxSpace |
NDArray[AnyType] |
srl.base.define.SpaceTypes |
TextSpace |
str |
DISCRETE |
MultiSpace |
list[SpaceBase] |
MULTI |
AnySpace |
any |
UNKNOWN |
RLBaseTypes(srl.base.define)
Name |
Type |
Space |
Note |
NONE |
copy |
No Change |
|
DISCRETE |
int |
DiscreteSpace |
|
ARRAY_DISCRETE |
list[int] |
ArrayDiscreteSpace |
|
CONTINUOUS |
float |
ContinuousSpace |
|
ARRAY_CONTINUOUS |
list[float] |
ArrayContinuousSpace |
|
NP_ARRAY |
NDArray[float] |
ArrayNpSpace |
float固定(configで変更可能) |
NP_ARRAY_UNTYPED |
NDArray[int] or NDArray[float] |
ArrayNpSpace |
dtype変換なし |
BOX |
NDArray[float] |
BoxSpace |
float固定(configで変更可能) |
BOX_UNTYPED |
NDArray[Any] |
BoxSpace |
dtype変換なし |
TEXT |
str |
TextSpace |
|
MULTI |
list |
MultiSpace |