Datasets:

ArXiv:
License:

Upload initial dataset

#1
by kxzxvbk - opened
Files changed (2) hide show
  1. READEME.md +27 -0
  2. pong-v4-expert.pkl +3 -0
READEME.md ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Dataset Card for Pong-v4-expert-MCTS
2
+ ## Table of Contents
3
+ - [Dataset Description](#dataset-description)
4
+ - [Dataset Structure](#dataset-structure)
5
+ - [Data Instances](#data-instances)
6
+ - [Data Fields](#data-fields)
7
+ - [Data Splits](#data-splits)
8
+ ## Dataset Description
9
+ This dataset includes 8 episodes of pong-v4 environment. The expert policy is EfficientZero, which is able to generate MCTS hidden states.
10
+
11
+ ## Dataset Structure
12
+ ### Data Instances
13
+ A data point comprises tuples of sequences of (observations, actions, hidden_states):
14
+ ```
15
+ {
16
+ "obs":datasets.Array2D(),
17
+ "actions":datasets.Array2D(),
18
+ "hidden_state":datasets.Array2D(),
19
+ }
20
+ ```
21
+ ### Data Fields
22
+ - `obs`: An Array2D containing observations from 8 trajectories of an evaluated agent.
23
+ - `actions`: An Array2D containing actions from 8 trajectories of an evaluated agent.
24
+ - `hidden_state`: An Array2D containing corresponding hidden states generated by EfficientZero, from 8 trajectories of an evaluated agent.
25
+
26
+ ### Data Splits
27
+ There is only a training set for this dataset, as evaluation is undertaken by interacting with a simulator.
pong-v4-expert.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc3b6c39eb1c5c5bd719f450f2dabe0b4ac98657d54671c22f70da711899df61
3
+ size 520535731