Loading
0 Follower
0 Following
orlov_alexander

Location

DE

Badges

0
0
0

Activity

Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Mon
Wed
Fri

Ratings Progression

Loading...

Challenge Categories

Loading...

Challenges Entered

Latest submissions

See All
graded 22438
graded 22217
graded 21638

Latest submissions

No submissions made in this challenge.

Sample-efficient reinforcement learning in Minecraft

Latest submissions

No submissions made in this challenge.

Multi Agent Reinforcement Learning on Trains.

Latest submissions

No submissions made in this challenge.

Robots that learn to interact with the environment autonomously

Latest submissions

No submissions made in this challenge.
Participant Rating
Participant Rating

NeurIPS 2019: Learn to Move - Walk Around

Constant baseline solution ~60 reward

Almost 5 years ago

Just guys if you are wondering - for baseline comparison
action = [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.7, 0.1, 0.1, 0.1, 0.886927] * 2
gives you about ~60 reward, more or less depends on starting position etc.

How I found it? I tried to make a symmetric vector of constant parameters and started from [0.1] * 22. Then I pulled each muscle and observed through visualisation the effect of it.
After founding that [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.7, 0.1, 0.1, 0.1, 0.8] stands a little longer than usual (e.g. 10 reward), I manually tuned 0.8 coefficient to make longest standing. I tuned 0.8 coefficient manually via binary search, that’s why it is so weird looking.

I spend like 1 hour for it, it is an easy one solution.

Different reward on local and remote environments

About 5 years ago

We have same issues. Our RL model could score 100+ on local, and score like 14 on remote.

Flatland Challenge

Any discord server?

About 5 years ago

Do you have a discord server?

orlov_alexander has not provided any information yet.