More flexible training - #16

Closed

bmielnicki wants to merge 15 commits into

HumanCompatibleAI:masterfrom

bmielnicki:more_agents

Commits on Feb 23, 2021

Moved to default bc params to separate file.
bmielnicki
committed
399ab55
Utils that gets attibute of overcooked mdp/env
bmielnicki
committed
726ae3b

Commits on Feb 24, 2021

More flexible bc data processing
bmielnicki
committed
14d4bc7

Commits on Feb 25, 2021

More flexible BC training
bmielnicki
committed
7f2179f
BC tests
bmielnicki
committed
f833d42

Commits on Feb 26, 2021

make PPO and rllib code more flexible; refactors
bmielnicki
committed
83df552

Commits on Feb 27, 2021

tests for changes in rllib and PPO code
bmielnicki
committed
a0e3eec

Commits on Mar 1, 2021

fix minor typos
bmielnicki
committed
24de495
refer to correct attr name in rllib/tests.py
bmielnicki
committed
c87381e
add submodule with compatible overcooked_ai version
bmielnicki
committed
9528ae5
Merge master from CHAI repo into more_agents (D2RL changes)
bmielnicki
committed
82437cc
d2rl test
bmielnicki
committed
5c88cd4
fix ppo lstm tests and ppo lstm code
bmielnicki
committed
8388c29
change overcooked_ai version to compatible with this version
bmielnicki
committed
0e6d19e

Commits on Mar 11, 2021

Upgrade of suggested tf version
bmielnicki
committed
9c2d4ea