issues
search
hr0nix
/
omega
A number of agents (PPO, MuZero) with a Perceiver-based NN architecture that can be trained to achieve goals in nethack/minihack environments.
GNU General Public License v3.0
38
stars
4
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Use MCTS-in-JAX
#16
ipsec
closed
2 years ago
3
PPO config for room_5x5 environment.
#15
hr0nix
closed
2 years ago
0
Fix rng usage
#14
hr0nix
closed
2 years ago
0
setup.py added: pip install -e ".[cpu]" working
#13
ipsec
opened
2 years ago
0
Gym environment
#12
ipsec
opened
2 years ago
1
Better gradient flow for memory and dynamics
#11
hr0nix
closed
2 years ago
0
The YAML file of PPO algorithm training
#10
xiami1234567890
closed
2 years ago
3
Episodes can now be rendered into gifs.
#9
hr0nix
closed
2 years ago
0
Terminal state fix
#8
hr0nix
closed
2 years ago
0
Stochastic muzero
#7
hr0nix
closed
2 years ago
0
Update env
#6
hr0nix
closed
2 years ago
0
Support for recurrent memory in MuZero
#5
hr0nix
closed
2 years ago
0
Support for GRU gating in transformers.
#4
hr0nix
closed
2 years ago
0
Replay buffer refactoring and prioritized replay implementation.
#3
hr0nix
closed
2 years ago
0
A bunch of changes to make MuZero really work
#2
hr0nix
closed
2 years ago
0
MuZero implementation added
#1
hr0nix
closed
2 years ago
0