Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Added configs from the paper #28

Merged
merged 1 commit into from
Jul 12, 2024
Merged

Added configs from the paper #28

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions .gitignore
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -170,8 +170,8 @@ scripts/*.pdf
scripts/*.jpg
scripts/*.png
src/xminigrid/envs/xland_tmp.py

g
# will remove later
scripts/*testing*
configs
configs/wip
scripts/eval_example.py
11 changes: 11 additions & 0 deletions configs/debug/meta.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,11 @@
# params for quick debug on the CPU, should be 0.8+ score
env_id: "XLand-MiniGrid-R1-9x9"
benchmark_id: "trivial-1m"
rnn_hidden_dim: 32
head_hidden_dim: 16
num_envs: 256
num_steps_per_env: 736
num_steps_per_update: 16
total_timesteps: 2_000_000
eval_num_envs: 64
eval_num_episodes: 3
6 changes: 6 additions & 0 deletions configs/debug/single.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,6 @@
# params for quick debug on the CPU, should be 0.8+ score
env_id: "MiniGrid-Empty-6x6"
rnn_hidden_dim: 16
head_hidden_dim: 16
num_envs: 64
total_timesteps: 100000
12 changes: 12 additions & 0 deletions configs/meta.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,12 @@
# params used in the paper for meta-RL experiments. Overwrite default benchmark_id if needed.
env_id: "XLand-MiniGrid-R4-13x13"
benchmark_id: "high-1m"
total_timesteps: 10_000_000_000
num_envs: 16384
num_steps_per_env: 12800
num_steps_per_update: 256
update_epochs: 1
num_minibatches: 32
eval_num_envs: 16384
eval_num_episodes: 25
train_seed: 5
14 changes: 14 additions & 0 deletions configs/meta_imgs.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
# params used in the paper for meta-RL with RGB observations experiments. Overwrite default benchmark_id if needed.
env_id: "XLand-MiniGrid-R1-13x13"
benchmark_id: "high-1m"
total_timesteps: 500_000_000
num_envs: 1024
num_steps_per_env: 12800
num_steps_per_update: 128
update_epochs: 1
num_minibatches: 32
eval_num_envs: 8192
eval_num_episodes: 25
img_obs: True
lr: 0.0001
train_seed: 5
13 changes: 13 additions & 0 deletions configs/meta_trillion.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
# params used in the paper for trillion transitions experiment.
# params optimized for 8xA100 GPU node
env_id: "XLand-MiniGrid-R4-13x13"
benchmark_id: "high-1m"
total_timesteps: 1_000_000_000_000
num_envs: 131072
num_steps_per_env: 12800
num_steps_per_update: 256
update_epochs: 1
num_minibatches: 32
eval_num_envs: 16384
eval_num_episodes: 25
train_seed: 5
11 changes: 11 additions & 0 deletions configs/single.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,11 @@
# params used in the paper for single-RL experiments.
env_id: "XLand-MiniGrid-R4-13x13"
benchmark_id: "trivial-1m"
ruleset_id: 0
total_timesteps: 1_000_000_000
num_envs: 16384
num_steps: 256
update_epochs: 1
num_minibatches: 8
eval_episodes: 512
train_seed: 5
Loading