Skip to content

Instantly share code, notes, and snippets.

@pengsun

pengsun/scalable_agent commands

Last active July 24, 2018 05:24
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save pengsun/3abd09ee8416674306e14c6eda81fa46 to your computer and use it in GitHub Desktop.
Save pengsun/3abd09ee8416674306e14c6eda81fa46 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
scalable_agent commands
# learner
python experiment.py \
--job_name=learner --task=0 \
--num_actors=1 \
--level_name=dmlab30 --batch_size=2 \
--entropy_cost=0.0033391318945337044 \
--learning_rate=0.00031866995608948655 \
--total_environment_frames=1000 \
--reward_clipping=soft_asymmetric
# actors
python experiment.py --job_name=actor --task=0 \
--num_actors=1 --level_name=dmlab30 \
--dataset_path=/home/psun/datasets/ObjectsAll
# learner
python experiment.py \
--job_name=learner --task=0 \
--num_actors=1 \
--batch_size=2 \
--entropy_cost=0.0033391318945337044 \
--learning_rate=0.00031866995608948655 \
--total_environment_frames=100000 \
--reward_clipping=soft_asymmetric
# actors
CUDA_VISIBLE_DEVICES= python experiment.py --job_name=actor --task=0 \
--num_actors=1
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment