-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun_test.sh
executable file
·21 lines (18 loc) · 1.41 KB
/
run_test.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
seed=$RANDOM
sample=1
export PYTORCH_ENABLE_MPS_FALLBACK=1
render_mode=rgb_array seed=$seed num_seed_steps=5000 num_unsup_steps=0 num_train_steps=50000 num_interact=4000 max_feedback=1000 reward_batch=2 reward_update=10 feed_type=$sample human_teacher=False agent.batch_size=256
python themis_train.py device=mps domain=ALE env=MsPacman-v5 render_mode=rgb_array seed=$seed num_seed_steps=5000 num_unsup_steps=0 num_train_steps=50000 num_interact=4000 max_feedback=1000 reward_batch=2 reward_update=10 feed_type=$sample human_teacher=False agent.batch_size=256 debug=True learn_reward=True
python themis_pretrain.py device=mps \
domain=ALE env=MsPacman-v5 render_mode=rgb_array seed=$seed \
num_seed_steps=5000 num_unsup_steps=0 num_train_steps=50000 \
replay_buffer_capacity=10000 reward_model_capacity=1000 \
num_interact=4000 max_feedback=1000 reward_batch=2 reward_update=10 feed_type=$sample segment=50 \
human_teacher=False debug=True learn_reward=True
python themis_train.py device=mps \
domain=ALE env=MsPacman-v5 render_mode=rgb_array seed=$seed \
num_seed_steps=5000 num_unsup_steps=0 num_train_steps=50000 \
replay_buffer_capacity=10000 reward_model_capacity=1000 \
num_interact=4000 max_feedback=1000 reward_batch=2 reward_update=10 feed_type=$sample segment=50\
human_teacher=False debug=True learn_reward=True \
xplain_action=False xplain_state=False