-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathPhase2c_gnn-ppo.sh
executable file
·38 lines (36 loc) · 1.1 KB
/
Phase2c_gnn-ppo.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
#!/bin/bash
# ## BATCH
#train_on="M3M5Mix"
#train_on="NWB_AMS"
#train_on="MixM5Met_T"
train_on="M3_012456"
#train_on="MixAll33"
#train_on="TEST"
emb_dim=64
itt=5
nfm="NFM_ev_ec_t_dt_at_um_us"
#qnet="s2v"
norm_agg="True"
optim='returns'
nstep=200000 #200000
train="True"
eval="False"
test="False"
numseeds=1
#seed0=1
solveselect='solvable'
edgeblock="True"
max_nodes=9
demoruns="False"
pursuit="Uon"
for seed0 in {"0",}
#"NFM_ec_t","NFM_ec_dtscaled"}
do
for qnet in {"gat2",}
do
tmux new-session -d -s "${qnet}-${seed0}"
tmux send-keys -t "${qnet}-${seed0}" "conda activate rlcourse-sb3c" Enter
tmux send-keys -t "${qnet}-${seed0}" "cd ~/testing/sim" Enter
tmux send-keys -t "${qnet}-${seed0}" "python Phase2c_gnn-ppo.py --train_on $train_on --emb_dim $emb_dim --emb_itT $itt --nfm_func $nfm --qnet $qnet --norm_agg $norm_agg --optim_target $optim --num_step $nstep --train $train --eval $eval --test $test --num_seeds $numseeds --seed0 $seed0 --solve_select $solveselect --edge_blocking $edgeblock --max_nodes $max_nodes --demoruns $demoruns --pursuit $pursuit" Enter
done
done