forked from parimarjan/learned-cardinalities
-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_beta.sh
36 lines (33 loc) · 1.26 KB
/
run_beta.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
#BETAS=(0.01 0.1 1.0)
#for i in "${!BETAS[@]}";
#do
#python3 main.py --algs nn --nn_type mscn --load_query_together 1 \
#--normalization_type mscn --max_epochs 30 --tfboard 0 \
#--save_gradients 1 --loss_func flow_loss2 \
#--hidden_layer_size 128 --weighted_mse ${BETAS[$i]} \
#--exp_prefix allTemplates-nli8-wMSE-${BETAS[$i]} --eval_epoch_plan_err 40 \
#--eval_epoch_flow_err 40 --eval_epoch_jerr 40 --eval_epoch 40 \
#--result_dir all_results/nli8_weighted_mse/normalized_8b \
#--normalize_flow_loss 1 \
#--losses qerr,join-loss,plan-loss,flow-loss \
#--join_loss_pool_num 40 --cost_model nested_loop_index8b
#done
HLS=(256 512)
LF=(flow_loss2 mse)
for i in "${!HLS[@]}";
do
for j in "${!LF[@]}";
do
python3 main.py --algs nn --nn_type mscn \
--normalization_type mscn --max_epochs 30 --tfboard 0 \
--test_diff_templates 0 \
--save_gradients 1 --loss_func ${LF[$j]} \
--hidden_layer_size ${HLS[$i]} \
--exp_prefix final_results --eval_epoch_plan_err 40 \
--eval_epoch_flow_err 40 --eval_epoch_jerr 40 --eval_epoch 40 \
--result_dir all_results/inl_fixed_scan_ops2/nested_loop_index7/final_results \
--normalize_flow_loss 1 \
--losses qerr,join-loss,plan-loss,flow-loss \
--join_loss_pool_num 40 --cost_model nested_loop_index7
done
done