forked from krishvishal/pgm_graph_inference
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathexp_specs.txt
37 lines (25 loc) · 928 Bytes
/
exp_specs.txt
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
Experiment specifications:
1) In-sample Experiments:
{star, path, fc}_small (train), 1300 graphs -> {star, path, fc}_small (test) 130 graphs, model_name {star, path, fc}_small
2) Out-of-sample Experiments:
TBA
---
You should be able to run
```bash
create_data.py --data_mode {train,val,test} [other settings for datasets' names]
```
(several times until all datasets required by dict from exp_helpers are generated), and then
```bash
train.py --train_set_name [...] --mode [...]
```
and
```bash
./experiments/experiments.py [TODO: arguments]
```
---
For example, to run an in-sample star_small experiment, do:
python create_data.py --graph_struct star --size_range 9_9 --num 1300 --data_mode train;
python create_data.py --graph_struct star --size_range 9_9 --num 1300 --data_mode test;
python train.py --train_set_name star_small;
python experiments/run_exps.py;
(Maybe I will make a bash runner for all of this.)