Issue about train_summary.csv about pgportfolio

[murphyyhuang]

Sorry I don't know much about your problem. But maybe you can try the following method I once used.

Under the path "PGportfolio/pgportfolio/autotrain/training.py", it has:
for dir in all_subdir: # train only if the log dir does not exist if not str.isdigit(dir): return # NOTE: logfile is for compatibility reason if not (os.path.isdir("./"+train_dir+"/"+dir+"/tensorboard") or os.path.isdir("./"+train_dir+"/"+dir+"/logfile")): p = Process(target=train_one, args=( "./" + train_dir + "/" + dir + "/netfile", load_config(dir), "./" + train_dir + "/" + dir + "/tensorboard", dir, logfile_level, console_level, device)) p.start() pool.append(p) else: continue
The first “if” check whether the names of folders under this directory are digital or not. However, for me I ran it on macOS, it would generate a hidden folder named ".DS_Store" automatically. And according to the order of ASCII, this folder would be checked first before those folders named digitally which we really need. Then this function returns directly without reading in any data.

What you can do, I think is:
(1) Delete the folder ".DS_Store" manually;
(2)Change the "return" under the first if into "continue".

[dexhunter]

@EstelleHuang666 Hi, there! What steps have you tried? The train_summary will be generated only after you finish the training.

An example operation should look something like:

1. config net_config.json under ./pgportfolio
2. python main.py --mode=generate --repeat=1
3. python main.py --mode=train --processes=1

Besides, if you are looking for the code. It is here

[ZhengyaoJiang]

Neither could I generate one by following your instruction. Look forward to your answer.

As @dexhunter said, if you finish training, it should be generated under the train_package folder.

[murphyyhuang]

Such a dumb - -#... Great thanks for your explanation!

[mertovun]

Hi. I am also having a problem with generating the train_summary.csv file after python main.py --mode=train --processes=1 step although it does not respond any errors.
net_config.json replicate folders seem to be successfully generated under train_package folder after python main.py --mode=generate --repeat=1 command. I also downloaded the data beforehand using python main.py --mode=download_data command as suggested. But I could not obtain the csv file.
Should I somehow modify the config file for the first run or is there something else missing?
Thanks.

[mertovun]

@EstelleHuang666 it turned out that it is the exact reason of my problem. Thank you so much! Problem solved.