Create nsphs_ml_qt.sif

[richelbilderbeek]

Now Sylabs has given me more space, I can use it with a new container :-)

[richelbilderbeek]

FATAL:   could not open image /home/richel/nsphs_ml_qt.sif: failed to retrieve path for /home/richel/nsphs_ml_qt.sif: lstat /castor/project/home/richel/nsphs_ml_qt.sif: no such file or directory

[richelbilderbeek]

I guess the path expansion is the problem:

[richel@sens2021565-bianca ~]$ cat data_issue_5/experiment_params.csv 
parameter,value
datadir,/home/richel/data_issue_5/
data,data_issue_5
superpops,/home/richel/data_issue_5/data_issue_5_labels.csv
model_id,M1
train_opts_id,ex3
data_opts_id,b_0_4
trainedmodeldir,~/data_issue_5_ae/
pheno_model_id,p0
analyse_epochs,"10,20,30,40,50,60,70,80,90,100,110,120,130,140,150,160,170,180,190,200,210,220,230,240,250,260,270,280,290,300,310,320,330,340,350,360,370,380,390,400,410,420,430,440,450,460,470,480,490,500,510,520,530,540,550,560,570,580,590,600,610,620,630,640,650,660,670,680,690,700,710,720,730,740,750,760,770,780,790,800,810,820,830,840,850,860,870,880,890,900,910,920,930,940,950,960,970,980,990,1000"
metrics,"f1_score_3,f1_score_5,f1_score_7,f1_score_9,f1_score_11"
gcae_folder,/opt/gcae_richel
ormr_folder_name,python3
gcae_version,1.0
python_version,3.6

[richelbilderbeek]

That was a step in the right direction. Still, the _ae folder does not exist. Is it really needed already?

Error in gcae_train_more(gcae_setup = gcae_experiment_params$gcae_setup,  : 
  'ae_out_subfolder' not found at path '/home/richel/data_issue_5_ae/ae.M1.ex3.b_0_4.data_issue_5.p0' 
'gcae_options$gcae_folder': /opt/gcae_richel
'gcae_options$ormr_folder_name': python3
'gcae_options$gcae_folder': /opt/gcae_richel
'gcae_options$python_version': 3.6'args': 'train --datadir /home/richel/data_issue_5/ --data data_issue_5 --model_id M1 --resume_from 0 --epochs 10 --save_interval 10 --train_opts_id ex3 --data_opts_id b_0_4 --trainedmodeldir /home/richel/data_issue_5_ae/ --pheno_model_id p0' 
Tip: you should be able to copy-paste the args :-)
Calls: <Anonymous> -> gcae_train_more
In addition: Warning message:
In system2(command = run_args[1], args = run_args[-1], stdout = TRUE,  :
  running command ''python3' /opt/gcae_richel/run_gcae.py train --datadir /home/richel/data_issue_5/ --data data_issue_5 --model_id M1 --resume_from 0 --epochs 10 --save_interval 10 --train_opts_id ex3 --data_opts_id b_0_4 --trainedmodeldir /home/richel/data_issue_5_ae/ --pheno_model_id p0 2>&1' had status 1
Execution halted

[richelbilderbeek]

Hmmm, locally gcaer does create the output folder, even if it is a sub-sub-sub folder:

  gcae_options <- create_gcae_options()
  gcae_setup <- create_gcae_setup(
    datadir = get_test_datadir(),
    data = "gcae_input_files_1",
    superpops = get_gcaer_filename("gcae_input_files_1_labels.csv"),
    model_id = "M1",
    data_opts_id = "b_0_4",
    train_opts_id = "ex3",
    pheno_model_id = "p2",
    trainedmodeldir = paste0(
      normalizePath(
        file.path(get_gcaer_tempfilename(), "sub", "sub", "gcae_input_files_1_ae"),
        mustWork = FALSE
      ), "/"
    )
  )

  # 2. Train, approx 3 mins
  Sys.time()
  train_filenames <- gcae_train_more(
    gcae_setup = gcae_setup,
    resume_from = 0,
    epochs = 1,
    save_interval = 1,
    verbose = TRUE
  )

with last lines of GCAE output:

-------- Save time: 0.015036 dir: /home/richel/.cache/gcaer/file43ad347905ae/sub/sub/gcae_input_files_1_ae/ae.M1.ex3.b_0_4.gcae_input_files_1.p2/weights/1
Done training. Wrote to /home/richel/.cache/gcaer/file43ad347905ae/sub/sub/gcae_input_files_1_ae/ae.M1.ex3.b_0_4.gcae_input_files_1.p2

[richelbilderbeek]

It is there and it works :-)