Closed louislau1129 closed 4 months ago
Hi, I think it may be an OS-specific problem, on which OS are you running this? Also you can try to change the backend if needed: here you can change the backend adding a parameter backend='multiprocessing'
for example reference.
Of course if it works, it should be done also for validation and test sets.
Hope this helps.
When I use
n_jobs> 1
, it shows: using backend locybackend withn_job
concurrent workers. fatal python error: init_sys_streams: can't initilalize sys standard streams, [OS Error] bad file descriptor.When n_jobs=1, everthing is ok. But I would like to download faster by parallel. Having tried increase the
ulimit
for allowing the increased open file descriptors. and reduce the verbose level, nothing works. Any idea to solve this parallel issue? Thanks in advance!