Removed multiprocessing during training

kundajelab / bpnet

Toolkit to train base-resolution deep neural networks on functional genomics data and to interpret them

http://bit.ly/bpnet-colab

MIT License

141 stars 33 forks source link

Removed multiprocessing during training #34

Closed albertjstanley closed 2 years ago

albertjstanley commented 2 years ago

According to the comment, the intention is to stop multiprocessing after loading the data into memory. However, setting num_workers = 1 results in a creation of an extra process later on when using the custom data loader.

Having the extra worker resulted in deadlocks when I worked with larger data.

Corrected by setting num_workers = 0.

Avsecz commented 2 years ago

Sorry for the delay. This looks good. Happy to merge this PR. Thanks!