Question on running the predictor for inference.

Great work, and it is nice to share codes with clean document! However, I tried to run Predictor/promoter_predictor.py for inference, but I got errors like:

FileNotFoundError: [Errno 2] No such file or directory: '../case8_consituent/case8_constituent_seqsresults.csv'

Did I missed something? BTW, I trained the model ;).

Another minor issue occured during the training of the predictor:

The model type is : 
denselstm
/data4/promoter/deepseed/Predictor/SeqRegressionModel.py:701: UserWarning: nn.init.kaiming_normal is now deprecated in favor of nn.init.kaiming_normal_.
  nn.init.kaiming_normal(m.weight.data)
Training iters
  0%|                                                                                                                                                                       | 0/88 [00:00<?, ?it/s]Could not load symbol cublasGetSmCountTarget from libcublas.so.11. Error: /usr/local/cuda/lib64/libcublas.so.11: undefined symbol: cublasGetSmCountTarget

I can not run the training code with the version suggested in the README file (I use a RTX 3070 therefore I have to use cuda11). I am not sure will this be harmless or bad? FYI, the training did converge:

100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 11/11 [00:00<00:00, 39.14it/s]
epoch:32train_loss y:1.185442566871643 test_loss y:5.481445789337158 test_coefs:0.7354892771702892
EarlyStopping counter: 10 out of 10

Thanks!

WangLabTHU / deepseed

Question on running the predictor for inference. #1