I am trying to profile my DL model to see the bottleneck. On using the following statement to profile it -
python -m torch.utils.bottleneck mycode.py
It keeps on running and running and not printing the expected profile summary. Below is the output-
Loaded pretrained weights for efficientnet-b3
Let's use 3 GPUs!
1536
Epoch: 1 Training Loss: 1.091230 Validation Loss: 1.110597
Valid Accuracy: 37% (203/539)
Validation loss decreased (inf --> 1.110597). Saving model ...
Epoch: 2 Training Loss: 1.092819 Validation Loss: 1.108480
Valid Accuracy: 38% (206/539)
Validation loss decreased (1.110597 --> 1.108480). Saving model ...
Running your script with the autograd profiler...
Loaded pretrained weights for efficientnet-b3
Let's use 3 GPUs!
1536
Epoch: 1 Training Loss: 1.091230 Validation Loss: 1.110597
Valid Accuracy: 37% (203/539)
Validation loss decreased (inf --> 1.110597). Saving model ...
Epoch: 2 Training Loss: 1.092819 Validation Loss: 1.108480