Xilinx / logicnets

Apache License 2.0
81 stars 26 forks source link

cannot train via remote ssh connection #29

Closed tsp6 closed 1 year ago

tsp6 commented 1 year ago

I have one small clarification, when I connect this docker container to vs code via remote ssh connection and try to debug this issue. When trying to run train.py file in vs code, it is not able to find or read the input dataset that is present in data folder in cybersecurity directory. But the dataset is located in the same folder and training worked in local docker container. when connected via ssh for training it is unble to read the dataset.

Exception has occurred: FileNotFoundError [Errno 2] No such file or directory: 'data/unsw_nb15_binarized.npz' File "/workspace/logicnets/examples/cybersecurity/dataset.py", line 21, in get_preqnt_dataset unsw_nb15_data = np.load(data_file) File "/workspace/logicnets/examples/cybersecurity/train.py", line 332, in dataset['train'] = get_preqnt_dataset(dataset_cfg['dataset_file'], split="train")

If any kind of solution or discussion is appreciated and will be helpful a lot.

tsp6 commented 1 year ago

For this you need to pass the absolute path in train.py file where the path for input dataset is defined. In detail where the --dataset-file is parsed in argument parser and then try to train. The problem will be solved.

nickfraser commented 1 year ago

Thank you reporting this issue and providing your solution.