search

ChuhuaW / SGNet.pytorch

Pytorch Implementation for Stepwise Goal-Driven Networks for Trajectory Prediction (RA-L/ICRA2022)
117 stars 16 forks source link

Run train_cave.py on win10, why there is no result? #5

Closed yanshuaibupt closed 2 years ago

ChuhuaW commented 2 years ago

Thank you for your interest in our paper! Could you please provide more details of the issue? What do you mean by no results?

yanshuaibupt commented 2 years ago

The code directory is as follows: -configs -data -- JAAD -lib -eval_cave.py -train_cave.py I put both configs and lib under tools/jaad, because I encountered an import error prompt in anaconda.
When I run python tools/jaad/train_cvae.py --gpu cuda --dataset data/JAAD --model SGNet_CVAE at the prompt, there is no printing information, and the program ends directly without an error.

ChuhuaW commented 2 years ago

For the import error, you may want to try `export PYTHONPATH=$PWD' under './SGNet.pytorch'. For '--gpu' flag, you should use your gpu index, since we need it to set variable 'CUDA_VISIBLE_DEVICES'. If the problem persists, I recommend evaluating the code with breakpoints (e.g., pdb). Thanks!

yanshuaibupt commented 2 years ago

Thanks，I found some python package call problems. Changing the relative references of SGNet.py/SGNet_CAVE.py and other files to absolute references can solve the reference problem on win10.