Run the following to install the requirements.
cd NeuReach
python3 -m pip install -r requirements.txt
Now you can follow the instructions below to reproduce the results in the paper.
First, in the NeuReach
directory, run the following command to train the NNs and evalaute them for each benchmark.
scripts/runall.sh
Once the above command finishes, you will see the results as in Fig. 2 (Left).
Next, run the following to generate the results in Fig. 1.
python3 scripts/plot_f16_single_timed.py --no_cuda --pretrained log/log_f16_GCAS_ours/checkpoint.pth.tar
python3 scripts/plot_jetengine.py --no_cuda --pretrained_ours log/log_jetengine_ours/checkpoint.pth.tar --pretrained_dryvr log/log_jetengine_dryvr/checkpoint.pth.tar
After it finshes, you should be able to find two images in the working directory: jetEngine.pdf
and f16_GCAS.pdf
.
Run the following command to generate the results in Fig. 2 (Right).
python3 scripts/run_jetE_lambda.py
python3 scripts/plot_lambdas.py
After it finishes, you should be able to find an image in the working directory: lambdas.pdf
. Now we have reproduced all the results in the submission.