Questions about the most relevant neurons

[JiachengXu123]

Hello, thanks for your excellent work.
I'm trying to reproduce the experiments. After reading through the source codes, I have two questions.
First, why does "np.argsort(totalR[subject_layer])[0][::-1][:num_rel]" refer to the most relevant neurons? When I set the num_rel to be 6, it seems that the output of "np.argsort(totalR[subject_layer])[0][::-1]" equals to the output of "np.argsort(totalR[subject_layer])[0][::-1][:num_rel]". The subject layers in this function is 3. The shape of "np.argsort(totalR[subject_layer])[0][::-1][:num_rel]" is (4, 4, 12).
Second, when calculating the IDC coverage, I got an error. The command ">python run.py -M lenet1 -DS mnist -L 1 -A idc -KS 10000 -KN 3 -RN 6 -C 0" is leveraged as input.
Traceback (most recent call last):
File "run.py", line 206, in
coverage, covered_combinations, max_comb = idc.test(X_test)
File "E:\githubAwesomeCode\1DLTesting\sadl_improve\coverages\idc.py", line 106, in test
coverage, covered_combinations, max_comb = measure_idc(self.model, self.model_name,
File "E:\githubAwesomeCode\1DLTesting\sadl_improve\coverages\idc.py", line 265, in measure_idc
lout.append(np.mean(test_layer_outs[subject_layer][test_idx][...,r]))
IndexError: index 8 is out of bounds for axis 2 with size 4.

Thank you very much for your kind consideration and I am looking forward to your early reply.