Benchmarking Causal Study to Interpret Large Language Models for Source Code | IEEE Conference Publication | IEEE Xplore