Add more benchmarks

[uditagarwal97]

To make LLTFI more robust and usable, we should add more benchmarks like PyTorch-based ML models, Caffe2-based ML models, and ML models from Apollo Baidu's autonomous vehicle simulator.