Model Loading on Pytorch

### Backstory
I have to load and run the models using pytorch. I dont have to run the direct model on the kinetics dataset, instead i have to tweak some layes, and then rerun it for some other task.

So to load the model in pytorch. Could you help with the same.
I tried loading using torch.load but that fails i guess, because the models have originally been saved through the caffe2 framework

I read all the other issues on this, so all the solutions include just changing the checkpointtype to caffe2 in the terminal command statement.
But the run_net.py is basically to ((frame model, load checkpoint, and run on data))
whereas the usecase i want is just to load the model into my python code and use it just another pytorch model.
I just want to load the caffe2 model, load its checkpoints into it, and just save it as a .pt model file, that i can use later.

### The main question
Can anyone help me with steps, to do the same.
I just need to load the pretrained models, in my pytorch python code and then run is accordingly later on.