There are 3 files here:
1. hindi_tokens.txt --> contains the map of tokens and their id's associated with the model
2. hindi_tl_demo.pt --> torch script model that needs to be made working on sherpa
3. hindi_tl_demo.nemo --> same as hindi_tl_demo.pt, but in it's orginal nvidia nemo format. Storing here, just in case needed for reference