TODO
- decide params
- 200 is enough? how many above 200?
- 20 is enough? how many nodes?
- prepare dataset
- run over the obj file
- generate code (if the vertex is dup, just remove the file!)
- also check the not 000 file, maybe some error happens? -> no out?
- now you have a pure code text file, as the dataset!
- run the train code?
- also the inference code!
- then ask mentor to distribute it to other machines!