Comment by vrighter

2 days ago

Which models release the full training dataset, the training code, the inference code, and any random seeds used? Without all of those, you can't "compile" the project. If you can't compile it yourself, then it's not open source.

Compiling the training or inference code does not give you a model.