personal code and scripts for deep learning work
Code is located in model-eval
and scripts to invoke the code are in scripts
. Models are located in model-eval/models_implementation
and that is where most the work is in -- importing a pre-defined model and tweaking parts of it.
The way I use it most of the time is creating a TOML document queueing up a few experiments of similar types with slightly different parameters, and invoking scripts/train2.ipy
on it.