ardamamur/EgoExOR ? reverse-engineered prompt

Reverse engineered prompt

Build me a working setup for this EgoExOR project so I can go from the raw dataset to training and evaluating the baseline model without having to figure out all the steps myself.

I want to be able to download the data, merge the HDF5 files, generate the training JSON, and then run the scene graph generation model on the EgoExOR benchmark. Please make the main commands easy to follow, and set things up so the config points to the right data and model paths. If anything is missing or depends on the latest docs, look it up online and keep the defaults sensible. I also want a simple way to run inference with the pretrained model from Hugging Face, and ideally some clear notes in the README for setup, data prep, training, and evaluation so a new user can reproduce the baseline end to end.

Are you gonna build this?

make sure you review the code using coderabbit

Try freeSponsored — opens CodeRabbit in a new tab