Hello, I am trying to reproduce metrics based on the Wasserstein distance, but I am missing one step of the evaluation pipeline.
- I prepared samples by the downloaded drugflow checkpoint
- I ran evaluation to get their
metrics_aggregated.csv and other files
- Now i need to run
scripts/python/postprocess_metrics.py. The script requires precomputed reference metrics for the dataset (metrics_detailed.csv and metrics_data.pkl), but there's no documented way to generate them.
Could you please provide a script to generate these files? Thank you!
Hello, I am trying to reproduce metrics based on the Wasserstein distance, but I am missing one step of the evaluation pipeline.
metrics_aggregated.csvand other filesscripts/python/postprocess_metrics.py. The script requires precomputed reference metrics for the dataset (metrics_detailed.csv and metrics_data.pkl), but there's no documented way to generate them.Could you please provide a script to generate these files? Thank you!