search

mle-infrastructure / mle-toolbox

Lightweight Tool to Manage Distributed ML Experiments 🛠
https://mle-infrastructure.github.io/mle_toolbox/toolbox/
MIT License
3 stars 1 forks source link

Enhance what is tracked in hyper_df (multi-objective, net ckpt) #13

Closed RobertTLange closed 3 years ago

RobertTLange commented 3 years ago

Store in the hyper_df generated from the search multiple objective/target values that summarize performance. Currently only single one is stored. In update print message simply print configuration for each one of the metrics.

Store in the hyper_df generated from the search a path pointer for the final saved checkpoint. This makes loading the trained agent back in smoother and allows for faster post-processing afterwards. If there are multiple random seeds only store one of them.

Store in the hyper_df generated from the search all different raw random seed ids on which we trained and the number of seeds that were aggregated for the scores.

RobertTLange commented 3 years ago

Addressed in 9a03f4a56021585d47e82979a6b0dd02cdbd865e and 737771a2704f0c58cd9c68d594359871972287a8