I will appreciate if anyone suggests a simple and efficient way to manage and save experiment results?
What would you like to save?
If you are referring to the model and optimizer, have a look at the Serialization docs.
Actually, my question was mainly about experiment management, how to monitor your model in organized way (e.g., hyper-parameters, loss, plots,). So far I use excel sheet to save my results and corresponding hyper parameters manually, I was wondering if there simple and effective tool to save results and help to do visualization also.
Would appreciate, if you as an expert, present your best practices for managing ML experiment.
Oh, I’m no expert in managing ML experiments, so let’s wait for the real experts to give their opinion, which tools to use.
https://wandb.ai is a great tool for that. Hope it helps.
A variety of tools is reported here: https://neptune.ai/blog/best-ml-experiment-tracking-tools
Facing the same problem, getting lost in too many experiments…
I would like to thank you, for the last 3 months I have been using wandb.ai, and all what I can say that it is really amazing
I’d like to put in a plug for runx, which is very lightweight, and we use it in many of my group’s research projects. https://github.com/NVIDIA/runx.