qlk1123

To optimize a given neural network model, we often apply multiple methods and carry out many trials, which produce a large amount of binaries. Even with git LFS support, handling them gracefully is still a headache.

Any good practices/frameworks to maintain the experimental models?

After finishing "Phoenix Project", I'm reading "DevOps Handbook" and "Effective DevOps." While I find the approach described in the books fascinating, it is also frustrating when trying to apply it, mainly because of cultural differences. Advocating for something and implementing openness are not necessarily virtue in non-western culture.

Any experiences to share?