#batchsize

[doyouknowkrwp]

하이퍼파라미터 튜닝 하나로 AI 성능 2배 - 같은 모델, 완전히 다른 결과의 비밀

AI 성능 30%는 하이퍼파라미터가 결정! Learning Rate 0.01 vs 0.001로 정확도 15% 차이. Grid Search < Random Search < Bayesian Optimization 비교, Optuna 실전 코드, 핵심 파라미터 10가지, Kaggle 상위 1% 전략, AutoML까지 완벽 가이드./ #AutoML #BatchSize #BayesianOptimization #Dropout #GridSearch #Hyperband #HyperparameterTuning #Kaggle #LearningRate #Optuna #RandomSearch #RayTune #WeightDecay #딥러닝최적화 #머신러닝 #모델튜닝 #하이퍼파라미터 Read the full article

[jonfazzaro]

Rocks Pebbles Sand: How to implement in practice

"When you happen across one of these four-leaf clovers, you obviously should do it, regardless of whether you consider it Sand (because it's fast) or a Pebble (because it has impact)."

[agilefictionorg]

In software development, one way to look at set-up time is to consider the time it takes to deploy software. Some organizations take weeks and months to deploy new software, and because of this they put as many features into a release as possible.

Mary and Tom Poppendieck