Speeding up deep learning training in practice Part I : local machineEfficient data pipeline are the core of model training and with deep understanding on the topic is crucial for scaling your projectMay 28, 2023A response icon1May 28, 2023A response icon1
[Paper review] Accelerating Large-Scale Inference with Anisotropic Vector QuantizationTable of ContentsDec 15, 2022A response icon1Dec 15, 2022A response icon1
[Paper review] PECOS: Predictions for Enormous and Correlated Output Spaces : Part Iminimizing search space for productionDec 5, 2022Dec 5, 2022
[Paper review] Monolith: TikTok’s Real-time Recommender System.TikTok’s flexible architecture that allows both real-time and batch recommender system.Nov 19, 2022A response icon2Nov 19, 2022A response icon2
Matrix Factorization Part III: Production phaseGoing over simple paper implementation, into what we need to do to be production ready.Nov 6, 2022Nov 6, 2022
Matrix Factorization Part II: SGD, optimized implementationOptimizing Matrix factorization with SGD.Nov 1, 2022Nov 1, 2022
Matrix Factorization Part I: Understanding all the processes and how stochastic gradient descent…One of traditional algorithm in recommender system, Matrix Factorization. It was first proposed by Simon Funk in 2006 and has been widely…Aug 20, 2022Aug 20, 2022
Customizing keras-tuner for flexibility and maintainabilityJul 5, 2022A response icon1Jul 5, 2022A response icon1
Hyperparameter tuning with keras-tuner full tutorialTuning with keras tuner basicsJun 19, 2022A response icon2Jun 19, 2022A response icon2
Linear Algebra — Singular Value Decomposition part IWho said mathematics is not practical subject? think again.Apr 23, 2022Apr 23, 2022