https://databricks.com/session_na20/end-to-end-deep-learning-with-horovod-on-apache-spark 介绍uber的深度学习整体流程，结合horovod和spark，如何基于spark构建端到端的DL https://databricks.com/session_na20/accelerating-mlflow-hyper-parameter-optimization-pipelines-with-rapids 使用MLFlow帮助AI开发者，利用hyperopt进行调参，RAPIDS加速 https://databricks.com/session_na20/how-not-to-scale-deep-learning-in-6-easy-steps sowen简单的例子介绍scale的手段，GPU、horovod、petastorm https://databricks.com/session_na21/enabling-vectorized-engine-in-apache-spark IBM vectorize https://databricks.com/session_na21/stage-level-scheduling-improving-big-data-and-ai-integration stage level sche HPC: [CENTOS AND HPC: IT’S OKAY, WE ARE MOVING ON](https://www.nextplatform.com/2020/12/21/centos-and-hpc-its-okay-we-are-moving-on/) [DOCKER VS SINGULARITY VS SHIFTER VS UGE CONTAINER EDITION](https://tin6150.github.io/psg/blogger_container_hpc.html) https://github.com/hpcng
https://databricks.com/session_na20/end-to-end-deep-learning-with-horovod-on-apache-spark
介绍uber的深度学习整体流程，结合horovod和spark，如何基于spark构建端到端的DL
https://databricks.com/session_na20/accelerating-mlflow-hyper-parameter-optimization-pipelines-with-rapids
使用MLFlow帮助AI开发者，利用hyperopt进行调参，RAPIDS加速
https://databricks.com/session_na20/how-not-to-scale-deep-learning-in-6-easy-steps
sowen简单的例子介绍scale的手段，GPU、horovod、petastorm
https://databricks.com/session_na21/enabling-vectorized-engine-in-apache-spark
IBM vectorize
https://databricks.com/session_na21/stage-level-scheduling-improving-big-data-and-ai-integration
stage level sche
HPC:
CENTOS AND HPC: IT’S OKAY, WE ARE MOVING ON
DOCKER VS SINGULARITY VS SHIFTER VS UGE CONTAINER EDITION
https://github.com/hpcng