Comments on: MXNet Distributed Training with Azure ML (Custom Configuration Sample)

By: Deploy Model with Custom Docker Image in Azure Machine Learning – tsmatz

Deploy Model with Custom Docker Image in Azure Machine Learning – tsmatz — Thu, 05 Aug 2021 09:56:55 +0000

[…] my early post, I have showed you how to bring your own custom docker image in training with Azure Machine […]

LikeLike

By: Jinpeng Wang

Jinpeng Wang — Mon, 13 May 2019 01:36:42 +0000

In reply to Tsuyoshi Matsuzaki.

Thank you.

LikeLike

By: Tsuyoshi Matsuzaki

Tsuyoshi Matsuzaki — Sat, 11 May 2019 20:49:09 +0000

Hi Jinpeng-san,
If you’re going to run pytorch with distributed training, it’s easy to use PyTorch Estimator in AML. (As I mentioned in /2018/11/20/azure-machine-learning-services/ , there exists high level API for only Tensorflow, PyTorch, and Chainner in AML.)
Please see https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-train-pytorch .

LikeLike

By: Jinpeng Wang

Jinpeng Wang — Thu, 09 May 2019 16:16:00 +0000

Hi Tsuyoshi,

Do you know how to run the following script in AzureML Service? I have read all the documents, but can’t figure out how to specify parameter “-m torch.distributed.launch –nproc_per_node=4” to python.

python -m torch.distributed.launch –nproc_per_node=4 /path_to_maskrcnn_benchmark/tools/train_net.py –config-file “path/to/config/file.yaml”

LikeLike

By: Azure Update (2019.01.24) | ブチザッキ

Azure Update (2019.01.24) | ブチザッキ — Wed, 23 Jan 2019 20:36:45 +0000

[…] MXNet Distributed Training with Azure ML service (Custom Configuration for Training) […]

LikeLike

By: 7 Reasons to Use Azure Machine Learning Service – tsmatz

7 Reasons to Use Azure Machine Learning Service – tsmatz — Thu, 17 Jan 2019 10:13:58 +0000

[…] : See “Customize Your Azure ML Compute for Training” for the same distributed training without built-in estimator. […]

LikeLike