Microsoft Azure Distributed Linear Learner Recipe #195
Conversation
| @@ -0,0 +1,27 @@ | |||
| ## MADL-CPU-OpenMPI Data Shredding | |||
| We included a python script that shows how to shred and deploy your training data prior to running an Azure training job. | |||
There was a problem hiding this comment.
minor:
prior to running a training job on Azure VMs via Open MPI.
recipes/MADL-CPU-OpenMPI/README.md
Outdated
|
|
||
| ### Pool Configuration | ||
| The pool configuration should enable the following properties: | ||
| * `vm_size` should be a CPU-only instance, 'STANDARD_D2_V2'. |
There was a problem hiding this comment.
be a CPU-Only instance, for example, STANDARD_D2_V2.
| * `-d` log global models to this directory at the host" | ||
| * `-b` location for the algorithm's binary" | ||
|
|
||
| * The training data will need to be shredded to match the number of VMs and the thread's count per VM, and then deployed to a mounted Azure blob that the VM docker images have read/write access. |
There was a problem hiding this comment.
You should provide the configuration example for this here, e.g.:
shared_data_volumesshould contain the shared data volume with anazureblobvolume driver as specified in the global configuration file.
There was a problem hiding this comment.
I included the part you added with a link to the full configuration file.
recipes/MADL-CPU-OpenMPI/README.md
Outdated
| * `multi_instance` property must be defined | ||
| * `num_instances` should be set to `pool_current_dedicated`, or | ||
| `pool_current_low_priority` | ||
| * `coordination_command` should be unset or `null`. |
There was a problem hiding this comment.
Recommend eliminating coordination_command and resource_files bullets since they are not needed.
| @@ -0,0 +1,27 @@ | |||
| ## MADL-CPU-OpenMPI Data Shredding | |||
| We included a python script that shows how to shred and deploy your training data prior to running a training job on Azure VMs via Open MPI. | |||
| Azure VMs via Open MPI. | |||
There was a problem hiding this comment.
Remove this line since you modified above.
| @@ -0,0 +1,27 @@ | |||
| ## MADL-CPU-OpenMPI Data Shredding | |||
| We included a python script that shows how to shred and deploy your training data prior to running a training job on Azure VMs via Open MPI. | |||
There was a problem hiding this comment.
Avoid using "We". Perhaps reword as:
This Data Shredding recipe shows how to shred...
There was a problem hiding this comment.
Please do a quick search on all your markdown files for use of "we" and replace with something else.
| #Dockerfile for MADL (Microsoft Distributed Learners) | ||
|
|
||
| FROM ubuntu:16.04 | ||
| MAINTAINER Saeed Maleki Todd Mytkowicz Madan Musuvathi Dany rouhana <https://github.com/Azure/batch-shipyard> |
| @@ -0,0 +1,27 @@ | |||
| ## MADL-CPU-OpenMPI Data Shredding | |||
| This Data Shredding recipe shows how to shred and deploy your training data prior to running a training job on Azure VMs via Open MPI. | |||
| Azure VMs via Open MPI. | |||
There was a problem hiding this comment.
Looks like you might have missed the prior comment: remove line 3.
No description provided.