feat: add documentation for mlflow autologging on website #1508
Conversation
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Codecov Report
@@ Coverage Diff @@
## master #1508 +/- ##
==========================================
- Coverage 84.30% 82.75% -1.55%
==========================================
Files 296 296
Lines 14906 14906
Branches 717 717
==========================================
- Hits 12566 12336 -230
- Misses 2340 2570 +230
Continue to review full report at Codecov.
|
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
website/docs/mlflow/autologging.md
Outdated
| 2. Upload your customized `log_model_allowlist.txt` file to dbfs by clicking File/Upload Data button on Databricks UI. | ||
| 3. Set Spark configuration: | ||
| ``` | ||
| spark.conf.set("spark.mlflow.pysparkml.autolog.logModelAllowlistFile", "/dbfs/FileStore/PATH_TO_YOUR_log_model_allowlist.txt") |
There was a problem hiding this comment.
does this accept URLs? Perhaps we can host a reasonable default in our blob!
There was a problem hiding this comment.
We might also want to mention this can be set in cluster configs too
There was a problem hiding this comment.
Thanks for this idea! I'll raise a PR to mlflow to make the URL work lol this sounds so reasonable. And I just tested that the above spark.conf.set doesn't work as the cluster is already started, I'll change it to add spark config inside cluster configuration.
There was a problem hiding this comment.
| ## Configuration process in Databricks as an example | ||
|
|
||
| 1. Install MLflow via `%pip install mlflow` | ||
| 2. Upload your customized `log_model_allowlist.txt` file to dbfs by clicking File/Upload Data button on Databricks UI. |
There was a problem hiding this comment.
if we can pass a URL in this param, perhaps we can make this platform agnostic or at least give advice for both Synapse and Databricks
There was a problem hiding this comment.
Sure, I'll add example in step 2 in section To enable autologging for SynapseML
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Summary
Add documentation for Mlflow autologgin on the website.
Tests
Rendered website and check
Dependency changes
None.
AB#1785118