[ML] Update trained model inference endpoint #556
Conversation
valeriy42
added
bug
README.md
Outdated
| [API](https://www.elastic.co/guide/en/elasticsearch/reference/current/start-trained-model-deployment.html) | ||
| you will be able to set the threading options to make best use of your | ||
| [API](https://www.elastic.co/guide/en/elasticsearch/reference/current/infer-trained-model.html) | ||
| you will be able to set the threading options to make the best use of your |
There was a problem hiding this comment.
This change isn't necessary. The start trained model deployment API still exists, and is the one where threading options are set.
|
|
||
| self.quantize = ( | ||
| True | ||
| if not (platform.system() == "Darwin" and platform.machine() == "arm64") |
There was a problem hiding this comment.
The changes in this file should go in a separate PR. It will be hard to search for them in the future if they're in a PR that's about updating the model inference endpoint.
README.md
Outdated
| @@ -244,8 +244,8 @@ command line and instead start the model using the ML UI in Kibana. | |||
| The `--start` argument will deploy the model with one allocation and one | |||
| thread per allocation, which will not offer good performance. When starting | |||
| the model deployment using the ML UI in Kibana or the Elasticsearch | |||
| [API](https://www.elastic.co/guide/en/elasticsearch/reference/current/start-trained-model-deployment.html) | |||
| you will be able to set the threading options to make best use of your | |||
| [API](https://www.elastic.co/guide/en/elasticsearch/reference/current/infer-trained-model.html) | |||
There was a problem hiding this comment.
| [API](https://www.elastic.co/guide/en/elasticsearch/reference/current/infer-trained-model.html) | |
| [API](https://www.elastic.co/guide/en/elasticsearch/reference/current/start-trained-model-deployment.html) |
This is referring to the start API
| ("n_node_samples", "<i8"), | ||
| ("weighted_n_node_samples", "<f8"), | ||
| ], | ||
| dtype={ |
There was a problem hiding this comment.
We should update the scikit learn version in requirements-dev.txt and setup.py otherwise these tests will fail for anyone on an older version
Currently it is:
scikit-learn>=0.22.1,<2
Suggest:
scikit-learn>=1.3,<2
|
Thank you for your comments. I moved the fix of unit tests into #558. Once merged, the changes here should be related to the deprecated API only. |
Infer trained model deployment API has been deprecated, so I changed the code to use the new one.