damccorm commented on code in PR #33406:
URL: https://github.com/apache/beam/pull/33406#discussion_r1895828517
##########
sdks/python/apache_beam/yaml/yaml_ml.py:
##########
@@ -134,17 +137,34 @@ def __init__(
project: str,
location: str,
preprocess: Dict[str, str],
+ postprocess: Optional[Dict[str, str]] = None,
experiment: Optional[str] = None,
network: Optional[str] = None,
private: bool = False,
min_batch_size: Optional[int] = None,
max_batch_size: Optional[int] = None,
max_batch_duration_secs: Optional[int] = None,
- env_vars: Optional[Dict[str, Any]] = None,
- postprocess: Optional[Dict[str, str]] = None):
+ env_vars: Optional[Dict[str, Any]] = None):
"""
ModelHandler for Vertex AI.
+ This Model Handler can be used with RunInference to load a model hosted
+ on VertexAI. Every model that is hosted on VertexAI should have three
+ distinct, required, parameters - `endpoint_id`, `project` and `location`.
+ These parameters tell the Model Handler how to access the model's endpoint
+ so that input data can be sent using an API request, and inferences can be
+ received as a response.
+
+ This Model Handler also required a `preprocess` function to be defined.
Review Comment:
```suggestion
This Model Handler also requires a `preprocess` function to be defined.
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
