yeandy commented on issue #21894:
URL: https://github.com/apache/beam/issues/21894#issuecomment-1183632885
1. Seems like an extra friction point as it will be an additional thing
users would have to learn and implement. Doable, but prefer not to do this over
choice 2 (see below).
2. I think a no-op would be the easiest to for others to use/understand. We
can add addition documentation explaining how this parameter is optional to use.
3. If it's moved into `base.py`, could you illustrate how would we make the
logic different for the different frameworks?
4. Would this cause compatibility issues with the existing interface for the
`run_inference` function since it expects `inference_args`?
```
def run_inference(
self,
batch: Sequence[ExampleT],
model: ModelT,
inference_args: Optional[Dict[str, Any]] = None) ->
Iterable[PredictionT]:
```
--
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]
