adds custom vectorizer support #156
Conversation
|
|
||
| super().__init__(model=model, dims=self._set_model_dims()) | ||
|
|
||
| def _validate_embed(self, func: Callable): |
There was a problem hiding this comment.
Given this is wrapping a pydantic BaseModel object, these validators should be implemented using the validator operators. (there should be some other examples of this here)
There was a problem hiding this comment.
Wrapping these with @validator decorators calls them before the embed methods are set in super().__init__. Pydantic enforces a certain order in initialization so we end up getting trapped in a circular dependency.
There was a problem hiding this comment.
Fine for now, but we will look at this later though (maybe for 0.3.0) because I believe we can find a cleaner way to use built-in validation, pre or post init.
|
@justin-cechmanek Can you work on rebasing these changes on top of main? Then we can get this in today. |
This PR introduces a new CustomTextVectorizer class that allows for users to wrap their own embedding functions and methods to be compatible with RedisVL.