[Community]: AzureSearchVectorStoreRetriever Fixed to account for search_kwargs #21572
Conversation
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎ 1 Ignored Deployment
|
dosubot
bot
added
size:M
|
@baskaryan please review |
| elif self.search_type == "similarity_score_threshold": | ||
| docs = [ | ||
| doc | ||
| for doc, _ in self.vectorstore.similarity_search_with_relevance_scores( | ||
| query, k=self.k, **kwargs | ||
| query, k=self.k, **self.search_kwargs, **kwargs |
There was a problem hiding this comment.
Same as above here (and below). Defining params before the if clause would be much cleaner.
Co-authored-by: Massimiliano Pronesti <massimiliano.pronesti@gmail.com>
|
@mspronesti I have made your changes. Please check |
|
@keenborder786 Thanks for taking my suggestion into consideration -- guess you forgot to apply these changes in the last 3 |
|
@baskaryan, @efriis, @eyurtsev, @hwchase17 please review |
…angchain into azure_search_fix
dosubot
bot
added
size:S
|
@keenborder786 There's a small conflict which arose from my latest PR being merged. Resolving it should only require replacing @ccurme Could you please review this as well? :) |
okay, I will resolve it. |
|
@mspronesti fixed |
|
@eyurtsev please see |
dosubot
bot
added
the
lgtm
| @@ -740,6 +739,18 @@ class AzureSearchVectorStoreRetriever(BaseRetriever): | |||
| or "semantic_hybrid_score_threshold".""" | |||
| k: int = 4 | |||
| """Number of documents to return.""" | |||
| search_kwargs: dict = Field( | |||
| default_factory=lambda: {"fetch_k": 20, "lambda_mult": 0.5} | |||
There was a problem hiding this comment.
can we just default to empty dict here and let vecstore set defaults?
There was a problem hiding this comment.
@baskaryan done
…r search_kwargs (langchain-ai#21572) - **Description:** Fixed `AzureSearchVectorStoreRetriever` to account for search_kwargs. More explanation is in the mentioned issue. - **Issue:** langchain-ai#21492 --------- Co-authored-by: MAC <mac@MACs-MacBook-Pro.local> Co-authored-by: Massimiliano Pronesti <massimiliano.pronesti@gmail.com> Co-authored-by: Bagatur <baskaryan@gmail.com>
AzureSearchVectorStoreRetrieverto account for search_kwargs. More explanation is in the mentioned issue.