Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

docs: small updates to astra vectorize docs #2497

Merged
merged 2 commits into from
Jul 4, 2024
Merged

docs: small updates to astra vectorize docs #2497

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e6d0e24
small updates to vectorize docs
jordanrfrazier Jul 2, 2024
cdb3e50
Merge branch 'main' into vectorize-doc-updates
ogabrielluiz Jul 3, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 14 additions & 13 deletions src/backend/base/langflow/components/embeddings/AstraVectorize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,37 +50,38 @@ class AstraVectorize(Component):
inputs = [
DropdownInput(
name="provider",
display_name="Provider name",
display_name="Provider",
options=VECTORIZE_PROVIDERS_MAPPING.keys(),
value="",
required=True,
),
MessageTextInput(
name="model_name",
display_name="Model name",
display_name="Model Name",
info=f"The embedding model to use for the selected provider. Each provider has a different set of models "
f"available (full list at https://docs.datastax.com/en/astra-db-serverless/databases/embedding-generation.html):\n\n{VECTORIZE_MODELS_STR}",
f"available (https://docs.datastax.com/en/astra-db-serverless/databases/embedding-generation.html):\n\n{VECTORIZE_MODELS_STR}",
required=True,
),
SecretStrInput(
name="provider_api_key",
display_name="Provider API Key",
info="An alternative to the Astra Authentication that passes an API key for the provider with each request to Astra DB. This may be used when Vectorize is configured for the collection, but no corresponding provider secret is stored within Astra's key management system.",
),
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I made this appear by default. The reasoning is that
a) This is the quickest path to testing out vectorize on a new collection
b) If the collection is already configured for vectorize, but the key was not stored in astra, this is (by my guess) the most common pattern for authentication.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think the most common pattern is to store keys on astra, that's why I've set api_key_name in the front page.

If you go through the documentation the steps are to import the key in astra and refer to it when creating the collection

MessageTextInput(
name="api_key_name",
display_name="API Key name",
display_name="Provider API Key Name",
info="The name of the embeddings provider API key stored on Astra. If set, it will override the 'ProviderKey' in the authentication parameters.",
advanced=True,
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there a common case where the providerKey is set differently than the default?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

what do you mean? there's no default value. The value is defined by the user

Copy link
Collaborator Author

@jordanrfrazier jordanrfrazier Jul 3, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I believe it does default just to providerKey, like so:

CollectionVectorServiceOptions(
           ...
            authentication={
                "providerKey": os.environ["OPENAI_API_KEY"],
           }
        )

But going through the UI flow, I see that you are correct in your other comment that the user-defined key name is required during creation. So I will undo these changes, and have Provider API Key Name shown by default, and Provider API Key in Advanced.

),
DictInput(
name="authentication",
display_name="Authentication parameters",
display_name="Authentication Parameters",
is_list=True,
advanced=True,
),
SecretStrInput(
name="provider_api_key",
display_name="Provider API Key",
info="An alternative to the Astra Authentication that let you use directly the API key of the provider.",
advanced=True,
),
DictInput(
name="model_parameters",
display_name="Model parameters",
display_name="Model Parameters",
advanced=True,
is_list=True,
),
Expand All @@ -96,7 +97,7 @@ def build_options(self) -> dict[str, Any]:
if api_key_name:
authentication["providerKey"] = api_key_name
return {
# must match exactly astra CollectionVectorServiceOptions
# must match astrapy.info.CollectionVectorServiceOptions
"collection_vector_service_options": {
"provider": provider_value,
"modelName": self.model_name,
Expand Down
36 changes: 19 additions & 17 deletions src/backend/base/langflow/components/vectorstores/AstraDB.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,6 +116,7 @@ class AstraVectorStoreComponent(LCVectorStoreComponent):
name="embedding",
display_name="Embedding or Astra Vectorize",
input_types=["Embeddings", "dict"],
info="Allows either an embedding model or an Astra Vectorize configuration. If Astra Vectorize is already configured for the collection, this field is not required.",
Copy link
Contributor

@nicoloboschi nicoloboschi Jul 3, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

did you try what happens if you don't set it ?
I believe the AstraDBVectorStore requires one of those.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ugh, yeah. I guess that's something to figure out. It's required by langchain-astradb to have one or the other at the moment.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Perhaps we could loosen that constraint iff the setup mode is SetupMode.OFF?

),
StrInput(
name="metadata_indexing_exclude",
Expand Down Expand Up @@ -163,6 +164,7 @@ class AstraVectorStoreComponent(LCVectorStoreComponent):
def _build_vector_store_no_ingest(self):
if self._cached_vectorstore:
return self._cached_vectorstore

try:
from langchain_astradb import AstraDBVectorStore
from langchain_astradb.utils.astradb import SetupMode
Expand Down Expand Up @@ -225,11 +227,6 @@ def _build_vector_store_no_ingest(self):

return vector_store

def build_vector_store(self):
vector_store = self._build_vector_store_no_ingest()
self._add_documents_to_vector_store(vector_store)
return vector_store

def _add_documents_to_vector_store(self, vector_store):
documents = []
for _input in self.ingest_data or []:
Expand All @@ -255,6 +252,18 @@ def _map_search_type(self):
else:
return "similarity"

def _build_search_args(self):
args = {
"k": self.number_of_results,
"score_threshold": self.search_score_threshold,
}

if self.search_filter:
clean_filter = {k: v for k, v in self.search_filter.items() if k and v}
if len(clean_filter) > 0:
args["filter"] = clean_filter
return args

def search_documents(self) -> list[Data]:
vector_store = self._build_vector_store_no_ingest()
self._add_documents_to_vector_store(vector_store)
Expand Down Expand Up @@ -282,21 +291,14 @@ def search_documents(self) -> list[Data]:
logger.debug("No search input provided. Skipping search.")
return []

def _build_search_args(self):
args = {
"k": self.number_of_results,
"score_threshold": self.search_score_threshold,
}

if self.search_filter:
clean_filter = {k: v for k, v in self.search_filter.items() if k and v}
if len(clean_filter) > 0:
args["filter"] = clean_filter
return args

def get_retriever_kwargs(self):
search_args = self._build_search_args()
return {
"search_type": self._map_search_type(),
"search_kwargs": search_args,
}

def build_vector_store(self):
vector_store = self._build_vector_store_no_ingest()
self._add_documents_to_vector_store(vector_store)
return vector_store
Loading