FAISS Vector Store Creator w/ Gemini Embedding Model Selector: ERROR

[sommer_sun]

Using “FAISS Vector Store Creator” in combination with “Gemini Embedding Model Selector” I get the following error message:

ERROR FAISS Vector Store Creator 4:1485 Execute failed: 1 validation error for VertexAIEmbeddings
Value error, ‘gemini-embedding-001’ is not a valid GoogleEmbeddingModelType [type=value_error, input_value={‘project’: 'knime-chatbo… ‘default_metadata’: ()}, input_type=dict]
For further information visit Redirecting...

Using “FAISS Vector Store Creator” with “OpenAI Embedding Model Selector” works fine.

I use KNIME 5.5.1

Can anybody help?

[k10shetty1]

Hi @sommer_sun,

Sorry to hear you’re running into this issue. Could you provide a bit more context on how you’re connecting to the Gemini Embedding Model Selector node?

In KNIME 5.6, I was able to reproduce the issue when using Google Authenticator → Vertex AI Connector → Gemini Embedding Model Selector. However, it works fine with Google AI Studio Authenticator → Gemini Embedding Model Selector.

At the moment, using Gemini embedding models via Google AI Studio should work as expected. We’ll investigate and address the Vertex AI issue soon.

Best,
Keerthan

[sommer_sun]

@k10shetty1

Could you provide a bit more context on how you’re connecting to the Gemini Embedding Model Selector node?

I connected to the Embedding Model Selector in the same was as it is shown in your workflow bottom part.

The embedding model selector is directly connected to the FAISS Vector Store Creator:

Regards
Peter

[k10shetty1]

As mentioned above, at the moment, using Gemini embedding models via Google AI Studio should work as expected. We’ll investigate and address the Vertex AI issue soon. Can you and check if the other way works for you?

Best,
Keerthan

[sommer_sun]

OK, I will try to use Google AI Studio Authenticator as proposed, thank you for support!

regards