Embedder error using vertexai embeddings

Crews
,
1

Hi I’m attempting to include memory in my agents using vertexai embedding models but keep running into validation error for the embedder parameter. I’ve try using a dictionary as well through google ai but run into a value error. Any tips to solve this?

image
image628×341 9.48 KB

image
image1200×140 16.3 KB

1 Like
2

I have same error with this. I don’t know why but may be the pydantic version issue that pydantic version 2 need dictionary for the input data. I also try with dictionary format but error occurs.

  embedding_config={
				"provider": "vertexai",
				"config": {
					"api_key": "my api key",
					"model_name": "text-multilingual-embedding-002",
     				"region": "us-central1",
                    "project_id": "myprojectid",
				}
			}
  #for the crew 
  return Crew(
			agents=self.agents,
			tasks=self.tasks,
			process=Process.sequential,
			verbose=True,
			knowledge_sources=[self.date_source],
			memory=True,
			embedder=embedding_config,
		)

[2025-01-31 08:49:25][ERROR]: Failed to upsert documents: Expected Embedings to be non-empty list or numpy array, got in upsert.

[2025-01-31 08:49:25][WARNING]: Failed to init knowledge: Expected Embedings to be non-empty list or numpy array, got in upsert.
ERROR:root:Error during short_term search: Expected Embedings to be non-empty list or numpy array, got in query.
ERROR:root:Error during entities search: Expected Embedings to be non-empty list or numpy array, got in query

I’m not sure but the document for the vertex ai can be outdated.