sqlalchemy.exc.StatementError: (builtins.ValueError) expected 4096 dimensions, not 768

General
1

from logging import Logger
from phi.llm.ollama import Ollama
from phi.embedder.ollama import OllamaEmbedder
from phi.vectordb.distance import Distance
from phi.agent import Agent
from phi.knowledge.pdf import PDFUrlKnowledgeBase
from phi.vectordb.pgvector import PgVector, SearchType

常量定义

DB_URL = “tmp/lancedb”
PDF_DIR = “tmp/pdfs”
MODEL_NAME = “qqwen2.5:14b”

创建 Ollama LLM 实例

ollama = Ollama(model=“qwen2.5:14b”)
embedder = OllamaEmbedder(model=MODEL_NAME)

db_url = “postgresql+psycopg://test:test@localhost:5432/test”
knowledge_base = PDFUrlKnowledgeBase(
# Read PDF from this URL
urls=[“https://phi-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf”],
# Store embeddings in the ai.recipes table
vector_db=PgVector(
schema=“test”, table_name=“recipes”, db_url=db_url, search_type=SearchType.hybrid,embedder = embedder, distance = Distance.cosine,
),

)

Load the knowledge base: Comment after first run

knowledge_base.load(upsert=True)

agent = Agent(
model=Ollama(id=“qwen2.5:3b”),
knowledge=knowledge_base,
# Enable RAG by adding references from AgentKnowledge to the user prompt.
enable_rag=True,
# Set as False because Agents default to search_knowledge=True
search_knowledge=False,
markdown=True,
debug_mode=True,
)

–
sqlalchemy.exc.StatementError: (builtins.ValueError) expected 4096 dimensions, not 768

2

Hello @zgli ! You are facing this error as in the recipes tables, you might have inserted some embeddings with 4096 dimensions and now using qqwen2.5:14b, the dimensions of the embedding is 786. Inserting new dimensions in a previously created table causes this error.

Can you please try creating a new table? Additionally, the dimensions of the Model have to be compatible with the dimensions of the Embedder.

1 Like
3

embedder = OllamaEmbedder(model=“nomic-embed-text:latest”, dimensions=768)

I set 768. But the result is still the same, 4096, which cannot be changed.

1 Like