TypeError: Query column vector must be a vector. Got list<item: double>.

[bhupender101]

Do you need to file an issue?

• I have searched the existing issues and this bug is not already filed.
• My model is hosted on OpenAI or Azure. If not, please look at the "model providers" issue and don't file a new one here.
• I believe this is a legitimate bug, not just a question. If this is a question, please use the Discussions area.

Describe the bug

TypeError Traceback (most recent call last)
Cell In[11], line 1
----> 1 graph.local_search(query="ok")

File ~/Desktop/Genai/bloomfire-backend/rm/msgraph.py:252, in MSGraphRAGSearch.local_search(self, query)
242 def local_search(self, query: str):
243 """
244 Perform a local search with the given query.
245
(...)
250 Dictionary containing search results and metadata
251 """
--> 252 result = self.local_search_engine.search(query=query)
253 return {
254 "response": convert_response_to_string(result.response),
255 "context_data": result.context_data,
(...)
259 "prompt_tokens": result.prompt_tokens,
260 }

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/graphrag/query/structured_search/local_search/search.py:158, in LocalSearch.search(self, query, conversation_history, **kwargs)
156 start_time = time.time()
157 search_prompt = ""
--> 158 context_text, context_records = self.context_builder.build_context(
159 query=query,
160 conversation_history=conversation_history,
161 **kwargs,
162 **self.context_builder_params,
163 )
164 log.info("GENERATE ANSWER: %d. QUERY: %s", start_time, query)
165 try:

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/graphrag/query/structured_search/local_search/mixed_context.py:140, in LocalSearchMixedContext.build_context(self, query, conversation_history, include_entity_names, exclude_entity_names, conversation_history_max_turns, conversation_history_user_turns_only, max_tokens, text_unit_prop, community_prop, top_k_mapped_entities, top_k_relationships, include_community_rank, include_entity_rank, rank_description, include_relationship_weight, relationship_ranking_attribute, return_candidate_context, use_community_summary, min_community_rank, community_context_name, column_delimiter, **kwargs)
135 pre_user_questions = "\n".join(
136 conversation_history.get_user_turns(conversation_history_max_turns)
137 )
138 query = f"{query}\n{pre_user_questions}"
--> 140 selected_entities = map_query_to_entities(
141 query=query,
142 text_embedding_vectorstore=self.entity_text_embeddings,
143 text_embedder=self.text_embedder,
144 all_entities_dict=self.entities,
145 embedding_vectorstore_key=self.embedding_vectorstore_key,
146 include_entity_names=include_entity_names,
147 exclude_entity_names=exclude_entity_names,
148 k=top_k_mapped_entities,
149 oversample_scaler=2,
150 )
152 # build context
153 final_context = liststr

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/graphrag/query/context_builder/entity_extraction.py:57, in map_query_to_entities(query, text_embedding_vectorstore, text_embedder, all_entities_dict, embedding_vectorstore_key, include_entity_names, exclude_entity_names, k, oversample_scaler)
53 matched_entities = []
54 if query != "":
55 # get entities with highest semantic similarity to query
56 # oversample to account for excluded entities
---> 57 search_results = text_embedding_vectorstore.similarity_search_by_text(
58 text=query,
59 text_embedder=lambda t: text_embedder.embed(t),
60 k=k * oversample_scaler,
61 )
62 for result in search_results:
63 if embedding_vectorstore_key == EntityVectorStoreKey.ID and isinstance(
64 result.document.id, str
65 ):

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/graphrag/vector_stores/lancedb.py:136, in LanceDBVectorStore.similarity_search_by_text(self, text, text_embedder, k, **kwargs)
134 query_embedding = text_embedder(text)
135 if query_embedding:
--> 136 return self.similarity_search_by_vector(query_embedding, k)
137 return []

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/graphrag/vector_stores/lancedb.py:115, in LanceDBVectorStore.similarity_search_by_vector(self, query_embedding, k, **kwargs)
101 docs = (
102 self.document_collection.search(
103 query=query_embedding, vector_column_name="vector"
(...)
107 .to_list()
108 )
109 else:
110 docs = (
111 self.document_collection.search(
112 query=query_embedding, vector_column_name="vector"
113 )
114 .limit(k)
--> 115 .to_list()
116 )
117 return [
118 VectorStoreSearchResult(
119 document=VectorStoreDocument(
(...)
127 for doc in docs
128 ]

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/lancedb/query.py:320, in LanceQueryBuilder.to_list(self)
312 def to_list(self) -> List[dict]:
313 """
314 Execute the query and return the results as a list of dictionaries.
315
(...)
318 fields are returned whether or not they're explicitly selected.
319 """
--> 320 return self.to_arrow().to_pylist()

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/lancedb/query.py:647, in LanceVectorQueryBuilder.to_arrow(self)
638 def to_arrow(self) -> pa.Table:
639 """
640 Execute the query and return the results as an
641 Apache Arrow Table.
(...)
645 vector and the returned vectors.
646 """
--> 647 return self.to_batches().read_all()

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/lancedb/query.py:678, in LanceVectorQueryBuilder.to_batches(self, batch_size)
664 vector = [v.tolist() for v in vector]
665 query = Query(
666 vector=vector,
667 filter=self._where,
(...)
676 offset=self._offset,
677 )
--> 678 result_set = self._table._execute_query(query, batch_size)
679 if self._reranker is not None:
680 rs_table = result_set.read_all()

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/lancedb/table.py:1742, in LanceTable._execute_query(self, query, batch_size)
1733 if len(query.vector) > 0:
1734 nearest = {
1735 "column": query.vector_column,
1736 "q": query.vector,
(...)
1740 "refine_factor": query.refine_factor,
1741 }
-> 1742 return ds.scanner(
1743 columns=query.columns,
1744 limit=query.k,
1745 filter=query.filter,
1746 prefilter=query.prefilter,
1747 nearest=nearest,
1748 full_text_query=query.full_text_query,
1749 with_row_id=query.with_row_id,
1750 batch_size=batch_size,
1751 offset=query.offset,
1752 ).to_reader()

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/lance/dataset.py:369, in LanceDataset.scanner(self, columns, filter, limit, offset, nearest, batch_size, batch_readahead, fragment_readahead, scan_in_order, fragments, full_text_query, prefilter, with_row_id, with_row_address, use_stats, fast_search, io_buffer_size)
367 builder = builder.full_text_search(**full_text_query)
368 if nearest is not None:
--> 369 builder = builder.nearest(**nearest)
370 return builder.to_scanner()

File ~/miniconda3/envs/genai/lib/python3.11/site-packages/lance/dataset.py:2449, in ScannerBuilder.nearest(self, column, q, k, metric, nprobes, refine_factor, use_index, ef)
2447 column_type = column_type.storage_type
2448 if not pa.types.is_fixed_size_list(column_type):
-> 2449 raise TypeError(
2450 f"Query column {column} must be a vector. Got {column_field.type}."
2451 )
2452 if len(q) != column_type.list_size:
2453 raise ValueError(
2454 f"Query vector size {len(q)} does not match index column size"
2455 f" {column_type.list_size}"
2456 )

TypeError: Query column vector must be a vector. Got list<item: double>.

Steps to reproduce

This is the Error when i am Doing the Local Searech and Global Seach

Expected Behavior

No response

GraphRAG Config Used

# Paste your config here

Logs and screenshots

No response

Additional Information

• GraphRAG Version:
• Operating System:
• Python Version:
• Related Issues:

[MBRSL]

Dup #1335

[donaldNtjana]

loading 'create_final_entities.parquet' and creating description_embedding worked for me when doing 'local search'.

# Concatenate name and description for embedding
entity_embedding_df["name_description"] = (
    entity_embedding_df["name"] + ":" + entity_embedding_df["description"]
)

# Run embedding
entity_embedding_df["description_embedding"] = embed_text(entity_embedding_df["name_description"]) 

[rivera2387]

@donaldNtjana What is embed_text? I'm getting error saying embed_text is not defined.

[donaldNtjana]

embed_text would be the function to call to embed your column/text.

Use the same embedding model you specified in your config. I used the Open AI embedding

from graphrag.query.llm.oai.embedding import OpenAIEmbedding
def embed_text(column):

    text_embedder = OpenAIEmbedding(
            api_key="your_api_key",
            api_base ="your_api_base",
            model = "your_model"
    )

    return column.apply(lambda x: text_embedder.embed(x))