Error when using ollama embeding

[DStarEpoch]

🐛 Describe the bug

env:

mem0ai 0.1.73
ollama 0.4.7
openai 1.68.0

code:

os.environ["OPENAI_API_KEY"] = "..."
openai_client = OpenAI()
config = {
"llm": {
"provider": "ollama",
"config": {
"model": "llama3.1:8b",
"max_tokens": 4000,
"ollama_base_url": "http://localhost:11434",
}
},
"embedder": {
"provider": "ollama",
"config": {
"model": "mxbai-embed-large:335m",
"ollama_base_url": "http://localhost:11434"
}
}
}
memory = Memory.from_config(config)

def chat_with_memories(message: str, user_id: str = "default_user") -> str:
# Retrieve relevant memories
relevant_memories = memory.search(query=message, user_id=user_id, limit=3)
memories_str = "\n".join(f"- {entry['memory']}" for entry in relevant_memories["results"])

# Generate Assistant response
system_prompt = f"You are a helpful AI. Answer the question based on query and memories.\nUser Memories:\n{memories_str}"
messages = [{"role": "system", "content": system_prompt}, {"role": "user", "content": message}]
response = openai_client.chat.completions.create(model="moonshot-v1-8k", messages=messages)
assistant_response = response.choices[0].message.content

# Create new memories from the conversation
messages.append({"role": "assistant", "content": assistant_response})
memory.add(messages, user_id=user_id)

return assistant_response

def main():
print("Chat with AI (type 'exit' to quit)")
while True:
user_input = input("You: ").strip()
if user_input.lower() == 'exit':
print("Goodbye!")
break
print(f"AI: {chat_with_memories(user_input)}")

if name == "main":
main()

running with error:

Chat with AI (type 'exit' to quit)
You: lets play a card game

ValueError Traceback (most recent call last)
Cell In[5], line 11
8 print(f"AI: {chat_with_memories(user_input)}")
10 if name == "main":
---> 11 main()

Cell In[5], line 8, in main()
6 print("Goodbye!")
7 break
----> 8 print(f"AI: {chat_with_memories(user_input)}")

Cell In[4], line 3, in chat_with_memories(message, user_id)
1 def chat_with_memories(message: str, user_id: str = "default_user") -> str:
2 # Retrieve relevant memories
----> 3 relevant_memories = memory.search(query=message, user_id=user_id, limit=3)
4 memories_str = "\n".join(f"- {entry['memory']}" for entry in relevant_memories["results"])
6 # Generate Assistant response

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/mem0/memory/main.py:462, in Memory.search(self, query, user_id, agent_id, run_id, limit, filters)
454 future_graph_entities = (
455 executor.submit(self.graph.search, query, filters, limit) if self.enable_graph else None
456 )
458 concurrent.futures.wait(
459 [future_memories, future_graph_entities] if future_graph_entities else [future_memories]
460 )
--> 462 original_memories = future_memories.result()
463 graph_entities = future_graph_entities.result() if future_graph_entities else None
465 if self.enable_graph:

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/concurrent/futures/_base.py:449, in Future.result(self, timeout)
447 raise CancelledError()
448 elif self._state == FINISHED:
--> 449 return self.__get_result()
451 self._condition.wait(timeout)
453 if self._state in [CANCELLED, CANCELLED_AND_NOTIFIED]:

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/concurrent/futures/_base.py:401, in Future.__get_result(self)
399 if self._exception:
400 try:
--> 401 raise self._exception
402 finally:
403 # Break a reference cycle with the exception in self._exception
404 self = None

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/concurrent/futures/thread.py:59, in _WorkItem.run(self)
56 return
58 try:
---> 59 result = self.fn(*self.args, **self.kwargs)
60 except BaseException as exc:
61 self.future.set_exception(exc)

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/mem0/memory/main.py:482, in Memory._search_vector_store(self, query, filters, limit)
480 def _search_vector_store(self, query, filters, limit):
481 embeddings = self.embedding_model.embed(query, "search")
--> 482 memories = self.vector_store.search(query=embeddings, limit=limit, filters=filters)
484 excluded_keys = {
485 "user_id",
486 "agent_id",
(...) 492 "id",
493 }
495 original_memories = [
496 {
497 **MemoryItem(
(...) 512 for mem in memories
513 ]

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/mem0/vector_stores/qdrant.py:143, in Qdrant.search(self, query, limit, filters)
131 """
132 Search for similar vectors.
133
(...) 140 list: Search results.
141 """
142 query_filter = self._create_filter(filters) if filters else None
--> 143 hits = self.client.query_points(
144 collection_name=self.collection_name,
145 query=query,
146 query_filter=query_filter,
147 limit=limit,
148 )
149 return hits.points

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/qdrant_client.py:558, in QdrantClient.query_points(self, collection_name, query, using, prefetch, query_filter, search_params, limit, offset, with_payload, with_vectors, score_threshold, lookup_from, consistency, shard_key_selector, timeout, **kwargs)
553 query = self._embed_models(query, is_query=True) if query is not None else None
554 prefetch = (
555 self._embed_models(prefetch, is_query=True) if prefetch is not None else None
556 )
--> 558 return self._client.query_points(
559 collection_name=collection_name,
560 query=query,
561 prefetch=prefetch,
562 query_filter=query_filter,
563 search_params=search_params,
564 limit=limit,
565 offset=offset,
566 with_payload=with_payload,
567 with_vectors=with_vectors,
568 score_threshold=score_threshold,
569 using=using,
570 lookup_from=lookup_from,
571 consistency=consistency,
572 shard_key_selector=shard_key_selector,
573 timeout=timeout,
574 **kwargs,
575 )

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/local/qdrant_local.py:466, in QdrantLocal.query_points(self, collection_name, query, using, prefetch, query_filter, search_params, limit, offset, with_payload, with_vectors, score_threshold, lookup_from, **kwargs)
463 query_filter = ignore_mentioned_ids_filter(query_filter, list(mentioned_ids))
465 prefetch = self._resolve_prefetches_input(prefetch, collection_name)
--> 466 return collection.query_points(
467 query=query,
468 prefetch=prefetch,
469 query_filter=query_filter,
470 using=using,
471 score_threshold=score_threshold,
472 limit=limit,
473 offset=offset or 0,
474 with_payload=with_payload,
475 with_vectors=with_vectors,
476 )

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/local/local_collection.py:703, in LocalCollection.query_points(self, query, prefetch, query_filter, limit, offset, with_payload, with_vectors, score_threshold, using, **kwargs)
690 scored_points = self._merge_sources(
691 sources=sources,
692 query=query,
(...) 699 score_threshold=score_threshold,
700 )
701 else:
702 # It is a base query
--> 703 scored_points = self._query_collection(
704 query=query,
705 using=using,
706 query_filter=query_filter,
707 limit=limit,
708 offset=offset,
709 with_payload=with_payload,
710 with_vectors=with_vectors,
711 score_threshold=score_threshold,
712 )
714 return types.QueryResponse(points=scored_points)

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/local/local_collection.py:840, in LocalCollection._query_collection(self, query, using, query_filter, limit, offset, with_payload, with_vectors, score_threshold)
838 return [record_to_scored_point(record) for record in records[offset:]]
839 elif isinstance(query, models.NearestQuery):
--> 840 return self.search(
841 query_vector=(using, query.nearest),
842 query_filter=query_filter,
843 limit=limit,
844 offset=offset,
845 with_payload=with_payload,
846 with_vectors=with_vectors,
847 score_threshold=score_threshold,
848 )
849 elif isinstance(query, models.RecommendQuery):
850 return self.recommend(
851 positive=query.recommend.positive,
852 negative=query.recommend.negative,
(...) 860 score_threshold=score_threshold,
861 )

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/local/local_collection.py:573, in LocalCollection.search(self, query_vector, query_filter, limit, offset, with_payload, with_vectors, score_threshold)
571 if isinstance(query_vector, np.ndarray):
572 if len(query_vector.shape) == 1:
--> 573 scores = calculate_distance(query_vector, vectors, distance)
574 else:
575 scores = calculate_multi_distance(query_vector, vectors, distance)

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/local/distances.py:152, in calculate_distance(query, vectors, distance_type)
149 assert not np.isnan(query).any(), "Query vector must not contain NaN"
151 if distance_type == models.Distance.COSINE:
--> 152 return cosine_similarity(query, vectors)
153 elif distance_type == models.Distance.DOT:
154 return dot_product(query, vectors)

File ~/software/miniconda3/envs/autogen0.4/lib/python3.12/site-packages/qdrant_client/local/distances.py:94, in cosine_similarity(query, vectors)
92 query_norm = np.linalg.norm(query)
93 query /= np.where(query_norm != 0.0, query_norm, EPSILON)
---> 94 return np.dot(vectors, query)
96 query_norm = np.linalg.norm(query, axis=-1)[:, np.newaxis]
97 query /= np.where(query_norm != 0.0, query_norm, EPSILON)

ValueError: shapes (0,1536) and (1024,) not aligned: 1536 (dim 1) != 1024 (dim 0)

[parshvadaftari]

@DStarEpoch This is because mem0 by default takes in 1536 embedding model dims. And the Embedding model used here has max embedding dimension as 1024. Try adding the embedding_model_dims in the config if it helps you.

[DStarEpoch]

@DStarEpoch This is because mem0 by default takes in 1536 embedding model dims. And the Embedding model used here has max embedding dimension as 1024. Try adding the embedding_model_dims in the config if it helps you.

Thanks for your instruction. It has been solved
I make change like this:
"embedder": {
"provider": "ollama",
"config": {
"model": "mxbai-embed-large:335m",
"ollama_base_url": "http://localhost:11434",
"embedding_dims": 1024,
}
},
"vector_store": {
"provider": "milvus",
"config": {
"collection_name": "my_interaction",
"url": "http://127.0.0.1:19530",
"embedding_model_dims": 1024,
},
},

It works!