short but very helpful tutorial. I am wondering whether it makes sense to delete all documents in the vector store while introducing new documents. Is there a way to add/replace certain documents? or what's the best way to track those documents since they were splitted from some long article or paragraphs?
Can you create endless indexes or something that separate a lot of data sets in the db? For a chatbot service where users can create custom bots with own data. Pinecone has only a very limited index quantity. Has mongo a limit?
Our indexes are very generous and capable of complex compound field aggregation. Here is a document outlining the limitations: www.mongodb.com/docs/manual/reference/limits/#indexes
Using OllamaEmbeddings and getting errorfull error: {'ok': 0.0, 'errmsg': 'PlanExecutor error during aggregation :: caused by :: vector field is indexed with 1536 dimensions but queried with 4096', 'code': 8, 'codeName': 'UnknownError', '$clusterTime': {'clusterTime': Timestamp(1713602616, 8), 'signature': {'hash': b'\xc7\xf0CZ\xe16he\x17k\xd7F\xa474T\xc6G5*', 'keyId': 7300890704606134274}}, 'operationTime': Timestamp(1713602616, 8)} Can you help me on this?
Hi there @MongoDB, I tried to follow the tutorial but got a key error = embedding at line docs = vectorStore.max_marginal_relevance_search(query, K = 1) Is there a fix for this? Look forward to hearing from you. Thanks
We’d love to help you build your RUclips channel :) lots of potential This video will do better if you start it with “how to” and categorize it as such - just like in the thumbnail you have right now - use that text for the title and watch this take off!
🔗 Written tutorial → mdb.link/ZvwUzcMvKiI-tutorial
🔗 GitHub repository → trymongodb.com/3H7kO3L
Awesome, thanks a lot! i enjoyed video about Vector Search and this one is a wonderful addition.
How do you connect the chunks created by text splitter to the original docs it came from?
short but very helpful tutorial.
I am wondering whether it makes sense to delete all documents in the vector store while introducing new documents.
Is there a way to add/replace certain documents? or what's the best way to track those documents since they were splitted from some long article or paragraphs?
Can you create endless indexes or something that separate a lot of data sets in the db? For a chatbot service where users can create custom bots with own data. Pinecone has only a very limited index quantity. Has mongo a limit?
Our indexes are very generous and capable of complex compound field aggregation. Here is a document outlining the limitations: www.mongodb.com/docs/manual/reference/limits/#indexes
can you please do for nodejs connecting atlas -openaiembeddings-langchain
Using OllamaEmbeddings and getting errorfull error: {'ok': 0.0, 'errmsg': 'PlanExecutor error during aggregation :: caused by :: vector field is indexed with 1536 dimensions but queried with 4096', 'code': 8, 'codeName': 'UnknownError', '$clusterTime': {'clusterTime': Timestamp(1713602616, 8), 'signature': {'hash': b'\xc7\xf0CZ\xe16he\x17k\xd7F\xa474T\xc6G5*', 'keyId': 7300890704606134274}}, 'operationTime': Timestamp(1713602616, 8)}
Can you help me on this?
Can we do this with mongodb compass?
Help do for Nodejs.
Hi there @MongoDB, I tried to follow the tutorial but got a key error = embedding at line
docs = vectorStore.max_marginal_relevance_search(query, K = 1)
Is there a fix for this? Look forward to hearing from you. Thanks
Used this instead:
docs = vectorStore.similarity_search_with_relevance_scores(query, k=3)
How is this compared to superduperdb?
i will keep using chroma thank you
Hi there, I tried to follow the tutorial but got a key error = embedding. Is there a fix for this? Look forward to hearing from you. Thanks
USe this instead:
docs = vectorStore.similarity_search_with_relevance_scores(query, k=3)
@@SahilKavitake Thanks Sahil 👍
We’d love to help you build your RUclips channel :) lots of potential
This video will do better if you start it with “how to” and categorize it as such - just like in the thumbnail you have right now - use that text for the title and watch this take off!
🙏🙏❤️