Find answers from the community

Updated 3 months ago

Nodes

Hi!
When creating a new index based on documents, how can I check the nodes it creates? I found a strange problem when nodes getting from Qdrant collection doesn't have the information it should have and show the information that the collection doesn't have. As a result, the querying doesn't return a correct result. So, firstly I'd like to check the nodes when creating index. Thanks!

5 comments

LLogan M

You can create the nodes beforehand and then use them in the constructor directly

VectorStoreIndex(nodes, storage_context=storage_context, service_context=service_context)

https://gpt-index.readthedocs.io/en/latest/core_modules/data_modules/node_parsers/usage_pattern.html

SSeaCat

Ok, interesting. Let me try. Do you have any idea why search could include a node with wrong information?

LLogan M

It's all based on vectors, so I could see it happening 🤔 tbh though I did see someone point out something weird about chroma on github, haven't had time to investigate yet

https://github.com/jerryjliu/llama_index/issues/7007

LLogan M

I wonder if a similar issue happens with the default vector store

SSeaCat

Well, okay, used the code you pointed me to, checked the nodes, everything looks fine. I don't know which glitch was it... thanks!

Add a reply