Find answers from the community

Updated last year

I m also wondering when setting the `id

I'm also wondering, when setting the id_ field for my documents (per the docs):

Plain Text

documents.append(Document(
        id_=str(s['id']),
        text=s['bodytext'],
        metadata={
          'id': s['id'],
          'headline': s['headline'],
          'published_time': s['date'],
...

Why in pinecone the id_ is still a random UID (screenshot attached).

Attachment

5 comments

LLogan M

So, the document object gets broken into nodes when inserted

Each node has a new id, but still contains a reference to the original document ID

LLogan M

https://github.com/jerryjliu/llama_index/blob/30d1ad35a568295d5ff145ba4a4eff66e5f02129/llama_index/vector_stores/utils.py#L48

LLogan M

We store that ID in the vectordb as well

vvedtam

aaaaaaa

vvedtam

that makes perfect sense, thanks!! 🍻

Add a reply