Find answers from the community

Updated 8 months ago

When creating an index in a jupyter

When creating an index in a jupyter notebook, the output is filled with Batches. Is there a way to disable these? The same "batch" spam is also present in RetrieverEvaluator.aevaluate_dataset() as well

Attachment

7 comments

WWhiteFang_Jr

Remove show_progress from Indexing step.

oogiorgil

The show_progress only removes the "Generating embeddings" line. That one I actually would like to keep.

The batches still shows

oogiorgil

Attachment

WWhiteFang_Jr

Do you have logging defined at the top?

Plain Text

import logging

logging.basicConfig(stream=sys.stdout, level=logging.INFO)
logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))

oogiorgil

I dont have this defined at the top. If I add this, the "batches" still show up

LLogan M

I have no idea where this is coming from actually -- I don't think this is coming from llama-index explicitly? Maybe something to do with the embed model you are using?

oogiorgil

This is the simplest example I can provide, which is basically the starter tutorial with local model. If it makes a difference, I am using Kaggle to run the code.

Attachment

Add a reply