search

nomic-ai / gpt4all

GPT4All: Run Local LLMs on Any Device. Open-source and available for commercial use.
https://nomic.ai/gpt4all
MIT License
70.15k stars 7.67k forks source link

hung procesing local docs #3119

Open ulifpl opened 1 day ago

ulifpl commented 1 day ago

teorias didactica de la matematica.pdf

Bug Report

hangs reading the file or is very slow, consuming 100% of the cpu, the ui works well

Steps to Reproduce

  1. just select the file in localdocs

Expected Behavior

read the pdf file teorias didactica de la matematica.pdf

Your Environment

manyoso commented 1 day ago

Do you see the words or embeddings increasing? This says it is either a hang or very slow... I need to know which. The difference is whether you see any progress at all or activity while creating the collection... does the word count increase? does the embedding count increase? do you see anything at all visually changing while the collection is being created? If you don't see anything visually changing at all.. have you tested it for minutes?

I just tried to reproduce a hang with this pdf and could not. It worked fine.

ulifpl commented 1 day ago

It detects all the words, about 90000. But the embedding reaches 30% and does not advance further but continues consuming 100% of CPU, the user interface continues working well. i tested for 20 minutes. When you close the app and reopen it, the embedding increases again.