Lyrics Search indexing is very slow about examples HOT 2 CLOSED

Thanks / Merci 👍

from examples.

Hello @canetcesc ,
Thank you for trying Jina out. It is definitely an expensive example to play around with your desktop.

Let me try to clarify yout questions:

1. It is unclear and very dependant on the hardware capabilities, but it is indeed very expensive for a desktop, since Bert model is very heavy.

2. If you check the contents of app.py you will see that it is the full dataset being indexed. Maybe you can try to index toy dataset in order to be able to play with it.

3. It should definitely improve to run the encoder in the GPU.

4. For now, it is not possible to run queries while indexing. It is something that we definitely want to cover in the future.

5. The docker image does not contain any indexing, so it is necessary to index.

I hope you find these insights helpful

from examples.