PaddleNLP icon indicating copy to clipboard operation
PaddleNLP copied to clipboard
verified publisher icon PaddlePaddle

I need ideas from your side

Open Dineshkumar-Anandan-ZS0367 opened this issue 1 year ago โ€ข 0 comments

่ฏทๆๅ‡บไฝ ็š„้—ฎ้ข˜

Hi Sir, I am using paddlenlp document intelligence model for image vqa. Please give ideas for these below 3 queries?

  1. Model getting two much of memory usage. How to use optimize this?
  2. How to send multiple question or how to process multiple question at a time?, [Ex, I will send 5 questions for a image, i need process parallel and get output with less memory consumption] Its there any way to solve this?
  3. How to load paddlenlp models from my local folder?

Dineshkumar-Anandan-ZS0367 avatar Feb 10 '24 19:02 Dineshkumar-Anandan-ZS0367