Generate low-quality OCR output

by nobita3921 - opened May 22

May 22

Hi, I tried running your quantized model with llama.cpp version b5453, but it generated a low-quality result and lost almost all information in the image because of early stop. I think, maybe the quantized model has some bugs and I tried to manually quantize the original model but llama.cpp throws the bug. Can you re-check your model ?

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment