PaddlePaddle/PaddleOCR-VL-1.6

<!-- SC_OFF --><div class="md"><p>Update your llama.cpp version. PR links have more details.</p> <ul> <li>DeepSeekOCR - <a

<!-- SC_OFF --><div class="md"><p>So far I have only tried Qwen and olmOCR. My biggest struggle at the moment has been extracting a date that is oriented in a title block, where

<table> <tr><td> <a href="https://www.reddit.com/r/LocalLLaMA/comments/1thkwgk/bytedance_released_an_open_source_model_that/"> <img alt="bytedance released an open source model

<!-- SC_OFF --><div class="md"><p>The way PaddleOCR designed their API, it moves memory too much back and forth between RAM and VRAM, which makes is too slow for my use case. Is

<table> <tr><td> <a href="https://www.reddit.com/r/LocalLLaMA/comments/1v479m5/paddlepaddlehpdparsing_hugging_face/"> <img alt="PaddlePaddle/HPD-Parsing · Hugging Face"