Repository Issue Activity (beta)

#39 Arxiv paper Link in the README.md leads to the PDF uploaded here in the Github repo

#58 [教程TUTORIAL] DeepSeek-OCR-2 Windows WSL本地安装部署避坑指南 Local Installation and Deployment Guide for Windows WSL

#1 [BUG] Environment Installation Failure

#57 Verify evals on Papers with Code

#30 阿里云测试失败：不能访问huggingface.co

#32 画的框是不对的

#18 Omnidocbench跑测疑问

#54 Mac-M5-Deepseek-OCR-2

#16 Analysis on Prompt + Document Type Compatibility

#55 OCR 结果大量重复

#53 Possible mistake in Table metric direction: R-order should be ↓ instead of ↑

#52 unsupported operand type(s) for +: 'NoneType' and 'int'

#50 cuda13.0, torch 2.9. 1 vllm==0.16.0, python 3.12, 可以直接用 vllm 启动

#28 Text repetition/looping issue with multi-page PDFs and dense documents

#51 官方建议的prompt有哪些

#49 6.7GB模型大小而16GB显存加载出现溢出的元凶：建议加载模型到GPU时，直接做数据类型转换！

#38 Analysis of causal query tokens

#6 怎么通过vllm部署

#47 How to finetune?

#46 Any prompt to return bounding box?

#40 Could you please upgrade the version of vllm?

#41 Unable to Install "pip install flash-attn==2.7.3 --no-build-isolation" on T4 google colab

#45 Errors during installation (Ubuntu 24.04)

#44 ImportError: cannot import name 'SamplingMetadata' from 'vllm.model_executor'

#42 The model may risk repeating the same word during name generation