mineru-4.0.4-released
QwenLM/Qwen3-VLmineru-4.0.4-releasedSep 19, 2026by myhloli
Upgrading?
AI Summary
MinerU 4.0.4 optimizes GPU memory utilization for high-VRAM devices, reduces logging noise by demoting performance logs to debug level, and fixes API server and WebUI issues.
Key Highlights
- GPU memory utilization logic optimized for high-VRAM devices
- Inference performance, model loading, and table priority logs lowered to debug level
- API server log-level scope restricted to server-side logs only
- WebUI filename stemming made idempotent for files over 120 bytes
Full Release Notes
## 更新内容 / What's Changed **修复 / Fixes** - VLM 引擎：高显存设备上 GPU memory utilization 逻辑优化 / engine: adjust GPU memory utilization logic for high-VRAM devices - 日志降噪：推理性能、模型加载与原生表格优先级日志由 info 降为 debug / Lower inference performance, model loading and native table priority logs to debug - API server：`--log-level` 仅作用于服务端日志，不再影响其他组件 / Scope `--log-level` to server-side logs only - WebUI：超长（>120 字节）文件名的收敛处理幂等化，重复保存不再截断 / Make filename stemming idempotent for >120-byte filenames **依赖 / Dependencies** - mineru-vl-utils 最低版本提升至 2.0.5（含 lmdeploy 引擎多帧响应不完整的修复）/ Require mineru-vl-utils>=2.0.5 (fixes incomplete lmdeploy multi-frame responses) **Full Changelog**: https://github.com/opendatalab/MinerU/compare/mineru-4.0.3-released...mineru-4.0.4-released
Related Repositories
rasbt/LLMs-from-scratch★ 107,078
Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
lyogavin/airllm★ 40,212
AirLLM 2.8T LLM inference with single 4GB GPU
NirDiamant/RAG_Techniques★ 29,961
This repository showcases various advanced techniques for Retrieval-Augmented Generation (RAG) systems. Each technique has a detailed notebook tutorial.
microsoft/OmniParser★ 25,539
A simple screen parsing tool towards pure vision based GUI agent
karpathy/nn-zero-to-hero★ 24,907
Neural Networks: Zero to Hero