### Install MinerU using pip

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/index.md

Installs or upgrades pip, installs uv, and then installs MinerU with all features using uv.

```bash
pip install --upgrade pip
pip install uv
uv pip install -U "mineru[all]"
```

--------------------------------

### Install MinerU from Source

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/index.md

Clones the MinerU repository and installs it in editable mode with all features using uv.

```bash
git clone https://github.com/opendatalab/MinerU.git
cd MinerU
uv pip install -e .[all]
```

--------------------------------

### Start OpenAI Compatible Server

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Starts an OpenAI compatible server for MinerU, typically used with vLLM or LMDeploy.

```APIDOC
## Start OpenAI Compatible Server
```bash
mineru-openai-server --port 30000
```

### Description
Launches a server compatible with the OpenAI API, allowing clients designed for OpenAI to interact with MinerU. This requires a `vllm` or `lmdeploy` environment.
```

--------------------------------

### Start OpenAI Compatible Server

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Starts an OpenAI compatible server using MinerU, requiring vLLM or lmdeploy. This server can be connected to by http clients.

```bash
# Start openai compatible server (requires vllm or lmdeploy environment)
mineru-openai-server --port 30000
```

--------------------------------

### Install MinerU Core Functionality

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/extension_modules.md

Install the core module for basic MinerU parsing features. This is a prerequisite for most other functionalities.

```bash
uv pip install "mineru[core]"
```

--------------------------------

### Install Lightweight Client for OpenAI-compatible Servers (vlm-http-client)

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/extension_modules.md

Install the basic MinerU package for a lightweight client to connect to OpenAI-compatible servers in VLM mode. Suitable for edge devices with CPU and network connectivity.

```bash
uv pip install mineru
mineru -p <input_path> -o <output_path> -b vlm-http-client -u http://127.0.0.1:30000
```

--------------------------------

### Example 4: Full Conversation Flow

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/plugin/Cherry_Studio.md

A complete example of a user request and the model's response using the MinerU-MCP tool.

```Plain
我有一份学术论文的 PDF，网址是 https://arxiv.org/pdf/2303.08774.pdf，能帮我转换成 Markdown 格式吗？
```

```Plain
我可以帮您将这份学术论文转换为 Markdown 格式。我将使用 MinerU-MCP 工具来处理这个任务。
[模型调用 parse_documents 工具，处理过程中...]
论文已成功转换为 Markdown 格式。转换结果如下：
# The Capacity of Diffusion Models to Memorize and Generate Training Data
## Abstract
Recent diffusion models can generate high-quality images that are nearly indistinguishable from real ones...
[显示论文内容...]
```

--------------------------------

### Start Web API Service with Docker Compose

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Starts the MinerU Web API service using Docker Compose. The API documentation can be accessed via a browser.

```bash
docker compose -f compose.yaml --profile api up -d
```

--------------------------------

### Start MinerU API Server

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Starts the MinerU API server using FastAPI. Access the interactive API documentation at http://127.0.0.1:8000/docs.

```APIDOC
## Start MinerU API Server
```bash
mineru-api --host 0.0.0.0 --port 8000
```

### Description
Starts the MinerU API server, providing endpoints for task submission, file parsing, and status checks. The API documentation is available at `/docs`.

### Endpoints
- `GET /health`: Returns system health information.
- `POST /tasks`: Submits tasks asynchronously.
- `POST /file_parse`: Parses files synchronously.
- `GET /tasks/{task_id}`: Queries the status of a specific task.
- `GET /tasks/{task_id}/result`: Fetches the result of a specific task.
```

--------------------------------

### Example 1: Parse Document via URL

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/plugin/Cherry_Studio.md

User prompt and the corresponding tool call parameters for parsing a PDF from a URL.

```Plain
请使用 MinerU MCP 将以下 URL 的 PDF 文档转换为 Markdown 格式：https://example.com/sample.pdf
```

```Plain
{"file_sources": "https://example.com/sample.pdf"}
```

--------------------------------

### Install Lightweight Client for OpenAI-compatible Servers (hybrid-http-client)

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/extension_modules.md

Install the MinerU pipeline extension package for a lightweight client to connect to OpenAI-compatible servers in hybrid mode. This package is suitable for devices with CPU and network connectivity and performs faster with GPU acceleration.

```bash
uv pip install "mineru[pipeline]"
mineru -p <input_path> -o <output_path> -b hybrid-http-client -u http://127.0.0.1:30000
```

--------------------------------

### Start MinerU Gradio WebUI

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Launches the MinerU Gradio WebUI for a visual frontend. Access http://127.0.0.1:7860 in your browser. Can start a local API or connect to an existing one.

```bash
mineru-gradio --server-name 0.0.0.0 --server-port 7860
```

--------------------------------

### Start OpenAI-compatible Server Service with Docker Compose

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Starts the OpenAI-compatible server service using Docker Compose. This profile is suitable for connecting via the vlm-http-client backend.

```bash
docker compose -f compose.yaml --profile openai-server up -d
```

--------------------------------

### Example 2: Parse Local Document

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/plugin/Cherry_Studio.md

User prompt and the corresponding tool call parameters for parsing a local PDF file.

```Plain
请使用 MinerU-MCP 将本地的 D://sample.pdf 文件转换为 Markdown 格式
```

```Plain
{"file_sources": "D://sample.pdf"}
```

--------------------------------

### Start MinerU API Server

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Launches the MinerU FastAPI server. Access http://127.0.0.1:8000/docs for API documentation. Supports asynchronous task submission, synchronous parsing, and task querying.

```bash
mineru-api --host 0.0.0.0 --port 8000
```

--------------------------------

### Start Gradio WebUI

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Launches the Gradio WebUI for a visual interface to MinerU.

```APIDOC
## Start Gradio WebUI
```bash
mineru-gradio --server-name 0.0.0.0 --server-port 7860
```

### Description
Starts the Gradio WebUI, accessible at `http://127.0.0.1:7860`. This UI allows users to upload files and interact with MinerU visually. It can either start a local `mineru-api` instance or connect to an existing one via `--api-url`.
```

--------------------------------

### Start MinerU Router

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Launches the MinerU router for multi-service and multi-GPU orchestration.

```APIDOC
## Start MinerU Router
```bash
mineru-router --host 0.0.0.0 --port 8002 --local-gpus auto
```

### Description
Starts the MinerU router, which acts as a unified entry point for multiple `mineru-api` services or local workers. It exposes the same API interface as `mineru-api` and supports aggregation of upstream services and automatic launching of local GPU workers.
```

--------------------------------

### Install MinerU S3 Extension Module

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/extension_modules.md

Install the S3 extension module to enable reading from or writing to S3 storage. Ensure you have the necessary S3 credentials configured.

```bash
uv pip install "mineru[s3]"
```

--------------------------------

### Start MinerU Router Service with Docker Compose

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Starts the MinerU Router service using Docker Compose. This service provides a unified entry point and can operate in local GPU auto-detection mode or aggregate existing services.

```bash
docker compose -f compose.yaml --profile router up -d
```

--------------------------------

### Install CUDA-enabled PyTorch for Blackwell GPUs on Windows

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/faq/index.md

Install the lmdeploy 0.11.1 + cu128 Windows wheel for Blackwell GPUs. Ensure PYTHON_VERSION is set to your Python version. This command installs the wheel and PyTorch with CUDA 12.8 support.

```powershell
$env:LMDEPLOY_VERSION = "0.11.1"
$env:PYTHON_VERSION = "312"

$wheel = "https://github.com/InternLM/lmdeploy/releases/download/v$($env:LMDEPLOY_VERSION)/lmdeploy-$($env:LMDEPLOY_VERSION)+cu128-cp$($env:PYTHON_VERSION)-cp$($env:PYTHON_VERSION)-win_amd64.whl"
pip install $wheel --extra-index-url https://download.pytorch.org/whl/cu128
```

--------------------------------

### Start Gradio WebUI Service with Docker Compose

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Starts the MinerU Gradio WebUI service using Docker Compose. This allows users to interact with MinerU through a web interface.

```bash
docker compose -f compose.yaml --profile gradio up -d
```

--------------------------------

### Install vLLM with ROCm Support

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/acceleration_cards/AMD.md

Manual installation of vLLM and its dependencies for ROCm. This involves cloning repositories, updating submodules, and setting the PYTORCH_ROCM_ARCH environment variable.

```bash
#手动安装aiter，vllm，amd-smi等，自行找一个位置clone，然后进入该目录吧
git clone --recursive https://github.com/ROCm/aiter.git
cd aiter
git submodule sync; git submodule update --init --recursive
python setup.py develop
cd ..
git clone https://github.com/vllm-project/vllm.git
cd vllm/ 
cp -r /opt/rocm/share/amd_smi ~/Pytorch/vllm/ 
pip install amd_smi/
pip install --upgrade numba \
    scipy \
    huggingface-hub[cli,hf_transfer] \
    setuptools_scm
pip install -r requirements/rocm.txt
export PYTORCH_ROCM_ARCH="gfx1100"   #根据自己的GPU架构 rocminfo | grep gfx
python setup.py develop
```

--------------------------------

### Install libGL library on Ubuntu 22.04 WSL2

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/faq/index.md

Resolves the 'ImportError: libGL.so.1' error on Ubuntu 22.04 WSL2 by installing the missing libgl1-mesa-glx package.

```bash
sudo apt-get install libgl1-mesa-glx
```

--------------------------------

### Install MinerU with vLLM for VLM Inference Acceleration

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/extension_modules.md

Install the vLLM module to accelerate VLM model inference on compatible GPUs. This module requires a Volta architecture GPU or later with at least 8GB VRAM. Note potential dependency conflicts if lmdeploy is also installed.

```bash
uv pip install "mineru[core,vllm]"
```

--------------------------------

### Install MinerU with lmdeploy for VLM Inference Acceleration

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/extension_modules.md

Install the lmdeploy module to accelerate VLM model inference on compatible GPUs. This module requires a Volta architecture GPU or later with at least 8GB VRAM. Note potential dependency conflicts if vLLM is also installed.

```bash
uv pip install "mineru[core,lmdeploy]"
```

--------------------------------

### Start MinerU Docker Container

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Starts a MinerU Docker container with specified port mappings and GPU access. This command launches an interactive bash session inside the container.

```bash
docker run --gpus all \
  --shm-size 32g \
  -p 30000:30000 -p 7860:7860 -p 8000:8000 -p 8002:8002 \
  --ipc=host \
  -it mineru:latest \
  /bin/bash
```

--------------------------------

### View mineru-gradio Help Information

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/cli_tools.md

This command displays the help message for the mineru-gradio tool, outlining all available options for configuring the Gradio interface, including example file enablement, HTTP client linking, and server settings.

```bash
mineru-gradio --help
```

--------------------------------

### Common CUDA_VISIBLE_DEVICES configurations

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/advanced_cli_parameters.md

Examples of how to set the CUDA_VISIBLE_DEVICES environment variable to control GPU visibility.

```bash
CUDA_VISIBLE_DEVICES=1  # Only device 1 will be seen
CUDA_VISIBLE_DEVICES=0,1  # Devices 0 and 1 will be visible
CUDA_VISIBLE_DEVICES="0,1"  # Same as above, quotation marks are optional
CUDA_VISIBLE_DEVICES=0,2,3  # Devices 0, 2, 3 will be visible; device 1 is masked
CUDA_VISIBLE_DEVICES=""  # No GPU will be visible
```

--------------------------------

### Install CJK Fonts on Ubuntu/Debian for PDF Text Rendering

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/faq/index.md

Installs Noto Core and CJK fonts on Ubuntu/Debian systems to resolve missing text information in PDF parsing results when using pypdfium2. Updates the font cache afterwards.

```bash
sudo apt update
sudo apt install fonts-noto-core
sudo apt install fonts-noto-cjk
fc-cache -fv
```

--------------------------------

### Example 3: Parse Document with OCR

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/plugin/Cherry_Studio.md

User prompt and the corresponding tool call parameters for parsing a scanned PDF with OCR enabled.

```Plain
请使用 MinerU-MCP 将以下 URL 的扫描版 PDF 文档转换为 Markdown 格式，并启用 OCR 功能：
https://example.com/scanned.pdf
```

```Plain
{"file_sources": "https://example.com/scanned.pdf","enable_ocr": true}
```

--------------------------------

### Install PyTorch with ROCm Support

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/acceleration_cards/AMD.md

Installs PyTorch and torchvision with ROCm 7.0 support using a Tsinghua mirror. Ensure your GPU architecture (e.g., gfx1100) is correctly specified.

```bash
uv venv --python python3.12
source .venv/bin/activate
uv pip install --pre torch torchvision   -i https://pypi.tuna.tsinghua.edu.cn/simple/   --extra-index-url https://download.pytorch.org/whl/nightly/rocm7.0
uv pip install pip
# 避免覆盖我们本地的pytorch，改用pip而没有继续使用uv pip
pip install -U "mineru[core]" -i https://pypi.mirrors.ustc.edu.cn/simple/
```

--------------------------------

### Start MinerU Router for Orchestration

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Launches the MinerU router for multi-service or multi-GPU orchestration. Exposes the same API interface as mineru-api and can aggregate services or launch local workers.

```bash
mineru-router --host 0.0.0.0 --port 8002 --local-gpus auto
```

--------------------------------

### Build MinerU Docker Image

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Builds the MinerU Docker image using the provided Dockerfile. Ensure you have Docker installed and the Dockerfile is accessible.

```bash
wget https://gcore.jsdelivr.net/gh/opendatalab/MinerU@master/docker/global/Dockerfile
docker build -t mineru:latest -f Dockerfile .
```

--------------------------------

### HTTP Client Usage

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Example of using an HTTP client to connect to a MinerU service, such as the OpenAI compatible server.

```APIDOC
## HTTP Client Usage

### Description
Connects to a MinerU service using an HTTP client. The example shows connecting to the `mineru-openai-server`.

### Command Example
```bash
mineru -p <input_path> -o <output_path> -b hybrid-http-client -u http://127.0.0.1:30000
```

### Notes
- `hybrid-http-client` requires local pipeline dependencies (`mineru[pipeline]`, `torch`).
- `vlm-http-client` is a lightweight option that does not require local `torch`.
```

--------------------------------

### Code Block Example

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/reference/output_files.md

Represents a code block, potentially with an optional caption. Includes bounding box, type, index, and a list of code body and caption blocks.

```json
{
  "type": "code",
  "bbox": [114,780,885,1231],
  "blocks": [
    {
      "bbox": [114,780,885,1231],
      "lines": [
        {
          "bbox": [114,780,885,1231],
          "spans": [
            {
              "bbox": [114,780,885,1231],
              "type": "text",
              "content": "1 // Fig. H.1: DivideByZeroNoExceptionHandling.java  \n2 // Integer division without exception handling.  \n3 import java.util.Scanner;  \n4  \n5 public class DivideByZeroNoExceptionHandling  \n6 {  \n7 // demonstrates throwing an exception when a divide-by-zero occurs  \n8 public static int quotient( int numerator, int denominator )  \n9 {  \n10 return numerator / denominator; // possible division by zero  \n11 } // end method quotient  \n12  \n13 public static void main(String[] args)  \n14 {  \n15 Scanner scanner = new Scanner(System.in); // scanner for input  \n16  \n17 System.out.print(\"Please enter an integer numerator: \");  \n18 int numerator = scanner.nextInt();  \n19 System.out.print(\"Please enter an integer denominator: \");  \n20 int denominator = scanner.nextInt();  \n21"
            }
          ]
        }
      ],
      "index": 17,
      "angle": 0,
      "type": "code_body"
    },
    {
      "bbox": [867,160,1280,189],
      "lines": [
        {
          "bbox": [867,160,1280,189],
          "spans": [
            {
              "bbox": [867,160,1280,189],
              "type": "text",
              "content": "Algorithm 1 Modules for MCTSteg"
            }
          ]
        }
      ],
      "index": 19,
      "angle": 0,
      "type": "code_caption"
    }
  ],
  "index": 17,
  "sub_type": "code"
}
```

--------------------------------

### List Block Example

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/reference/output_files.md

Represents a list of text blocks, often used for section headings or enumerated items. Includes bounding box, type, angle, index, and a list of blocks with their own details.

```json
{
  "bbox": [174,155,818,333],
  "type": "list",
  "angle": 0,
  "index": 11,
  "blocks": [
    {
      "bbox": [174,157,311,175],
      "type": "text",
      "angle": 0,
      "lines": [
        {
          "bbox": [174,157,311,175],
            "spans": [
              {
                "bbox": [174,157,311,175],
                "type": "text",
                "content": "H.1 Introduction"
              }
            ]
        }
      ],
      "index": 3
    },
    {
      "bbox": [175,182,464,229],
      "type": "text",
      "angle": 0,
      "lines": [
        {
          "bbox": [175,182,464,229],
          "spans": [
            {
              "bbox": [175,182,464,229],
              "type": "text",
              "content": "H.2 Example: Divide by Zero without Exception Handling"
            }
          ]
        }
      ],
      "index": 4
    }
  ],
  "sub_type": "text"
}
```

--------------------------------

### Install lmdeploy wheel without re-downloading PyTorch

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/faq/index.md

If you have already installed the cu128 version of PyTorch, use this command to install the lmdeploy wheel without re-downloading PyTorch dependencies. Define the $wheel variable first.

```powershell
pip install $wheel --no-dependencies
```

--------------------------------

### View mineru-api Help Information

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/cli_tools.md

Use this command to view the available options and their descriptions for the mineru-api tool. This is useful for understanding how to configure and run the API server.

```bash
mineru-api --help
```

--------------------------------

### View mineru-router Help Information

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/cli_tools.md

Execute this command to see the help information for the mineru-router tool. It lists options for configuring the router, including host and port settings, upstream URLs, and local GPU worker specifications.

```bash
mineru-router --help
```

--------------------------------

### View MinerU CLI Help

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/cli_tools.md

To access comprehensive help information for the MinerU CLI, append the `--help` flag to the `mineru` command. This displays all available options and their descriptions.

```bash
mineru --help
Usage: mineru [OPTIONS]

Options:
  -v, --version                   Show version and exit
  -p, --path PATH                 Input file path or directory (required)
  -o, --output PATH               Output directory (required)
  --api-url TEXT                  MinerU FastAPI base URL; if omitted, `mineru` starts a temporary local `mineru-api`
  -m, --method [auto|txt|ocr]     Parsing method: auto (default), txt, ocr (pipeline and hybrid* backend only)
  -b, --backend [pipeline|vlm-engine|hybrid-engine|vlm-http-client|hybrid-http-client]
                                  Parsing backend (default: hybrid-engine)
  --effort [medium|high]          Hybrid parsing effort (default: medium)
  -l, --lang [ch|ch_server|korean|ta|te|ka|th|el|arabic|east_slavic|cyrillic|devanagari]
                                  Specify document language (improves OCR accuracy, pipeline backend only)
  -u, --url TEXT                  OpenAI-compatible backend URL passed through to the server when using http-client
  -s, --start INTEGER             Starting page number for parsing (0-based)
  -e, --end INTEGER               Ending page number for parsing (0-based)
  -f, --formula BOOLEAN           Enable formula parsing (default: enabled)
  -t, --table BOOLEAN             Enable table parsing (default: enabled)
  --image-analysis BOOLEAN        Enable image/chart analysis for VLM and hybrid
                                  backends. Hybrid medium effort automatically
                                  disables image/chart analysis (default: enabled)
  --client-side-output-generation BOOLEAN
                                  Generate Markdown and content lists locally
                                  from server-returned middle JSON, images, and
                                  original files (default: disabled)
  --help                          Show help information
```

--------------------------------

### Download Models with CLI Help

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/model_source.md

View help information for the mineru-models-download command. This command is used to download models to local storage.

```bash
mineru-models-download --help
```

--------------------------------

### Deploy MinerU using Docker

Source: https://github.com/opendatalab/mineru/blob/master/README.md

Provides instructions for deploying MinerU using Docker. This method is supported on Linux and Windows with WSL2.

```bash
# Docker deployment instructions are provided via a link to external documentation.
```

--------------------------------

### Sample JSON for Common Fields

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/reference/output_files.md

Illustrates the structure of common fields in output files, including type, content, and bounding box.

```json
[
  [
    {
      "type": "title",
      "content": {
        "title_content": [
          {
            "type": "text",
            "content": "1 Introduction"
          }
        ],
        "level": 1
      },
      "bbox": [83, 121, 917, 156]
    },
    {
      "type": "page_footnote",
      "content": {
        "page_footnote_content": [
          {
            "type": "text",
            "content": "* Corresponding author"
          }
        ]
      },
      "bbox": [71, 815, 915, 841]
    }
  ]
]
```

--------------------------------

### Download Models Interactively

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/model_source.md

Use the interactive command line tool to select and download models. The download process will output the model path and update `mineru.json`.

```bash
mineru-models-download
```

--------------------------------

### Connect to OpenAI Server via HTTP Client

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/quick_usage.md

Connects to a running MinerU OpenAI compatible server using a hybrid http client. Requires local pipeline dependencies like mineru[pipeline] and torch.

```bash
# Start openai compatible server (requires vllm or lmdeploy environment)
mineru-openai-server --port 30000
```

```bash
# In another terminal, connect to openai server via http client
 mineru -p <input_path> -o <output_path> -b hybrid-http-client -u http://127.0.0.1:30000
```

--------------------------------

### Configure MinerU-MCP Environment Variables

Source: https://github.com/opendatalab/mineru/blob/master/docs/zh/usage/plugin/Cherry_Studio.md

Required environment variables for the MinerU-MCP server configuration in Cherry Studio.

```Plain
MINERU_API_BASE=https://mineru.net
MINERU_API_KEY=您的API密钥
OUTPUT_DIR=./downloads
USE_LOCAL_API=false
LOCAL_MINERU_API_BASE=http://localhost:8888
```

--------------------------------

### Download Docker Compose File

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Downloads the compose.yaml file for deploying MinerU services using Docker Compose. This file defines configurations for multiple services.

```bash
wget https://gcore.jsdelivr.net/gh/opendatalab/MinerU@master/docker/compose.yaml
```

--------------------------------

### Launch mineru-router across multiple GPUs

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/advanced_cli_parameters.md

Use the router to manage services across a specified set of GPUs.

```bash
CUDA_VISIBLE_DEVICES=0,1,2,3 mineru-router --host 127.0.0.1 --port 8002
```

--------------------------------

### Connect to OpenAI Server via HTTP Client

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/docker_deployment.md

Connects to the running OpenAI-compatible server using the mineru CLI. This client requires only CPU and network access.

```bash
mineru -p <input_path> -o <output_path> -b vlm-http-client -u http://<server_ip>:30000
```

--------------------------------

### Run multiple openai-server instances on different GPUs

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/advanced_cli_parameters.md

Launch separate openai-server services on specific GPUs by setting the environment variable per terminal session.

```bash
# In terminal 1
CUDA_VISIBLE_DEVICES=0 mineru-openai-server --engine vllm --port 30000
# In terminal 2
CUDA_VISIBLE_DEVICES=1 mineru-openai-server --engine vllm --port 30001
```

--------------------------------

### Sample Data for VLM Backend Output

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/reference/output_files.md

Provides a sample JSON structure for the VLM backend's model inference results, including header and title blocks.

```json
[
  [
    {
      "type": "header",
      "bbox": [0.077, 0.095, 0.18, 0.181],
      "angle": 0,
      "score": null,
      "block_tags": null,
      "content": "ELSEVIER",
      "format": null,
      "content_tags": null
    },
    {
      "type": "title",
      "bbox": [0.157, 0.228, 0.833, 0.253],
      "angle": 0,
      "score": null,
      "block_tags": null,
      "content": "The response of flow duration curves to afforestation",
      "format": null,
      "content_tags": null
    }
  ]
]
```

--------------------------------

### Parse Documents with GPU Acceleration

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/index.md

Parses documents from an input path to an output path using MinerU with GPU acceleration.

```bash
mineru -p <input_path> -o <output_path>
```

--------------------------------

### Run multiple fastapi services on different GPUs

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/advanced_cli_parameters.md

Launch separate mineru-api services on specific GPUs by setting the environment variable per terminal session.

```bash
# In terminal 1
CUDA_VISIBLE_DEVICES=0 mineru-api --host 127.0.0.1 --port 8000
# In terminal 2
CUDA_VISIBLE_DEVICES=1 mineru-api --host 127.0.0.1 --port 8001
```

--------------------------------

### Parse Documents with CPU Only

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/quick_start/index.md

Parses documents from an input path to an output path using MinerU in a pure CPU environment by specifying the 'pipeline' backend.

```bash
mineru -p <input_path> -o <output_path> -b pipeline
```

--------------------------------

### Configure GPU visibility for MinerU

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/usage/advanced_cli_parameters.md

Use the CUDA_VISIBLE_DEVICES environment variable to specify which GPU devices are accessible to the MinerU command.

```bash
CUDA_VISIBLE_DEVICES=1 mineru -p <input_path> -o <output_path>
```

--------------------------------

### Sample Model Inference Results (Pipeline Backend)

Source: https://github.com/opendatalab/mineru/blob/master/docs/en/reference/output_files.md

This JSON sample shows the model inference results for the pipeline backend, including classification ID, label, confidence score, bounding box, and index for detected elements.

```json
[
    {
        "cls_id": 12,
        "label": "header",
        "score": 0.93,
        "bbox": [
            1217,
            104,
            1516,
            134
        ],
        "index": 2
    },
    {
        "cls_id": 6,
        "label": "doc_title",
        "score": 0.9751,
        "bbox": [
            275,
            181,
            1512,
            292
        ],
        "index": 3
    },
    {
        "cls_id": 22,
        "label": "text",
        "score": 0.9217,
        "bbox": [
            275,
            330,
            524,
            370
        ],
        "index": 4
    }
]
```