MarkEverythingDown

+ MarkEverythingDown - 你的全能文档Markdown转换神器！🚀 一键将PDF/Office/图片/代码等文件转换为结构清晰的Markdown，专为LLM优化设计。结合Qwen2.5 VL视觉模型，连扫描件都能智能解析！

✨ 优势

✅ AI超能力 - 深度集成Qwen2.5 VL模型，完美保留表情符号和图像描述
✅ 格式全覆盖 - 从微信截图到学术论文统统搞定
✅ 双模处理 - 本地/云端自由切换，隐私与性能兼得
✅ 小白友好 - 无需代码，拖拽文件立即转换
✅ 智能分批 - 优化处理大型PDF文档，自动调整批次大小

MarkEverythingDown is a versatile document conversion tool that transforms various file formats into clean, structured markdown. Whether you're working with PDFs, Office documents, images, code files, or notebooks, MarkEverythingDown provides a unified interface to convert them all.

The tool is specifically designed to leverage Qwen2.5 VL models through OpenAI-compatible APIs, supporting both local inference engines like LMStudio and cloud API providers like DashScope. This design enables high-quality processing of visual content while maintaining flexibility in deployment options.

I developed this tool to streamline the conversion of documents into markdown format, which is both LLM-friendly and easy for human to read. The goal is to make document processing as seamless as possible, allowing users to easily convert their files for RAG applications or SFT dataset preparations.

Roadmap

Recently Implemented (April 2025)

Enhanced Processing Options

• ✅ Temperature Control: Added temperature parameter (0.0-1.0) for controlling the determinism of AI output
• ✅ Max Tokens Setting: Implemented customizable token limits for generation
• ✅ Multi-Page Processing: Added support for processing multiple PDF pages in a single API call
• ✅ Dynamic Batch Sizing: Implemented intelligent adjustment of batch sizes based on page complexity
• ✅ Optimized Token Management: Added max_tokens_per_batch option to prevent token limit issues

Improved Document Support

• ✅ Enhanced Table Handling in Word Documents: Better preservation of table structure and formatting in DOCX files
• ✅ Excel Spreadsheet Support: Full support for XLSX files with proper table formatting
• ✅ Better Visual Elements Preservation: Improved handling of emojis and image descriptions

Interface Improvements

• ✅ Enhanced UI Tooltips: Clearer explanations of processing options
• ✅ Improved Error Handling: Better feedback for processing issues
• ✅ Progress Indicators: Added visual feedback during processing

Planned Features

Near-term

• 🔜 CSV and TSV Support: Native support for tabular data files
• 🔜 Custom Templates: User-defined output formats for different document types
• 🔜 Batch Processing Improvements: Enhanced management of large document collections

Long-term

• 🔜 Multi-model Support: Integration with additional vision-language models
• 🔜 Advanced Document Analysis: Improved extraction of complex structures like footnotes and citations
• 🔜 API Mode: Headless operation for integration with other applications
• 🔜 Collaborative Editing: Real-time collaborative editing of converted documents

Features

• Multi-format support: Convert PDFs, DOCX, PPTX, XLSX, images, code files, notebooks, and markdown variants
• Intelligent processing: Automatically selects the appropriate processor for each file type
• Vision AI support: Optimized for Qwen2.5 VL models with OpenAI-compatible interface
• Dual processing options: Support local inference APIs and cloud APIs
• Batch processing: Process multiple files at once with a simple interface
• User-friendly UI: Easy-to-use web UI with Gradio and helpful tooltips
• Command line interface: Quick conversions from the terminal

Supported Formats

Category	Formats
Documents	PDF, DOCX, PPTX, XLSX
Images	PNG, JPG, JPEG, BMP
Code	Python, R, and other programming languages
Notebooks	Jupyter Notebooks (ipynb)
Markdown	MD, RMD (R Markdown)
Text	TXT

Installation

# Clone the repository
git clone https://github.com/RoffyS/MarkEverythingDown.git
cd MarkEverythingDown

# Set up a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Usage

Web UI (Recommended)

# Launch the web interface
python main.py --ui

The MarkEverythingDown web interface provides an intuitive way to convert your documents to markdown:

1. Upload Files: Drag and drop single or multiple files into the upload area
2. Configure Output: Specify where you want your converted markdown files to be saved
3. Processing Options:
4. Concurrent Processing: Control how many API calls are made at once
5. Pages Per Batch: Set the maximum number of PDF pages to send in each API call
6. Dynamic Batching: Enable automatic adjustment of Pages Per Batch based on page complexity and token limits
7. Temperature: Adjust the creativity level of the AI (0.0 for deterministic results)

8. Max Tokens: Set token limits for generation (blank uses model default)

9. API Configuration: Configure API settings for your vision model:

10. API Key: Your API key (default: "lmstudio" for local inference)
11. API URL: The base URL for your API endpoint (default: "http://localhost:1234/v1")
12. Model Name: The model to use for processing (default: "qwen2.5-vl-7b-instruct")

Command Line

python main.py sample_pdf.pdf # path to input file \
    --api-key lm_studio \
    --base-url http://localhost:1234/v1 \
    --model qwen2.5-vl-32b-instruct \
    --force-vision \
    --max-concurrent 1 \
    --output test \
    --images-per-batch 1 \
    --dynamic-batching \
    --max-tokens-per-batch 8192

Command Line Options

Option	Description	Default
--output, -o	Output directory for markdown files	output
--ui	Launch the graphical user interface	-
--force-vision	Use vision model for PDFs instead of text extraction	False
--max-concurrent	Maximum concurrent workers for PDF page processing	2
--images-per-batch	Maximum number of PDF pages per API call	1
--dynamic-batching	Automatically adjust images-per-batch based on page complexity and maximum tokens per batch	True
--no-dynamic-batching	Disable dynamic batching	-
--max-tokens-per-batch	Maximum tokens per batch for dynamic batching	4000
--temperature	Temperature for generation (0.0-1.0)	0.0
--max-tokens	Maximum tokens for generation	Model default
--api-key	API key for vision processor	lmstudio
--base-url	Base URL for API endpoint	http://localhost:1234/v1
--model	Model name to use	qwen2.5-vl-7b-instruct

Example Use Cases

Below are several examples of converting images, PDFs, and Office documents into markdown format. You are welcome to try it out with your own documents, either through the web UI or the command line. You can also play around with the prompt templates in processors/vision/vision_processor.py to customize the output format of PDFs and images.

1. Image Processing

Course Slides with Images

Input:

Output (test_output/test_image1.md):

# 2018 Turing Award for deep learning

The most prestigious technical award, given to individuals who have made major 
contributions of lasting importance to computing.

## Recipients

- **Geoffrey Hinton**
- **Yoshua Bengio**
- **Yann LeCun**

## Lecture Details
- **Lecture 1 - Slide 27**
- **Date:** April 4, 2023
- **Presenters:** Fei-Fei Li, Yunzhu Li, Ruohan Gao

Course Slides with Code

Input:

Output (test_output/test_image3.md):

# Basic Data Types in R: Numeric

## Numeric: Default Data Type in R Representing Decimal Values

- **Numeric:** The default data type in R for representing decimal values.
  - Assign a decimal value:
    ```R
    x <- 3.14
    ```
  - Print the value of `x`:
    ```R
    x
    # [1] 3.14
    ```
  - Print the class name of `x`:
    ```R
    class(x)
    # [1] "numeric"
    ```
  - Assign an integer value:
    ```R
    k <- 3
    ```
  - Print the value of `k`:
    ```R
    k
    # [1] 3
    ```
  - Print the class name of `k`:
    ```R
    class(k)
    # [1] "numeric"
    ```
- Even integer values are stored as numeric unless explicitly declared:
    ```R
    class(k)
    # [1] "numeric"
    ```
  - Check if `k` is an integer:
    ```R
    is.integer(k)
    # [1] FALSE
    ```

## Try it Yourself:
- [Link to Practice](https://campus.datacamp.com/courses/r-short-and-sweet/hello-r?ex=2)

WeChat Screenshot

Input:

Output (test_output/test_image2.md):

# WeChat Transcript

**Sender:** User 1  
> Can't believe I'm using a random WeChat history generator to create a test case

**Sender:** User 2  
> Guess they will never know

**Sender:** User 1  
> yea alright

2. PDF Processing (Two Methods)

Text Extraction (Default)

Input: sample_pdf.pdf

Output (test_output/sample_pdf_noVision.md):

As you can tell, PDF is a really tricky format to process. The output is not very clean, and the formatting is not preserved.

```markdown

Page 1

March 5, 2025 Qwen2.5-VL Technical Report Qwen Team, Alibaba Group https://chat.qwenlm.aihttps://huggingface.co/Qwen https://modelscope.cn/organization/qwenhttps://github.com/QwenLM/Qwen2.5-VL Abstract We introduce Qwen2.5-VL, the latest flagship model of Qwen vision-language series, which demonstrates significant advancements in both foundational capabilities and innovative functionalities. Qwen2.5-VL achieves a major leap forward in understanding and interacting with the world through enhanced visual recognition, precise object local- ization, robust document parsing, and long-video comprehension. A standout feature of Qwen2.5-VL is its ability to localize objects using bounding boxes or points accurately. It provides robust structured data extraction from invoices, forms, and tables, as well as detailed analysis of charts, diagrams, and layouts. To handle complex inputs, Qwen2.5- VL introduces dynamic resolution processing and absolute time encoding, enabling it to process images of varying sizes and videos of extended durations (up to hours) with second-level event localization. This allows the model to natively perceive spatial scales and temporal dynamics without relying on traditional normalization techniques. By training a native dynamic-resolution Vision Transformer (ViT) from scratch and incorpo- rating Window Attention, we have significantly reduced computational overhead while maintaining native resolution. As a result, Qwen2.5-VL excels not only in static image and document understanding but also as an interactive visual agent capable of reasoning, tool usage, and task execution in real-world scenarios such as operating computers and mobile devices. The model achieves strong generalization across domains without requir- ing task-specific fine-tuning. Qwen2.5-VL is available in three sizes, addressing diverse use cases from edge AI to high-performance computing. The flagship Qwen2.5-VL-72B model matches state-of-the-art models like GPT-4o and Claude 3.5 Sonnet, particularly excelling in document and diagram understanding. The smaller Qwen2.5-VL-7B and Qwen2.5-VL-3B models outperform comparable competitors, offering strong capabilities even in resource-constrained environments. Additionally, Qwen2.5-VL maintains robust linguistic performance, preserving the core language competencies of the Qwen2.5 LLM. 1arXiv:2502.13923v1 [cs.CV] 19 Feb 2025

Page 2

1Introduction Large vision-language models ( LVLMs ) ( OpenAI ,2024;Anthropic ,2024a ;Team et al. ,2023;Wang et al. , 2024f ) represent a pivotal breakthrough in artificial intelligence, signaling a transformative approach to multimodal understanding and interaction. By seamlessly integrating visual perception with natural language processing, these advanced models are fundamentally reshaping how machines interpret and analyze complex information across diverse domains. Despite significant advancements in multimodal large language models, the current capabilities of these models can be likened to the middle layer of a sandwich cookie—competent across various tasks but falling short of exceptional performance. Fine- grained visual tasks form the foundational layer of this analogy. In this iteration of Qwen2.5-VL, we are committed to exploring fine-grained perception capabilities, aiming to establish a robust foundation for LVLMs and create an agentic amplifier for real-world applications. The top layer of this framework is multi-modal reasoning, which is enhanced by leveraging the latest Qwen2.5 LLM and employing multi-modal QA data construction. A spectrum of works have promoted the development of multimodal large models, characterized by architectural design, visual input processing, and data curation. One of the primary drivers of progress in LVLMs is the continuous innovation in architecture. The studies presented in ( Alayrac et al. ,2022; Li et al. ,2022a ;2023b ;Liu et al. ,2023b ;a;Wang et al. ,2024i ;Zhang et al. ,2024b ;Wang et al. ,2023) have incrementally shaped the current paradigm, which typically consists of a visual encoder, a cross-modal projector, and LLM. Fine-grained perception models have emerged as another crucial area. Models like (Xiao et al. ,2023;Liu et al. ,2023c ;Ren et al. ,2024;Zhang et al. ,2024a ;d;Peng et al. ,2023;Deitke et al. , 2024) have pushed the bound