---
comments: true
---

# Document Visual Language Model Module Tutorial

## I. Overview

Document visual language models are a cutting-edge multimodal processing technology aimed at addressing the limitations of traditional document processing methods. Traditional methods are often limited to processing document information in specific formats or predefined categories, whereas document visual language models can integrate visual and linguistic information to understand and handle diverse document content. By combining computer vision and natural language processing technologies, these models can recognize images, text, and their relationships within documents, and even understand semantic information within complex layout structures. This makes document processing more intelligent and flexible, with stronger generalization capabilities, showing broad application prospects in automated office work, information extraction, and other fields.

## II. Supported Model List

<table>
<tr>
<th>Model</th><th>Model Download Link</th>
<th>Model Storage Size (GB)</th>
<th>Total Score</th>
<th>Description</th>
</tr>
<tr>
<td>PP-DocBee-2B</td><td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocBee-2B_infer.tar">Inference Model</a></td>
<td>4.2</td>
<td>765</td>
<td rowspan="2">PP-DocBee is a self-developed multimodal large model by the PaddlePaddle team, focusing on document understanding, and it performs excellently in Chinese document understanding tasks. The model is fine-tuned and optimized using nearly 5 million multimodal datasets for document understanding, including general VQA, OCR, charts, text-rich documents, mathematics and complex reasoning, synthetic data, and pure text data, with different training data ratios set. On several authoritative English document understanding evaluation lists in academia, PP-DocBee has basically achieved SOTA for models of the same parameter scale. In terms of internal business Chinese scenario indicators, PP-DocBee also outperforms the current popular open-source and closed-source models.</td>
</tr>
<tr>
<td>PP-DocBee-7B</td><td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocBee-7B_infer.tar">Inference Model</a></td>
<td>15.8</td>
<td>-</td>
</tr>
<tr>
<td>PP-DocBee2-3B</td><td><a href="https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0.0/PP-DocBee2-3B_infer.tar">Inference Model</a></td>
<td>7.6</td>
<td>852</td>
<td>PP-DocBee2 is a self-developed multimodal large model by the PaddlePaddle team, further optimizing the base model on the foundation of PP-DocBee and introducing a new data optimization scheme to improve data quality. Using a small amount of 470,000 data generated by a self-developed data synthesis strategy, PP-DocBee2 performs better in Chinese document understanding tasks. In terms of internal business Chinese scenario indicators, PP-DocBee2 improves by about 11.4% compared to PP-DocBee, and also outperforms the current popular open-source and closed-source models of the same scale.</td>
</tr>
</table>

<b>Note: The total scores of the above models are test results from an internal evaluation set, where all images have a resolution (height, width) of (1680, 1204), with a total of 1196 data entries, covering scenarios such as financial reports, laws and regulations, scientific and technical papers, manuals, humanities papers, contracts, research reports, etc. There are no plans for public release at the moment.</b>

## III. Quick Start

> ❗ Before starting quickly, please install the PaddleOCR wheel package. For details, please refer to the [Installation Guide](../installation.en.md).

You can quickly experience it with one line of command:

```bash
paddleocr doc_vlm -i "{'image': 'https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/medal_table.png', 'query': '识别这份表格的内容, 以markdown格式输出'}"
```

You can also integrate the model inference from the open document visual language model module into your project. Before running the following code, please download the [sample image](https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/medal_table.png) locally.

```python
from paddleocr import DocVLM
model = DocVLM(model_name="PP-DocBee2-3B")
results = model.predict(
    input={"image": "medal_table.png", "query": "识别这份表格的内容, 以markdown格式输出"},
    batch_size=1
)
for res in results:
    res.print()
    res.save_to_json(f"./output/res.json")
```

After running, the result is:

```bash
{'res': {'image': 'medal_table.png', 'query': '识别这份表格的内容, 以markdown格式输出', 'result': '| 名次 | 国家/地区 | 金牌 | 银牌 | 铜牌 | 奖牌总数 |\n| --- | --- | --- | --- | --- | --- |\n| 1 | 中国（CHN） | 48 | 22 | 30 | 100 |\n| 2 | 美国（USA） | 36 | 39 | 37 | 112 |\n| 3 | 俄罗斯（RUS） | 24 | 13 | 23 | 60 |\n| 4 | 英国（GBR） | 19 | 13 | 19 | 51 |\n| 5 | 德国（GER） | 16 | 11 | 14 | 41 |\n| 6 | 澳大利亚（AUS） | 14 | 15 | 17 | 46 |\n| 7 | 韩国（KOR） | 13 | 11 | 8 | 32 |\n| 8 | 日本（JPN） | 9 | 8 | 8 | 25 |\n| 9 | 意大利（ITA） | 8 | 9 | 10 | 27 |\n| 10 | 法国（FRA） | 7 | 16 | 20 | 43 |\n| 11 | 荷兰（NED） | 7 | 5 | 4 | 16 |\n| 12 | 乌克兰（UKR） | 7 | 4 | 11 | 22 |\n| 13 | 肯尼亚（KEN） | 6 | 4 | 6 | 16 |\n| 14 | 西班牙（ESP） | 5 | 11 | 3 | 19 |\n| 15 | 牙买加（JAM） | 5 | 4 | 2 | 11 |\n'}}
```

The meaning of the result parameters is as follows:
- `image`: Indicates the path of the input image to be predicted
- `query`: Represents the input text information to be predicted
- `result`: Information of the model's prediction result

The visualization of the prediction result is as follows:

```bash
| 名次 | 国家/地区 | 金牌 | 银牌 | 铜牌 | 奖牌总数 |
| --- | --- | --- | --- | --- | --- |
| 1 | 中国（CHN） | 48 | 22 | 30 | 100 |
| 2 | 美国（USA） | 36 | 39 | 37 | 112 |
| 3 | 俄罗斯（RUS） | 24 | 13 | 23 | 60 |
| 4 | 英国（GBR） | 19 | 13 | 19 | 51 |
| 5 | 德国（GER） | 16 | 11 | 14 | 41 |
| 6 | 澳大利亚（AUS） | 14 | 15 | 17 | 46 |
| 7 | 韩国（KOR） | 13 | 11 | 8 | 32 |
| 8 | 日本（JPN） | 9 | 8 | 8 | 25 |
| 9 | 意大利（ITA） | 8 | 9 | 10 | 27 |
| 10 | 法国（FRA） | 7 | 16 | 20 | 43 |
| 11 | 荷兰（NED） | 7 | 5 | 4 | 16 |
| 12 | 乌克兰（UKR） | 7 | 4 | 11 | 22 |
| 13 | 肯尼亚（KEN） | 6 | 4 | 6 | 16 |
| 14 | 西班牙（ESP） | 5 | 11 | 3 | 19 |
| 15 | 牙买加（JAM） | 5 | 4 | 2 | 11 |
```

Explanations of related methods, parameters, etc., are as follows:

* `DocVLM` instantiates the document visual language model (taking `PP-DocBee-2B` as an example), with specific explanations as follows:
<table>
<thead>
<tr>
<th>Parameter</th>
<th>Description</th>
<th>Type</th>
<th>Default</th>
</tr>
</thead>
<tbody>
<tr>
<td><code>model_name</code></td>
<td>Model name</td>
<td><code>str</code></td>
<td><code>PP-DocBee-2B</code></td>
</tr>
<tr>
<td><code>model_dir</code></td>
<td>Model storage path</td>
<td><code>str</code></td>
<td><code>None</code></td>
</tr>
<tr>
<td><code>device</code></td>
<td>Device(s) to use for inference.<br/>
<b>Examples:</b> <code>cpu</code>, <code>gpu</code>, <code>npu</code>, <code>gpu:0</code>, <code>gpu:0,1</code>.<br/>
If multiple devices are specified, inference will be performed in parallel. Note that parallel inference is not always supported.<br/>
By default, GPU 0 will be used if available; otherwise, the CPU will be used.
</td>
<td><code>str</code></td>
<td><code>None</code></td>
</tr>
<tr>
<td><code>enable_hpi</code></td>
<td>Whether to use the high performance inference.</td>
<td><code>bool</code></td>
<td><code>False</code></td>
</tr>
<tr>
<td><code>use_tensorrt</code></td>
<td>Whether to use the Paddle Inference TensorRT subgraph engine.</br>
For Paddle with CUDA version 11.8, the compatible TensorRT version is 8.x (x>=6), and it is recommended to install TensorRT 8.6.1.6.</br>
For Paddle with CUDA version 12.6, the compatible TensorRT version is 10.x (x>=5), and it is recommended to install TensorRT 10.5.0.18.
</td>
<td><code>bool</code></td>
<td><code>False</code></td>
</tr>
<tr>
<td><code>min_subgraph_size</code></td>
<td>Minimum subgraph size for TensorRT when using the Paddle Inference TensorRT subgraph engine.</td>
<td><code>int</code></td>
<td><code>3</code></td>
</tr>
<tr>
<td><code>precision</code></td>
<td>Precision for TensorRT when using the Paddle Inference TensorRT subgraph engine.<br/><b>Options:</b> <code>fp32</code>, <code>fp16</code>, etc.</td>
<td><code>str</code></td>
<td><code>fp32</code></td>
</tr>
<tr>
<td><code>enable_mkldnn</code></td>
<td>
Whether to enable MKL-DNN acceleration for inference. If MKL-DNN is unavailable or the model does not support it, acceleration will not be used even if this flag is set.
</td>
<td><code>bool</code></td>
<td><code>True</code></td>
</tr>
<tr>
<td><code>cpu_threads</code></td>
<td>Number of threads to use for inference on CPUs.</td>
<td><code>int</code></td>
<td><code>10</code></td>
</tr>
</tbody>
</table>


* Among them, `model_name` must be specified. After specifying `model_name`, the default PaddleX built-in model parameters will be used. On this basis, when specifying `model_dir`, user-defined models will be used.

* Call the `predict()` method of the document visual language model for inference prediction. This method will return a result list. Additionally, this module also provides the `predict_iter()` method. Both are completely consistent in terms of parameter acceptance and result return, the difference being that `predict_iter()` returns a `generator`, capable of gradually processing and obtaining prediction results, suitable for handling large datasets or scenarios where memory saving is desired. You can choose to use either of these methods based on actual needs. The `predict()` method parameters include `input`, `batch_size`, with specific explanations as follows:

<table>
<thead>
<tr>
<th>Parameter</th>
<th>Description</th>
<th>Type</th>
<th>Default</th>
</tr>
</thead>
<tr>
<td><code>input</code></td>
<td>Input data. Required. Since multimodal models have different input requirements, please refer to the specific model for the correct format.<br/>
For example, for the PP-DocBee series models, the input format should be: <code>{'image': image_path, 'query': query_text}</code>
</td>
<td><code>dict</code></td>
<td>None</td>
</tr>
<tr>
<td><code>batch_size</code></td>
<td>Batch size, positive integer.</td>
<td><code>int</code></td>
<td>1</td>
</tr>
</table>

* Process the prediction results. The prediction result for each sample is the corresponding Result object, and it supports operations such as printing and saving as `json` file:

<table>
<thead>
<tr>
<th>Method</th>
<th>Description</th>
<th>Parameter</th>
<th>Type</th>
<th>Description</th>
<th>Default</th>
</tr>
</thead>
<tr>
<td rowspan = "3"><code>print()</code></td>
<td rowspan = "3">Print results to terminal</td>
<td><code>format_json</code></td>
<td><code>bool</code></td>
<td>Whether to format the output content using <code>JSON</code> indentation</td>
<td><code>True</code></td>
</tr>
<tr>
<td><code>indent</code></td>
<td><code>int</code></td>
<td>Specify the indentation level to beautify the output <code>JSON</code> data, making it more readable, effective only when <code>format_json</code> is <code>True</code></td>
<td>4</td>
</tr>
<tr>
<td><code>ensure_ascii</code></td>
<td><code>bool</code></td>
<td>Control whether non-<code>ASCII</code> characters are escaped to <code>Unicode</code>. When set to <code>True</code>, all non-<code>ASCII</code> characters will be escaped; <code>False</code> retains the original characters, effective only when <code>format_json</code> is <code>True</code></td>
<td><code>False</code></td>
</tr>
<tr>
<td rowspan = "3"><code>save_to_json()</code></td>
<td rowspan = "3">Save the result as a json format file</td>
<td><code>save_path</code></td>
<td><code>str</code></td>
<td>Path of the file to be saved. When it is a directory, the naming of the saved file is consistent with the input file type.</td>
<td>None</td>
</tr>
<tr>
<td><code>indent</code></td>
<td><code>int</code></td>
<td>Specify the indentation level to beautify the output <code>JSON</code> data, making it more readable, effective only when <code>format_json</code> is <code>True</code></td>
<td>4</td>
</tr>
<tr>
<td><code>ensure_ascii</code></td>
<td><code>bool</code></td>
<td>Control whether non-<code>ASCII</code> characters are escaped to <code>Unicode</code>. When set to <code>True</code>, all non-<code>ASCII</code> characters will be escaped; <code>False</code> retains the original characters, effective only when <code>format_json</code> is <code>True</code></td>
<td><code>False</code></td>
</tr>
</table>

* Additionally, it also supports obtaining prediction results through attributes, as follows:

<table>
<thead>
<tr>
<th>Attribute</th>
<th>Description</th>
</tr>
</thead>
<tr>
<td rowspan = "1"><code>json</code></td>
<td rowspan = "1">Get the prediction result in <code>json</code> format</td>
</tr>
</table>

## IV. Secondary Development

The current module does not support fine-tuning training temporarily, only inference integration is supported. The fine-tuning training of this module is planned to be supported in the future.

## V. FAQ