Instructions to use kp-forks/TeleOCR with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kp-forks/TeleOCR with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="kp-forks/TeleOCR", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("kp-forks/TeleOCR", trust_remote_code=True)
model = AutoModelForMultimodalLM.from_pretrained("kp-forks/TeleOCR", trust_remote_code=True, device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use kp-forks/TeleOCR with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kp-forks/TeleOCR"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kp-forks/TeleOCR",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/kp-forks/TeleOCR

SGLang

How to use kp-forks/TeleOCR with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "kp-forks/TeleOCR" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kp-forks/TeleOCR",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "kp-forks/TeleOCR" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kp-forks/TeleOCR",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use kp-forks/TeleOCR with Docker Model Runner:
```
docker model run hf.co/kp-forks/TeleOCR
```

parikhkadam commited on 16 days ago

Commit

cdac4fa

1 Parent(s): 9e6159d

Update README.md

Browse files

Original commit: c394c1f5d481e836aea7b8651b76d71fe3bb7dc0

Files changed (1) hide show

README.md +39 -3

README.md CHANGED Viewed

@@ -29,8 +29,14 @@ NaviDC-OCR: Navigating Document Parsing Across Digital and Camera-Captured Docum
   <img src="https://raw.githubusercontent.com/caipeng328/NaviDC-OCR/refs/heads/main/assets/score.png" width="800">
 </div>
-## Introduction
-**Navigating Document Parsing Across Digital and Camera-Captured Documents**
 **NaviDC-OCR** is a lightweight (~1.2B parameters), open-source Vision-Language Model designed specifically for document parsing.
@@ -47,6 +53,8 @@ Compared with previous document parsing models, NaviDC-OCR introduces
 These techniques enable NaviDC-OCR to achieve state-of-the-art performance on both digital and camera-captured document benchmarks while remaining lightweight enough for practical deployment.
 NaviDC-OCR  achieves state-of-the-art performance on multiple public document parsing benchmarks.
 ### Layout Visualization of Distorted Documents
@@ -56,6 +64,28 @@ To evaluate the model's ability to understand complex document deformations, we
 <img src="https://raw.githubusercontent.com/caipeng328/NaviDC-OCR/refs/heads/main/assets/dir300.png" width="800" alt="Parsing evaluation on the DIR300 dataset.">
 <img src="https://raw.githubusercontent.com/caipeng328/NaviDC-OCR/refs/heads/main/assets/docunet.png" width="800" alt="Parsing evaluation on the DocUNet dataset.">
 </div>
 ### [OmniDocBench v1.6](https://github.com/opendatalab/OmniDocBench)
 | Model Type | Methods | Param | Overall ↑ | Text Edit ↓ | Formula CDM ↑ | Table TEDS ↑ | Table TEDS-S ↑ | Read Order Edit ↓ |
@@ -86,6 +116,9 @@ To evaluate the model's ability to understand complex document deformations, we
 | | GPT-5.2 | -- | 86.52 | 0.114 | 88.00 | 82.95 | 87.93 | 0.193 |
 | | InternVL3.5-241B | 241B | 83.61 | 0.130 | 89.52 | 74.35 | 79.78 | 0.215 |
 ### [Wild_OmniDocBench](https://github.com/VirtualLUOUCAS/Wild_OmniDocBench)
 | Model Type          | Methods           | Param |    Overall ↑ | Text Edit ↓ | Formula CDM ↑ | Table TEDS ↑ | Table TEDS-S ↑ | Read Order Edit ↓ |
 | ------------------- | ----------------- | ----: | -----------: | ----------: | ------------: | -----------: | -------------: | ----------------: |
@@ -99,6 +132,7 @@ To evaluate the model's ability to understand complex document deformations, we
 |                     | HunyuanOCR-1.5    |    1B |        77.62 |      0.1979 |         85.12 |        67.54 |          70.67 |            0.2750 |
 |                     | Logics-Parsing-v2 |    4B |        77.10 |      0.4029 |          91.4 |        80.19 |          87.16 |            0.2355 |
 ### [PureDocBench](https://github.com/zhihengli-casia/puredocbench/)
@@ -122,6 +156,7 @@ To evaluate the model's ability to understand complex document deformations, we
 |                    | Gemini-3.1-Pro    |           70.04 |        0.306 |           65.63 |         75.08 |                      69.28 |                   0.322 |                      65.81 |                    74.24 |               **71.98** |                0.300 |                   68.62 |                 77.26 |
 |                    | Qwen3.5-397B-A17B |           69.12 |        0.233 |           65.26 |         65.40 |                      68.34 |                   0.244 |                      63.91 |                    65.53 |                   62.70 |                0.287 |                   60.70 |                 56.12 |
 ### [ICDAR2026 Sci-ImageMiner](https://sites.google.com/view/sci-imageminer/)
@@ -134,8 +169,9 @@ To evaluate the model's ability to understand complex document deformations, we
 |     5 | DocMiner        |     12.67 |     53.72 |     33.19 |
 |     6 | Qwen3 VL 8B     |     14.08 |     57.86 |     35.97 |
-## Installation
 ```bash
 pip install transformers torch pillow

   <img src="https://raw.githubusercontent.com/caipeng328/NaviDC-OCR/refs/heads/main/assets/score.png" width="800">
 </div>
+## 🔥 News
+- **2026/09/01** - We noticed that EMNLP 2026 is hosting the [Dr.DocBench Challenge](https://eval.ai/web/challenges/challenge-page/2717/overview), a document parsing competition. We evaluated NaviDC-OCR with its native weights, achieving better results than MinerU 2.5 Pro and PaddleOCR-VL 1.6. Detailed results are shown below **dr.docbench-challenge**. We welcome the use of NaviDC‑OCR for competitions. Going forward, we will continue to deliver competitive parsing models for the community.
+- **2026/08/29** — Thanks to Nandraj for the GGUF conversion and llama.cpp support!  [🔗 NaviDC-OCR-GGUF](https://huggingface.co/nandraj/NaviDC-OCR-GGUF)
+- **2026/08/17** — NaviDC-OCR [model weights](https://huggingface.co/StarDoc-AI/NaviDC-OCR) and [technical report](https://arxiv.org/pdf/2608.12898) have been released.
+## 📖 Introduction
 **NaviDC-OCR** is a lightweight (~1.2B parameters), open-source Vision-Language Model designed specifically for document parsing.
 These techniques enable NaviDC-OCR to achieve state-of-the-art performance on both digital and camera-captured document benchmarks while remaining lightweight enough for practical deployment.
+## 📊 Experimental Results
 NaviDC-OCR  achieves state-of-the-art performance on multiple public document parsing benchmarks.
 ### Layout Visualization of Distorted Documents
 <img src="https://raw.githubusercontent.com/caipeng328/NaviDC-OCR/refs/heads/main/assets/dir300.png" width="800" alt="Parsing evaluation on the DIR300 dataset.">
 <img src="https://raw.githubusercontent.com/caipeng328/NaviDC-OCR/refs/heads/main/assets/docunet.png" width="800" alt="Parsing evaluation on the DocUNet dataset.">
 </div>
+---
+## Dr.DocBench Challenge
+|模型|overall ↑|Text edit ↓|formula cdm ↑|Table teds ↑|order edit ↓|
+|---|---|---|---|---|---|
+|**Specialized VLMs**| | | | | |
+|Navidc‑OCR|67.96|0.1903|0.02|64.97|0.398|
+|Mineru 2.5 pro|62.26|0.3402|0.04|67.75|0.356|
+|Mineru 2.5|54.37|0.33|24.15|55.85|0.30|
+|PaddleOCRvl 1.6|55.11 | 0.4364 | 0.21 | 51.34 |0.412 |
+|PaddleOCRvl 1.5|34.78|0.73|30.73|51.79|0.71|
+|**General VLMs**| | | | | |
+|Qwen3.5‑Flash|57.51|0.26|35.69|46.24|0.26|
+|Qwen3.5‑122B‑A10B|56.32|0.23|32.54|38.31|0.23|
+|Kimi‑K2.5|60.38|0.19|27.04|51.98|0.18|
+|Doubao‑Seed‑1.6‑Vision|53.14|0.34|41.28|3.93|0.28|
+|GPT‑5.5|61.94|0.19|34.55|48.90|0.17|
+---
 ### [OmniDocBench v1.6](https://github.com/opendatalab/OmniDocBench)
 | Model Type | Methods | Param | Overall ↑ | Text Edit ↓ | Formula CDM ↑ | Table TEDS ↑ | Table TEDS-S ↑ | Read Order Edit ↓ |
 | | GPT-5.2 | -- | 86.52 | 0.114 | 88.00 | 82.95 | 87.93 | 0.193 |
 | | InternVL3.5-241B | 241B | 83.61 | 0.130 | 89.52 | 74.35 | 79.78 | 0.215 |
+---
 ### [Wild_OmniDocBench](https://github.com/VirtualLUOUCAS/Wild_OmniDocBench)
 | Model Type          | Methods           | Param |    Overall ↑ | Text Edit ↓ | Formula CDM ↑ | Table TEDS ↑ | Table TEDS-S ↑ | Read Order Edit ↓ |
 | ------------------- | ----------------- | ----: | -----------: | ----------: | ------------: | -----------: | -------------: | ----------------: |
 |                     | HunyuanOCR-1.5    |    1B |        77.62 |      0.1979 |         85.12 |        67.54 |          70.67 |            0.2750 |
 |                     | Logics-Parsing-v2 |    4B |        77.10 |      0.4029 |          91.4 |        80.19 |          87.16 |            0.2355 |
+---
 ### [PureDocBench](https://github.com/zhihengli-casia/puredocbench/)
 |                    | Gemini-3.1-Pro    |           70.04 |        0.306 |           65.63 |         75.08 |                      69.28 |                   0.322 |                      65.81 |                    74.24 |               **71.98** |                0.300 |                   68.62 |                 77.26 |
 |                    | Qwen3.5-397B-A17B |           69.12 |        0.233 |           65.26 |         65.40 |                      68.34 |                   0.244 |                      63.91 |                    65.53 |                   62.70 |                0.287 |                   60.70 |                 56.12 |
+---
 ### [ICDAR2026 Sci-ImageMiner](https://sites.google.com/view/sci-imageminer/)
 |     5 | DocMiner        |     12.67 |     53.72 |     33.19 |
 |     6 | Qwen3 VL 8B     |     14.08 |     57.86 |     35.97 |
+---
+## 🚀 Installation
 ```bash
 pip install transformers torch pillow