Instructions to use strangervisionhf/deepseek-ocr-2-transformers-v4.57.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use strangervisionhf/deepseek-ocr-2-transformers-v4.57.1 with Transformers:

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

pipe = pipeline("image-text-to-text", model="strangervisionhf/deepseek-ocr-2-transformers-v4.57.1", trust_remote_code=True)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("strangervisionhf/deepseek-ocr-2-transformers-v4.57.1", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use strangervisionhf/deepseek-ocr-2-transformers-v4.57.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "strangervisionhf/deepseek-ocr-2-transformers-v4.57.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "strangervisionhf/deepseek-ocr-2-transformers-v4.57.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/strangervisionhf/deepseek-ocr-2-transformers-v4.57.1

SGLang

How to use strangervisionhf/deepseek-ocr-2-transformers-v4.57.1 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 "strangervisionhf/deepseek-ocr-2-transformers-v4.57.1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "strangervisionhf/deepseek-ocr-2-transformers-v4.57.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "strangervisionhf/deepseek-ocr-2-transformers-v4.57.1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "strangervisionhf/deepseek-ocr-2-transformers-v4.57.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use strangervisionhf/deepseek-ocr-2-transformers-v4.57.1 with Docker Model Runner:
```
docker model run hf.co/strangervisionhf/deepseek-ocr-2-transformers-v4.57.1
```

This is a copy of the model weights from the https://huggingface.co/deepseek-ai/DeepSeek-OCR-2 model. These weights cannot be used for other purposes. If you wish to do so, please visit the original model page.

Previously, inference with the model https://huggingface.co/deepseek-ai/DeepSeek-OCR-2 ran smoothly on transformers==4.46.3. However, running it on newer versions of transformers caused compatibility issues. We have identified and fixed the problem, and the model now runs smoothly with the latest transformers (v4.57.1) and other compatible versions.

This page includes the updated model weights and corrected configuration, which resolve the issue and allow transformers inference to run smoothly.

Last updated: 11:50 AM (IST), February 12, 2026.

Quick Start with Transformers

Install the required packages

torch
torchvision
transformers==4.57.1
accelerate
matplotlib 
einops
addict 
easydict

Usage

from transformers import AutoModel, AutoTokenizer
import torch
import os
os.environ["CUDA_VISIBLE_DEVICES"] = '0'
model_name = 'strangervisionhf/deepseek-ocr-2-transformers-v4.57.1'

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModel.from_pretrained(model_name,
                                  #_attn_implementation='flash_attention_2',
                                  trust_remote_code=True,
                                  use_safetensors=True)
model = model.eval().cuda().to(torch.bfloat16)

# prompt = "<image>\nFree OCR. "
prompt = "<image>\n<|grounding|>Convert the document to markdown. "
image_file = 'your_image.jpg'
output_path = 'your/output/dir'


res = model.infer(tokenizer, prompt=prompt, image_file=image_file, output_path = output_path, base_size = 1024, image_size = 768, crop_mode=True, save_results = True)