Instructions to use jinaai/ReaderLM-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jinaai/ReaderLM-v2 with Transformers:

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

pipe = pipeline("text-generation", model="jinaai/ReaderLM-v2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jinaai/ReaderLM-v2")
model = AutoModelForCausalLM.from_pretrained("jinaai/ReaderLM-v2", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use jinaai/ReaderLM-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jinaai/ReaderLM-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jinaai/ReaderLM-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/jinaai/ReaderLM-v2

SGLang

How to use jinaai/ReaderLM-v2 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 "jinaai/ReaderLM-v2" \
    --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": "jinaai/ReaderLM-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "jinaai/ReaderLM-v2" \
        --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": "jinaai/ReaderLM-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use jinaai/ReaderLM-v2 with Docker Model Runner:
```
docker model run hf.co/jinaai/ReaderLM-v2
```

Optimizing Model Inference Speed for Large HTML Inputs

#12

by chiralzhan - opened Feb 21, 2025

Discussion

chiralzhan

Feb 21, 2025

•

edited Feb 21, 2025

Hello,

I'm currently utilizing the H20 , ReaderLM model with the max_new_tokens parameter set to 4096, adhering to the official guidelines. My typical input consists of HTML documents ranging from 20 to 40 kb. The inference process generally takes about 4 to 5 minutes. Is this duration considered normal? If so, could you recommend strategies to enhance the inference speed?

Thank you for your assistance.

numb3r3

Jina AI org Feb 26, 2025

I believe you can do some simple cleaning of HTML input. Basically, it works.

chiralzhan

Feb 27, 2025

Thank you for your reply. The HTML I input has been thoroughly cleaned, even more so than the official cleaning logic provided. Since I'm working on real-time search and web-based RAG, some web pages, after cleaning, do become that large.

Currently, after replacing Transformer inference with VLLM, the speed has increased to five times faster than before, and in the best cases, even ten times faster. However, some HTMLs cause errors when using VLLM's tokenizer. Although no error is reported, the phenomenon is that tens of thousands of tokens load per second, and the return is empty. The specific reason is still unclear, but at least it's certain that using VLLM inference can significantly improve speed.

Once again, sincere thanks for your reply.

eric-23fe2

Mar 10, 2025

I believe you can do some simple cleaning of HTML input. Basically, it works.

is it possible to train the model to inference multiple tokens in one predict step? instead of just inferece one token each time.

numb3r3

Jina AI org Mar 10, 2025

That's a good idea which demonstrate a good boost on decoding stage. Actually, we are investigating some advanced decoding strategy. We hope we can bring a more efficiency model later at some time point.

chaochaoli

Apr 28, 2025

mark

devops724

Dec 2, 2025

Currently, after replacing Transformer inference with VLLM, the speed has increased to five times faster than before, and in the best cases, even ten times faster. However, some HTMLs cause errors when using VLLM's tokenizer. Although no error is reported, the phenomenon is that tens of thousands of tokens load per second, and the return is empty. The specific reason is still unclear, but at least it's certain that using VLLM inference can significantly improve speed.

same issue on vllm
cause endless repetitive content output

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment