Instructions to use RUC-AIBOX/Qwen-7B-SimpleDeepSearcher with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use RUC-AIBOX/Qwen-7B-SimpleDeepSearcher with Transformers:

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

pipe = pipeline("text-generation", model="RUC-AIBOX/Qwen-7B-SimpleDeepSearcher")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("RUC-AIBOX/Qwen-7B-SimpleDeepSearcher")
model = AutoModelForCausalLM.from_pretrained("RUC-AIBOX/Qwen-7B-SimpleDeepSearcher")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use RUC-AIBOX/Qwen-7B-SimpleDeepSearcher with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "RUC-AIBOX/Qwen-7B-SimpleDeepSearcher"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "RUC-AIBOX/Qwen-7B-SimpleDeepSearcher",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/RUC-AIBOX/Qwen-7B-SimpleDeepSearcher

SGLang

How to use RUC-AIBOX/Qwen-7B-SimpleDeepSearcher 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 "RUC-AIBOX/Qwen-7B-SimpleDeepSearcher" \
    --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": "RUC-AIBOX/Qwen-7B-SimpleDeepSearcher",
		"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 "RUC-AIBOX/Qwen-7B-SimpleDeepSearcher" \
        --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": "RUC-AIBOX/Qwen-7B-SimpleDeepSearcher",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use RUC-AIBOX/Qwen-7B-SimpleDeepSearcher with Docker Model Runner:
```
docker model run hf.co/RUC-AIBOX/Qwen-7B-SimpleDeepSearcher
```

nielsr HF Staff commited on Oct 9, 2025

Commit

6543860

verified ·

1 Parent(s): 82d3f1f

Improve model card: add metadata, links, and detailed description

Browse files

This PR significantly improves the model card for `Online-Searcher-Qwen-7B` by:

- Adding `pipeline_tag: text-generation` to the metadata, allowing the model to be discovered more easily (e.g., at https://huggingface.co/models?pipeline_tag=text-generation).
- Adding `license: mit` to the metadata, clearly stating the model's license as found in the GitHub repository.
- Adding a link to the original paper [SimpleDeepSearcher: Deep Information Seeking via Web-Powered Reasoning Trajectory Synthesis](https://huggingface.co/papers/2505.16834).
- Adding a link to the associated [GitHub repository](https://github.com/RUCAIBox/SimpleDeepSearcher).
- Populating the "Model description", "Intended uses & limitations", and "Training and evaluation data" sections with information extracted from the paper abstract and GitHub README.
- Adding a dedicated "License" section within the model card content.
- Removing the auto-generated comment.

These changes provide a more comprehensive and informative model card for users.

Files changed (1) hide show

README.md +16 -7

README.md CHANGED Viewed

@@ -1,5 +1,7 @@
 ---
 library_name: transformers
 tags:
 - generated_from_trainer
 model-index:
@@ -7,24 +9,27 @@ model-index:
   results: []
 ---
-<!-- This model card has been generated automatically according to the information the Trainer had access to. You
-should probably proofread and complete it, then remove this comment. -->
 # Online-Searcher-Qwen-7B
-This model was trained from scratch on an unknown dataset.
 ## Model description
-More information needed
 ## Intended uses & limitations
-More information needed
 ## Training and evaluation data
-More information needed
 ## Training procedure
@@ -45,3 +50,7 @@ The following hyperparameters were used during training:
 - Pytorch 2.5.1+cu124
 - Datasets 2.19.0
 - Tokenizers 0.20.3

 ---
 library_name: transformers
+pipeline_tag: text-generation
+license: mit
 tags:
 - generated_from_trainer
 model-index:
   results: []
 ---
 # Online-Searcher-Qwen-7B
+This model is a Qwen2-7B based deep search agent, presented in the paper [SimpleDeepSearcher: Deep Information Seeking via Web-Powered Reasoning Trajectory Synthesis](https://huggingface.co/papers/2505.16834).
+The code for the paper and model training is available on the [Github repository](https://github.com/RUCAIBox/SimpleDeepSearcher).
 ## Model description
+SimpleDeepSearcher is a lightweight yet effective framework for enhancing large language models (LLMs) in complex deep search scenarios requiring multi-step reasoning and iterative information retrieval. Unlike traditional Retrieval-Augmented Generation (RAG) or Reinforcement Learning (RL)-based methods, SimpleDeepSearcher strategically synthesizes high-quality reasoning trajectories in real-world web environments. This approach enables efficient Supervised Fine-Tuning (SFT) using only a small amount of curated data, resulting in strong performance with significantly reduced computational cost and development complexity.
+Key contributions of the framework include:
+-   A real web-based data synthesis framework that simulates realistic user search behaviors, generating multi-turn reasoning and search trajectories.
+-   A multi-criteria data curation strategy that jointly optimizes both input question selection and output response filtering through orthogonal filtering dimensions.
 ## Intended uses & limitations
+This model is intended for deep information seeking tasks and complex search scenarios that demand multi-step reasoning and iterative information retrieval from the web. It aims to provide practical insights for efficient deep search systems by leveraging its ability to synthesize high-quality reasoning trajectories.
 ## Training and evaluation data
+The model was fine-tuned using Supervised Fine-Tuning (SFT) on a small, high-quality dataset consisting of only 871 curated samples. This data was strategically synthesized by simulating realistic user interactions in live web search environments, coupled with a multi-criteria curation strategy that optimizes diversity and quality.
 ## Training procedure
 - Pytorch 2.5.1+cu124
 - Datasets 2.19.0
 - Tokenizers 0.20.3
+## License
+This project is released under the [MIT License](https://github.com/RUCAIBox/SimpleDeepSearcher/blob/main/LICENSE).