Instructions to use yujianing/Qwen3-0.6B-ZJU with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use yujianing/Qwen3-0.6B-ZJU with Transformers:

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

pipe = pipeline("text-generation", model="yujianing/Qwen3-0.6B-ZJU")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("yujianing/Qwen3-0.6B-ZJU")
model = AutoModelForCausalLM.from_pretrained("yujianing/Qwen3-0.6B-ZJU", 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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use yujianing/Qwen3-0.6B-ZJU with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "yujianing/Qwen3-0.6B-ZJU"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "yujianing/Qwen3-0.6B-ZJU",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/yujianing/Qwen3-0.6B-ZJU

SGLang

How to use yujianing/Qwen3-0.6B-ZJU 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 "yujianing/Qwen3-0.6B-ZJU" \
    --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": "yujianing/Qwen3-0.6B-ZJU",
		"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 "yujianing/Qwen3-0.6B-ZJU" \
        --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": "yujianing/Qwen3-0.6B-ZJU",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use yujianing/Qwen3-0.6B-ZJU with Docker Model Runner:
```
docker model run hf.co/yujianing/Qwen3-0.6B-ZJU
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Qwen3-0.6B ZJU Cyber Safety Math A+

This is the final Model A+ v2 candidate prepared for the Zhejiang University cybersecurity summer camp project evaluation.

Model Summary

Base model: Qwen/Qwen3-0.6B
Method: 4bit QLoRA SFT
Final version: Model A+ v2
Task focus: safety alignment, mathematical reasoning, and general ability preservation
Format: merged full 0.6B model, not a LoRA adapter

The LoRA adapter was merged back into the original unquantized Qwen3-0.6B base model. This directory is intended to be loadable directly with Hugging Face Transformers.

Intended Use

This model is intended for authorized academic evaluation in the Zhejiang University cybersecurity summer camp setting. It is designed to improve refusal behavior for harmful requests, preserve useful answers for legitimate defensive cybersecurity questions, and improve math reasoning while keeping general instruction-following ability stable.

Loading

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_path = "yujianing/qwen3-0.6b-zju-cyber-safety-math-a-plus"

tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    torch_dtype=torch.float16,
    device_map="auto",
    trust_remote_code=True,
)

Limitations

This model is not guaranteed to be safe in all adversarial settings. It should not be used to provide harmful instructions, offensive code, unauthorized exploitation guidance, or other unsafe operational details. Further evaluation is recommended before deployment beyond the project evaluation context.

Downloads last month: 8

Safetensors

Model size

0.6B params

Tensor type

BF16

Model tree for yujianing/Qwen3-0.6B-ZJU

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Finetuned

(1173)

this model