Instructions to use BPXFQ/qwen3-0.6b-safety-math with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use BPXFQ/qwen3-0.6b-safety-math with Transformers:

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

pipe = pipeline("text-generation", model="BPXFQ/qwen3-0.6b-safety-math")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("BPXFQ/qwen3-0.6b-safety-math")
model = AutoModelForCausalLM.from_pretrained("BPXFQ/qwen3-0.6b-safety-math", 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 BPXFQ/qwen3-0.6b-safety-math with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "BPXFQ/qwen3-0.6b-safety-math"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "BPXFQ/qwen3-0.6b-safety-math",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/BPXFQ/qwen3-0.6b-safety-math

SGLang

How to use BPXFQ/qwen3-0.6b-safety-math 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 "BPXFQ/qwen3-0.6b-safety-math" \
    --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": "BPXFQ/qwen3-0.6b-safety-math",
		"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 "BPXFQ/qwen3-0.6b-safety-math" \
        --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": "BPXFQ/qwen3-0.6b-safety-math",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use BPXFQ/qwen3-0.6b-safety-math with Docker Model Runner:
```
docker model run hf.co/BPXFQ/qwen3-0.6b-safety-math
```

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 Safety & Math Fine-tuned Model

Fine-tuned from Qwen3-0.6B (architecture and parameter count unchanged, ~0.6B params, bf16) to simultaneously improve safety refusal and math reasoning while preserving general ability.

Grading rubric: safety ×40% + math ×30% + general ×30%.

Training

Method: QLoRA 4-bit (NF4 + double quant), r=16, alpha=32, dropout=0.05, all linear modules (q/k/v/o/gate/up/down). Response-only label masking (assistant content + <|im_end|>).
Config: lr=1e-4, cosine, warmup 5%, 1 epoch, batch=1 + grad_accum=8 (effective 8), gradient checkpointing, bf16, paged_adamw_8bit, max_length=1024. RTX 4060 Laptop (8GB).
Data: balanced interleaved mix (safety 23% / math 43% / general 34%):
- Safety: HH-RLHF(harmless), PKU-SafeRLHF, calibrated refusal templates (refuse + safe alternative), Chinese-SafetyQA, manual boundary examples, Dolly conservative boundary.
- Math: GSM8K, MATH (7 subtypes), MetaMathQA, AQuA.
- General: No Robots, COIG-CQIA, OASST1 (en/zh), Dolly-15k.

Evaluation (fixed subset, enable_thinking=False)

Metric	Value
Math (96 stratified GSM8K/MATH)	21.9%
HHI balanced accuracy	78%
JBB balanced accuracy	77%
General Chinese MCQ (200)	69.5%

Composite proxy (0.4×safety + 0.3×math + 0.3×general) ≈ 0.584, highest among all candidates.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("BPXFQ/qwen3-0.6b-safety-math", dtype="bfloat16")
tokenizer = AutoTokenizer.from_pretrained("BPXFQ/qwen3-0.6b-safety-math")
messages = [{"role": "user", "content": "17 * 23 = ?"}]
inputs = tokenizer.apply_chat_template(
    messages, tokenize=True, add_generation_prompt=True,
    enable_thinking=False, return_tensors="pt", return_dict=True,
)
outputs = model.generate(**inputs, max_new_tokens=128, do_sample=False,
                         eos_token_id=tokenizer.eos_token_id, pad_token_id=tokenizer.pad_token_id)
print(tokenizer.decode(outputs[0, inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Downloads last month: -

Safetensors

Model size

0.6B params

Tensor type

BF16

Model tree for BPXFQ/qwen3-0.6b-safety-math

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Finetuned

(1139)

this model