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

Libraries

How to use broadfield-dev/Qwen3-0.6B-onnx with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("broadfield-dev/Qwen3-0.6B-onnx")
model = AutoModelForCausalLM.from_pretrained("broadfield-dev/Qwen3-0.6B-onnx")
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 broadfield-dev/Qwen3-0.6B-onnx with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/broadfield-dev/Qwen3-0.6B-onnx

SGLang

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

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

ONNX Export: Qwen/Qwen3-0.6B

This is a version of Qwen/Qwen3-0.6B that has been converted to ONNX and optimized.

Model Details

Base Model: Qwen/Qwen3-0.6B
Task: text-generation
Opset Version: 17
Optimization: FP32 (No Quantization)

Usage

Installation

For a lightweight mobile/serverless setup, you only need onnxruntime and tokenizers.

pip install onnxruntime tokenizers optimum

Python Example


from tokenizers import Tokenizer
import onnxruntime as ort
import numpy as np

# 1. Load the lightweight tokenizer (No Transformers dependency needed)
tokenizer = Tokenizer.from_pretrained("broadfield-dev/Qwen3-0.6B-onnx")

# 2. Load the ONNX model
# For Generative/Chat models, use: optimum.onnxruntime.ORTModelForCausalLM
session = ort.InferenceSession("model.onnx")

# 3. Preprocess (Simple text encoding)
text = "Run inference on mobile!"
encoding = tokenizer.encode(text)

# Prepare inputs (Exact names vary by model, usually input_ids + attention_mask)
inputs = {
    "input_ids": np.array([encoding.ids], dtype=np.int64),
    "attention_mask": np.array([encoding.attention_mask], dtype=np.int64)
}

# 4. Run Inference
outputs = session.run(None, inputs)
print("Output logits shape:", outputs[0].shape)

About this Export

This model was exported using Optimum. It includes the FP32 (No Quantization) quantization settings and a pre-compiled tokenizer.json for fast loading.

Downloads last month: 5

Model tree for broadfield-dev/Qwen3-0.6B-onnx

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Quantized

(304)

this model