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

Libraries

How to use WindyLab/Qwen3-0.6B-cybertown-SFT with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WindyLab/Qwen3-0.6B-cybertown-SFT")
model = AutoModelForCausalLM.from_pretrained("WindyLab/Qwen3-0.6B-cybertown-SFT")
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 WindyLab/Qwen3-0.6B-cybertown-SFT with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/WindyLab/Qwen3-0.6B-cybertown-SFT

SGLang

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

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

Qwen3-0.6B-cybertown-SFT

This model is a Cybertown task-planning SFT checkpoint based on Qwen3-0.6B.

The supervised data was collected from gemini-3-flash-preview and filtered to successful validator runs. The dataset contains 3,423 prompt-response examples, all with validator_valid=True.

Training Data Distribution

By Source

source	count	ratio
first_plan_success	1083	31.6%
replan_success	2340	68.4%

By Goal Type

goal_type	count	ratio
assembly	1145	33.5%
transport	1010	29.5%
guidance	356	10.4%
emergency_response	355	10.4%
target_following	331	9.7%
traffic_enforcement	226	6.6%

Intended Use

This model is intended for Cybertown semantic task planning and replanning experiments. It outputs structured planning responses for downstream validation and reinforcement learning.

Loading

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "WindyLab/Qwen3-0.6B-cybertown-SFT"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

Downloads last month: 5

Safetensors

Model size

0.6B params

Tensor type

BF16

Model tree for WindyLab/Qwen3-0.6B-cybertown-SFT

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Finetuned

(1037)

this model

Finetunes

1 model

WindyLab
/

Qwen3-0.6B-cybertown-SFT