Instructions to use Ciaranshu/decor-qwen3-4b-dse with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Ciaranshu/decor-qwen3-4b-dse with Transformers:

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

pipe = pipeline("text-generation", model="Ciaranshu/decor-qwen3-4b-dse")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Ciaranshu/decor-qwen3-4b-dse")
model = AutoModelForCausalLM.from_pretrained("Ciaranshu/decor-qwen3-4b-dse")
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 Ciaranshu/decor-qwen3-4b-dse with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Ciaranshu/decor-qwen3-4b-dse"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Ciaranshu/decor-qwen3-4b-dse",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Ciaranshu/decor-qwen3-4b-dse

SGLang

How to use Ciaranshu/decor-qwen3-4b-dse 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 "Ciaranshu/decor-qwen3-4b-dse" \
    --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": "Ciaranshu/decor-qwen3-4b-dse",
		"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 "Ciaranshu/decor-qwen3-4b-dse" \
        --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": "Ciaranshu/decor-qwen3-4b-dse",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Ciaranshu/decor-qwen3-4b-dse with Docker Model Runner:
```
docker model run hf.co/Ciaranshu/decor-qwen3-4b-dse
```

DecoR: Qwen3-4B Fine-tuned on DSE-Cleaned LIMO

This model is Qwen3-4B fine-tuned on the DSE-cleaned (Dead Step Elimination) version of the LIMO dataset.

DSE removes structurally dead reasoning steps — steps that are unreachable from the conclusion via dependency DAG analysis. This produces cleaner training data that leads to more efficient reasoning.

Key Results

Benchmark	Base Qwen3-4B	Original LIMO SFT	DSE LIMO SFT
MATH-500	56.6%	69.6%	72.8%
AIME 2025	13.3%	40.0%	43.3%
AIME 2026	36.7%	46.7%	53.3%
GPQA Diamond	43.4%	55.6%	49.0%

+3.2% on MATH-500 vs Original LIMO SFT
28% faster training (6h26m vs 8h56m)
11.5% lower training loss (0.240 vs 0.271)
18.5% shorter output traces (more efficient reasoning)

Training Details

Base model: Qwen/Qwen3-4B
Training data: LIMO-DSE (817 samples, DSE-cleaned)
Framework: LLaMA-Factory + DeepSpeed ZeRO-2
Hardware: 2× NVIDIA A100 80GB (CSD3 HPC)
Hyperparameters:
- Learning rate: 5e-6 (cosine schedule, 10% warmup)
- Batch size: 8 (effective)
- Epochs: 15
- Max sequence length: 16384
- Full fine-tuning (no LoRA)

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("Ciaranshu/decor-qwen3-4b-dse")
tokenizer = AutoTokenizer.from_pretrained("Ciaranshu/decor-qwen3-4b-dse")

Citation

@article{decor2026,
  title={The Structural Reward Gap: How Outcome-Based RL Creates Superstitious Reasoning},
  author={Shu, Chang},
  year={2026}
}

Framework versions

Transformers 4.52.4
Pytorch 2.5.1+cu124
Datasets 3.6.0
Tokenizers 0.21.1

Downloads last month: 6

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for Ciaranshu/decor-qwen3-4b-dse

Base model

Qwen/Qwen3-4B-Base

Finetuned

Qwen/Qwen3-4B

Finetuned

(690)

this model

Paper for Ciaranshu/decor-qwen3-4b-dse

LIMO: Less is More for Reasoning

Paper • 2502.03387 • Published Feb 5, 2025 • 63