Instructions to use wingchiuloong/DCore-DPace-Qwen3-4B-Budget8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use wingchiuloong/DCore-DPace-Qwen3-4B-Budget8 with Transformers:

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

pipe = pipeline("text-generation", model="wingchiuloong/DCore-DPace-Qwen3-4B-Budget8", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModel

tokenizer = AutoTokenizer.from_pretrained("wingchiuloong/DCore-DPace-Qwen3-4B-Budget8", trust_remote_code=True)
model = AutoModel.from_pretrained("wingchiuloong/DCore-DPace-Qwen3-4B-Budget8", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use wingchiuloong/DCore-DPace-Qwen3-4B-Budget8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/wingchiuloong/DCore-DPace-Qwen3-4B-Budget8

SGLang

How to use wingchiuloong/DCore-DPace-Qwen3-4B-Budget8 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 "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use wingchiuloong/DCore-DPace-Qwen3-4B-Budget8 with Docker Model Runner:
```
docker model run hf.co/wingchiuloong/DCore-DPace-Qwen3-4B-Budget8
```

DCore-DPace-Qwen3-4B-Budget8

Draft model for speculative decoding with Qwen/Qwen3-4B as the target model, trained with a draft budget of 8 tokens per block.

D-Pace is the training recipe of DCore. This checkpoint applies D-Pace to the original DFlash architecture (recipe ablation).

This is a draft model: it is conditioned on hidden states of the target model and is not meant to be used as a standalone language model.

Paper: arXiv:2610.24993

Model Details

Item	Value
Target model	Qwen/Qwen3-4B (36 layers)
Draft architecture	`DFlashDraftModel`
Draft layers	5
Hidden size / heads / KV heads	2560 / 32 / 8
Block size	8
Target layers used as context features	1, 9, 17, 25, 33
Vocabulary	151936 (Qwen3 tokenizer)
Precision	bfloat16

Files

config.json
dflash.py
model.safetensors
registry.py

Usage

from transformers import AutoModel

draft = AutoModel.from_pretrained(
    "wingchiuloong/DCore-DPace-Qwen3-4B-Budget8",
    trust_remote_code=True,
    dtype="bfloat16",
)

The draft model must be paired with Qwen/Qwen3-4B as the target model in a speculative decoding framework that supports DFlash-style block drafting.