README.md

---
license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
base_model:
  - Qwen/Qwen3.8-27B
inference: false
tags:
  - dflash2
  - speculative-decoding
  - block-diffusion
  - draft-model
  - sglang
  - vllm
  - mirror
  - veloGB10
---

# Qwen3.8-27B-DFlash2 (mirror)

> **This repository is a mirror of [z-lab/Qwen3.8-27B-DFlash2](https://huggingface.co/z-lab/Qwen3.8-27B-DFlash2).**
> We mirror it here so it can be pulled from a stable location. The model, weights, and the
> original model card below are **unmodified** from the upstream source — full credit to **z-lab / Inco AI**.
> This mirror is provided for the [veloGB10](https://github.com/sf-stav/veloGB10) engine and workflow.
>
> **License: Apache-2.0** (the upstream card declares `apache-2.0`; a copy of the license is
> included in this repository as `LICENSE`). The upstream [README](https://huggingface.co/z-lab/Qwen3.8-27B-DFlash2/raw/main/README.md)
> is reproduced in full below.

---

# Qwen3.8-27B-DFlash2

[Blog](https://inco.ai/blog/dflash2/) | [GitHub](https://github.com/z-lab/dflash)

This repository contains the DFlash 2 draft model for
[`Qwen/Qwen3.8-27B`](https://huggingface.co/Qwen/Qwen3.8-27B).
It is not a standalone language model: it runs inside a speculative
decoding server and drafts tokens for the target model to verify. The checkpoint is also
mirrored at [`z-lab/Qwen3.8-27B-DFlash2`](https://huggingface.co/z-lab/Qwen3.8-27B-DFlash2).

DFlash 2 is a block-diffusion drafter for speculative decoding. It predicts
a whole block of tokens in a single pass and keeps the top candidates at
every position. A lightweight selector then traces one coherent path through them.
Two-tap dynamic convolutions in the backbone keep the draft from decaying
toward the end of the block. Decoding is lossless: greedy output
matches the target model exactly, and sampling preserves its distribution.

<div align="center">
  <img src="assets/dflash2-figure.png" alt="DFlash 2: parallel block drafting with a candidate path selector" width="100%">
</div>

## Quick Start

Serve with [SGLang](https://github.com/sgl-project/sglang):

```bash
pip install "sglang[all] @ git+https://github.com/sgl-project/sglang.git#subdirectory=python"

python -m sglang.launch_server \
  --model-path Qwen/Qwen3.8-27B \
  --speculative-algorithm DFLASH \
  --speculative-draft-model-path incoai/Qwen3.8-27B-DFlash2 \
  --speculative-num-draft-tokens 8
```

Or with [vLLM](https://github.com/vllm-project/vllm):

```bash
pip install -U "vllm @ git+https://github.com/vllm-project/vllm.git@refs/pull/52816/head"

vllm serve Qwen/Qwen3.8-27B \
  --speculative-config '{
    "method": "dflash",
    "model": "incoai/Qwen3.8-27B-DFlash2",
    "num_speculative_tokens": 7
  }'
```

See the [blog post](https://inco.ai/blog/dflash2/) for other engines and more details.

## Evaluation

- Runtime: SGLang on one NVIDIA H200, with FlashAttention 3 for target and draft attention
- Speculation block size: 8 (7 draft tokens per verification step)
- Sampling: Qwen3.8's officially recommended parameters (temperature 1.0, top-p 0.95, top-k 20), with `xhigh` reasoning effort
- Maximum new tokens: 4096
- Prompts: benchmark formatting from [`z-lab/dflash`](https://github.com/z-lab/dflash)

We compare autoregressive decoding, Qwen3.8's built-in seven-token MTP,
a community DSpark drafter
([`RadixArk/Qwen3.8-27B-DSpark`](https://huggingface.co/RadixArk/Qwen3.8-27B-DSpark)),
and DFlash 2. All speculative methods propose seven draft tokens per
verification step.

### Acceptance Length

Acceptance length is the per-request mean of completion tokens divided by verification steps.
Higher is better.

| Task | MTP | DSpark | DFlash 2 |
| :--- | ---: | ---: | ---: |
| GSM8K | 5.02 | 4.36 | **5.46** |
| MATH-500 | 4.72 | 3.92 | **5.28** |
| HumanEval | 3.91 | 3.30 | **4.39** |
| MBPP | 3.99 | 3.51 | **4.79** |
| MT-Bench | 3.74 | 3.01 | **4.10** |

### Throughput

Throughput is total output tokens divided by end-to-end wall time.
Each cell shows `output tok/s (speedup vs. autoregressive)`.

#### Concurrency 1

| Task | Autoregressive | MTP | DSpark | DFlash 2 |
| :--- | ---: | ---: | ---: | ---: |
| GSM8K | 68.9 | 178.5 (2.59×) | 185.3 (2.69×) | **236.1 (3.43×)** |
| MATH-500 | 69.0 | 172.8 (2.51×) | 174.5 (2.53×) | **230.7 (3.34×)** |
| HumanEval | 69.0 | 151.9 (2.20×) | 159.9 (2.32×) | **214.6 (3.11×)** |
| MBPP | 69.0 | 153.1 (2.22×) | 163.3 (2.37×) | **226.9 (3.29×)** |
| MT-Bench | 68.9 | 134.9 (1.96×) | 137.6 (2.00×) | **184.0 (2.67×)** |

#### Concurrency 8

| Task | Autoregressive | MTP | DSpark | DFlash 2 |
| :--- | ---: | ---: | ---: | ---: |
| GSM8K | 467.2 | 1,022.1 (2.19×) | 1,040.8 (2.23×) | **1,328.7 (2.84×)** |
| MATH-500 | 480.0 | 1,023.5 (2.13×) | 1,025.8 (2.14×) | **1,368.3 (2.85×)** |
| HumanEval | 483.4 | 934.2 (1.93×) | 956.5 (1.98×) | **1,291.5 (2.67×)** |
| MBPP | 478.0 | 938.1 (1.96×) | 974.1 (2.04×) | **1,328.0 (2.78×)** |
| MT-Bench | 480.5 | 835.2 (1.74×) | 802.3 (1.67×) | **1,090.2 (2.27×)** |

#### Concurrency 32

| Task | Autoregressive | MTP | DSpark | DFlash 2 |
| :--- | ---: | ---: | ---: | ---: |
| GSM8K | 1,329.8 | 1,381.1 (1.04×) | 1,506.5 (1.13×) | **1,922.5 (1.45×)** |
| MATH-500 | 1,505.8 | 1,415.6 (0.94×) | 1,429.0 (0.95×) | **1,951.8 (1.30×)** |
| HumanEval | 1,546.5 | 1,296.8 (0.84×) | 1,330.1 (0.86×) | **1,799.0 (1.16×)** |
| MBPP | 1,507.7 | 1,314.9 (0.87×) | 1,361.3 (0.90×) | **1,886.8 (1.25×)** |
| MT-Bench | 1,507.4 | 1,159.7 (0.77×) | 1,115.5 (0.74×) | **1,525.3 (1.01×)** |

## Citation

If you find DFlash 2 useful, please cite:

```bibtex
@misc{inco2026dflash2,
  title  = {{DFlash 2: Keep Drafting Parallel}},
  author = {{Inco AI}},
  year   = {2026},
  month  = {August},
  url    = {https://inco.ai/blog/dflash2/}
}
```

Please also cite the original DFlash paper:

```bibtex
@inproceedings{chen2026dflash,
  title     = {{DFlash: Block Diffusion for Flash Speculative Decoding}},
  author    = {Chen, Jian and Liang, Yesheng and Liu, Zhijian},
  booktitle = {International Conference on Machine Learning (ICML)},
  year      = {2026}
}
```