Skip to content
Other

SDAR-TargetSFT-1.7B-b16-ep3

A drafter model for accelerating Qwen3-8B generation

1 min read

This model is built as a drafter (supporting model) for speculative decoding (a method for speeding up token generation in LLMs), partnering with the target model Qwen/Qwen3-8B. It is trained on top of JetLM/SDAR-1.7B-Chat-b16 using the TargetSFT method (block diffusion optimization of SDAR on responses generated by Qwen3-8B). It helps speed up larger language models during text generation. The model will be useful for developers and researchers working on optimizing inference speed (model runtime process) for large language models.

Key features

  • Designed for speculative decoding paired with Qwen3-8B
  • Based on the 1.7B model JetLM/SDAR-1.7B-Chat-b16
  • Trained on a dataset with response regeneration from Qwen3-8B
  • Uses the SDAR block diffusion architecture

What you can do

  • Apply the model in a speculative decoding pipeline to accelerate Qwen3-8B inference
  • Research multi-step denoising and language model acceleration methods

Example

The model is loaded similarly to the base SDAR model with the parameter trust_remote_code=True, and it shares the tokenizer with Qwen.

Read moreDetails
Base model: JetLM/SDAR-1.7B-Chat-b16
Parameters: 2.03 billion
Training: 5,931 steps on a single RTX PRO 6000 card
Data: 99,032 dialogues from mlabonne/open-perfectblend
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 3.8 GB · ready files4.7 GB
  • FP8 / INT8recommendedweights 1.9 GB2.6 GB
  • 4-bitweights 1.1 GB1.7 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Mye0n9/SDAR-TargetSFT-1.7B-b16-ep3
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Mye0n9/SDAR-TargetSFT-1.7B-b16-ep3 --local-dir ~/models/SDAR-TargetSFT-1.7B-b16-ep3
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of SDAR-1.7B-Chat-b16 (quants, finetunes, LoRAs) will show up under the bell.

Fine-tunes 1All on Hugging Face

About the model

Mye0n9/SDAR-TargetSFT-1.7B-b16-ep3
Author
Mye0n9
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments