AC

scragvae

AceGGUF by scragnogUpdated 2026-04-19
Download on Hugging Face
Parameters169M
Context—
Layers—
Architectureacestep-vae
Smallest quant0.3 GB
Quick answer

scragvae needs about 0.9 GB of memory at BF16 (a 0.3 GB file) with an 8K context window, so it fits on a 4 GB GPU.

ScragVAE – improved audio VAE for ACE‑Step 1.5 music generation

ScragVAE is a fine‑tuned AutoencoderOobleck decoder that enhances high‑frequency reconstruction for ACE‑Step 1.5. It is drop‑in compatible with existing DiT checkpoints and preserves more detail above 6kHz. The model is released under the MIT license.

Good at

  • High‑frequency audio fidelity
  • Clearer vocals and transients
  • Compatible with ACE‑Step pipelines

Limits

  • No vision support
  • Requires 1920x downsampling ratio
  • No language support

Best for

  • Music generation with ACE‑Step
  • Audio post‑processing for higher clarity
  • Fine‑tuning VAE decoders

Who it’s for. Developers and researchers working on music generation pipelines that use ACE‑Step 1.5. The model is ideal for those who need improved audio fidelity without changing the encoder or DiT components.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
BF160.3 GB 0.9 GB
Fits Get file →

The only available quant is BF16. The file size is 0.34 GB and requires approximately 0.9 GB of VRAM at 8 k context. For 8 GB VRAM, BF16 is suitable. If more memory is available, no other quant is provided.

What GPU do you need for scragvae?

These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is BF16 (0.3 GB file, about 0.9 GB in use).

Which scragvae quant should you choose?

BF16 (0.3 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureacestep-vae
Parameters169M

Source

Quantized byscragnog
FamilyAce
Licensemit
Downloads (HF)5,260
Last updated2026-04-19

Frequently asked questions about scragvae

How much VRAM does scragvae need?

About 0.9 GB at BF16 with an 8K context window. That is the 0.3 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run scragvae on an 8 GB GPU?

Yes, with BF16: the 0.3 GB file needs about 0.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of scragvae GGUF?

It depends on the quant: BF16 is 0.3 GB.

Where can I download scragvae GGUF?

The files are on Hugging Face in the scragnog/Ace-Step-1.5-ScragVAE repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does scragvae use?

Hugging Face lists the license as mit. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run ScragVAE on 8 GB VRAM?

Yes, BF16 requires about 0.9 GB of VRAM at 8 k context, well within an 8 GB budget.

Does ScragVAE support vision tasks?

No, ScragVAE is an audio VAE and has no vision capabilities.

Is ScragVAE compatible with ACE‑Step DiT models?

Yes, it is drop‑in compatible with all existing ACE‑Step DiT checkpoints.

Similar models

Browse all →

More from scragnog

All by scragnog →

More Ace models

All Ace →

Other tiny models

All tiny models →