xl_regrind_v1 needs about 4.6 GB of memory at Q4_K_M (a 3.0 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 7 quants, from 0.3 GB (BF16) to 5.3 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.
xl_regrind_v1 – 4.99B parameter DiT model with regrind VAE for artifact suppression
xl_regrind_v1 is a 4.99B parameter DiT model based on ACE-Step acestep-v15-xl-turbo. It includes a fine‑tuned VAE decoder and DiT LoRA to reduce the Oobleck hum artifact. The model supports a 32,768 token context and is licensed under cc‑by‑nc‑sa‑4.0.
Good at
- High‑quality audio generation in 8 steps
- Artifact suppression via regrind VAE
- Large context handling
Limits
- No vision support
- License restricts commercial use
- Requires 8k context for memory estimates
Best for
- Audio generation with low artifact
- Fine‑tuning LoRA on DiT models
- Research on DiT architecture
Who it’s for. Developers and researchers building audio generation pipelines who need a 4B DiT model with artifact suppression. The model is suitable for non‑commercial projects due to its license.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| BF16 | 0.3 GB | 1.9 GB | Fits | Get file → | |
| BF16 | 0.3 GB | 1.9 GB | Fits | Get file → | |
| BF16 | 0.3 GB | 1.9 GB | Fits | Get file → | |
| BF16 | 0.3 GB | 1.9 GB | Fits | Get file → | |
| Q4_K_M | 3.0 GB | 4.6 GB | Fits | Get file → | |
| Q6_K | 4.1 GB | 5.7 GB | Fits | Get file → | |
| Q8_0 | 5.3 GB | 6.9 GB | Fits | Get file → |
For 8 GB VRAM, use Q4_K_M (file size 2.99 GB, estimated memory 4.6 GB). For 12 GB VRAM, Q6_K (4.1 GB, 5.7 GB). For 16 GB VRAM, Q8_0 (5.31 GB, 6.9 GB). BF16 variants are 0.34 GB but require 1.9 GB memory.
What GPU do you need for xl_regrind_v1?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) 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 1.9 GB in use).
- 6 GB VRAM: the best quant that fits is Q6_K (4.1 GB file, about 5.7 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (5.3 GB file, about 6.9 GB in use).
Which xl_regrind_v1 quant should you choose?
Q4_K_M is the usual starting point for xl_regrind_v1: it balances file size and quality well. Here it is a 3.0 GB file.
The Q8_0 file is 5.3 GB, about 1.8 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is BF16 at 0.3 GB, with the quality trade-offs that come with fewer bits. 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
Source
Frequently asked questions about xl_regrind_v1
How much VRAM does xl_regrind_v1 need?
About 4.6 GB at Q4_K_M with an 8K context window. That is the 3.0 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 xl_regrind_v1 on an 8 GB GPU?
Yes, with Q8_0: the 5.3 GB file needs about 6.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of xl_regrind_v1 GGUF?
It depends on the quant: BF16 is 0.3 GB, BF16 is 0.3 GB, BF16 is 0.3 GB, BF16 is 0.3 GB, Q4_K_M is 3.0 GB, Q6_K is 4.1 GB and 1 more.
What context length does xl_regrind_v1 support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.
Where can I download xl_regrind_v1 GGUF?
The files are on Hugging Face in the mdmachine/ACEStep-XL-Regrind-V1 repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does xl_regrind_v1 use?
Hugging Face lists the license as cc-by-nc-sa-4.0. 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 xl_regrind_v1 on 8 GB VRAM?
Yes, using the Q4_K_M quantized file which estimates 4.6 GB memory at 8k context.
Does the model support vision inputs?
No, the model has no vision capabilities.
Is the model suitable for commercial use?
No, the cc‑by‑nc‑sa‑4.0 license prohibits commercial use.