NI

Ninja V3

NinjaGGUF by TuitouUpdated 2024-08-13
Download on Hugging Face
Parameters12.9B
Context32K
Layers32
Architecturellama
Smallest quant6.9 GB
Quick answer

Ninja V3 needs about 9.4 GB of memory at Q4_K_M (a 7.8 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 6 quants, from 6.9 GB (IQ4_XS) to 25.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

Ninja V3 is a 12.9B-parameter Ninja model in GGUF format, quantized by Tuitou. Its trained context length is 32K tokens. The repository has 6 quant files, from 6.9 GB (IQ4_XS) to 25.8 GB (BF16).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ4_XS6.9 GB 8.5 GB
Too big Get file →
Q4_K_M7.8 GB 9.4 GB
Too big Get file →
Q5_K9.1 GB 10.7 GB
Too big Get file →
Q6_K10.6 GB 12.1 GB
Too big Get file →
Q8_013.7 GB 15.3 GB
Too big Get file →
BF1625.8 GB 27.3 GB
Too big Get file →

What GPU do you need for Ninja V3?

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: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: the best quant that fits is Q5_K (9.1 GB file, about 10.7 GB in use).
  • 16 GB VRAM: the best quant that fits is Q8_0 (13.7 GB file, about 15.3 GB in use).
  • 24 GB VRAM: the best quant that fits is Q8_0 (13.7 GB file, about 15.3 GB in use).
  • 32 GB VRAM: the best quant that fits is BF16 (25.8 GB file, about 27.3 GB in use).

Which Ninja V3 quant should you choose?

Q4_K_M is the usual starting point for Ninja V3: it balances file size and quality well. Here it is a 7.8 GB file.

The Q8_0 file is 13.7 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 IQ4_XS at 6.9 GB, with the quality trade-offs that come with fewer bits. The usual sweet spot between file size and quality, and the quant most people start with.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters12.9B (approximate, mixture of experts)
Layers32
Trained context32,768 tokens
KV cache (fp16)128 KB per token

Source

Quantized byTuitou
FamilyNinja
Licenseapache-2.0
Downloads (HF)22
Last updated2024-08-13

Frequently asked questions about Ninja V3

How much VRAM does Ninja V3 need?

About 9.4 GB at Q4_K_M with an 8K context window. That is the 7.8 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 Ninja V3 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 8.5 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Ninja V3 GGUF?

It depends on the quant: IQ4_XS is 6.9 GB, Q4_K_M is 7.8 GB, Q5_K is 9.1 GB, Q6_K is 10.6 GB, Q8_0 is 13.7 GB, BF16 is 25.8 GB.

What context length does Ninja V3 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 Ninja V3 GGUF?

The files are on Hugging Face in the Tuitou/Ninja-v3-Vecteus-v1-MoE-2x7B repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Ninja V3 use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Similar models

Browse all →

More from Tuitou

All by Tuitou →

More Ninja models

All Ninja →

Other medium models

All medium models →