ZE

zenz v3.1 small mixed input

ZenzGGUF by motchii709Updated 2026-10-16
Download on Hugging Face
Parameters91M
Context1K
Layers12
Architecturegpt2
Smallest quant0.1 GB
Quick answer

zenz v3.1 small mixed input needs about 0.9 GB of memory at Q5_K_M (a 0.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 0.1 GB (Q5_K_M) to 0.2 GB (F16).

A 90.5M‑parameter GPT‑2 fine‑tuned for Japanese IME with ASCII handling

zenz v3.1 small mixed input is a 90.5M‑parameter GPT‑2 model fine‑tuned for Japanese kana‑kanji conversion. It outputs ASCII words in canonical Latin spelling and preserves katakana loanwords in context. The model is released under CC BY‑SA 4.0.

Good at

  • ASCII‑aware Japanese IME
  • Preserves katakana in context
  • Outputs Latin spelling for unseen loanwords

Limits

  • No vision support
  • Context length limited to 1024 tokens
  • License requires attribution

Best for

  • Japanese IME development
  • Text correction tools
  • Language learning apps

Who it’s for. Developers building Japanese input methods who need accurate handling of mixed ASCII and kana. Researchers testing fine‑tuned language models for mixed‑script scenarios.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q5_K_M0.1 GB 0.6 GB
Fits Get file →
F160.2 GB 0.7 GB
Fits Get file →

For systems with about 1 GB VRAM, use the Q5_K_M quantized file (0.07 GB, ~0.9 GB memory at 8 k context). If more memory is available, the F16 quantized file (0.18 GB, ~1 GB memory) offers higher precision.

What GPU do you need for zenz v3.1 small mixed input?

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 F16 (0.2 GB file, about 1.0 GB in use).

Which zenz v3.1 small mixed input quant should you choose?

Q5_K_M (0.1 GB) is the middle option in this repository. Closer to the original model than 4-bit, for a moderate increase in size. A good pick when you have memory to spare.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegpt2
Parameters91M
Layers12
Trained context1,024 tokens
KV cache (fp16)36 KB per token

Source

Quantized bymotchii709
FamilyZenz
Downloads (HF)880
Last updated2026-10-16

Frequently asked questions about zenz v3.1 small mixed input

How much VRAM does zenz v3.1 small mixed input need?

About 0.9 GB at Q5_K_M with an 8K context window. That is the 0.1 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 zenz v3.1 small mixed input on an 8 GB GPU?

Yes, with F16: the 0.2 GB file needs about 1.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of zenz v3.1 small mixed input GGUF?

It depends on the quant: Q5_K_M is 0.1 GB, F16 is 0.2 GB.

What context length does zenz v3.1 small mixed input support?

The GGUF header lists a trained context length of 1,024 tokens (1K). The KV cache grows with the context you use: about 0.3 GB at 8K and 0.0 GB at 1K.

Where can I download zenz v3.1 small mixed input GGUF?

The files are on Hugging Face in the motchii709/zenz-v3.1-small-mixed-input repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does zenz v3.1 small mixed input use?

Hugging Face lists the license as cc-by-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 this model on 8 GB VRAM?

Yes, both Q5_K_M and F16 quantizations fit comfortably within 8 GB VRAM.

Does this model work with llama.cpp?

Yes, the GGUF files are compatible with llama.cpp for inference.

Does it support vision or image input?

No, the model is purely text‑based and has no vision capabilities.

Similar models

Browse all →

More from motchii709

All by motchii709 →

More Zenz models

All Zenz →

Other tiny models

All tiny models →