KI

Rose Kimiko 20B

KimiGGUF by TheBlokeUpdated 2023-12-14
Download on Hugging Face
Parameters20B
Context4K
Layers62
Architecturellama
Smallest quant8.3 GB
Quick answer

Rose Kimiko 20B needs about 22.9 GB of memory at Q4_K_M (a 12.0 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 12 quants, from 8.3 GB (Q2_K) to 21.2 GB (Q8_0).

Rose Kimiko 20B – fine‑tuned 20B LLaMA model for spelling correction

Rose Kimiko 20B is a 20B‑parameter LLaMA model fine‑tuned on the Kimiko dataset. It uses ShareGPT/Vicuna prompt formats and supports a 4096‑token context window.

Good at

  • Spelling correction
  • ShareGPT/Vicuna prompt handling
  • Fine‑tuned on Kimiko dataset

Limits

  • No vision support
  • Requires 8k context for memory estimates
  • License: llama2

Best for

  • Local spelling correction
  • Chat applications needing accurate spelling
  • Fine‑tuned LLaMA inference on GPU

Who it’s for. Developers building local LLM chatbots that need improved spelling, or researchers testing fine‑tuned LLaMA models.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K8.3 GB 14.0 GB
Too big Get file →
Q3_K_S8.7 GB 14.4 GB
Too big Get file →
Q3_K_M9.7 GB 15.4 GB
Too big Get file →
Q3_K_L10.6 GB 16.3 GB
Too big Get file →
Q4_011.3 GB 17.0 GB
Too big Get file →
Q4_K_S11.3 GB 17.0 GB
Too big Get file →
Q4_K_M12.0 GB 17.7 GB
Too big Get file →
Q5_013.8 GB 19.5 GB
Too big Get file →
Q5_K_S13.8 GB 19.5 GB
Too big Get file →
Q5_K_M14.2 GB 19.9 GB
Too big Get file →
Q6_K16.4 GB 22.1 GB
Too big Get file →
Q8_021.2 GB 26.9 GB
Too big Get file →

For 8GB VRAM the smallest quant (Q2_K) still needs about 19.2GB memory at 8k context, so it is not feasible. With 16GB VRAM, Q4_0 (11.29GB file, 22.2GB memory) is the lowest memory option. For 32GB VRAM, Q8_0 (21.25GB file, 32.1GB memory) provides the highest quality.

What GPU do you need for Rose Kimiko 20B?

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: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: the best quant that fits is Q4_K_M (12.0 GB file, about 22.9 GB in use).
  • 32 GB VRAM: the best quant that fits is Q6_K (16.4 GB file, about 27.3 GB in use).
  • 48 GB VRAM: the best quant that fits is Q8_0 (21.2 GB file, about 32.1 GB in use).

Which Rose Kimiko 20B quant should you choose?

Q4_K_M is the usual starting point for Rose Kimiko 20B: it balances file size and quality well. Here it is a 12.0 GB file.

The Q8_0 file is 21.2 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 Q2_K at 8.3 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters20B
Layers62
Trained context4,096 tokens
KV cache (fp16)1240 KB per token

Source

Quantized byTheBloke
FamilyKimi
Licensellama2
Downloads (HF)1,256
Last updated2023-12-14

Frequently asked questions about Rose Kimiko 20B

How much VRAM does Rose Kimiko 20B need?

About 22.9 GB at Q4_K_M with an 8K context window. That is the 12.0 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 Rose Kimiko 20B on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 19.2 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Rose Kimiko 20B GGUF?

It depends on the quant: Q2_K is 8.3 GB, Q3_K_S is 8.7 GB, Q3_K_M is 9.7 GB, Q3_K_L is 10.6 GB, Q4_0 is 11.3 GB, Q4_K_S is 11.3 GB and 6 more.

What context length does Rose Kimiko 20B support?

The GGUF header lists a trained context length of 4,096 tokens (4K). The KV cache grows with the context you use: about 10.4 GB at 8K and 5.2 GB at 4K.

Where can I download Rose Kimiko 20B GGUF?

The files are on Hugging Face in the TheBloke/Rose-Kimiko-20B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Rose Kimiko 20B use?

Hugging Face lists the license as llama2. 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 Rose Kimiko 20B on 8GB VRAM?

No, the smallest quant requires about 19.2GB memory at 8k context, exceeding 8GB VRAM.

What context length does the model support?

The model supports a 4096‑token context length.

Is the model licensed for commercial use?

The model is licensed under llama2; check the license terms for commercial use.

Similar models

Browse all →

More from TheBloke

All by TheBloke →

More Kimi models

All Kimi →

Other medium models

All medium models →