README.md

---
license: mit
pipeline_tag: text-generation
language:
  - ja
tags:
  - japanese
  - qwen2
  - gguf
base_model:
  - cyberagent/DeepSeek-R1-Distill-Qwen-14B-Japanese
datasets:
  - TFMC/imatrix-dataset-for-japanese-llm 
---

# DeepSeek-R1-Distill-Qwen-14B-Japanese GGUF

## Model Description

This repository provides a GGUF quantized version of the [cyberagent/DeepSeek-R1-Distill-Qwen-14B-Japanese](https://huggingface.co/cyberagent/DeepSeek-R1-Distill-Qwen-14B-Japanese) model from [CyberAgent](https://cyberagent.ai).

## Choosing the Right Quantized Version

The choice of which quantized version to use depends your avilable VRAM:


  - >= 8GB VRAM: `DeepSeek-R1-Distill-Qwen-14B-Japanese-IQ3_XS.gguf`
  - >= 12GB VRAM: `DeepSeek-R1-Distill-Qwen-14B-Japanese-IQ4_XS.gguf`
  - >= 12GB VRAM: `DeepSeek-R1-Distill-Qwen-14B-Japanese-Q4_K_M.gguf`
  - >= 16GB VRAM: `DeepSeek-R1-Distill-Qwen-14B-Japanese-Q6_K.gguf`

## License
[MIT License](https://github.com/deepseek-ai/DeepSeek-R1/blob/main/LICENSE)