Model Intelligence Sheet
gghfez/DeepSeek-V3-0324-Opus3-Styletune-GGUF overview
gghfez/DeepSeek V3 0324 Opus3 Styletune GGUF Trained the lm head output tensor on 991 sampls from the retired Opus 3 model. Based on: Gryphe/Gemma 4 31B StyleT…
Runs locally from ~8.32 GB disk (12 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
1 GGUF files detected
Direct downloads for local inference
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| DeepSeek-V3-0324-IQ3_KS.gguf-00007-of-00007.gguf | GGUF | IQ3_KS | 8.32 GB | Download |
Model Details
Model README
---
license: cc-by-nc-4.0
base_model:
- deepseek-ai/DeepSeek-V3-0324
datasets:
- NoSlop4U/opus-3-1000x
tags:
- styletune
- ik_llama.cpp
---
gghfez/DeepSeek-V3-0324-Opus3-Styletune-GGUF
Trained the lm_head output tensor on 991 sampls from the retired Opus-3 model.
Based on: Gryphe/Gemma-4-31B-StyleTune, see their model card for a more detailed explanation.
How to use:
- Download this repo
- Download gghfez/DeepSeek-V3-0324-IQ3_KS
- Symlink shards 00001-of-00007 - 00001-of-00006 from gghfez/DeepSeek-V3-0324-IQ3_KS but not
ln -s $(realpath DeepSeek-V3-0324-IQ3_KS/DeepSeek-V3-0324-IQ3_KS.gguf-00001-of-00007.gguf) $(realpath DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00001-of-00007.gguf)
ln -s $(realpath DeepSeek-V3-0324-IQ3_KS/DeepSeek-V3-0324-IQ3_KS.gguf-00002-of-00007.gguf) $(realpath DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00002-of-00007.gguf)
ln -s $(realpath DeepSeek-V3-0324-IQ3_KS/DeepSeek-V3-0324-IQ3_KS.gguf-00003-of-00007.gguf) $(realpath DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00003-of-00007.gguf)
ln -s $(realpath DeepSeek-V3-0324-IQ3_KS/DeepSeek-V3-0324-IQ3_KS.gguf-00004-of-00007.gguf) $(realpath DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00004-of-00007.gguf)
ln -s $(realpath DeepSeek-V3-0324-IQ3_KS/DeepSeek-V3-0324-IQ3_KS.gguf-00005-of-00007.gguf) $(realpath DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00005-of-00007.gguf)
ln -s $(realpath DeepSeek-V3-0324-IQ3_KS/DeepSeek-V3-0324-IQ3_KS.gguf-00006-of-00007.gguf) $(realpath DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00006-of-00007.gguf) llama-server -m DeepSeek-V3-0324-Opus3-Styletune-GGUF/DeepSeek-V3-0324-IQ3_KS.gguf-00001-of-00007.gguf ...rest of your cli flags
Run gghfez/DeepSeek-V3-0324-Opus3-Styletune-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models