BrainDelay commited on
Commit
13e81df
·
verified ·
1 Parent(s): fc87258

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -0
README.md CHANGED
@@ -139,6 +139,18 @@ By default, the baked-in system prompt includes a instruction to keep answers co
139
 
140
  ![Siren Standing Upright 1](https://cdn-uploads.huggingface.co/production/uploads/69afe899440366f1cb4c0293/boVeMw6nzmt8cKxfHNjrU.png)
141
 
 
 
 
 
 
 
 
 
 
 
 
 
142
 
143
  ![Siren Posing 1](https://cdn-uploads.huggingface.co/production/uploads/69afe899440366f1cb4c0293/6CuaBpbUwzk_yns0OUoLo.png)
144
 
 
139
 
140
  ![Siren Standing Upright 1](https://cdn-uploads.huggingface.co/production/uploads/69afe899440366f1cb4c0293/boVeMw6nzmt8cKxfHNjrU.png)
141
 
142
+ ## 📥 Available Quantizations
143
+
144
+ ⚠️ **Recommendation:** Please use at least the **Q4** quantization. Lower bits often struggle to maintain the specific *italics* and **BOLD** formatting styles that make this model unique.
145
+
146
+ | File Name | Size | Recommended VRAM | Quality Level |
147
+ | :--- | :--- | :--- | :--- |
148
+ | **Mistral-Siren_V1_Q2_K.gguf** | ~4.7 GB | 4 - 5 GB+ | **Low** (Lossy) |
149
+ | **Mistral-Siren_V1_Q3_K_M.gguf** | ~5.9 GB | 6 GB | **Good** (Mobile/Old PCs) |
150
+ | **Mistral-Siren_V1_Q4_K_M.gguf** | ~7.3 GB | 8 GB | **Standard** (Recommended) |
151
+ | **Mistral-Siren_V1_Q5_K_M.gguf** | ~8.5 GB | 10 GB | **Very Good** |
152
+ | **Mistral-Siren_V1_Q6_K.gguf** | ~9.8 GB | 12 GB | **Excellent** (Sweet Spot) |
153
+ | **Mistral-Siren_V1_Q8_0.gguf** | ~12.7 GB | 16 GB | **Maximum** (No loss) |
154
 
155
  ![Siren Posing 1](https://cdn-uploads.huggingface.co/production/uploads/69afe899440366f1cb4c0293/6CuaBpbUwzk_yns0OUoLo.png)
156