Commit
•
757d4d8
1
Parent(s):
0b5891c
auto-patch README.md
Browse files
README.md
CHANGED
|
@@ -29,12 +29,30 @@ more details, including on how to concatenate multi-part files.
|
|
| 29 |
|
| 30 |
| Link | Type | Size/GB | Notes |
|
| 31 |
|:-----|:-----|--------:|:------|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q2_K.gguf) | i1-Q2_K | 4.1 | IQ3_XXS probably better |
|
| 33 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 4.3 | lower quality |
|
|
|
|
|
|
|
|
|
|
| 34 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ3_M.gguf) | i1-IQ3_M | 4.9 | |
|
| 35 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q3_K_M.gguf) | i1-Q3_K_M | 5.3 | IQ3_S probably better |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_K_S.gguf) | i1-Q4_K_S | 6.2 | optimal size/speed/quality |
|
| 37 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_K_M.gguf) | i1-Q4_K_M | 6.6 | fast, recommended |
|
|
|
|
|
|
|
|
|
|
| 38 |
|
| 39 |
Here is a handy graph by ikawrakow comparing some lower-quality quant
|
| 40 |
types (lower is better):
|
|
|
|
| 29 |
|
| 30 |
| Link | Type | Size/GB | Notes |
|
| 31 |
|:-----|:-----|--------:|:------|
|
| 32 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ1_S.gguf) | i1-IQ1_S | 2.5 | for the desperate |
|
| 33 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ1_M.gguf) | i1-IQ1_M | 2.7 | mostly desperate |
|
| 34 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 3.0 | |
|
| 35 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ2_XS.gguf) | i1-IQ2_XS | 3.3 | |
|
| 36 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ2_S.gguf) | i1-IQ2_S | 3.5 | |
|
| 37 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ2_M.gguf) | i1-IQ2_M | 3.8 | |
|
| 38 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q2_K.gguf) | i1-Q2_K | 4.1 | IQ3_XXS probably better |
|
| 39 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 4.3 | lower quality |
|
| 40 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ3_XS.gguf) | i1-IQ3_XS | 4.5 | |
|
| 41 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q3_K_S.gguf) | i1-Q3_K_S | 4.8 | IQ3_XS probably better |
|
| 42 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ3_S.gguf) | i1-IQ3_S | 4.8 | beats Q3_K* |
|
| 43 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ3_M.gguf) | i1-IQ3_M | 4.9 | |
|
| 44 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q3_K_M.gguf) | i1-Q3_K_M | 5.3 | IQ3_S probably better |
|
| 45 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q3_K_L.gguf) | i1-Q3_K_L | 5.8 | IQ3_M probably better |
|
| 46 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-IQ4_XS.gguf) | i1-IQ4_XS | 5.9 | |
|
| 47 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_0_4_4.gguf) | i1-Q4_0_4_4 | 6.2 | fast on arm, low quality |
|
| 48 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_0_4_8.gguf) | i1-Q4_0_4_8 | 6.2 | fast on arm+i8mm, low quality |
|
| 49 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_0_8_8.gguf) | i1-Q4_0_8_8 | 6.2 | fast on arm+sve, low quality |
|
| 50 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_0.gguf) | i1-Q4_0 | 6.2 | fast, low quality |
|
| 51 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_K_S.gguf) | i1-Q4_K_S | 6.2 | optimal size/speed/quality |
|
| 52 |
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q4_K_M.gguf) | i1-Q4_K_M | 6.6 | fast, recommended |
|
| 53 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q5_K_S.gguf) | i1-Q5_K_S | 7.5 | |
|
| 54 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q5_K_M.gguf) | i1-Q5_K_M | 7.7 | |
|
| 55 |
+
| [GGUF](https://huggingface.co/mradermacher/SOLAR-10.7b-Instruct-dpo-i1-GGUF/resolve/main/SOLAR-10.7b-Instruct-dpo.i1-Q6_K.gguf) | i1-Q6_K | 8.9 | practically like static Q6_K |
|
| 56 |
|
| 57 |
Here is a handy graph by ikawrakow comparing some lower-quality quant
|
| 58 |
types (lower is better):
|