mradermacher commited on
Commit
c05e2b3
1 Parent(s): 648a0c1

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +19 -0
README.md CHANGED
@@ -37,8 +37,27 @@ more details, including on how to concatenate multi-part files.
37
 
38
  | Link | Type | Size/GB | Notes |
39
  |:-----|:-----|--------:|:------|
 
 
 
 
 
 
40
  | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q2_K.gguf) | i1-Q2_K | 17.4 | IQ3_XXS probably better |
 
 
 
 
41
  | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ3_M.gguf) | i1-IQ3_M | 21.5 | |
 
 
 
 
 
 
 
 
 
42
 
43
  Here is a handy graph by ikawrakow comparing some lower-quality quant
44
  types (lower is better):
 
37
 
38
  | Link | Type | Size/GB | Notes |
39
  |:-----|:-----|--------:|:------|
40
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ1_S.gguf) | i1-IQ1_S | 9.9 | for the desperate |
41
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ1_M.gguf) | i1-IQ1_M | 10.9 | mostly desperate |
42
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 12.7 | |
43
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ2_XS.gguf) | i1-IQ2_XS | 14.0 | |
44
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ2_S.gguf) | i1-IQ2_S | 14.2 | |
45
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ2_M.gguf) | i1-IQ2_M | 15.6 | |
46
  | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q2_K.gguf) | i1-Q2_K | 17.4 | IQ3_XXS probably better |
47
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 18.3 | lower quality |
48
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ3_XS.gguf) | i1-IQ3_XS | 19.5 | |
49
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ3_S.gguf) | i1-IQ3_S | 20.5 | beats Q3_K* |
50
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q3_K_S.gguf) | i1-Q3_K_S | 20.5 | IQ3_XS probably better |
51
  | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ3_M.gguf) | i1-IQ3_M | 21.5 | |
52
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q3_K_M.gguf) | i1-Q3_K_M | 22.6 | IQ3_S probably better |
53
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q3_K_L.gguf) | i1-Q3_K_L | 24.3 | IQ3_M probably better |
54
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-IQ4_XS.gguf) | i1-IQ4_XS | 25.2 | |
55
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q4_0.gguf) | i1-Q4_0 | 26.7 | fast, low quality |
56
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q4_K_S.gguf) | i1-Q4_K_S | 26.8 | optimal size/speed/quality |
57
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q4_K_M.gguf) | i1-Q4_K_M | 28.5 | fast, recommended |
58
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q5_K_S.gguf) | i1-Q5_K_S | 32.3 | |
59
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q5_K_M.gguf) | i1-Q5_K_M | 33.3 | |
60
+ | [GGUF](https://huggingface.co/mradermacher/dolphin-2.6-mixtral-8x7b-i1-GGUF/resolve/main/dolphin-2.6-mixtral-8x7b.i1-Q6_K.gguf) | i1-Q6_K | 38.5 | practically like static Q6_K |
61
 
62
  Here is a handy graph by ikawrakow comparing some lower-quality quant
63
  types (lower is better):