mradermacher commited on
Commit
47830fb
·
verified ·
1 Parent(s): 7a71ae4

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -0
README.md CHANGED
@@ -63,12 +63,22 @@ more details, including on how to concatenate multi-part files.
63
  | Link | Type | Size/GB | Notes |
64
  |:-----|:-----|--------:|:------|
65
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own qwuants) |
 
 
 
 
 
66
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q2_K.gguf) | i1-Q2_K | 7.1 | IQ3_XXS probably better |
67
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 7.4 | lower quality |
 
68
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ3_M.gguf) | i1-IQ3_M | 8.5 | |
69
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 9.1 | IQ3_S probably better |
 
 
70
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ4_NL.gguf) | i1-IQ4_NL | 10.7 | prefer IQ4_XS |
71
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 10.8 | optimal size/speed/quality |
 
 
72
 
73
  Here is a handy graph by ikawrakow comparing some lower-quality quant
74
  types (lower is better):
 
63
  | Link | Type | Size/GB | Notes |
64
  |:-----|:-----|--------:|:------|
65
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own qwuants) |
66
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ1_M.gguf) | i1-IQ1_M | 4.6 | mostly desperate |
67
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 5.2 | |
68
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ2_XS.gguf) | i1-IQ2_XS | 5.7 | |
69
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ2_M.gguf) | i1-IQ2_M | 6.4 | |
70
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q2_K_S.gguf) | i1-Q2_K_S | 6.6 | very low quality |
71
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q2_K.gguf) | i1-Q2_K | 7.1 | IQ3_XXS probably better |
72
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 7.4 | lower quality |
73
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q3_K_S.gguf) | i1-Q3_K_S | 8.3 | IQ3_XS probably better |
74
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ3_M.gguf) | i1-IQ3_M | 8.5 | |
75
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 9.1 | IQ3_S probably better |
76
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q3_K_L.gguf) | i1-Q3_K_L | 9.9 | IQ3_M probably better |
77
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ4_XS.gguf) | i1-IQ4_XS | 10.2 | |
78
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-IQ4_NL.gguf) | i1-IQ4_NL | 10.7 | prefer IQ4_XS |
79
  | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 10.8 | optimal size/speed/quality |
80
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 11.4 | fast, recommended |
81
+ | [GGUF](https://huggingface.co/mradermacher/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B-i1-GGUF/resolve/main/Qwen3-MOE-6Bx4-Almost-Human-XMEN-X3-X4-X2-X1-24B.i1-Q6_K.gguf) | i1-Q6_K | 15.4 | practically like static Q6_K |
82
 
83
  Here is a handy graph by ikawrakow comparing some lower-quality quant
84
  types (lower is better):