Transformers
GGUF
Mixture of Experts
mixtral
sharegpt
axolotl
Inference Endpoints
imatrix
conversational
mradermacher commited on
Commit
cfd2e92
1 Parent(s): eed9be9

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +14 -0
README.md CHANGED
@@ -44,9 +44,23 @@ more details, including on how to concatenate multi-part files.
44
 
45
  | Link | Type | Size/GB | Notes |
46
  |:-----|:-----|--------:|:------|
 
 
47
  | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q2_K.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q2_K.gguf.part2of2) | i1-Q2_K | 52.2 | IQ3_XXS probably better |
48
  | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_XXS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_XXS.gguf.part2of2) | i1-IQ3_XXS | 55.0 | lower quality |
 
 
 
 
 
 
 
 
49
  | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_K_S.gguf.part2of2) | i1-Q4_K_S | 80.6 | optimal size/speed/quality |
 
 
 
 
50
 
51
  Here is a handy graph by ikawrakow comparing some lower-quality quant
52
  types (lower is better):
 
44
 
45
  | Link | Type | Size/GB | Notes |
46
  |:-----|:-----|--------:|:------|
47
+ | [GGUF](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 38.0 | |
48
+ | [GGUF](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ2_M.gguf) | i1-IQ2_M | 46.8 | |
49
  | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q2_K.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q2_K.gguf.part2of2) | i1-Q2_K | 52.2 | IQ3_XXS probably better |
50
  | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_XXS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_XXS.gguf.part2of2) | i1-IQ3_XXS | 55.0 | lower quality |
51
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_XS.gguf.part2of2) | i1-IQ3_XS | 58.3 | |
52
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_S.gguf.part2of2) | i1-IQ3_S | 61.6 | beats Q3_K* |
53
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q3_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q3_K_S.gguf.part2of2) | i1-Q3_K_S | 61.6 | IQ3_XS probably better |
54
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ3_M.gguf.part2of2) | i1-IQ3_M | 64.6 | |
55
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q3_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q3_K_M.gguf.part2of2) | i1-Q3_K_M | 67.9 | IQ3_S probably better |
56
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q3_K_L.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q3_K_L.gguf.part2of2) | i1-Q3_K_L | 72.7 | IQ3_M probably better |
57
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ4_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-IQ4_XS.gguf.part2of2) | i1-IQ4_XS | 75.6 | |
58
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_0.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_0.gguf.part2of2) | i1-Q4_0 | 80.0 | fast, low quality |
59
  | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_K_S.gguf.part2of2) | i1-Q4_K_S | 80.6 | optimal size/speed/quality |
60
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q4_K_M.gguf.part2of2) | i1-Q4_K_M | 85.7 | fast, recommended |
61
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q5_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q5_K_S.gguf.part2of2) | i1-Q5_K_S | 97.1 | |
62
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q5_K_M.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q5_K_M.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q5_K_M.gguf.part3of3) | i1-Q5_K_M | 100.1 | |
63
+ | [PART 1](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q6_K.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q6_K.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/Goku-8x22B-v0.2-i1-GGUF/resolve/main/Goku-8x22B-v0.2.i1-Q6_K.gguf.part3of3) | i1-Q6_K | 115.6 | practically like static Q6_K |
64
 
65
  Here is a handy graph by ikawrakow comparing some lower-quality quant
66
  types (lower is better):