Hugging Face's logo

mradermacher
/
Hermes-2-Pro-Mixtral-4x7B-i1-GGUF

Transformers
GGUF
English
Mixture of Experts
Merge
mergekit
lazymergekit
NousResearch/Hermes-2-Pro-Mistral-7B
Mixtral
instruct
finetune
chatml
DPO
RLHF
gpt4
synthetic data
distillation
function calling
json mode
imatrix
conversational
Model card Files Community
mradermacher commited on
Commit
2a0509f
·
verified ·
1 Parent(s): e6911cb

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +18 -0
README.md CHANGED
@@ -46,9 +46,27 @@ more details, including on how to concatenate multi-part files.
46
 
47
  | Link | Type | Size/GB | Notes |
48
  |:-----|:-----|--------:|:------|
 
 
 
 
 
 
49
  | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q2_K.gguf) | i1-Q2_K | 8.9 | IQ3_XXS probably better |
 
 
 
 
50
  | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ3_M.gguf) | i1-IQ3_M | 10.7 | |
 
 
 
 
51
  | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 13.8 | optimal size/speed/quality |
 
 
 
 
52
 
53
  Here is a handy graph by ikawrakow comparing some lower-quality quant
54
  types (lower is better):
 
46
 
47
  | Link | Type | Size/GB | Notes |
48
  |:-----|:-----|--------:|:------|
49
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ1_S.gguf) | i1-IQ1_S | 5.1 | for the desperate |
50
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ1_M.gguf) | i1-IQ1_M | 5.6 | mostly desperate |
51
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 6.5 | |
52
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ2_XS.gguf) | i1-IQ2_XS | 7.2 | |
53
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ2_S.gguf) | i1-IQ2_S | 7.4 | |
54
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ2_M.gguf) | i1-IQ2_M | 8.1 | |
55
  | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q2_K.gguf) | i1-Q2_K | 8.9 | IQ3_XXS probably better |
56
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 9.4 | lower quality |
57
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ3_XS.gguf) | i1-IQ3_XS | 10.0 | |
58
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q3_K_S.gguf) | i1-Q3_K_S | 10.5 | IQ3_XS probably better |
59
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ3_S.gguf) | i1-IQ3_S | 10.6 | beats Q3_K* |
60
  | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ3_M.gguf) | i1-IQ3_M | 10.7 | |
61
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 11.7 | IQ3_S probably better |
62
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q3_K_L.gguf) | i1-Q3_K_L | 12.6 | IQ3_M probably better |
63
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-IQ4_XS.gguf) | i1-IQ4_XS | 13.0 | |
64
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q4_0.gguf) | i1-Q4_0 | 13.8 | fast, low quality |
65
  | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 13.8 | optimal size/speed/quality |
66
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 14.7 | fast, recommended |
67
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q5_K_S.gguf) | i1-Q5_K_S | 16.7 | |
68
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q5_K_M.gguf) | i1-Q5_K_M | 17.2 | |
69
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-2-Pro-Mixtral-4x7B-i1-GGUF/resolve/main/Hermes-2-Pro-Mixtral-4x7B.i1-Q6_K.gguf) | i1-Q6_K | 19.9 | practically like static Q6_K |
70
 
71
  Here is a handy graph by ikawrakow comparing some lower-quality quant
72
  types (lower is better):