Hugging Face's logo

mradermacher
/
Hermes-4.3-36B-i1-GGUF

Transformers
GGUF
English
Bytedance Seed
instruct
finetune
reasoning
hybrid-mode
chatml
function calling
tool use
json mode
structured outputs
atropos
dataforge
long context
roleplaying
chat
imatrix
conversational
Model card Files
xet
Community
mradermacher commited on
Commit
5609e7b
·
verified ·
1 Parent(s): 5dcb88a

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +16 -0
README.md CHANGED
@@ -54,13 +54,29 @@ more details, including on how to concatenate multi-part files.
54
  | Link | Type | Size/GB | Notes |
55
  |:-----|:-----|--------:|:------|
56
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own quants) |
 
 
 
 
 
57
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ2_M.gguf) | i1-IQ2_M | 12.6 | |
 
58
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q2_K.gguf) | i1-Q2_K | 13.7 | IQ3_XXS probably better |
59
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 14.2 | lower quality |
 
 
 
60
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ3_M.gguf) | i1-IQ3_M | 16.6 | |
61
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 17.7 | IQ3_S probably better |
 
 
 
62
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 20.8 | optimal size/speed/quality |
63
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 21.9 | fast, recommended |
 
 
 
 
64
 
65
  Here is a handy graph by ikawrakow comparing some lower-quality quant
66
  types (lower is better):
 
54
  | Link | Type | Size/GB | Notes |
55
  |:-----|:-----|--------:|:------|
56
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own quants) |
57
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ1_S.gguf) | i1-IQ1_S | 8.2 | for the desperate |
58
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ1_M.gguf) | i1-IQ1_M | 8.8 | mostly desperate |
59
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 10.0 | |
60
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ2_XS.gguf) | i1-IQ2_XS | 11.0 | |
61
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ2_S.gguf) | i1-IQ2_S | 11.7 | |
62
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ2_M.gguf) | i1-IQ2_M | 12.6 | |
63
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q2_K_S.gguf) | i1-Q2_K_S | 12.7 | very low quality |
64
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q2_K.gguf) | i1-Q2_K | 13.7 | IQ3_XXS probably better |
65
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 14.2 | lower quality |
66
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ3_XS.gguf) | i1-IQ3_XS | 15.2 | |
67
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q3_K_S.gguf) | i1-Q3_K_S | 16.0 | IQ3_XS probably better |
68
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ3_S.gguf) | i1-IQ3_S | 16.0 | beats Q3_K* |
69
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ3_M.gguf) | i1-IQ3_M | 16.6 | |
70
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 17.7 | IQ3_S probably better |
71
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q3_K_L.gguf) | i1-Q3_K_L | 19.2 | IQ3_M probably better |
72
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-IQ4_XS.gguf) | i1-IQ4_XS | 19.6 | |
73
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q4_0.gguf) | i1-Q4_0 | 20.7 | fast, low quality |
74
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 20.8 | optimal size/speed/quality |
75
  | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 21.9 | fast, recommended |
76
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q4_1.gguf) | i1-Q4_1 | 22.9 | |
77
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q5_K_S.gguf) | i1-Q5_K_S | 25.1 | |
78
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q5_K_M.gguf) | i1-Q5_K_M | 25.7 | |
79
+ | [GGUF](https://huggingface.co/mradermacher/Hermes-4.3-36B-i1-GGUF/resolve/main/Hermes-4.3-36B.i1-Q6_K.gguf) | i1-Q6_K | 29.8 | practically like static Q6_K |
80
 
81
  Here is a handy graph by ikawrakow comparing some lower-quality quant
82
  types (lower is better):