mradermacher commited on
Commit
0fa24d8
·
verified ·
1 Parent(s): 0e9568a

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -0
README.md CHANGED
@@ -35,8 +35,14 @@ more details, including on how to concatenate multi-part files.
35
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q3_K_S.gguf) | Q3_K_S | 3.3 | |
36
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q3_K_M.gguf) | Q3_K_M | 3.6 | lower quality |
37
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q3_K_L.gguf) | Q3_K_L | 3.9 | |
 
38
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q4_K_S.gguf) | Q4_K_S | 4.2 | fast, recommended |
 
 
 
39
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q6_K.gguf) | Q6_K | 6.0 | very good quality |
 
 
40
 
41
  Here is a handy graph by ikawrakow comparing some lower-quality quant
42
  types (lower is better):
 
35
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q3_K_S.gguf) | Q3_K_S | 3.3 | |
36
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q3_K_M.gguf) | Q3_K_M | 3.6 | lower quality |
37
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q3_K_L.gguf) | Q3_K_L | 3.9 | |
38
+ | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.IQ4_XS.gguf) | IQ4_XS | 4.0 | |
39
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q4_K_S.gguf) | Q4_K_S | 4.2 | fast, recommended |
40
+ | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q4_K_M.gguf) | Q4_K_M | 4.5 | fast, recommended |
41
+ | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q5_K_S.gguf) | Q5_K_S | 5.1 | |
42
+ | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q5_K_M.gguf) | Q5_K_M | 5.2 | |
43
  | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q6_K.gguf) | Q6_K | 6.0 | very good quality |
44
+ | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.Q8_0.gguf) | Q8_0 | 7.8 | fast, best quality |
45
+ | [GGUF](https://huggingface.co/mradermacher/Ice0.67-25.01-RP-GGUF/resolve/main/Ice0.67-25.01-RP.f16.gguf) | f16 | 14.6 | 16 bpw, overkill |
46
 
47
  Here is a handy graph by ikawrakow comparing some lower-quality quant
48
  types (lower is better):