Hugging Face's logo

mradermacher
/
dagger-4B_SFT-GGUF

Transformers
GGUF
Bengali
English
math
reasoning
computational-graph
bangla
low-resource
distractor-aware
sft
small-model
conversational
Model card Files
xet
Community
mradermacher commited on
Commit
79322dc
·
verified ·
1 Parent(s): 3d48629

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -0
README.md CHANGED
@@ -53,6 +53,18 @@ more details, including on how to concatenate multi-part files.
53
  |:-----|:-----|--------:|:------|
54
  | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.mmproj-Q8_0.gguf) | mmproj-Q8_0 | 0.7 | multi-modal supplement |
55
  | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.mmproj-f16.gguf) | mmproj-f16 | 1.0 | multi-modal supplement |
 
 
 
 
 
 
 
 
 
 
 
 
56
 
57
  Here is a handy graph by ikawrakow comparing some lower-quality quant
58
  types (lower is better):
 
53
  |:-----|:-----|--------:|:------|
54
  | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.mmproj-Q8_0.gguf) | mmproj-Q8_0 | 0.7 | multi-modal supplement |
55
  | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.mmproj-f16.gguf) | mmproj-f16 | 1.0 | multi-modal supplement |
56
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q2_K.gguf) | Q2_K | 1.8 | |
57
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q3_K_S.gguf) | Q3_K_S | 2.0 | |
58
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q3_K_M.gguf) | Q3_K_M | 2.2 | lower quality |
59
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q3_K_L.gguf) | Q3_K_L | 2.3 | |
60
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.IQ4_XS.gguf) | IQ4_XS | 2.4 | |
61
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q4_K_S.gguf) | Q4_K_S | 2.5 | fast, recommended |
62
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q4_K_M.gguf) | Q4_K_M | 2.6 | fast, recommended |
63
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q5_K_S.gguf) | Q5_K_S | 2.9 | |
64
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q5_K_M.gguf) | Q5_K_M | 2.9 | |
65
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q6_K.gguf) | Q6_K | 3.3 | very good quality |
66
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.Q8_0.gguf) | Q8_0 | 4.2 | fast, best quality |
67
+ | [GGUF](https://huggingface.co/mradermacher/dagger-4B_SFT-GGUF/resolve/main/dagger-4B_SFT.f16.gguf) | f16 | 7.9 | 16 bpw, overkill |
68
 
69
  Here is a handy graph by ikawrakow comparing some lower-quality quant
70
  types (lower is better):