Hugging Face's logo

mradermacher
/
ChineseErrorCorrector2-7B-SFT-GRPO-GGUF

Transformers
GGUF
English
conversational
Model card Files Community
mradermacher commited on
Commit
35dfc24
·
verified ·
1 Parent(s): fe670b5

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -0
README.md CHANGED
@@ -32,10 +32,14 @@ more details, including on how to concatenate multi-part files.
32
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q3_K_S.gguf) | Q3_K_S | 3.6 | |
33
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q3_K_M.gguf) | Q3_K_M | 3.9 | lower quality |
34
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q3_K_L.gguf) | Q3_K_L | 4.2 | |
 
35
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q4_K_S.gguf) | Q4_K_S | 4.6 | fast, recommended |
36
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q4_K_M.gguf) | Q4_K_M | 4.8 | fast, recommended |
 
 
37
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q6_K.gguf) | Q6_K | 6.4 | very good quality |
38
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q8_0.gguf) | Q8_0 | 8.2 | fast, best quality |
 
39
 
40
  Here is a handy graph by ikawrakow comparing some lower-quality quant
41
  types (lower is better):
 
32
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q3_K_S.gguf) | Q3_K_S | 3.6 | |
33
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q3_K_M.gguf) | Q3_K_M | 3.9 | lower quality |
34
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q3_K_L.gguf) | Q3_K_L | 4.2 | |
35
+ | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.IQ4_XS.gguf) | IQ4_XS | 4.4 | |
36
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q4_K_S.gguf) | Q4_K_S | 4.6 | fast, recommended |
37
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q4_K_M.gguf) | Q4_K_M | 4.8 | fast, recommended |
38
+ | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q5_K_S.gguf) | Q5_K_S | 5.4 | |
39
+ | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q5_K_M.gguf) | Q5_K_M | 5.5 | |
40
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q6_K.gguf) | Q6_K | 6.4 | very good quality |
41
  | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.Q8_0.gguf) | Q8_0 | 8.2 | fast, best quality |
42
+ | [GGUF](https://huggingface.co/mradermacher/ChineseErrorCorrector2-7B-SFT-GRPO-GGUF/resolve/main/ChineseErrorCorrector2-7B-SFT-GRPO.f16.gguf) | f16 | 15.3 | 16 bpw, overkill |
43
 
44
  Here is a handy graph by ikawrakow comparing some lower-quality quant
45
  types (lower is better):