mradermacher commited on
Commit
ddba875
·
verified ·
1 Parent(s): 6e30507

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +13 -0
README.md CHANGED
@@ -30,16 +30,29 @@ more details, including on how to concatenate multi-part files.
30
 
31
  | Link | Type | Size/GB | Notes |
32
  |:-----|:-----|--------:|:------|
 
33
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ1_M.gguf) | i1-IQ1_M | 4.7 | mostly desperate |
 
 
 
34
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ2_M.gguf) | i1-IQ2_M | 5.8 | |
35
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q2_K.gguf) | i1-Q2_K | 5.9 | IQ3_XXS probably better |
36
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 6.4 | lower quality |
 
 
37
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q3_K_S.gguf) | i1-Q3_K_S | 6.9 | IQ3_XS probably better |
38
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ3_M.gguf) | i1-IQ3_M | 7.5 | |
39
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q3_K_M.gguf) | i1-Q3_K_M | 7.8 | IQ3_S probably better |
40
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ4_XS.gguf) | i1-IQ4_XS | 7.9 | |
 
 
 
 
 
41
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_K_S.gguf) | i1-Q4_K_S | 8.7 | optimal size/speed/quality |
42
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_K_M.gguf) | i1-Q4_K_M | 9.5 | fast, recommended |
 
 
43
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q6_K.gguf) | i1-Q6_K | 12.4 | practically like static Q6_K |
44
 
45
  Here is a handy graph by ikawrakow comparing some lower-quality quant
 
30
 
31
  | Link | Type | Size/GB | Notes |
32
  |:-----|:-----|--------:|:------|
33
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ1_S.gguf) | i1-IQ1_S | 4.5 | for the desperate |
34
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ1_M.gguf) | i1-IQ1_M | 4.7 | mostly desperate |
35
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 5.0 | |
36
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ2_XS.gguf) | i1-IQ2_XS | 5.3 | |
37
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ2_S.gguf) | i1-IQ2_S | 5.5 | |
38
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ2_M.gguf) | i1-IQ2_M | 5.8 | |
39
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q2_K.gguf) | i1-Q2_K | 5.9 | IQ3_XXS probably better |
40
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 6.4 | lower quality |
41
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ3_XS.gguf) | i1-IQ3_XS | 6.7 | |
42
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ3_S.gguf) | i1-IQ3_S | 6.9 | beats Q3_K* |
43
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q3_K_S.gguf) | i1-Q3_K_S | 6.9 | IQ3_XS probably better |
44
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ3_M.gguf) | i1-IQ3_M | 7.5 | |
45
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q3_K_M.gguf) | i1-Q3_K_M | 7.8 | IQ3_S probably better |
46
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-IQ4_XS.gguf) | i1-IQ4_XS | 7.9 | |
47
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q3_K_L.gguf) | i1-Q3_K_L | 8.1 | IQ3_M probably better |
48
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_0_4_4.gguf) | i1-Q4_0_4_4 | 8.3 | fast on arm, low quality |
49
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_0_4_8.gguf) | i1-Q4_0_4_8 | 8.3 | fast on arm+i8mm, low quality |
50
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_0_8_8.gguf) | i1-Q4_0_8_8 | 8.3 | fast on arm+sve, low quality |
51
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_0.gguf) | i1-Q4_0 | 8.3 | fast, low quality |
52
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_K_S.gguf) | i1-Q4_K_S | 8.7 | optimal size/speed/quality |
53
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q4_K_M.gguf) | i1-Q4_K_M | 9.5 | fast, recommended |
54
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q5_K_S.gguf) | i1-Q5_K_S | 10.1 | |
55
+ | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q5_K_M.gguf) | i1-Q5_K_M | 11.0 | |
56
  | [GGUF](https://huggingface.co/mradermacher/ShieldLM-14B-qwen-i1-GGUF/resolve/main/ShieldLM-14B-qwen.i1-Q6_K.gguf) | i1-Q6_K | 12.4 | practically like static Q6_K |
57
 
58
  Here is a handy graph by ikawrakow comparing some lower-quality quant