Hugging Face's logo

tensorblock
/
llama3.1_korean_v0.1_sft_by_aidx-GGUF

Text Generation
GGUF
Korean
English
ktds
ko_leaderboard
korean
TensorBlock
GGUF
Inference Endpoints
conversational
Model card Files Community
morriszms commited on
Commit
bc08312
·
verified ·
1 Parent(s): b905442

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +20 -12
README.md CHANGED
@@ -33,8 +33,16 @@ This repo contains GGUF format model files for [SEOKDONG/llama3.1_korean_v0.1_sf
33
 
34
  The files were quantized using machines provided by [TensorBlock](https://tensorblock.co/), and they are compatible with llama.cpp as of [commit b4011](https://github.com/ggerganov/llama.cpp/commit/a6744e43e80f4be6398fc7733a01642c846dce1d).
35
 
 
 
 
 
 
 
 
36
  ## Prompt template
37
 
 
38
  ```
39
  <|begin_of_text|><|start_header_id|>system<|end_header_id|>
40
 
@@ -47,18 +55,18 @@ The files were quantized using machines provided by [TensorBlock](https://tensor
47
 
48
  | Filename | Quant type | File Size | Description |
49
  | -------- | ---------- | --------- | ----------- |
50
- | [llama3.1_korean_v0.1_sft_by_aidx-Q2_K.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q2_K.gguf) | Q2_K | 2.961 GB | smallest, significant quality loss - not recommended for most purposes |
51
- | [llama3.1_korean_v0.1_sft_by_aidx-Q3_K_S.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q3_K_S.gguf) | Q3_K_S | 3.413 GB | very small, high quality loss |
52
- | [llama3.1_korean_v0.1_sft_by_aidx-Q3_K_M.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q3_K_M.gguf) | Q3_K_M | 3.743 GB | very small, high quality loss |
53
- | [llama3.1_korean_v0.1_sft_by_aidx-Q3_K_L.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q3_K_L.gguf) | Q3_K_L | 4.025 GB | small, substantial quality loss |
54
- | [llama3.1_korean_v0.1_sft_by_aidx-Q4_0.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q4_0.gguf) | Q4_0 | 4.341 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
55
- | [llama3.1_korean_v0.1_sft_by_aidx-Q4_K_S.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q4_K_S.gguf) | Q4_K_S | 4.370 GB | small, greater quality loss |
56
- | [llama3.1_korean_v0.1_sft_by_aidx-Q4_K_M.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q4_K_M.gguf) | Q4_K_M | 4.583 GB | medium, balanced quality - recommended |
57
- | [llama3.1_korean_v0.1_sft_by_aidx-Q5_0.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q5_0.gguf) | Q5_0 | 5.215 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
58
- | [llama3.1_korean_v0.1_sft_by_aidx-Q5_K_S.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q5_K_S.gguf) | Q5_K_S | 5.215 GB | large, low quality loss - recommended |
59
- | [llama3.1_korean_v0.1_sft_by_aidx-Q5_K_M.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q5_K_M.gguf) | Q5_K_M | 5.339 GB | large, very low quality loss - recommended |
60
- | [llama3.1_korean_v0.1_sft_by_aidx-Q6_K.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q6_K.gguf) | Q6_K | 6.143 GB | very large, extremely low quality loss |
61
- | [llama3.1_korean_v0.1_sft_by_aidx-Q8_0.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/tree/main/llama3.1_korean_v0.1_sft_by_aidx-Q8_0.gguf) | Q8_0 | 7.954 GB | very large, extremely low quality loss - not recommended |
62
 
63
 
64
  ## Downloading instruction
 
33
 
34
  The files were quantized using machines provided by [TensorBlock](https://tensorblock.co/), and they are compatible with llama.cpp as of [commit b4011](https://github.com/ggerganov/llama.cpp/commit/a6744e43e80f4be6398fc7733a01642c846dce1d).
35
 
36
+
37
+ <div style="text-align: left; margin: 20px 0;">
38
+ <a href="https://tensorblock.co/waitlist/client" style="display: inline-block; padding: 10px 20px; background-color: #007bff; color: white; text-decoration: none; border-radius: 5px; font-weight: bold;">
39
+ Run them on the TensorBlock client using your local machine ↗
40
+ </a>
41
+ </div>
42
+
43
  ## Prompt template
44
 
45
+
46
  ```
47
  <|begin_of_text|><|start_header_id|>system<|end_header_id|>
48
 
 
55
 
56
  | Filename | Quant type | File Size | Description |
57
  | -------- | ---------- | --------- | ----------- |
58
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q2_K.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q2_K.gguf) | Q2_K | 2.961 GB | smallest, significant quality loss - not recommended for most purposes |
59
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q3_K_S.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q3_K_S.gguf) | Q3_K_S | 3.413 GB | very small, high quality loss |
60
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q3_K_M.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q3_K_M.gguf) | Q3_K_M | 3.743 GB | very small, high quality loss |
61
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q3_K_L.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q3_K_L.gguf) | Q3_K_L | 4.025 GB | small, substantial quality loss |
62
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q4_0.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q4_0.gguf) | Q4_0 | 4.341 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
63
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q4_K_S.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q4_K_S.gguf) | Q4_K_S | 4.370 GB | small, greater quality loss |
64
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q4_K_M.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q4_K_M.gguf) | Q4_K_M | 4.583 GB | medium, balanced quality - recommended |
65
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q5_0.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q5_0.gguf) | Q5_0 | 5.215 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
66
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q5_K_S.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q5_K_S.gguf) | Q5_K_S | 5.215 GB | large, low quality loss - recommended |
67
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q5_K_M.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q5_K_M.gguf) | Q5_K_M | 5.339 GB | large, very low quality loss - recommended |
68
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q6_K.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q6_K.gguf) | Q6_K | 6.143 GB | very large, extremely low quality loss |
69
+ | [llama3.1_korean_v0.1_sft_by_aidx-Q8_0.gguf](https://huggingface.co/tensorblock/llama3.1_korean_v0.1_sft_by_aidx-GGUF/blob/main/llama3.1_korean_v0.1_sft_by_aidx-Q8_0.gguf) | Q8_0 | 7.954 GB | very large, extremely low quality loss - not recommended |
70
 
71
 
72
  ## Downloading instruction