RichardErkhov commited on
Commit
d556a9e
β€’
1 Parent(s): 5fb6bed

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +92 -0
README.md ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Quantization made by Richard Erkhov.
2
+
3
+ [Github](https://github.com/RichardErkhov)
4
+
5
+ [Discord](https://discord.gg/pvy7H8DZMG)
6
+
7
+ [Request more models](https://github.com/RichardErkhov/quant_request)
8
+
9
+
10
+ ft-solar-10.7b-v2.1-dpo - GGUF
11
+ - Model creator: https://huggingface.co/ifuseok/
12
+ - Original model: https://huggingface.co/ifuseok/ft-solar-10.7b-v2.1-dpo/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [ft-solar-10.7b-v2.1-dpo.Q2_K.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q2_K.gguf) | Q2_K | 3.73GB |
18
+ | [ft-solar-10.7b-v2.1-dpo.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.IQ3_XS.gguf) | IQ3_XS | 4.14GB |
19
+ | [ft-solar-10.7b-v2.1-dpo.IQ3_S.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.IQ3_S.gguf) | IQ3_S | 4.37GB |
20
+ | [ft-solar-10.7b-v2.1-dpo.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q3_K_S.gguf) | Q3_K_S | 4.34GB |
21
+ | [ft-solar-10.7b-v2.1-dpo.IQ3_M.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.IQ3_M.gguf) | IQ3_M | 4.51GB |
22
+ | [ft-solar-10.7b-v2.1-dpo.Q3_K.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q3_K.gguf) | Q3_K | 4.84GB |
23
+ | [ft-solar-10.7b-v2.1-dpo.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q3_K_M.gguf) | Q3_K_M | 4.84GB |
24
+ | [ft-solar-10.7b-v2.1-dpo.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q3_K_L.gguf) | Q3_K_L | 5.26GB |
25
+ | [ft-solar-10.7b-v2.1-dpo.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.IQ4_XS.gguf) | IQ4_XS | 5.43GB |
26
+ | [ft-solar-10.7b-v2.1-dpo.Q4_0.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q4_0.gguf) | Q4_0 | 5.66GB |
27
+ | [ft-solar-10.7b-v2.1-dpo.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.IQ4_NL.gguf) | IQ4_NL | 5.72GB |
28
+ | [ft-solar-10.7b-v2.1-dpo.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q4_K_S.gguf) | Q4_K_S | 5.7GB |
29
+ | [ft-solar-10.7b-v2.1-dpo.Q4_K.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q4_K.gguf) | Q4_K | 6.02GB |
30
+ | [ft-solar-10.7b-v2.1-dpo.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q4_K_M.gguf) | Q4_K_M | 6.02GB |
31
+ | [ft-solar-10.7b-v2.1-dpo.Q4_1.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q4_1.gguf) | Q4_1 | 6.27GB |
32
+ | [ft-solar-10.7b-v2.1-dpo.Q5_0.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q5_0.gguf) | Q5_0 | 6.89GB |
33
+ | [ft-solar-10.7b-v2.1-dpo.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q5_K_S.gguf) | Q5_K_S | 6.89GB |
34
+ | [ft-solar-10.7b-v2.1-dpo.Q5_K.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q5_K.gguf) | Q5_K | 7.08GB |
35
+ | [ft-solar-10.7b-v2.1-dpo.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q5_K_M.gguf) | Q5_K_M | 7.08GB |
36
+ | [ft-solar-10.7b-v2.1-dpo.Q5_1.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q5_1.gguf) | Q5_1 | 7.51GB |
37
+ | [ft-solar-10.7b-v2.1-dpo.Q6_K.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q6_K.gguf) | Q6_K | 8.2GB |
38
+ | [ft-solar-10.7b-v2.1-dpo.Q8_0.gguf](https://huggingface.co/RichardErkhov/ifuseok_-_ft-solar-10.7b-v2.1-dpo-gguf/blob/main/ft-solar-10.7b-v2.1-dpo.Q8_0.gguf) | Q8_0 | 10.62GB |
39
+
40
+
41
+
42
+
43
+ Original model description:
44
+ ---
45
+ language:
46
+ - ko
47
+ pipeline_tag: text-generation
48
+ datasets:
49
+ - nlpai-lab/databricks-dolly-15k-ko
50
+ - kyujinpy/KOR-OpenOrca-Platypus-v3
51
+ - KETI-AIR/kor_boolq
52
+ - heegyu/open-korean-instructions
53
+ license: cc-by-nc-sa-4.0
54
+ ---
55
+
56
+ **Input** Models input text only.
57
+
58
+ **Output** Models generate text only.
59
+
60
+ **Base Model** [yanolja/KoSOLAR-10.7B-v0.1](https://huggingface.co/yanolja/KoSOLAR-10.7B-v0.1-deprecated)
61
+
62
+ **Training Dataset**
63
+ - [nlpai-lab/databricks-dolly-15k-ko](https://huggingface.co/datasets/nlpai-lab/databricks-dolly-15k-ko)
64
+ - [kyujinpy/KOR-OpenOrca-Platypus-v3](https://huggingface.co/datasets/kyujinpy/KOR-OpenOrca-Platypus-v3)
65
+ - [heegyu/open-korean-instructions](heegyu/open-korean-instructions)
66
+ - [KETI-AIR/kor_boolq](https://huggingface.co/datasets/KETI-AIR/kor_boolq)
67
+ - [AIhub μ˜ν•œ λ²ˆμ—­ 데이터 일뢀](https://aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&dataSetSn=71593)
68
+
69
+ # Implementation Code
70
+ ```python
71
+ from transformers import AutoModelForCausalLM, AutoTokenizer
72
+ import torch
73
+ repo = "ifuseok/sft-solar-10.7b-v2.1-dpo"
74
+ OpenOrca = AutoModelForCausalLM.from_pretrained(
75
+ repo,
76
+ return_dict=True,
77
+ torch_dtype=torch.float16,
78
+ device_map='auto'
79
+ )
80
+ OpenOrca_tokenizer = AutoTokenizer.from_pretrained(repo)
81
+ ```
82
+
83
+ # Prompt Example
84
+ ```
85
+ ### System:
86
+ μ‹œμŠ€ν…œ λ©”μ‹œμ§€ μž…λ‹ˆλ‹€.
87
+ ### User:
88
+ μœ μ € μž…λ‹ˆλ‹€.
89
+ ### Assistant
90
+ μ–΄μ‹œμŠ€ν„΄νŠΈ μž…λ‹ˆλ‹€.
91
+ ```
92
+