Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,76 @@
|
|
1 |
---
|
2 |
license: cc-by-nc-4.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: cc-by-nc-4.0
|
3 |
+
datasets:
|
4 |
+
- kyujinpy/KOR-gugugu-platypus-set
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
- ko
|
8 |
+
base_model:
|
9 |
+
- yanolja/KoSOLAR-10.7B-v0.2
|
10 |
+
pipeline_tag: text-generation
|
11 |
---
|
12 |
+
|
13 |
+
# KoSOLAR-v0.2-gugutypus-10.7B
|
14 |
+
|
15 |
+
<img src="logo.png" height=350, width=350>
|
16 |
+
|
17 |
+
|
18 |
+
---
|
19 |
+
|
20 |
+
|
21 |
+
## Model Details
|
22 |
+
|
23 |
+
**Model Developers**
|
24 |
+
- DongGeon Lee ([oneonlee](https://huggingface.co/oneonlee))
|
25 |
+
|
26 |
+
**Model Architecture**
|
27 |
+
- **KoSOLAR-v0.2-gugutypus-10.7B** is a instruction fine-tuned auto-regressive language model, based on the [SOLAR](https://huggingface.co/upstage/SOLAR-10.7B-v1.0) transformer architecture.
|
28 |
+
|
29 |
+
**Base Model**
|
30 |
+
- [yanolja/KoSOLAR-10.7B-v0.2](https://huggingface.co/yanolja/KoSOLAR-10.7B-v0.2)
|
31 |
+
|
32 |
+
**Training Dataset**
|
33 |
+
- [kyujinpy/KOR-gugugu-platypus-set](https://huggingface.co/datasets/kyujinpy/KOR-gugugu-platypus-set)
|
34 |
+
|
35 |
+
|
36 |
+
---
|
37 |
+
|
38 |
+
|
39 |
+
## Model comparisons
|
40 |
+
|
41 |
+
- **Ko-LLM leaderboard (YYYY/MM/DD)** [[link]](https://huggingface.co/spaces/upstage/open-ko-llm-leaderboard)
|
42 |
+
|
43 |
+
| Model | Average | Ko-ARC | Ko-HellaSwag | Ko-MMLU | Ko-TruthfulQA | Ko-CommonGen V2 |
|
44 |
+
| --------------------- | ------- | ------ | ------------ | ------- | ------------- | --------------- |
|
45 |
+
| **KoSOLAR-gugutypus** | NaN | NaN | NaN | NaN | NaN | NaN |
|
46 |
+
|
47 |
+
<br>
|
48 |
+
|
49 |
+
- **AI-Harness evaluation** [[link]](https://github.com/Beomi/ko-lm-evaluation-harness)
|
50 |
+
|
51 |
+
|
52 |
+
| Model | Copa | Copa | HellaSwag | HellaSwag | BoolQ | BoolQ | Sentineg | Sentineg |
|
53 |
+
| --------------------- | ------ | ------ | --------- | --------- | ------ | ------ | -------- | -------- |
|
54 |
+
| | 0-shot | 5-shot | 0-shot | 5-shot | 0-shot | 5-shot | 0-shot | 5-shot |
|
55 |
+
| **KoSOLAR-gugutypus** | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
|
56 |
+
|
57 |
+
|
58 |
+
---
|
59 |
+
|
60 |
+
|
61 |
+
## Implementation Code
|
62 |
+
|
63 |
+
```python
|
64 |
+
### KoSOLAR-gugutypus
|
65 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
66 |
+
import torch
|
67 |
+
|
68 |
+
repo = "oneonlee/KoSOLAR-v0.2-gugutypus-10.7B"
|
69 |
+
model = AutoModelForCausalLM.from_pretrained(
|
70 |
+
repo,
|
71 |
+
return_dict=True,
|
72 |
+
torch_dtype=torch.float16,
|
73 |
+
device_map='auto'
|
74 |
+
)
|
75 |
+
tokenizer = AutoTokenizer.from_pretrained(repo)
|
76 |
+
```
|