MarkrAI/kyujin-CoTy-platypus-ko-12.8b

(주)미디어그룹사람과숲과 (주)마커의 LLM 연구 컨소시엄에서 개발된 모델입니다
The license is cc-by-nc-sa-4.0.

CoTy-platypus-ko

Poly-platypus-ko + CoT = CoTy-platypus-ko

Model Details

Model Developers Kyujin Han (kyujinpy)

Input Models input text only.

Output Models generate text only.

Model Architecture
CoTy-platypus-ko is an auto-regressive language model based on the polyglot-ko transformer architecture.

Repo Link
Github CoTy-platypus-ko: CoTy-platypus-ko

Base Model
Polyglot-ko-12.8b

Fine-tuning method
Methodology by KO-Platypus2+CoT-llama2-ko

Training Dataset
I use KoCoT_2000.
I use A100 GPU 40GB and COLAB, when trianing.

---

Model Bechmark1

KO-LLM leaderboard

• Follow up as Open KO-LLM LeaderBoard.

Model	Average	Ko-ARC	Ko-HellaSwag	Ko-MMLU	Ko-TruthfulQA	Ko-CommonGen V2
CoTy-platypus-ko-12.8b(ours)	46.44	34.98	49.11	25.68	37.59	84.86
hyunseoki/ko-en-llama2-13b	46.68	42.15	54.23	38.90	40.74	57.39
momo/polyglot-ko-12.8b-Chat-QLoRA-Merge	45.71	35.49	49.93	25.97	39.43	77.70
KoT-platypus2-7B	45.62	38.05	49.63	34.68	37.69	68.08
DopeorNope/COLA3-7B	45.61	39.16	50.98	35.21	37.81	64.91

Compare with Top 4 SOTA models. (update: 10/03)

---

Implementation Code

### KO-Platypus
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

repo = "MarkrAI/kyujin-CoTy-platypus-ko-12.8b"
CoT-llama = AutoModelForCausalLM.from_pretrained(
        repo,
        return_dict=True,
        torch_dtype=torch.float16,
        device_map='auto'
)
CoT-llama_tokenizer = AutoTokenizer.from_pretrained(repo)

Readme format: kyujinpy/KoT-platypus2-7B

---