README.md

metadata

language:
  - en
pipeline_tag: text-generation
datasets:
  - nlpai-lab/databricks-dolly-15k-ko
  - kyujinpy/KOR-OpenOrca-Platypus-v3
  - KETI-AIR/kor_boolq
  - heegyu/open-korean-instructions

Input Models input text only.

Output Models generate text only.

Base Model upstage/SOLAR-10.7B-Instruct-v1.0

Training Dataset

• nlpai-lab/databricks-dolly-15k-ko
• kyujinpy/KOR-OpenOrca-Platypus-v3
• heegyu/open-korean-instructions
• KETI-AIR/kor_boolq
• AIhub 영한 번역 데이터 일부

Implementation Code

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
repo = "ifuseok/sft-solar-10.7b-v1.1"
OpenOrca = AutoModelForCausalLM.from_pretrained(
        repo,
        return_dict=True,
        torch_dtype=torch.float16,
        device_map='auto'
)
OpenOrca_tokenizer = AutoTokenizer.from_pretrained(repo)

Prompt Example

### System:
시스템 메시지 입니다. 
### User:
유저  입니다.
### Assistant
어시스턴트 입니다.