TwinDoc's picture
Update README.md
e95b62f verified
metadata
language:
  - ko
  - en
library_name: transformers
license: cc-by-nc-sa-4.0
pipeline_tag: text-generation
tags:
  - pytorch

Model Description

S-B 고객사 ν”„λ‘œμ νŠΈ μ‹œ μƒμ„±ν•œ RAG 데이터셋을 ν™œμš©ν•˜μ—¬ Supervised Fine-Tuning(a.k.a SFT) ν•™μŠ΅ν•œ λͺ¨λΈμž…λ‹ˆλ‹€. ν•™μŠ΅ 데이터셋은 λ³΄μ•ˆμ— μ˜ν•΄ κ³΅κ°œν•˜μ§€ μ•ŠμŠ΅λ‹ˆλ‹€.

About the Model

  • Name: TwinDoc/RedWhale-tv-10.8B-sft-s

  • Finetuned from model: TwinDoc/RedWhale-tv-10.8B-v1.0

  • Train Datasets: private

  • Developed by: μ• μžμΌμ†Œλ‹€ (AGILESODA)

  • Model type: llama

  • Language(s) (NLP): ν•œκ΅­μ–΄

  • License: cc-by-nc-sa-4.0

  • train setting

    • Lora r, alpha : 4, 16
    • Dtype : bf16
    • Epoch : 7
    • Learning rate : 1e-4
    • Global batch : 4
    • Context length : 4096
  • inference setting

    • BOS id : 1
    • EOS id : 2
    • Top-p : 0.95
    • Temperature : 0.01

prompt template

Human: ##원문##κ³Ό ##질문##이 주어지면, ##원문##에 μžˆλŠ” 정보λ₯Ό λ°”νƒ•μœΌλ‘œ κ³ ν’ˆμ§ˆμ˜ ##λ‹΅λ³€##을 λ§Œλ“€μ–΄μ£Όμ„Έμš”. ##원문##μ—μ„œ ##질문##에 λŒ€ν•œ λͺ…ν™•ν•œ 닡을 찾을 수 없을 경우 "닡변을 찾을 수 μ—†μŠ΅λ‹ˆλ‹€."둜 ##λ‹΅λ³€##을 μž‘μ„±ν•΄μ•Όν•˜λ©° ##원문##에 μ—†λŠ” λ‚΄μš©μ€ ##λ‹΅λ³€##에 ν¬ν•¨ν•˜μ§€ μ•Šμ•„μ•Ό ν•©λ‹ˆλ‹€.
    
##원문##
{CONTEXT}
##질문##
{QUESTION}
 Assistant: {ANSWER}

License

The content of this project, created by AGILESODA, is licensed under the Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).

Citation

@misc{vo2024redwhaleadaptedkoreanllm,
      title={RedWhale: An Adapted Korean LLM Through Efficient Continual Pretraining}, 
      author={Anh-Dung Vo and Minseong Jung and Wonbeen Lee and Daewoo Choi},
      year={2024},
      eprint={2408.11294},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2408.11294}, 
}

Built with:

AgileSoda TwinDoc Icon