K-HyunmooGPT

K-Data Science Hackaton 3rd Award

Development and use of Korean large-scale generative language models

20231128~20231129

• LLM serving

  • Local server (1* RTX 6000)
  • Docker
  • HuggingFace pipeline

• LLM Finetuning

  • Original model from KoAlpaca-Polyglot-5.8B
  • Finetuned using Peft+LoRA

• RAG

  • ChromaDB
  • FastAPI
  • Docker

• Gradio

  • Original code from Baize

• Prompoting

...