Skip to content

최종 결과용 실험 계획 #22

Description

@GirinMan

실험 계획

  • 어떤 tuning 방법을 사용했을때 memory efficient한가?
  • 어떤 quantization 방법을 사용했을 때 inference time에서 정확도가 높은가?

Finetuning 과정에서 메모리 사용량 비교군

  1. Full finetuning
  2. LoRA tuning
  3. llm.int8() + LoRA tuning

Inference 시점에서 정확도 비교군

  1. Full finetuning + dymamic quantization
  2. LoRA + dynamic quantization
  3. llm.int8() + LoRA tuning

Task 종류

image
image

  1. GLUE: MRPC, RTE, COLA, STS-B
  2. SAMsum(요약)

Task별 프롬프트 구성

https://www.promptingguide.ai/introduction/examples

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Labels

help wantedExtra attention is needed

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions