Dongwookss/q_a_korean_futsal
Viewer • Updated • 2.33k • 22
How to use Dongwookss/futfut_by_zephyr7b_gguf with Transformers:
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Dongwookss/futfut_by_zephyr7b_gguf", device_map="auto")How to use Dongwookss/futfut_by_zephyr7b_gguf with llama.cpp:
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf Dongwookss/futfut_by_zephyr7b_gguf:F16 # Run inference directly in the terminal: llama cli -hf Dongwookss/futfut_by_zephyr7b_gguf:F16
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf Dongwookss/futfut_by_zephyr7b_gguf:F16 # Run inference directly in the terminal: llama cli -hf Dongwookss/futfut_by_zephyr7b_gguf:F16
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf Dongwookss/futfut_by_zephyr7b_gguf:F16 # Run inference directly in the terminal: ./llama-cli -hf Dongwookss/futfut_by_zephyr7b_gguf:F16
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf Dongwookss/futfut_by_zephyr7b_gguf:F16 # Run inference directly in the terminal: ./build/bin/llama-cli -hf Dongwookss/futfut_by_zephyr7b_gguf:F16
docker model run hf.co/Dongwookss/futfut_by_zephyr7b_gguf:F16
How to use Dongwookss/futfut_by_zephyr7b_gguf with Ollama:
ollama run hf.co/Dongwookss/futfut_by_zephyr7b_gguf:F16
How to use Dongwookss/futfut_by_zephyr7b_gguf with Docker Model Runner:
docker model run hf.co/Dongwookss/futfut_by_zephyr7b_gguf:F16
How to use Dongwookss/futfut_by_zephyr7b_gguf with Lemonade:
# Download Lemonade from https://lemonade-server.ai/ lemonade pull Dongwookss/futfut_by_zephyr7b_gguf:F16
lemonade run user.futfut_by_zephyr7b_gguf-F16
lemonade list
Base Model : 'HuggingFaceH4/zephyr-7b-beta'
Purpose : '얼마든지 물어보세요~! 풋풋!'이 말 끝에 붙으며 '해요'체를 사용하는 챗봇을 구현하려고 한다. 프로젝트 목적상 RAG를 통해 풋살 도메인에 대한 정보를 제공하는 '풋풋이' 컨셉이기에 말투 설정이 이와 같다.
Method : Unsloth 패키지를 사용하여 gpu 자원이 초과되지 않도록 하였으며 SFTrainer를 사용하여 모델훈련을 진행함.
Environ : Colab L4 GPU를 사용하여 진행하였습니다.
About gguf : 풋풋이 최종 버전을 만들기 위해 각 단계별 모델을 저장하였습니다. 모델 양자화(Quantize) 종류로 F16, Q8_0, Q5_K_M 3가지로 저장하였습니다.
5-bit
8-bit
16-bit
Base model
mistralai/Mistral-7B-v0.1