
아래의 doc을 참고하여 우선 GPU 서버가 없으므로 CPU에 작은 모델을 서빙 해보려고 한다. https://github.com/triton-inference-server/server GitHub - triton-inference-server/server: The Triton Inference Server provides an optimized cloud and edge inferencing solution. The Triton Inference Server provides an optimized cloud and edge inferencing solution. - GitHub - triton-inference-server/server: The Triton Inference Server provides..

사용자의 쿼리에 대하여 토큰 비용을 줄일 수 있는 방법 중에 캐싱 기법을 적용해보았다. 물론 토큰 비용이 얼마 되지 않아 그냥 해도 되지만 응답 속도는 확연하게 체감이 될 정도로 빨랐다. 랭체인에서 제공하는 라이브러리를 사용하면 캐싱 구현은 정말 간단하다. from langchain.cache import RedisCache from langchain.globals import set_llm_cache set_llm_cache(RedisCache(redis_=Redis(host='redis', port=6379, db=1))) response = rag_chain_with_source.invoke(text.question) 질문을 할때 마다 Key가 생성이 되었다. Key는 Hash로 이루어 져있는데 어..
FastAPI + Sqlalchemy를 사용하여 개발하던 중 insert 실패 후 다른 데이터를 insert를 할때 아래 와 같은 애러가 발생하였다. sqlalchemy.exc.PendingRollbackError: This Session's transaction has been rolled back due to a previous exception during flush. To begin a new transaction with this Session, first issue Session.rollback(). Original exception was: (psycopg2.errors.UniqueViolation) duplicate key value violates unique constraint 위의 애..
- Total
- Today
- Yesterday
- 다이나믹프로그래밍
- 코딩테스트
- 자료구조
- java
- 보안
- synflooding
- 리눅스
- 분산시스템
- 보안기사
- 정보보안
- 백준
- Ai
- linux
- huggingface
- 프로그래머스
- 시간초과
- springboot
- lightsail
- Python
- 파이썬
- llm
- FastAPI
- 카카오페이면접후기
- 딥러닝
- 우선순위큐
- 정보보안기사
- t-test
- 그리디
- 카카오페이
- LangChain
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 | 31 |