Notice
Recent Posts
Recent Comments
Link
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 |
Tags
- GPT
- 파이썬 #기초 #응애
- python
- onnx #openvino #tf2onnx #openvinokeras
- CentOS
- GPU
- NLP
- Mecab
- OpenAI
- pytorch
- BitCoin
- Python #for문 #while문 #응애
- overloaed
- ML
- leetcode
- GCP
- vram
- 매매봇
- GPT API
- Backoff
- Anaconda
- ML #머신러닝
- Python #open #read #파이썬
- UUID
- Flask
- docker
- Gunicorn
- 고유값
- MySQL
- DataLoader
Archives
- Today
- Total
람쥐썬더
[PYTHON] GPU VRAM(메모리) 초기화 (Winodows) 본문
tensorflow 학습 중 자주 보게 되는 OOM (Out Of Memory) 에러 ..
내부 모듈이 아닌 역전파로 학습 시키려다 보니까 어김없이 나타나고 말았다

뭐 하는것도 아닌데 메모리 드립다 먹고 있는 녀석
사실 한두번 이러면 그냥 재부팅 해주면 되는 문제긴 한데 계속 재부팅하기 귀찮아서 코드로 안되나 찾아봤음
Linux 환경이면 sudo kill 로 메모리를 죽일 수 있다는데 코드에서 해결 할 수 없나 찾아봤고 역시나 스택오버플로우에 의인들이 있었다 재부팅은 아니더라도 런타임은 다시 실행시켜줘야댐 ..............
해결 :
pip install numba
from numba import cuda
device = cuda.get_current_device()
device.reset()

조와
'파이썬' 카테고리의 다른 글
[Leetcode] 377. Combination Sum IV (0) | 2023.09.10 |
---|---|
[Leetcode] 118. Pascal's Triangle (0) | 2023.09.08 |
[PYTHON] IndexError: single positional indexer is out-of-bounds 해결 (0) | 2022.10.27 |
[PYTHON] Tensorflow op [~] is not supported (0) | 2022.10.24 |
[PYTHON] API 연동 [POST] Token 받기 (0) | 2022.07.20 |