| 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 |
- 수학
- cnn
- 딥러닝 #deep learning #machine learning #computer vision #object detection
- 고유값 분해
- chagpt
- mobile deep learning
- domain adaptation
- 3d
- 딥러닝
- zero shot
- deep learning mesh
- 머신러닝
- 쿠버플로우
- semi supervised semantic segmentation
- depth map
- 파이썬
- GPT 논문
- convolution neural network
- 통계
- semi supervised learning
- GPT
- deep learning on mobile
- Python
- sudo 권한
- Computer Vision
- Deep Learning
- GPT 1
- segmentation
- psuedo label
- depth data
- Today
- Total
목록Python (2)
튼튼한 나무
import pynvml import torch def get_memory_free_MiB(gpu_index): pynvml.nvmlInit() handle = pynvml.nvmlDeviceGetHandleByIndex(int(gpu_index)) mem_info = pynvml.nvmlDeviceGetMemoryInfo(handle) total= mem_info.total // 1024 ** 2 free = mem_info.free // 1024 ** 2 return free, total print('GPU Memory : ',{f'GPU-{i}':f'{get_memory_free_MiB(i)[0]:,d}/{get_memory_free_MiB(i)[1]:,d} MiB' for i in range(to..
필자는 서버 굉장히 많이(아니 대부분) 사용하는데, tensorboard로 scalar, image, audio, weight 등 다양한 값을 저장하고 서버에 연결하여 코딩도 하지만 juyter notebook으로 다양한 작업을 한다. 이런 경우에 tensorboard를 원격으로 접속가능하게끔 띄워도 되지만 뭔가 굉장히 귀찮다.. 이걸 또 안하려니 매번 tensorboard file을 로컬에 다운받아서 tensorboard를 로컬에 띄워야 한다. (매우 귀찮다..) 그래서 jupyter notebook에서 tensorboard를 실시간으로 띄워보고자 한다. 먼저 port를 뚫어준다. (아래 예시는 6006~6020 port open) 1) ubuntu server sudo ufw allow 6006:60..