mirrors
/
yolov5
mirror of https://github.com/ultralytics/yolov5.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Update utils.py strip_optimizer() (#509)

pull/514/head
Glenn Jocher 2020-07-24 19:26:56 -07:00 committed by GitHub
parent 1d17b9af0f
commit 0032af2980
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 7 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Dockerfile
View File

@ -43,7 +43,7 @@ COPY . /usr/src/app
# sudo docker commit 092b16b25c5b usr/resume && sudo docker run -it --gpus all --ipc=host -v "$(pwd)"/coco:/usr/src/coco --entrypoint=sh usr/resume
# Send weights to GCP
# python -c "from utils.utils import *; create_pretrained('path/last.pt')" && gsutil cp weights/pretrained.pt gs://*
# python -c "from utils.utils import *; strip_optimizer('runs/exp0/weights/last.pt', 'temp.pt')" && gsutil cp temp.pt gs://*
# Clean up
# docker system prune -a --volumes

20
utils/utils.py
View File

@ -645,26 +645,18 @@ def non_max_suppression(prediction, conf_thres=0.1, iou_thres=0.6, merge=False,
return output
def strip_optimizer(f='weights/best.pt'): # from utils.utils import *; strip_optimizer()
# Strip optimizer from *.pt files for lighter files (reduced by 1/2 size)
x = torch.load(f, map_location=torch.device('cpu'))
x['optimizer'] = None
x['model'].half() # to FP16
torch.save(x, f)
print('Optimizer stripped from %s, %.1fMB' % (f, os.path.getsize(f) / 1E6))
def create_pretrained(f='weights/best.pt', s='weights/pretrained.pt'): # from utils.utils import *; create_pretrained()
# create pretrained checkpoint 's' from 'f' (create_pretrained(x, x) for x in glob.glob('./*.pt'))
def strip_optimizer(f='weights/best.pt', s=''): # from utils.utils import *; strip_optimizer()
# Strip optimizer from 'f' to finalize training, optionally save as 's'
x = torch.load(f, map_location=torch.device('cpu'))
x['optimizer'] = None
x['training_results'] = None
x['epoch'] = -1
x['model'].half() # to FP16
for p in x['model'].parameters():
p.requires_grad = True
torch.save(x, s)
print('%s saved as pretrained checkpoint %s, %.1fMB' % (f, s, os.path.getsize(s) / 1E6))
p.requires_grad = False
torch.save(x, s or f)
mb = os.path.getsize(s or f) / 1E6 # filesize
print('Optimizer stripped from %s,%s %.1fMB' % (f, (' saved as %s,' % s) if s else '', mb))
def coco_class_count(path='../coco/labels/train2014/'):