Improved model+EMA checkpointing 2 (#2295)

pull/2300/head
Glenn Jocher 2021-02-24 21:03:21 -08:00 committed by GitHub
parent ec1d8496ba
commit 71dd2768f2
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 4 additions and 4 deletions

View File

@ -269,6 +269,7 @@ def test(data,
print(f'pycocotools unable to run: {e}') print(f'pycocotools unable to run: {e}')
# Return results # Return results
model.float() # for training
if not training: if not training:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else '' s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}") print(f"Results saved to {save_dir}{s}")

View File

@ -4,6 +4,7 @@ import math
import os import os
import random import random
import time import time
from copy import deepcopy
from pathlib import Path from pathlib import Path
from threading import Thread from threading import Thread
@ -381,8 +382,8 @@ def train(hyp, opt, device, tb_writer=None, wandb=None):
ckpt = {'epoch': epoch, ckpt = {'epoch': epoch,
'best_fitness': best_fitness, 'best_fitness': best_fitness,
'training_results': results_file.read_text(), 'training_results': results_file.read_text(),
'model': (model.module if is_parallel(model) else model).half(), 'model': deepcopy(model.module if is_parallel(model) else model).half(),
'ema': (ema.ema.half(), ema.updates), 'ema': (deepcopy(ema.ema).half(), ema.updates),
'optimizer': optimizer.state_dict(), 'optimizer': optimizer.state_dict(),
'wandb_id': wandb_run.id if wandb else None} 'wandb_id': wandb_run.id if wandb else None}
@ -392,8 +393,6 @@ def train(hyp, opt, device, tb_writer=None, wandb=None):
torch.save(ckpt, best) torch.save(ckpt, best)
del ckpt del ckpt
model.float(), ema.ema.float()
# end epoch ---------------------------------------------------------------------------------------------------- # end epoch ----------------------------------------------------------------------------------------------------
# end training # end training