Browse Source

train with multi-gpu half test bug fix #99

5.0
Glenn Jocher 4 years ago
parent
commit
d1e5716610
1 changed files with 1 additions and 1 deletions
  1. +1
    -1
      test.py

+ 1
- 1
test.py View File

@@ -46,7 +46,7 @@ def test(data,
else: # called by train.py
training = True
device = next(model.parameters()).device # get model device
half = device.type != 'cpu' # half precision only supported on CUDA
half = device.type != 'cpu' and torch.cuda.device_count() == 1 # half precision only supported on single-GPU
if half:
model.half() # to FP16


Loading…
Cancel
Save