Unverified 提交 1ff43702 authored 作者: Glenn Jocher's avatar Glenn Jocher 提交者: GitHub

Validate with 2x `--workers` single-GPU/CPU fix (#6659)

Fix for #6658 for single-GPU and CPU training use cases
上级 67690219
......@@ -110,7 +110,7 @@ def create_dataloader(path, imgsz, batch_size, stride, single_cls=False, hyp=Non
batch_size = min(batch_size, len(dataset))
nd = torch.cuda.device_count() # number of CUDA devices
nw = min([2 * os.cpu_count() // max(nd, 1), batch_size if batch_size > 1 else 0, workers]) # number of workers
nw = min([os.cpu_count() // max(nd / 2, 1), batch_size if batch_size > 1 else 0, workers]) # number of workers
sampler = None if rank == -1 else distributed.DistributedSampler(dataset, shuffle=shuffle)
loader = DataLoader if image_weights else InfiniteDataLoader # only DataLoader allows for attribute updates
return loader(dataset,
......
Markdown 格式
0%
您添加了 0 到此讨论。请谨慎行事。
请先完成此评论的编辑!
注册 或者 后发表评论