When I run dist training in a long time.
DistGraph obj is NoneType occur and the training broken.
why DistGraph will be NoneType ?
When I run dist training in a long time.
DistGraph obj is NoneType occur and the training broken.
why DistGraph will be NoneType ?
I dont know Why DistGraph auto recycle by python when in my training loop 。 for example : I run 20000 steps later the distgraph is NoneType
sorry I make a mistake