Why is closure not supported in GradScaler ?
|
|
4
|
1745
|
March 12, 2022
|
AMP twice as slow when using a different GPU
|
|
1
|
875
|
March 9, 2022
|
Use tensorcore explicitly on non-DL code
|
|
2
|
580
|
March 8, 2022
|
Torch.save numerical differences
|
|
5
|
1373
|
March 1, 2022
|
Gradient clipping for one of two losses when using AMP
|
|
0
|
627
|
February 25, 2022
|
Loss function precision with AMP
|
|
4
|
1323
|
February 22, 2022
|
Overflow on CPU, but not GPU
|
|
1
|
548
|
February 9, 2022
|
Register_hook with GradScalar
|
|
1
|
572
|
February 9, 2022
|
WGAN-GP with Mixed Precision forces Scaler to 0
|
|
0
|
721
|
February 5, 2022
|
.hal() or the use of mixed precision increases model size
|
|
1
|
899
|
January 30, 2022
|
WARNING:root:Torch AMP is not available on this platform
|
|
6
|
3267
|
January 20, 2022
|
Deterministic training when using mixed-precision
|
|
3
|
894
|
January 18, 2022
|
Does NCCL allreduce use fp16?
|
|
7
|
1531
|
January 14, 2022
|
Optimizer.step() -- ok; scaler.step(optimizer): No inf checks were recorded for this optimizer
|
|
2
|
6541
|
January 13, 2022
|
Override AMP casting during bfloat16 training
|
|
1
|
1215
|
January 12, 2022
|
Switching between mixed-precision training and full-precision training after training is started
|
|
7
|
1371
|
January 4, 2022
|
Increased GPU memory usage on GPU 0 when using AMP
|
|
3
|
1643
|
December 23, 2021
|
The performance gap between torch.cuda.amp and nviddia-apex
|
|
10
|
2420
|
December 23, 2021
|
AMP for two optimizers
|
|
1
|
1232
|
December 10, 2021
|
Mixed precision training is so slow when deterministic=True
|
|
1
|
1154
|
December 10, 2021
|
RuntimeError: expected scalar type Float but found Half in deform_conv2d
|
|
3
|
5938
|
December 4, 2021
|
Increased memory usage with AMP
|
|
3
|
4047
|
November 29, 2021
|
Where I can see whether and operation autocasts or not?
|
|
2
|
558
|
November 26, 2021
|
Torch.cuda.amp blocks gradient computation
|
|
3
|
738
|
November 16, 2021
|
Does amp autocast cache fp16 copies of model parameters?
|
|
1
|
618
|
November 8, 2021
|
amp_C fused kernels unavailable
|
|
3
|
2450
|
November 5, 2021
|
Do I need to save the state_dict oof GradScaler?
|
|
5
|
2276
|
November 5, 2021
|
Training GANs using automatic mixed precision?
|
|
1
|
1079
|
November 1, 2021
|
How to fix NaN in the Bert layer?
|
|
3
|
1139
|
October 31, 2021
|
Mixed precision with generative adversarial network(GAN)
|
|
0
|
768
|
October 20, 2021
|