Suggestions for backpropagating DSP code
|
|
3
|
118
|
February 28, 2024
|
Propagate gradient to parameters not directly involved in computation
|
|
4
|
98
|
February 28, 2024
|
Implementing custom backward for banded system solver
|
|
6
|
160
|
February 27, 2024
|
Local modifications of the backpropagation in Pytorch
|
|
2
|
112
|
February 27, 2024
|
Why is this compuation graph failling?
|
|
6
|
92
|
February 27, 2024
|
Analyzing PyTorch Profiling with Tensorboard Intergration
|
|
0
|
69
|
February 27, 2024
|
Dynamically change a models forward function during runtime
|
|
2
|
124
|
February 26, 2024
|
Custom backward step for convolutions
|
|
6
|
160
|
February 24, 2024
|
How to train a parameter initialized outside of nn.Module
|
|
0
|
87
|
February 24, 2024
|
Model not training, gradients are None
|
|
8
|
1415
|
February 23, 2024
|
Backpropagation with model ensembling
|
|
0
|
81
|
February 23, 2024
|
Error at loss.backward() when pretraining Llama model from scratch "TRYING TO BACKWARD SECOND TIME"
|
|
2
|
199
|
February 22, 2024
|
Why does pytorch prompt "[W accumulate_grad.h:170] Warning: grad and param do not obey the gradient layout contract. This is not an error, but may impair performance."?
|
|
21
|
11264
|
February 22, 2024
|
Speeding up gradient computation; instead of using a for loop
|
|
5
|
158
|
February 21, 2024
|
Get the underlying function calls for a function call and locate them in PyTorch code base
|
|
3
|
160
|
February 19, 2024
|
How to obtain a result of 0 from CrossEntropyLoss?
|
|
3
|
115
|
February 14, 2024
|
Slow aten::fill_ and aten::add_
|
|
0
|
84
|
February 18, 2024
|
optimizer.step() Not updating Model Weights/Parameters
|
|
5
|
319
|
February 17, 2024
|
Speedup forward/backward propagation
|
|
0
|
118
|
February 15, 2024
|
Custom layer weights do not move to 'mps' device
|
|
3
|
134
|
February 15, 2024
|
Accumulating model output blows up cuda memory?
|
|
5
|
101
|
February 15, 2024
|
Behavior explain - parameters deepcopied from model 1 do not update in model 2 where they are part of the computation
|
|
0
|
70
|
February 14, 2024
|
Why is my gradient accumulation failing?
|
|
5
|
116
|
February 14, 2024
|
Unexpected hook behavior with 3D tensor and inplace operation
|
|
2
|
105
|
February 12, 2024
|
What's the difference between torch.autograd.grad and backward()?
|
|
9
|
6350
|
February 12, 2024
|
Passing custom tensor using __torch_dispatch__ to nn.Parameter
|
|
0
|
92
|
February 11, 2024
|
Implement custom LayerNormalization layer for channel-wise normalization
|
|
1
|
850
|
February 9, 2024
|
Implementing calculation of the Laplacian
|
|
5
|
3441
|
February 9, 2024
|
Pruning nn.Linear weights causing unexpected errors
|
|
0
|
81
|
February 7, 2024
|
Function 'Scaled Dot Product Efficient Attention Backward0' returned nan values in its 0th output
|
|
11
|
611
|
February 6, 2024
|