I build a decompiler to convert bytecode generated by dynamo into readable source code!
|
|
4
|
514
|
August 29, 2023
|
CUDAGraphs in Pytorch 2.0
|
|
3
|
2316
|
August 11, 2023
|
Supporting Dynamo in Python 3.11 - CPython Frame Evaluation
|
|
0
|
816
|
August 8, 2023
|
How are guards installed on frames that are transient objects?
|
|
9
|
443
|
August 6, 2023
|
How to capture NCCL communication ops in FakeTensorMode?
|
|
3
|
552
|
August 3, 2023
|
Min-cut optimal(*) recomputation (i.e. activation checkpointing) with AOTAutograd
|
|
10
|
12238
|
July 28, 2023
|
Supporting Dynamo in Python 3.11 - NULL
|
|
0
|
1101
|
July 24, 2023
|
Can we add a default backend when OpenMP is not available?
|
|
5
|
582
|
July 21, 2023
|
Questions about Inductor code generation for gemm on Nvidia device
|
|
4
|
906
|
July 21, 2023
|
[Fatal Bug] changed nn.Module.training does not trigger recompilation
|
|
0
|
276
|
July 20, 2023
|
What is the relation between PrimTorch and Decompostions?
|
|
6
|
1496
|
July 7, 2023
|
Why is silu_backward decomposed in fx graph trace?
|
|
4
|
959
|
June 21, 2023
|
Dynamo allow in graph fail on tensor.view(dtype)
|
|
1
|
449
|
June 17, 2023
|
Backward module does not contains weight's gradients calculation
|
|
0
|
346
|
June 16, 2023
|
Lengths' argument should be a 1D CPU int64 tensor, but got 1D meta Long tensor
|
|
1
|
442
|
June 12, 2023
|
Inductor updates
|
|
0
|
987
|
June 1, 2023
|
Add a new compiler backend for jit.trace
|
|
1
|
507
|
May 25, 2023
|
How to differentiate target devices in dynamo backends?
|
|
1
|
467
|
May 24, 2023
|
Torch.compile support for Python 3.11 completed
|
|
0
|
1101
|
May 23, 2023
|
A pure Python implementation of roi_align that looks just like its CUDA kernel
|
|
4
|
1624
|
May 14, 2023
|
What is the split strategy for large-dimensional reduction operations in triton code generation?
|
|
1
|
410
|
May 14, 2023
|
viewOp not getting functionalized for the LTC backend
|
|
2
|
518
|
May 8, 2023
|
Major updates to the after AOT accuracy minifier!
|
|
0
|
807
|
May 3, 2023
|
Is new torch.compile() able to compile training model?
|
|
1
|
505
|
April 25, 2023
|
How to read the parameter info
|
|
1
|
385
|
April 20, 2023
|
PrimTorch: How backend/compiler writers interact with various IRs
|
|
2
|
1536
|
April 11, 2023
|
Functionalization in PyTorch: Everything You Wanted To Know
|
|
2
|
4181
|
April 7, 2023
|
Debugging story: The case of the garbage text generation
|
|
0
|
1848
|
March 30, 2023
|
How is triton_mm called
|
|
1
|
545
|
March 29, 2023
|
TorchInductor CPP Backend Vectorization Status Analysis
|
|
0
|
741
|
March 27, 2023
|