|
The forward graphs captured by torch.export and aot_export_module are different
|
|
2
|
269
|
November 17, 2024
|
|
Dynamic slicing torch.export
|
|
2
|
863
|
November 16, 2024
|
|
Dynamo Trace with Parameter Lifting
|
|
1
|
271
|
November 16, 2024
|
|
How to limit torch.compile to CPU only?
|
|
3
|
575
|
November 1, 2024
|
|
Slow convolutions in triton
|
|
1
|
922
|
October 25, 2024
|
|
Does dynamo not support nested if?
|
|
3
|
151
|
October 17, 2024
|
|
Saving a model compiled with Torch-TensorRT with dynamic inputs
|
|
0
|
86
|
October 2, 2024
|
|
Fatal error: pyconfig.h: No such file or directory
|
|
3
|
2062
|
September 17, 2024
|
|
Unsupported FX node: <built-in function mod>
|
|
1
|
268
|
September 12, 2024
|
|
AssertionError: Dict types must use ConstDictVariable
|
|
1
|
282
|
September 9, 2024
|
|
Understanding dynamic shapes and guards and when it does/does not cause graph breaks
|
|
1
|
301
|
August 28, 2024
|
|
How to gracefully mask CompositeImplicitAutograd for different backends
|
|
0
|
174
|
August 26, 2024
|
|
Torch_tensorrt compile for a range of sizes with steps
|
|
1
|
109
|
August 17, 2024
|
|
Is there any way to get traced GraphModule without creating torch dynamo's constraint guards?
|
|
0
|
82
|
August 1, 2024
|
|
Any chance to torch.export TorchScript?
|
|
1
|
476
|
July 29, 2024
|
|
NvFuser with torch.compile
|
|
8
|
2008
|
July 11, 2024
|
|
Windows support timeline for torch.compile
|
|
6
|
7995
|
July 8, 2024
|
|
Issue with max-autotune mode in torch.compile
|
|
1
|
2090
|
May 30, 2024
|
|
Disk Quota Exceeded error
|
|
4
|
1497
|
May 30, 2024
|
|
Serving Torch_TersorRT converted model with Triton
|
|
0
|
208
|
April 25, 2024
|
|
Le: "b8[64, 2048, 7, 7]" = torch.ops.aten.le.Scalar(relu_48, 0); relu_48 = None
|
|
1
|
360
|
April 24, 2024
|
|
Torch.compile cache_size_limit best practice
|
|
1
|
1495
|
April 12, 2024
|
|
Export and save problem（torch-tensorrt）
|
|
0
|
372
|
April 11, 2024
|
|
How to achieve Torch-MLIR compatibility with Dynamo backend?
|
|
1
|
561
|
April 9, 2024
|
|
[dynamo] [onnx] Models exported with torch.onnx.dynamo_export show worse performance during inference
|
|
0
|
434
|
April 8, 2024
|
|
Unique node name across graph re-compilations and graph breaks
|
|
0
|
310
|
April 3, 2024
|
|
Timing model modules
|
|
0
|
262
|
March 6, 2024
|
|
Avoid loop unrolling with TorchDynamo for memory savings
|
|
0
|
383
|
March 4, 2024
|
|
Can save optimized_model created with torch_tensorrt.compile
|
|
3
|
1807
|
February 26, 2024
|
|
Torch.compile regularly fails with Triton Error [CUDA]: device kernel image is invalid
|
|
2
|
4427
|
February 23, 2024
|