compile

torch._inductor The default compiler in torch.compile, builds loops schedules and codegens triton or C++ kernels. FX (Functional Transformations) FX is a toolkit for developers to use to transform nn.Module instances. FX consists of three main components: a symbolic tracer, an intermediate representation , and Python code generation . torch._dynamo PyTorch’s compiler frontend which can output subgraphs when programs aren’t traceable.

Topic	Replies	Views	Activity
About the compile category compile	0	253	January 9, 2023
Model trained with PyTorch 2.0 and Multi GPUs does not work with one GPU compile	4	39	June 8, 2023
How to pass include (-I) options for torch.compile compile	1	28	June 8, 2023
PyTorch compile requires fixed input naming? compile	3	61	June 6, 2023
How to use `torch.compile` with CUDA graphs when using gradient activation checkpointing compile	3	159	June 5, 2023
What can cause "[WARNING] DeviceCopy in input program"? compile	9	721	June 2, 2023
Graph breaks or No graph breaks in torchdynamo torch._dynamo	1	49	June 1, 2023
Torch.compile's deployment story to non-Python host processes torch._dynamo	3	73	May 30, 2023
Invalid use of incomplete type ‘PyFrameObject’ {aka ‘struct _frame’} compile	1	81	May 30, 2023
Any wheel for CUDA 12 coming? compile	8	791	May 27, 2023
Is there a way to get QR supported through magma for torch+rocm5.4.2? compile	0	48	May 26, 2023
Where is bmm source code? When using Inductor backend, it is not generating Triton kernels? compile	1	41	May 26, 2023
Ways to pass compilation target device to dynamo backend compile	1	43	May 24, 2023
[TorchDynamo] How to generate fx graph with nn.* layers? torch._dynamo	2	67	May 23, 2023
Dynamic sequence length and recurrent layers compile	3	142	May 23, 2023
Does `torch.func.functionalize` support `fx.GraphModule` from `dynamo.export` w/ aten graph? compile	2	149	May 23, 2023
How to check PyTorch2 model is compiled compile	4	91	May 22, 2023
Torch won't compile with venv python installs compile	0	57	May 22, 2023
State_dict after compile and onnx.export compile	2	93	May 19, 2023
Can torch.compile support a model with dict holding data from hooks? compile	1	113	May 18, 2023
Example where PyTorch substantially slower than C++/CUDA compile	4	181	May 17, 2023
Tensors of variable size compile	1	64	May 16, 2023
Compile pytorch 2.0 with backend support for opencl opengl and vulkan compile	0	135	May 15, 2023
FSDP + compile says "RuntimeError: assigned grad has data of a different size" compile	1	66	May 13, 2023
Assertion Error with torch.compile torch._dynamo	3	122	May 12, 2023
How to copy weights from torch.nn.sequential weights from source class to target class compile	0	52	May 10, 2023
Torch compile change the module name compile	3	79	May 9, 2023
How should I use torch.compile properly? compile	5	289	May 8, 2023
What's the recommended way to achieve asynchronicity for dynamo backend? FX (Functional Transformations)	2	109	May 8, 2023
torch.cuda.DeferredCudaCallError: A strange bug after building a cuda extension compile	3	81	May 8, 2023