bin/python: relocation error: /lib/python3.11/site-packages/torch/lib/libcudnn_cnn_infer.so.8: symbol _ZN15TracebackLoggerC1EPKc version libcudnn_ops_infer.so.8 not defined in file libcudnn_ops_infer.so.8 with link time reference
|
|
1
|
279
|
January 23, 2024
|
How to only make a change to a submodule and then update Pytorch
|
|
3
|
268
|
January 22, 2024
|
C++/Cuda / aot.compile and cuda graph
|
|
0
|
172
|
January 22, 2024
|
Cannot Find ptxas
|
|
6
|
593
|
January 21, 2024
|
How to customize leaf node in dynamo trace?
|
|
0
|
236
|
January 19, 2024
|
How can Pytorch 2.x support different backends?
|
|
2
|
188
|
January 19, 2024
|
Graph breaks or No graph breaks in torchdynamo
|
|
3
|
1457
|
January 18, 2024
|
Where to PrimTorch's operator list?
|
|
2
|
217
|
January 18, 2024
|
Ignore a inner function with @torch.compile
|
|
1
|
324
|
January 18, 2024
|
UserDefinedObjectVariable should not be break renson
|
|
2
|
580
|
January 18, 2024
|
torch._dynamo.exc.Unsupported: Tensor.backward
|
|
6
|
1552
|
January 17, 2024
|
Easy way to check which guard fails in check_fn()
|
|
0
|
140
|
January 17, 2024
|
If I use torch.compile to compile the whole graph,in the my own compiler, how to manage the memory in my own compiler?
|
|
0
|
189
|
January 16, 2024
|
Dynamic sequence length and recurrent layers
|
|
4
|
568
|
January 16, 2024
|
Graph exported by torch.onnx.dynamo_export
|
|
1
|
302
|
January 13, 2024
|
How to get the node name mapping between FX and ONNX as far as possible?
|
|
0
|
237
|
January 12, 2024
|
Type information for arguments of ATen operations
|
|
0
|
160
|
January 11, 2024
|
Torch.compile max-autotune To prevent overwriting, clone the tensor outside of torch.compile() or call torch.compiler.cudagraph_mark_step_begin() before each model invocation
|
|
0
|
375
|
January 9, 2024
|
Torch.compile: AttributeError: module 'z3' has no attribute 'ExprRef'
|
|
6
|
601
|
January 8, 2024
|
Lib conflicts with ray and spacy
|
|
3
|
190
|
January 8, 2024
|
How to get whole graph when subgraph_compile triggle
|
|
2
|
701
|
January 8, 2024
|
How to avoid graph break when i use `UserDefinedClassVariable`
|
|
0
|
201
|
January 8, 2024
|
Stitching together graph breaks for large "compilation units"
|
|
4
|
391
|
January 4, 2024
|
What's the best way to serialize and reuse torch.compile'd functions?
|
|
3
|
320
|
January 4, 2024
|
Skip a submodule that cannot be compile
|
|
3
|
255
|
January 4, 2024
|
What functions like torch._C._DisablePythonDispatcher do?
|
|
2
|
268
|
January 4, 2024
|
Exact code line for checking if node op is instance of torch.ops.OpOverload
|
|
3
|
211
|
January 4, 2024
|
Torch.compile error for recurrent model
|
|
1
|
361
|
January 3, 2024
|
Torch.Compile different fx graph for inductor and custom backend
|
|
0
|
252
|
January 3, 2024
|
What's the relation between torch.export.export and torch._dynamo.eval_frame.export?
|
|
0
|
164
|
December 28, 2023
|