-
### š Describe the bug
I have a custom attribute defined in the saved tensors. But after unpacking, the attribute is deleted by the hook. How can I obtain the attribute?
```python
import torch
ā¦
-
### Your current environment
```text
PyTorch version: 2.2.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS: Ubuntu 22.04.4 LTS (x86_64)
GCC veā¦
-
### Your current environment information
Collecting environment information...
PyTorch version: 2.1.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/ā¦
-
### š Describe the bug
I am getting the following error when training LlamaForCausalLM with torch 2.1 and FSDP (mixed precision) and torch.compile. Same exact code works when torch.compile disabledā¦
-
### š Describe the bug
Need the `make_fx` to make it work with that autograd call but it's forcing a specialization on the shapes which hurts downstream deployment where shapes are varying . Any suā¦
-
### Your current environment
PyTorch version: 2.3.1+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS: Ubuntu 22.04.4 LTS (x86_64)
GCC version: ā¦
-
### š Describe the bug
[Recent change for `torch.cuda.is_bf16_compatible()`](https://github.com/pytorch/pytorch/commit/fc5fda14bcc41008e5610bf39c53eb066933ea4e) is now labeling Turing (sm_75) and Vā¦
-
### Your current environment
```text
Collecting environment information...
PyTorch version: 2.4.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
ā¦
-
### Your current environment
```text
PyTorch version: 2.3.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS: Ubuntu 22.04.2 LTS (x86_64)
GCC veā¦
-
### š Describe the bug
Thank you for all the dedication and effort put into this influential library.
I'm part of a team at NVIDIA, and we encountered some issues when intending to use the PyTorch Nā¦