-
### 🐛 Describe the bug
Calling `.generate` on a HuggingFace model that has been FSDP wrapped results in an error. I was able to work around this error by summoning full params without recurse, which …
-
### 🐛 Describe the bug
The optimization scheduler `StopOnPlateau` does not break the optimization loop correctly when the linear solver failed.
Code
```python
kernel = Huber(delta=.1)
…
-
### 🐛 Describe the bug
When reading PyTorch documentation (and the base `LRScheduler` class's implementation of `state_dict`), it is clear that the intention behind a learning rate scheduler's `sta…
-
### 🐛 Describe the bug
Original complain is from internal users of torch.compile:
```
import torch
def fn(x):
return x.to(device="cuda", non_blocking=True)
inp = torch.randn(3, 4)
tor…
-
### 🐛 Describe the bug
When using compile mode, i am getting graph breaks due with following errors
"[rank0]:I0702 10:23:29.307000 139872906987520 torch/_dynamo/variables/higher_order_ops.py:468] …
-
### 🐛 Describe the bug
##Code
Encoder:
```
import torch
import torch.nn as nn
import torch.optim as optim
class Encoder(nn.Module):
def __init__(self, input_dim, emb_dim, hid_di…
-
### 🐛 Describe the bug
Hello, when I attempted to utilize the `torch_geometric.loader.DataLoader` object to load my dataset, I encountered the following user warning:
> /root/miniconda3/envs/chem/…
-
### 🐛 Describe the bug
I am using PyTorch's execution trace observer to collect traces about a basic MNIST program. While the program **does** produce valid data (everything is there), it **does not*…
-
### 🐛 Describe the bug
amp static shape cpp wrapper
suite
name
thread
batch_size_new
speed_up_new
inductor_new
eager_new
compilation_…
-
## 🐛 Bug
The base classes for `torch.utils.data.Dataset` and `torch.utils.data.Sampler` do not define a `__len__` method, which means that downstream code relying on PyTorch (with typehints) cannot…