-
### 🐛 Describe the bug
import numpy as np
import torch
data = torch.tensor(np.array([[1, 2], [3, 4], [5, 6]], dtype=np.float32))
q1 = torch.quantile(data, 0.25, dim=0) # works
data = torch.tens…
-
### 🐛 Describe the bug
We observed a NaN regression with 2.5.0, and traced it to CUDNN attention.
2.5.0:
:
class M(torch.nn.Module):
def forward(self, x, y, z):
a = y.shape[0]
…
-
### 🐛 Describe the bug
The process is working correctly with DDP world size 1 but then with world size > 1 is going to hang with GPU 0 at 0% and GPU 1 fixed to max occupancy. I've replicated this bot…
-
### Your current environment
ValueError:Could not broadcast input array from shape (542,) into shape (512,)
```text
Collecting environment information...
WARNING 11-03 10:23:50 _custom_ops.py:…
-
### 🐛 Describe the bug
```
File "/mnt/clusterstorage/workspace/kevin/ml-monorepo/chadfusion/train_fsdp.py", line 363, in fsdp_train
scaler.step(opt)
File "/usr/local/lib/python3.10/dist-…
-
### 🐛 Describe the bug
`output = model(input_data)` has no output, don't return.
```
import os
import torch
import torch.nn as nn
from models import encoders
os.environ['CUDA_VISIBLE_DEVICE…