-
pytorch-captcha-recognition git:(master) ✗ python captcha_train.py
init net
('epoch:', 0, 'step:', 0, 'loss:', 0.7110539674758911)
('epoch:', 1, 'step:', 0, 'loss:', 0.20382478833198547)
('epoch:…
-
### 🐛 Describe the bug
python code
``` python
class Net(nn.Module):
def __init__(self, device="cuda"):
super(Net, self).__init__()
self.expert_layer1 = nn.ModuleList([nn.Se…
-
### 🐛 Describe the bug
I have been trying to JIT trace Whisper, using this code:
```
from transformers import WhisperProcessor, WhisperForConditionalGeneration
from datasets import load_datase…
-
### 🐛 Describe the bug
Casting torch.sparse_coo_tensor to torch.sparse_csr_tensor with .to_sparse_csr() causes an integer overflow in nnz for sufficiently large tensors. See below example where COO t…
jvend updated
11 months ago
-
**MNIST Dataset**
The MNIST dataset is a large collection of handwritten digits that's widely used for training and testing in machine learning, especially in the area of image processing. Here's a…
-
### 🐛 Describe the bug
Flex attention on FSDP works without compile, but not with compile. The key error seems to be `ValueError: Pointer argument (at 2) cannot be accessed from Triton (cpu tensor?)`…
-
### 🐛 Describe the bug
The following program shows that torch.export.save doesn't support torch.float8_e4m3fn type, though torch.save of model and model state_dict is supported -
```
import torch…
-
### Describe the issue
Hi,
I have successfully exported a pytorch model via torch.onnx.dynamo_export.
Now i would like to start an onnxruntime Inferencesession to compare the results between the…
-
## 🐛 Bug
In some cases, torch.cdist returns non-zero (i.e. far from machine epsilon) diagonal values with CUDA.
Behaviour is as expected on CPU.
The issue seems more severe on Ampere GPUs.
## …
-
### 🐛 Describe the bug
As it says in the title, using `ParameterDict.items()` will break `torch.jit.script` as seen below
```
import torch
from torch import nn
class MyModule(nn.Module):
…
bigd4 updated
5 months ago