-
### Your current environment
Note: The vLLM is built from source by the latest main branch.
```text
Collecting environment information...
PyTorch version: 2.3.0+cu121
Is debug build: False
C…
-
### 🐛 Describe the bug
I am training in cuda with large batchsize=5000 and observed that my cpu memory is increasing when monitored via htop. Gpu memory is stable with nvidia-smi.
Since my simple …
-
### 🐛 Describe the bug
new_perf_regression in 2024-03-23
suite
name
thread
batch_size_new
speed_up_new
inductor_new
eager_new
compila…
-
### 🐛 Describe the bug
I am getting large discrepancies between PyTorch inference and ONNXRuntime inference for a number of models. I am unsure why this is the case. See below:
| model …
-
### Your current environment
The output of `python collect_env.py`
```text
Collecting environment information...
PyTorch version: 2.4.0+cu121
Is debug build: False
CUDA used to build PyTorch…
-
### 🐛 Describe the bug
```python
input = torch.tensor([-48.4])
other = 1.1
print(torch.remainder(input, other))
```
Here result is coming as
```
tensor([1.1000])
```
-44 * 1.1 = -48.4, he…
-
### Description of the bug | 错误描述
### How to reproduce the bug | 如何复现
pip install magic_pdf-0.7.0a1-p…
dt-yy updated
3 months ago
-
I'm getting an error each time I try to run the training process. For context, I'm using Windows 11 with an RTX 3090, and both A111 and ComfyUI work fine on my machine.
OneTrainer installs fine wit…
-
### 🐛 Describe the bug
In compile mode, following error is seen for torch.norm_except_dim() operator
``` python
import torch
def norm_except_dim(v, pow, dim):
return torch.norm_except_dim…
-
### 🐛 Describe the bug
We aim to apply `pt2e quantization` for LLM models, but encounter challenges in capturing graphs with inputs of dynamic shape despite setting constraints in `capture_pre_aut…