-
CLDR collations are configured per-locale (typically per-language in reality) in a set of [configuration files](https://github.com/unicode-org/cldr/tree/main/common/collation). These files need to be…
-
Hello,
I have an issue with multiple GPU performance.
- I use the recipe `lora_finetune_single_device` with the config `mini_lora_single_device.yaml` on 6000ADA, I got ~5it/s
- I use the recipe `lo…
-
I was running NeMo on a 1 hour wav file, with stemming turned on(demcs). whisper and alignment runs fine but when it enters diarization, I encounter the below error.
The same file runs fine end-to-e…
-
(sam3d) a@a-Super-Server:/media/a/DATA/ssw-baselines/SAM-Med3D$ python3 train.py
Loaded checkpoint from ckpt/sam_med3d.pth (epoch 0)
Epoch: 0/199
0%| …
-
Hello there,
Thanks for the great work!
I believe there is a bug for multi-aspect ratio training. When I train on multiGPUs and batch != 1, here is the error msg:
for step, batch in enumer…
-
### 🐛 Describe the bug
When we use default collate function in https://github.com/pytorch/pytorch/blob/217aac96d779841666527402fc113493b6bd6323/torch/utils/data/_utils/collate.py#L171
If we have two…
-
To shorten the feedback loop between alignment, upload, and reviewing of data in case there are faulty observations, @phraenquex suggests that we need some UI to be able to see the paths to the origin…
-
I'm working on a series of data augmentations for text and have implemented their use via the `torch.utils.data.DataLoader.collate_fn` but `jiant` appears to be using this function already. Is there a…
-
```
What steps will reproduce the problem?
1. Allow users the option to send data to yourselves on the first three values
of the MAC address of attacked AP's with the variables used to attack that AP…
-
```
What steps will reproduce the problem?
1. Allow users the option to send data to yourselves on the first three values
of the MAC address of attacked AP's with the variables used to attack that AP…