-
-
## 🚀 Feature
Expose dataloaders to the `LightningModule`'s `setup` method.
### Motivation
This will allow for a real dynamic setup, meaning that some layers' size can be set up correctly than…
-
Hello,
As per [this line](https://github.com/aqlaboratory/openfold/blob/main/openfold/model/evoformer.py#L649), when neither Deepspeed nor LMA is selected, the custom memory-efficient [kernel](http…
-
Hi again,
I've noticed that are some visual effects and sound effects are missing in your port.
I've recorded 2 videos showing that I say.
- This happen with the old build too (before cvar fix)…
-
I actually encountered a similar scenario.
The standard Huggingface [bert-base-cased](https://huggingface.co/bert-base-cased/blob/main/config.json) model trained with 16 bit mixed precision (using …
-
This one is really confusing to me. When using PyTorch lightning and loading checkpoints from OCI, I occasionally hit this weird error:
```
Traceback (most recent call last):
File "", line 198,…
-
### Description & Motivation
I'm trying to get the NeMo multimodel imagen example to use thunder, but Lightning itself does not support Thunder:
```
Error executing job with overrides: ['traine…
-
## 🚀 Feature
### Motivation
After reading through PyTorch codebase, I came across this code example:
https://github.com/pytorch/pytorch/blob/538647fe1fb94b7822ea3b8bbbd6901961431d60/torch/f…
-
In PyTorch, we know that Torch.Compile will bring us a lot of benefits, and the TransformerEngine also brings performance improvements through strategies such as Transformer fusion optimization, so do…
-
## 🚀 Feature
Integrate https://github.com/pytorch/torchsnapshot
### Motivation
The library is design with composition in mind and is very modular.
The distributed training benchmarks look ve…