-
While running a llama2 pretraining script with specific configurations, I encountered an illegal memory access error. The detailed error message is as follows:
```
[2023-08-09 07:56:18,503] [INFO]…
-
### 🐛 Describe the bug
When I have some network before main model, my memory of gpu has not reduced. As follows,
with torch.no_grad():
z = autoencoder.encode(img)
engine.zero_grad()
…
-
The requirement is like if I am receiving 2 packets with interval of 10 micro second (example value), then xdp(kernel) has to be time stamped the packet with 10 micro second gap, what is currently obs…
-
Since fullloading is very time consuming and often expensive, it is neccesary to introduce Autoloader.
-
### Problem description
This way you could potentially completely stream a dataframe. Even when it's larger than life (/memory) itself.
-
Hey, we really appreciate this library, it resolves a number of issues we experienced when trying the default kinesis connector in Glue. We have a question about the behaviour of the library for a hig…
-
### 🐛 Describe the bug
I am trying to reproduce the resnet50-pipeline parallel demo in this page: https://colossalai.org/docs/features/pipeline_parallel
I can go well with the code in this page.…
-
Hi there, I have about 10 bracelets of EU model Cement 1.1 that I'm trying to test this library with (fresh batteries by the way). I'm using a Mega 2560 with no level shifter.
I'm having some small…
-
Hi, I am looking for GPipe source code, but only found this: https://github.com/tensorflow/lingvo/blob/master/lingvo/core/gpipe.py
Maybe I am wrong but I think the source code of GPipe should inclu…
-
@fsalem
I am using your spark-kafka writer for my spark streaming application, and I am getting an error with "too many open files" problem.
What is the proper way to close kafka producers?
`…