Bing Xie
Bing Xie
@yerimChoi, can you please provide more details about your code and the software/hardware environment you're running on? thanks.
@lpty , I see you and @tjruwase are discussing the errors in https://github.com/microsoft/DeepSpeed/issues/2977. And seems the OOM issue only showed on WSL2. Please confirm these, if it is only on...
close for no response from the user, will reopen if needed.
> Can you please add some unit tests for this? I am happy to chat about them. changed printout format and add a unit test
@molly-smith, seems you started working on this issue, any update?
@Modas-Li , are you using 1 GPU or multiple GPUs? If you're using 1 GPU, please try to increase the number of GPUs to see if it helps.
the software issue is closed and will open again if necessary.
the software issue has been solved.
please check [tutorial of zero optimizers] (https://www.deepspeed.ai/tutorials/zero/).
Yes, My question is: can I generate KJT on GPUs directly? The reason I want to do so is I want to reduce the data size transferred via PCIe, I...