Masahiro Wada
Masahiro Wada
Hi! I want to take the followings. * pytorch_lightning/callbacks/finetuning.py * pytorch_lightning/tuner/batch_size_scaling.py
Is there anyone to look into it? If no one, may I try to look into it?
@krshrimali Thank for your kindness. I'm going to get to this issue.
Hi, after I looked into this issue, I found the cause of it. It doesn’t depend on lightning-flash. It occurs with only pytorch-lightning. And there is some issue like this...
Hi, after my investigation, I found that finetuning using pytorch-lightning can not work with DeepSpeed correctly. Concretely, loading its parameters doesn't work when resuming training. As I commented in [this...
@ethanwharris Thanks for your suggestion! I'm so interested in creating a PR about the `no_freeze_deepspeed` strategy. And the `freeze` strategy doesn't work because of the same reason as `no_freeze`. So...
I don't know how to fix `Lightning-AI.lightning-flash (Examples)` jobs. Could you give me some help?
Hi @krshrimali Thanks for your review and suggestions! It's so helpful for me. Because my English is poor. I'm so glad that this PR was approved.
I checked the reason for `Lightning-AI.lightning-flash (Examples)` and found the follwoings. ``` RuntimeError: CUDA error: the launch timed out and was terminated CUDA kernel errors might be asynchronously reported at...
Hi @oneandonlyoddo Thanks for your comment. I'm sorry for unhelpful document about this functions. I ran simple_event and resize_image and then they worked correctly. Please check the following steps. 1....