Afrin Dange comments

Repositories
Issues
Comments

Results 4 comments of


                                            Afrin Dange

Distributed Training Gets Stuck?

@muellerzr, did you find a workaround for this? I am facing the same issue.

Shared tensors not correctly saved.

Hi @GabPrato, I had the same issue with Accelerate while using for single GPU. Using, `safe_serialization=False` in accelerate.save_state() resolved it.

RuntimeError: 'weight' must be 2-D while training Flan-T5 models with stage 3

I am facing the same issue with the DPR and GPT2 models. I am using the latest torch version to use FullyShardedDataParallel for distributed training. The training works fine (regardless...

Pretrained models missing

Hi, I was looking for the same. Checkout this issue https://github.com/malllabiisc/EmbedKGQA/issues/69