Comparing training log [Shared my training log]

[charliezjw]

I am currently training the model, and I am hoping to compare it with others. I am only using only 2 A100-80G. Here is my wanb log: https://wandb.ai/charliezjw/huggingface/runs/hil1q6lt

[chavinlo]

https://wandb.ai/peruano/huggingface/runs/ei57qbzm/overview?workspace=user-peruano

[charliezjw]

I was using only 2 GPUs, the estimated total time is ~10hrs. I think yours is abnormally slow.

https://wandb.ai/peruano/huggingface/runs/ei57qbzm/overview?workspace=user-peruano

[ZeyuTeng96]

Hi, I do not why I ran into some problems about training cuz the trained model cannot generate relevant response in training set by providing similar instruction. Did you face this problem? this is my training flow: https://github.com/tatsu-lab/stanford_alpaca/issues/116