congrats on great work
I am a master's student researching depth estimation. Since our group is not well-funded and only has eight 4090 GPUs, I need to share one 4090 GPU with another team member. Because his model training only takes five hours, I occasionally pause my training, let him run his five-hour training session, and then resume my own training.
I want to ask whether frequently pausing training and then resuming from checkpoints affects the final training results.
congrats on great work
I am a master's student researching depth estimation. Since our group is not well-funded and only has eight 4090 GPUs, I need to share one 4090 GPU with another team member. Because his model training only takes five hours, I occasionally pause my training, let him run his five-hour training session, and then resume my own training.
I want to ask whether frequently pausing training and then resuming from checkpoints affects the final training results.