C2W4:Issues with gradient calculation in Lab 2

The loss calculated is different for each device therefore so are the gradients doesn’t this mean that all the models across devices train differently here.

Multi GPU Mirrored Strategy | Coursera

is this a graded assignment link? @Aaryan2

ungraded Lab 2

Thank you for your response. @Aaryan2 Yes you have the correct understanding that every models trains differently based on GPU based on its processor configuration and capabilities.