Probably not with the same amount of training time, but I'd imagine a recent MBP GPU could handle GPT-2 training. The biggest challenge is that the training would need to be reimplemented for Metal instead of CUDA.
M3 Max is 28 TFLOPS at FP16.
Based on FLOPS alone, it would be more like a year or two.
https://www.amd.com/en/products/accelerators/instinct/mi300/...