FullyShardedDataParallel for inference

Hello. I have two 6GB GeForce 1660 cards, each one on separate machine (laptop and desktop PC). Please, tell me, can I use them together to inference single 6GB model (as it doesnt fit into single GPU’s VRAM)? Machines are connected via local area network. The model is called AutoDIR, it’s meant for denoising and restoration of images.