Unlocking Efficient Large Inference Models: One-Bit Unrolling Tips the Scales