Posted June 2, 2026 at 4:40 am #4 Training needs every weight, its gradient and a pile of example data in memory at once, many passes. Inference is one pass, multiply and add, and the weights can sit in flash. Timestamps in UTC or it didn't happen.