lora

Separate packed base weights, quantization metadata and trainable adapter state. Count the real projection shapes, then leave room for activations and temporary work before calling a one-card configuration feasible.

Follow adapter-specific token rows through a shared base, calculate stored adapter bytes, and separate the resident set from the active batch.

Group compatible adapters by pinned base, derive their size from target matrices and ranks, and benchmark mixed batches. Catalogue size, CPU residency, GPU slots and active adapters are separate limits; base sharing does not make fifty variants operationally free.

Separate frozen weights from trainable state, count rank-16 adapters from matrix shapes, and check peak memory before choosing a fine-tuning layout.

Choose the tenant boundary before planning base-model reuse. Account for frozen weights, adapter state and activations separately, then schedule by compatibility, fairness and the real cost of restarting training.

Follow the matrix shapes through a low-rank update. Verify merged inference numerically and show why multiplying sums of adapter factors gives the wrong weights.