Latency

Hardware resources used for training/inference; constrained by memory bandwidth, FLOPs, and parallelism.

Techniques that fine-tune small additional components rather than all weights to reduce compute and storage.

PEFT method injecting trainable low-rank matrices into layers, enabling efficient fine-tuning.

Reducing numeric precision of weights/activations to speed inference and reduce memory with acceptable accuracy loss.

Structured dataset documentation covering collection, composition, recommended uses, biases, and maintenance.

Central system to store model versions, metadata, approvals, and deployment state.

Ability to replicate results given same code/data; harder in distributed training and nondeterministic ops.

Tracking where data came from and how it was transformed; key for debugging and compliance.