Amazon rolled out fine-tuning and distillation for Vision LLMs like Nova Lite via Bedrock and SageMaker. Translation: better doc parsing—think messy tax forms, receipts, invoices.
Developers get two tuning paths: PEFT or full fine-tune. Then choose how to ship: on-demand inference (ODI) or Provisioned Throughput. Real-world results? Up to 39% better precision, with full recall on key fields.
