AWS published a two-part guide on preparing and improving datasets for supervised fine-tuning
Other Provisional 72% confidence first seen
AWS Machine Learning released two articles describing how to audit raw data quality, format supervised fine-tuning examples (e.g., as JSONL), and apply strategies such as filtering, learning-curve analysis, and augmentation to improve training effectiveness. The coverage focuses on dataset curation and experiment planning rather than reporting a new model or product release.