BWhat changed
Scaled filtered web and synthetic training data to produce compact general-purpose language models with on-device deployment experiments and documented post-training.
Scientific scope
Scaled filtered web and synthetic training data to produce compact general-purpose language models with on-device deployment experiments and documented post-training. Builds on the already-covered Phi-1 textbook-data approach rather than inventing synthetic-data training. Generalization, phone latency and benchmark parity are configuration-specific. Later Phi3.5/MoE additions are excluded from the original v1 record.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Builds on the already-covered Phi-1 textbook-data approach rather than inventing synthetic-data training. Generalization, phone latency and benchmark parity are configuration-specific. Later Phi3.5/MoE additions are excluded from the original v1 record.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
- Microsoft: Phi-3 data-curated compact language-model system — author/company record
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction