Phi-3

2024-04 · text · generative · anchor

Small language model family trained on heavily filtered and synthetic data, cited in the corpus at the Medium size and in the 3.5 point release at Mini and mixture-of-experts sizes.

Note
the 3.5 releases are carried as variants rather than as a separate model, because the citing papers reference them against this same technical report
Variants
Phi-3 Medium, Phi-3.5, Phi-3.5 Mini, Phi-3.5 MoE, Phi-3-Medium-128K-Instruct, Phi-3-Medium-4K-Instruct, Phi-3-mini Instruct, Phi-3-mini-128k-instruct, Phi-3-small, Phi-3-Vision-Instruct, Phi-3-3.8B-Instruct, Phi-3-3.8B-IT

Findings

Shared mechanisms