gbnam8/jp-music-title-encoder-teacher

This model was mostly vibecoded by Claude (Anthropic's AI assistant). The data pipeline, training scripts, evaluation and this card were largely written by Claude under light human direction. The numbers come from real runs, but nobody has carefully audited the code, the data or the methodology. Treat it as an experiment, not a vetted model.

JP music title encoder — teacher (v7)

The 12-layer teacher of gbnam8/jp-music-title-encoder: multilingual-e5-small fine-tuned for music entity retrieval (JP/EN titles, artist names, platform wrappers, artist-conditioned covers). Use the 6-layer student for inference; it scores the same or higher (artists recall@20 @511k: student 0.828, teacher 0.817) at half the cost. This checkpoint is for further fine-tuning or re-distillation.

Input format and evaluation: see the student's card.