SheetSage-A2LS model checkpoints

Model checkpoints for Audio-to-Lead-Sheet Transcription (A2LS), the retrained models described in our paper (under review). Each checkpoint greedily decodes a Humdrum **kern lead sheet (melody **kern, chords **cdata, lyrics **text) from pre-computed MuQ features.

file	lyrics tokeniser (`--lyrics_tokeniser`)
`cummins_retrained.ckpt`	`none`
`word_tok.ckpt`	`word`
`char_tok.ckpt`	`character`
`bpe_tok.ckpt`	`bpe` (GPT-2)

Inference and evaluation code: GitHub repository.

Weights are released under CC-BY-NC-SA 4.0.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support