-
Training and data
Sabich v2.3: a fifth more recordings, and how it scored
Sabich v2.3 is the same recipe as v2.2 on 1,074.3 minutes of the speaker's takes. The training data, the held-out results against v2.2, and where the data and eval files are.
By Daniel Rosehill and Opus 5.5 · Read → -
Training and data
Code-switching and domain vocabulary in one ASR fine-tune: can you have both?
Does adding tech vocabulary to a code-switching ASR fine-tune cost accuracy on the Hebrew words? A controlled comparison on the same data, and why it set the Sabich formula.
By Daniel Rosehill and Opus 5.5 · Read → -
Explainers
Live dictation without a streaming model: how Parakeet gets sentence boundaries right
Why the apps recommend Parakeet-based Sabich over the Nemotron streaming model for dictation, and the work at the joins that fixed sentence boundaries, with measurements.
By Daniel Rosehill and Opus 5.5 · Read → -
Training and data
Sabich v2.1 on the new benchmark: what changed, and what it scored
Stock Parakeet, Jachnun and every Sabich version on the new domain-balanced test set v2: what changed in v2.1's training, the results, and which differences are within noise.
By Daniel Rosehill and Opus 5.5 · Read → -
Explainers
What "word error rate" means, and why we also count Hebrew words
A plain-English guide to the two numbers on the model card: word error rate, and how often the Hebrew words come out right.
By Daniel Rosehill · Read →
Follow along: RSS feed · or get email updates.