YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Talk grammar + paragraphs (Core ML)

PUSHTYPE's on-device grammar and paragraph model for Talk. A GECToR-style edit tagger (Omelianchuk et al. 2020, our own reimplementation) on RoBERTa-base (MIT), with a paragraph-break head on the same encoder, one pass for both. Trained by PUSHTYPE on C4_200M edits (CC BY 4.0; text ODC-BY) and synthetic spoken-style pairs; paragraph breaks learned from C4 documents. Edit vocabulary and verb forms from GECToR (Apache 2.0).

Files: TalkGrammar.mlmodelc (int8 weights; inputs input_ids and attention_mask, 1 x 128; outputs best_id, best_p, p_incorrect, p_paragraph per piece), vocab.json and merges.txt (RoBERTa byte-level BPE plus $START = 50265), labels.txt (tag per id), verbs.tsv (verb-form transforms).

License: MIT.

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. ๐Ÿ™‹ Ask for provider support