Create config.json

Add full metadata: base_model, tags, pipeline_tag

Update README.md

Create config.json

Simplify benchmark table

Fix benchmark results - 560ms is 2.12% WER not 0.49%

Update README with benchmark results for all chunk sizes

Clean up 1120ms: remove .mlpackage and float32 encoder

Replace chunk variants with int8 quantized .mlmodelc files

Add chunk-size variants: 560ms, 160ms, 80ms

Rename models to nemotron_coreml_1120ms (1.12s chunk window)

Upload 54 files

initial commit