Modelpublic

aoiandroid/nllb200-coreml-256-ane

sourceHugging Facecc-by-nc-4.0updated 6mo agoView on Hugging Face

0likes55downloads

Model Card

NLLB-200 distilled 600M — Core ML (seq 256, ANE, float32)

Core ML .mlpackage bundles for encoder and KV-cache decoder (init + step), exported from `facebook/nllb-200-distilled-600M`. Fixed sequence length 256, ANE-oriented graph, float32 weights (no palettization in this revision).

Artifact	Role
`NLLB_Encoder_256.mlpackage`	Encode source tokens
`NLLB_Decoder_256_init.mlpackage`	Decoder first step (no past KV)
`NLLB_Decoder_256_step.mlpackage`	Autoregressive step with past KV
`config.json`	Architecture metadata (`seq_len`, `compute_units`, etc.)
`tokenizer/`	Reference tokenizer files for the app pipeline

Companion variant

—Palettized (pal8) weights (smaller on disk, different quality/latency tradeoffs): `aoiandroid/nllb200-coreml-256-ane-pal8`

Config highlights (`config.json`)

—model_type: m2m_100 / M2M100ForConditionalGeneration
—seq_len: 256
—ane_optimized: true
—compute_units: CPU_AND_NE
—dtype: float32

Base model and license

—Source weights: NLLB-200 distilled 600M (Meta / FAIR).
—Original terms are CC-BY-NC 4.0. This repository only ships a derived Core ML export for on-device use; the same license restrictions apply.

Intended use

—macOS / iOS on-device translation via MLModel (or coremltools in Python for validation).
—Language direction is controlled in your app: tokenizer src_lang, forced target BOS (e.g. eng_Latn), and your beam/greedy loop around Decoder_256_step.
—Validate quality and performance on your target hardware (ANE vs CPU/GPU behavior can differ).

Limitations

—Fixed max sequence length 256 at export time; longer inputs require truncation or a different export.
—Not a drop-in Hugging Face transformers checkpoint; this is a split Core ML deployment (encoder + decoder packages).

Maintainer

`aoiandroid`