Add files using upload-large-folder tool

README.md

@@ -1,48 +1,38 @@
 ---
-license: other
-license_name: cc-by-nc-4.0
-license_link: https://creativecommons.org/licenses/by-nc/4.0/
-base_model: facebook/nllb-200-distilled-600M
+license: cc-by-nc-4.0
+language:
+  - multilingual
 tags:
   - coreml
   - translation
-  - multilingual
+  - nllb
+  - on-device
   - apple-silicon
-  - quantized
+library_name: coreml
 ---
 
-# NLLB-200 distilled 600M — Core ML (seq 256, palettized / pal8, ANE-oriented)
+# NLLB-200 distilled 600M — Core ML KV (seq 256, palettized / pal8)
 
-Core ML packages for on-device translation with **KV-cache decoders**, exported from [`facebook/nllb-200-distilled-600M`](https://huggingface.co/facebook/nllb-200-distilled-600M).
+Core ML packages for **encoder** and **KV-cache decoder** (`init` + `step`) exported from `facebook/nllb-200-distilled-600M`, **sequence length 256**, **8-bit palettization (pal8)** for Apple Neural Engine–friendly deployment.
 
 ## Contents
 
-| Path | Role |
-|------|------|
-| `NLLB_Encoder_256.mlpackage` | Encoder, max sequence length **256** |
-| `NLLB_Decoder_256_init.mlpackage` | Decoder first step (EOS prefix) |
-| `NLLB_Decoder_256_step.mlpackage` | Decoder autoregressive step with KV |
-| `tokenizer/` | Tokenizer assets aligned with NLLB |
-| `config.json` | Architecture metadata (`seq_len: 256`, `ane_optimized: true`) |
-
-Weights use **8-bit palettization (pal8)** for smaller size and Apple Neural Engine–friendly deployment.
-
-## License
-
-The original NLLB distilled checkpoint is released under **CC-BY-NC 4.0** (non-commercial). This derivative export inherits the same restrictions. See the [NLLB model card](https://huggingface.co/facebook/nllb-200-distilled-600M).
-
-## Intended use
+- `NLLB_Encoder_256.mlpackage`
+- `NLLB_Decoder_256_init.mlpackage`
+- `NLLB_Decoder_256_step.mlpackage`
+- `config.json`, bundled `tokenizer/` (reference)
 
-- iOS / macOS apps via `CoreML` / `MLModel`
-- Offline or on-device NMT with FLORES language codes (e.g. `jpn_Jpan` → `eng_Latn`)
+## Base model
 
-Not for training or redistribution that violates the base model license.
+- [facebook/nllb-200-distilled-600M](https://huggingface.co/facebook/nllb-200-distilled-600M)  
+- Original NLLB-200 licensing applies (**CC-BY-NC**). This repo only redistributes a derived Core ML export for on-device use.
 
-## Hub snapshot
+## Usage notes
 
-Published as: [aoiandroid/nllb200-coreml-256-ane-pal8](https://huggingface.co/aoiandroid/nllb200-coreml-256-ane-pal8) (same token cannot create repos under the `PoliteAI` namespace until that org grants write access or you upload from the PoliteAI account).
+- Intended for **macOS / iOS** inference via `MLModel` (or Python `coremltools` for tests).
+- Source language is selected via tokenizer / `src_lang`; target via forced BOS (e.g. `eng_Latn`) in your app pipeline.
+- Quality and parity should be validated on your target devices (ANE vs CPU/GPU paths may differ).
 
-## Provenance
+## Organization
 
-- Source weights: `facebook/nllb-200-distilled-600M`
-- Format: Core ML `.mlpackage` (TranslateBlue export pipeline)
+Published by [PoliteAI](https://huggingface.co/PoliteAI).

config.json

@@ -40,6 +40,7 @@
     "LABEL_1": 1
   },
   "problem_type": null,
+  "_name_or_path": "/Users/am/.cache/huggingface/hub/models--facebook--nllb-200-distilled-600M/snapshots/f8d333a098d19b4fd9a8b18f94170487ad3f821d",
   "transformers_version": "5.3.0",
   "model_type": "m2m_100",
   "tokenizer_class": "NllbTokenizer",
@@ -47,4 +48,4 @@
   "seq_len": 256,
   "ane_optimized": true,
   "compute_units": "CPU_AND_NE"
-}
+}