grammar-en is an open-weight model from Kotoshu, released under Creative Commons Attribution-NonCommercial-ShareAlike 4.0. Its published files total 87.3 MB.
An 85.8 MB int8 ONNX token tagger that detects and corrects grammatical errors fully locally — no data leaves the machine. GECToR architecture (Omelianchuk et al.
Model Card
An 85.8 MB int8 ONNX token tagger that detects and corrects grammatical errors fully locally — no data leaves the machine. GECToR architecture (Omelianchuk et al. 2020): distilroberta encoder, 5,000-label closed vocabulary ($KEEP / $REPLACE / $APPEND / $DELETE / $TRANSFORM / $MERGE), iterative apply-and-retag decoding (4 rounds). Scored with the official M2 scorer, noalt gold, --maxunchangedwords 2, 4-round iterative decode: 2.9x LanguageTool; runs in ~6 ms/sentence on a laptop CPU through onnxruntime. labels.json (id -> label map — do not rebuild), vocab.json + merges.txt (distilroberta byte-level BPE). label vocab = top 5,000 from the training mix. real-time (~6 ms/sentence), 85.8 MB int8…
Excerpt from the card by Kotoshu, licensed cc-by-nc-sa-4.0.
Identity and Version
- Repository
- kotoshu/grammar-en
- Publisher
- Kotoshu
- Task
- Not stated by the source
- Modality
- Other
- Library
- onnxruntime
- Parameters
- Not stated by the source
- Languages
- en
- Revision
- f7dd722b1e42c62e290920ed4f22a3f5d0bdca79
- First published
- 2026-10-05
- Last updated
- 2026-10-12
Files and Weights
7 files, 87.3 MB in total. The weights are 1 file totalling 85.8 MB in onnx.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| gec-tagger.int8.onnx | Weights | 85.8 MB | aeaf388ec564 |
| catalog_index.json | Configuration | 864 B | — |
| labels.json | Configuration | 141.1 KB | — |
| README.md | Documentation | 3.4 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
| merges.txt | Tokenizer | 456.3 KB | — |
| vocab.json | Tokenizer | 898.8 KB | — |
License and Download
- License
- cc-by-nc-sa-4.0
- Access
- Open weights, no gate
- Download size
- 85.8 MB
Released by Kotoshu through its official repository on Hugging Face.
Built From
- Trained on (disclosed) c4
- Trained on (disclosed) clang8
- Trained on (disclosed) coedit
- Trained on (disclosed) fce
- Trained on (disclosed) nucle
- Trained on (disclosed) wi-locness
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 85.8 MB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About grammar-en
Can I use grammar-en commercially?
Not without separate permission. grammar-en is released under Creative Commons Attribution-NonCommercial-ShareAlike 4.0. CC BY-NC-SA 4.0 permits non-commercial sharing and adapting with credit, and requires adaptations to use the same license. Commercial use needs separate permission.