Meyssa/gector-large-2024

GECToR Large 2024 (ONNX)

ONNX quantized version of Grammarly's GECToR 2024 model for browser-based grammatical error correction with Transformers.js.

Original Model

• Source: Grammarly Pillars of GEC
• Paper: Pillars of Grammatical Error Correction (2024)
• Architecture: RoBERTa-Large + token classification head
• Parameters: ~355M

Conversion Details

• Format: ONNX
• Quantization: INT8 (dynamic quantization)
• Size: ~350MB
• Converted by: Manual export from PyTorch

How It Works

GECToR uses a token classification approach - instead of generating corrected text, it predicts edit operations for each token:

• $KEEP - Keep token unchanged
• $DELETE - Remove token
• $REPLACE_word - Replace with specific word
• $APPEND_word - Append word after token
• $TRANSFORM_* - Apply transformation (case, verb form, etc.)

The model runs iteratively (typically 2-3 passes) until no more edits are predicted.

Usage with Transformers.js

import { pipeline } from '@huggingface/transformers';

const classifier = await pipeline(
  'token-classification',
  'YOUR_USERNAME/gector-large-2024',
  { dtype: 'q8' }
);

const result = await classifier('He go to school yesterday.');
// Returns token predictions with edit tags

Performance

Best accuracy among GECToR variants. Recommended for quality-critical applications.

License

Apache 2.0 (following original model license)