OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Gector

Gector is a public Python repository containing the official PyTorch implementation of GECToR, a tagging-based grammatical error-correction approach, along with code for text simplification by tagging. It provides data preprocessing, model training, inference and evaluation workflows for researchers and developers.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Gector — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Gector when you need a modifiable research codebase for training, testing or running tagging-based text correction or simplification models, rather than a hosted end-user editing service.

Who it is for

  • Researchers developing or evaluating grammatical error-correction models.
  • Developers who want to train or run tagging-based correction models on text files.
  • Teams exploring text simplification using a tagging-based model workflow.

Documented uses

  • Preprocess paired source and target data for training.
  • Train grammatical error-correction models with training and development sets.
  • Run a trained model on an input file and save predictions.
  • Apply the repository’s related preprocessing and training interface to text simplification.
  • Evaluate correction or simplification outputs with the named evaluation packages.
Before you choose

Requirements & limits

The repository is public and identifies its license as Apache-2.0. Installation instructions use pip and a requirements file; the project says it was tested with Python 3.7.

  • The method predicts edits through tagging rather than generating a rewritten passage.
  • Training data must be preprocessed into the repository’s required format.
  • The README lists several encoders and training parameters, so results depend on model and configuration choices.
  • The README notes that reported scores differ from paper results because of library-version changes and, for simplification, averaging across models.
  • The repository names separate evaluation tools: M²Scorer and ERRANT for correction, and EASSE for simplification.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare grammatical error-correction training data

Convert source and target text files into the special format required for training.

Before you start

  • Python environment with the repository requirements installed.
  • Source and target datasets; the README points to public grammatical error-correction datasets and synthetic data resources.

Steps

  1. Install the required packages with the command below.

    Documented command
    pip install -r requirements.txt

  2. Run the command below with the source, target and output paths.

    Documented command
    python utils/preprocess_data.py -s SOURCE -t TARGET -o OUTPUT_FILE

Expected outcome

A preprocessed data file in the format described for model training.

Guide limitations

  • This covers data preparation only; it does not train or run a model.
  • The project says it was tested using Python 3.7.
Documentation 1 ↗
Source-based practical guide

Run prediction with a trained correction model

Apply a trained model to an input file and write predictions to an output file.

Before you start

  • A trained model path and vocabulary path.
  • An input text file and an installed repository environment.

Steps

  1. Install the required packages with the command below.

    Documented command
    pip install -r requirements.txt

  2. Run the command below, substituting the relevant paths.

    Documented command
    python predict.py --model_path MODEL_PATH [MODEL_PATH ...] --vocab_path VOCAB_PATH --input_file INPUT_FILE --output_file OUTPUT_FILE

  3. Set documented inference options such as min_error_probability or additional_confidence when needed.

Expected outcome

Predictions written to the specified output file.

Guide limitations

  • The guide does not cover training a model or configuring a particular deployment.
  • Evaluation is a separate stage; the README names M²Scorer and ERRANT for grammatical error correction.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
GectorPreprocess paired source and target data for training.Linux, ServerThe repository is public and identifies its license as Apache-2.0. Installation instructions use pip and a requirements file; the project says it was tested with Python 3.7.
Pulse Of MotionEstimate a video's physical frame rate from its visual dynamics.LinuxThe repository is public. Its README describes a released checkpoint that downloads from Hugging Face on first inference, with a manual download option. The strongest internal model is not fully released; the README directs interested users to contact the project about access. No license information is provided in the repository evidence.
Neural Image AssessmentTrain the model using AVA images and rating annotations.Cloud, Linux, macOS, Windows
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
grammarly/gector · Official homepage title · retained in its source language
How we review sourcesSuggest a correction