Trained models & code to predict toxic comments on all 3 Jigsaw Toxic Comment Challenges. Built using ⚡ Pytorch Lightning and 🤗 Transformers. For access to our API, please email us at contact@unitary.ai.

The world's largest social media toxicity dataset.

SFU Opinion and Comments Corpus

A Python Implementation of Measures for Order Flow Risk, e.g. VPIN

Detect text toxicity in a simple way, using React. Based in a Keras model, loaded with Tensorflow.js.

A static code analysis tool for JavaScript and TypeScript.

A Python package to compute HONEST, a score to measure hurtful sentence completions in language models. Published at NAACL 2021.

The toxEval R-package includes a set of functions to analyze, visualize, and organize measured concentration data as it relates to chosen biological effects benchmarks. See https://doi-usgs.github.io/toxEval/ for more details

Highlight refactors and efforts to keep code base lean

Smash molecule and obtain significant fragments

Data from "Crowdsourcing of Parallel Corpora: the Case of Style Transfer for Detoxification" paper

The official code to reproduce results from the NACCL2019 paper: White-to-Black: Efficient Distillation of Black-Box Adversarial Attacks

Toxic text (comments/posts etc) detection library

Datasets used in the tox21 challenge

A collection of transformer-based models and developmental scripts presented in the publication "Transformers enable accurate prediction of acute and chronic chemical toxicity in aquatic organisms".

In silico platform to analize MD trajectories using metrics, clustering and machine&deep learning techniques

Module for predicting toxicity messages in Russian and English

Prediction and mechanistic analysis of Drug-Induced Liver Injury (DILI) based on chemical structure

Telegram bot that detects toxic comments based on Perspective API