Name		Name	Last commit message	Last commit date
Latest commit History 26 Commits
embeddings/glove		embeddings/glove
Davidson_augmented_GPT.csv		Davidson_augmented_GPT.csv
Founta_augmented_GPT_25K.csv		Founta_augmented_GPT_25K.csv
Hasoc2021.csv		Hasoc2021.csv
Hasoc_2021_aspect_words.bin		Hasoc_2021_aspect_words.bin
README.md		README.md
Toxnet_model.ipynb		Toxnet_model.ipynb
davidson_aug_GPT_aspect_words.bin		davidson_aug_GPT_aspect_words.bin
founta_aspect_words.bin		founta_aspect_words.bin
founta_aug_GPT_25K_aspect_words.bin		founta_aug_GPT_25K_aspect_words.bin
founta_dataset.csv		founta_dataset.csv
hasoc_augmented_GPT.csv		hasoc_augmented_GPT.csv
hasoc_augmented_GPT_aspect_words.bin		hasoc_augmented_GPT_aspect_words.bin
hasoc_upsampled.csv		hasoc_upsampled.csv
hasoc_upsampled_aspect_words.bin		hasoc_upsampled_aspect_words.bin
labeled_data(t-davison).csv		labeled_data(t-davison).csv
labeled_data_t_davidson.csv		labeled_data_t_davidson.csv
t_davison_aspect_words.bin		t_davison_aspect_words.bin

Repository files navigation

toxnet_research_code

About

ToxNet: A Multi-layer Network Approach to Toxicity Analysis Graph Neural Networks | PyTorch | Transformers | NLP Developed a multi-layer GNN model for hate speech detection in tweets, achieving a 60% F1 score improvement over existing models. Designed a GPT-2-based augmentation technique to balance datasets, further boosting F1 scores by 15-20%

Readme