Gradient Masters at BLP-2025 Task 1: Advancing Low-Resource NLP for Bengali using Ensemble-Based Adversarial Training for Hate Speech Detection

TL;DR

This paper presents 'Gradient Masters', an ensemble-based adversarial training method for low-resource Bengali hate speech detection, achieving competitive results in the BLP-2025 shared task.

Contribution

It introduces a hybrid ensemble fine-tuning approach on Bangla Language Models, improving hate speech classification performance in low-resource settings.

Findings

Outperformed baseline models in subtask 1A and 1B

Achieved 6th place in hate-type classification with 73.23% F1 score

Secured 3rd place in target group classification with 73.28% F1 score

Abstract

This paper introduces the approach of "Gradient Masters" for BLP-2025 Task 1: "Bangla Multitask Hate Speech Identification Shared Task". We present an ensemble-based fine-tuning strategy for addressing subtasks 1A (hate-type classification) and 1B (target group classification) in YouTube comments. We propose a hybrid approach on a Bangla Language Model, which outperformed the baseline models and secured the 6th position in subtask 1A with a micro F1 score of 73.23% and the third position in subtask 1B with 73.28%. We conducted extensive experiments that evaluated the robustness of the model throughout the development and evaluation phases, including comparisons with other Language Model variants, to measure generalization in low-resource Bangla hate speech scenarios and data set coverage. In addition, we provide a detailed analysis of our findings, exploring misclassification patterns in the detection of hate speech.