The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
banglamentalBERT Corpuses
Project: Detecting Mental Health and Suicidal Tendencies on Social Media using Multimodal NLP
Institution: BRAC University, Department of CSE
Authors: Md. Nazim Hossain, Kazi Tarif Rahman, Raiseen Jahan Ritu, Ahmad Sameer
This repository contains all the raw, translated, and synthetic datasets used for Domain-Adaptive Pretraining (DAPT) and Task-Adaptive Pretraining (TAPT) in the thesis project.
Datasets Included
angla_dapt_corpus_clean.txt
- Purpose: Domain-Adaptive Pretraining (DAPT)
- Contents: 250,154 lines of translated mental health Reddit posts (via NLLB-200-3.3B).
** rain_augmented_qwen3-32b.xlsx**
- Purpose: Task-Adaptive Pretraining (TAPT) - Fix-A/Fix-B experiments
- Contents: 17,131 synthetic and real Bangla mental health posts generated using Qwen3-32B.
** apt_qwen_data/tapt_corpus_clean.txt**
- Purpose: Mega-TAPT experiments
- Contents: 24,699 synthetic Bangla mental health posts.
** apt_qwen_data/mega_tapt_corpus.txt**
- Purpose: Mega-TAPT experiments
- Contents: ~45,000 synthetic Bangla mental health posts.
** llb_backtranslated_data/train_backtranslated_cleaned.xlsx**
- Purpose: Data augmentation baseline comparisons
- Contents: Back-translated variations of the dataset.
dataset.xlsx
- Purpose: Fine-tuning and evaluation
- Contents: 4,898 samples, the original annotated depression severity corpus from Kabir et al.
Citation
Part of the thesis: Detecting Mental Health and Suicidal Tendencies on Social Media using Multimodal NLP (2026).
- Downloads last month
- 11