This text classification model predicts whether a sequence of words are suicidal (1) or non-suicidal (0).

The model was trained on the Suicide and Depression Dataset obtained from Kaggle. The dataset was scraped from Reddit and consists of 232,074 rows equally distributed between 2 classes - suicide and non-suicide.

The model fine-tuning was conducted on 1 epoch, with batch size of 6, and learning rate of 0.00001. Due to limited computing resources and time, we were unable to scale up the number of epochs and batch size.

The model has achieved the following results after fine-tuning on the aforementioned dataset:

Load the model via the transformers library:

from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("gooohjy/suicidal-electra")
model = AutoModel.from_pretrained("gooohjy/suicidal-electra")

For more resources, including the source code, please refer to the GitHub repository gohjiayi/suicidal-text-detection.