Downloads · 30 days
78
0% of all-time downloads
suicidal-electra
gohjiayi/suicidal-electra
suicidal-electra is a text classification model from gohjiayi. Use it when you need a label for a piece of text. It is set up for transformers. The card lists the license as mit.
This text classification model predicts whether a sequence of words are suicidal (1) or non-suicidal (0).
All-time downloads
24.2K
Public
Repo size
876 MB
Likes
5
Public
Repo makeup
Click a slice to open those files.
.bin438 MB · 100%
At a glance
- Task
- Text Classification
- Library
- transformers
- License
- mit
- Model type
- electra
- Access
- Public
- Created
- Mar 2, 2022
- Updated
- Jan 5, 2025
- SHA
- 3eb0da3c
Try a prompt
Base models
Spaces
Model card
From the Hugging Face model README
Suicidal-ELECTRA
This text classification model predicts whether a sequence of words are suicidal (1) or non-suicidal (0).
Data
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.
Parameters
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.
Performance
The model has achieved the following results after fine-tuning on the aforementioned dataset:
- Accuracy: 0.9792
- Recall: 0.9788
- Precision: 0.9677
- F1 Score: 0.9732
How to Use
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")
Resources
For more resources, including the source code, please refer to the GitHub repository gohjiayi/suicidal-text-detection.
- Task
- Text Classification
- Library
- transformers
- Type
- electra
- License
- mit
- Languages
- en
- Created
- Mar 2, 2022
- Updated
- Jan 5, 2025