---
base_model: google/paligemma2-10b-pt-224
library_name: peft
pipeline_tag: image-text-to-text
tags:
- base_model:adapter:google/paligemma2-10b-pt-224
- paligemma
- paligemma2
- gemma
- vision-language
- image-text-to-text
- multimodal
- transformers
- lora
- image-classification
- meme-classification
- toxicity-detection
- content-moderation
- harmful-content
- fine-tuned
- stemtox
- toxictags
license: mit
datasets:
- swainsubhankar/ToxicTags
language:
- en
---
## STEMTOX: From Collaborative Tags to Fine-Grained Toxic Meme Detection via Entropy-Guided Multi-Task Learning

🎉 **Accepted at Transactions of the Association for Computational Linguistics (TACL 2026).**
---
## Links

- **Paper:** https://arxiv.org/abs/2508.04166
- **Dataset:** https://huggingface.co/datasets/swainsubhankar/ToxicTags
- **Code:** https://github.com/hate-alert/STEMTOX
---
## STEMTOX Architecture

<p align="center">
  <img src="architecture.png" alt="STEMTOX Architecture" width="800">
</p>


## Citation
```bibtex
@misc{swain2026stemtoxsocialtagsfinegrained,
  title={STEMTOX: From Social Tags to Fine-Grained Toxic Meme Detection via Entropy-Guided Multi-Task Learning},
  author={Subhankar Swain and Naquee Rizwan and Vishwa Gangadhar S and Nayandeep Deb and Animesh Mukherjee},
  year={2026},
  eprint={2508.04166},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2508.04166}
}
```