The authors assembled a large, realistic dataset of videos, images, and audio from Instagram, TikTok, YouTube, and Facebook, then benchmarked several popular deepfake detection models. Across these platforms, detection accuracy fell by 15‑20%, with fast models still missing many fakes, highlighting the need for more robust, adaptable solutions and occasional human review.
A Hugging Face dataset provides the world’s largest collection of UFC fight data, hourly odds, and PostgreSQL dumps, plus a pretrained AutoGluon model that reportedly outperforms Vegas odds. The artifacts include training CSVs and scripts for easy reproduction or custom retraining.
A new Hugging Face dataset, Tamazight/Tifinagh-OCR-39K, provides over 39,000 labeled images of Tifinagh script for OCR training. It enriches the scarce digital resources for Amazigh (Berber) languages, enabling better document understanding and language technology development.
You’re caught up.
That was all 3 stories in the 7 June issue.