Explore how leveraging pre-trained vision-language models (PVLMs) revolutionizes hateful meme detection, achieving new milestones without fine-tuning.

Learn about Pro-Cap, a new method that enhances hateful meme detection by leveraging frozen Vision-Language Models (PVLMs) in a zero-shot learning approach.

Explore the technical nuances behind implementing and optimizing models for hateful meme detection

Delve into the evaluation and analysis of a probing-based approach for detecting hateful memes.

Explore a probing-based approach for eliciting critical image details crucial for hateful content detection.

Learn about zero-shot visual question answering (VQA) and the use of pre-trained vision-language models (PVLMs) like BLIP-2 in bridging the modality gap.

Delve into hateful meme detection methodologies, from fine-tuning PVLMs and model ensembling to leveraging pre-trained models like BERT and CLIP.

RLHF is an innovative approach to mitigating bias in LLMs. It incorporates human input in the training process to reduce bias and improve fairness.