Abstract: Multimodal Large Language Models (MLLMs) have shown promising capabilities in Audio-Video Question-Answering (AVQA) tasks. However, during training and inference, they often suffer from ...
We introduce TokLIP, a visual tokenizer that enhances comprehension by semanticizing vector-quantized (VQ) tokens and incorporating CLIP-level semantics while enabling end-to-end multimodal ...
Abstract: Recently, deep generative models have greatly advanced the progress of face video coding towards promising rate-distortion performance and diverse application functionalities. Beyond ...
Safe Codebook (SaCo) is a training-free safety framework designed for Visual Autoregressive (VAR) text-to-image models. SaCo enhances safety by leveraging the model’s discrete codebook rather than ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results