Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation
MP3•Beranda episode
Manage episode 434172414 series 2954468
Konten disediakan oleh Rob. Semua konten podcast termasuk episode, grafik, dan deskripsi podcast diunggah dan disediakan langsung oleh Rob atau mitra platform podcast mereka. Jika Anda yakin seseorang menggunakan karya berhak cipta Anda tanpa izin, Anda dapat mengikuti proses yang diuraikan di sini https://id.player.fm/legal.
While Large Language Models (LLMs) are the dominant models for generative tasks in language, they do not perform as well as diffusion models on image and video generation. To effectively use LLMs for visual generation, one crucial component is the visual tokenizer that maps pixel-space inputs to discrete tokens appropriate for LLM learning. In this paper, we introduce MAGVIT-v2, a video tokenizer designed to generate concise and expressive tokens for both videos and images using a common token vocabulary. Equipped with this new tokenizer, we show that LLMs outperform diffusion models on standard image and video generation benchmarks including ImageNet and Kinetics. In addition, we demonstrate that our tokenizer surpasses the previously top-performing video tokenizer on two more tasks: (1) video compression comparable to the next-generation video codec (VCC) according to human evaluations, and (2) learning effective representations for action recognition tasks.
2023: Lijun Yu, José Lezama, N. B. Gundavarapu, Luca Versari, Kihyuk Sohn, David C. Minnen, Yong Cheng, Agrim Gupta, Xiuye Gu, Alexander G. Hauptmann, Boqing Gong, Ming-Hsuan Yang, Irfan Essa, David A. Ross, Lu Jiang
https://arxiv.org/pdf/2310.05737
…
continue reading
2023: Lijun Yu, José Lezama, N. B. Gundavarapu, Luca Versari, Kihyuk Sohn, David C. Minnen, Yong Cheng, Agrim Gupta, Xiuye Gu, Alexander G. Hauptmann, Boqing Gong, Ming-Hsuan Yang, Irfan Essa, David A. Ross, Lu Jiang
https://arxiv.org/pdf/2310.05737
298 episode