JPEG-LM: LLMs as Image Generators with Canonical Codec Representations
- URL: http://arxiv.org/abs/2408.08459v2
- Date: Wed, 21 Aug 2024 00:24:53 GMT
- Title: JPEG-LM: LLMs as Image Generators with Canonical Codec Representations
- Authors: Xiaochuang Han, Marjan Ghazvininejad, Pang Wei Koh, Yulia Tsvetkov,
- Abstract summary: Discretization represents continuous data like images and videos as discrete tokens.
Common methods of discretizing images and videos include modeling raw pixel values.
We show that using canonical representations can help lower the barriers between language generation and visual generation.
- Score: 51.097213824684665
- License: http://creativecommons.org/licenses/by-nc-sa/4.0/
- Abstract: Recent work in image and video generation has been adopting the autoregressive LLM architecture due to its generality and potentially easy integration into multi-modal systems. The crux of applying autoregressive training in language generation to visual generation is discretization -- representing continuous data like images and videos as discrete tokens. Common methods of discretizing images and videos include modeling raw pixel values, which are prohibitively lengthy, or vector quantization, which requires convoluted pre-hoc training. In this work, we propose to directly model images and videos as compressed files saved on computers via canonical codecs (e.g., JPEG, AVC/H.264). Using the default Llama architecture without any vision-specific modifications, we pretrain JPEG-LM from scratch to generate images (and AVC-LM to generate videos as a proof of concept), by directly outputting compressed file bytes in JPEG and AVC formats. Evaluation of image generation shows that this simple and straightforward approach is more effective than pixel-based modeling and sophisticated vector quantization baselines (on which our method yields a 31% reduction in FID). Our analysis shows that JPEG-LM has an especial advantage over vector quantization models in generating long-tail visual elements. Overall, we show that using canonical codec representations can help lower the barriers between language generation and visual generation, facilitating future research on multi-modal language/image/video LLMs.
Related papers
- QLIP: Text-Aligned Visual Tokenization Unifies Auto-Regressive Multimodal Understanding and Generation [101.28446308930367]
Quantized Language-Image Pretraining (QLIP) combines state-of-the-art reconstruction quality with state-of-the-art zero-shot image understanding.
QLIP trains a binary-spherical-quantization-based autoencoder with reconstruction and language-image alignment objectives.
We demonstrate that QLIP enables a unified mixed-modality auto-regressive model for understanding and generation.
arXiv Detail & Related papers (2025-02-07T18:59:57Z) - TS-LLaVA: Constructing Visual Tokens through Thumbnail-and-Sampling for Training-Free Video Large Language Models [52.590072198551944]
Recent advances in multimodal Large Language Models (LLMs) have shown great success in understanding multi-modal contents.
For video understanding tasks, training-based video LLMs are difficult to build due to the scarcity of high-quality, curated video-text paired data.
In this work, we explore the limitations of the existing compression strategies for building a training-free video LLM.
arXiv Detail & Related papers (2024-11-17T13:08:29Z) - VideoGPT+: Integrating Image and Video Encoders for Enhanced Video Understanding [15.959757105308238]
Video LMMs rely on either image or video encoders to process visual inputs, each of which has its own limitations.
We introduce VideoGPT+, which combines the complementary benefits of the image encoder (for detailed spatial understanding) and the video encoder (for global temporal context modeling)
Our architecture showcases improved performance across multiple video benchmarks, including VCGBench, MVBench and Zero-shot question-answering.
arXiv Detail & Related papers (2024-06-13T17:59:59Z) - Video-LaVIT: Unified Video-Language Pre-training with Decoupled Visual-Motional Tokenization [52.63845811751936]
Video pre-training is challenging due to the modeling of its dynamics video.
In this paper, we address such limitations in video pre-training with an efficient video decomposition.
Our framework is both capable of comprehending and generating image and video content, as demonstrated by its performance across 13 multimodal benchmarks.
arXiv Detail & Related papers (2024-02-05T16:30:49Z) - Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation [122.63617171522316]
Large Language Models (LLMs) are the dominant models for generative tasks in language.
We introduce MAGVIT-v2, a video tokenizer designed to generate concise and expressive tokens for both videos and images.
arXiv Detail & Related papers (2023-10-09T14:10:29Z) - Frozen CLIP Models are Efficient Video Learners [86.73871814176795]
Video recognition has been dominated by the end-to-end learning paradigm.
Recent advances in Contrastive Vision-Language Pre-training pave the way for a new route for visual recognition tasks.
We present Efficient Video Learning -- an efficient framework for directly training high-quality video recognition models.
arXiv Detail & Related papers (2022-08-06T17:38:25Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.