Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation

We introduce LlamaGen, a new family of image generation models that apply original ``next-token prediction'' paradigm of large language models to visual generation domain. It is an affirmative answer to whether vanilla autoregressive models, e.g., Llama, without inductive biases on visual signals can achieve state-of-the-art image generation performance if scaling properly. We reexamine design spaces of image tokenizers, scalability properties of image generation models, and their training data quality. The outcome of this exploration consists of: (1) An image tokenizer with downsample ratio of 16, reconstruction quality of 0.94 rFID and codebook usage of 97% on ImageNet benchmark. (2) A series of class-conditional image generation models ranging from 111M to 3.1B parameters, achieving 2.18 FID on ImageNet 256x256 benchmarks, outperforming the popular diffusion models such as LDM, DiT. (3) A text-conditional image generation model with 775M parameters, from two-stage training on LAION-COCO and high aesthetics quality images, demonstrating competitive performance of visual quality and text alignment. (4) We verify the effectiveness of LLM serving frameworks in optimizing the inference speed of image generation models and achieve 326% - 414% speedup. We release all models and codes to facilitate open-source community of visual generation and multimodal foundation models.

Efficiency for Free: Ideal Data Are Transportable Representations

Data, the seminal opportunity and challenge in modern machine learning, currently constrains the scalability of representation learning and impedes the pace of model evolution. Existing paradigms tackle the issue of learning efficiency over massive datasets from the perspective of self-supervised learning and dataset distillation independently, while neglecting the untapped potential of accelerating representation learning from an intermediate standpoint. In this work, we delve into defining the ideal data properties from both optimization and generalization perspectives. We propose that model-generated representations, despite being trained on diverse tasks and architectures, converge to a shared linear space, facilitating effective linear transport between models. Furthermore, we demonstrate that these representations exhibit properties conducive to the formation of ideal data. The theoretical/empirical insights therein inspire us to propose a Representation Learning Accelerator (ReLA), which leverages a task- and architecture-agnostic, yet publicly available, free model to form a dynamic data subset and thus accelerate (self-)supervised learning. For instance, employing a CLIP ViT B/16 as a prior model for dynamic data generation, ReLA-aided BYOL can train a ResNet-50 from scratch with 50% of ImageNet-1K, yielding performance surpassing that of training on the full dataset. Additionally, employing a ResNet-18 pre-trained on CIFAR-10 can enhance ResNet-50 training on 10% of ImageNet-1K, resulting in a 7.7% increase in accuracy.

Exploring Extreme Quantization in Spiking Language Models

Despite the growing prevalence of large language model (LLM) architectures, a crucial concern persists regarding their energy and power consumption, which still lags far behind the remarkable energy efficiency of the human brain. Recent strides in spiking language models (LM) and transformer architectures aim to address this concern by harnessing the spiking activity of biological neurons to enhance energy/power efficiency. Doubling down on the principles of model quantization and energy efficiency, this paper proposes the development of a novel binary/ternary (1/1.58-bit) spiking LM architecture. Achieving scalability comparable to a deep spiking LM architecture is facilitated by an efficient knowledge distillation technique, wherein knowledge from a non-spiking full-precision "teacher" model is transferred to an extremely weight quantized spiking "student" LM. Our proposed model represents a significant advancement as the first-of-its-kind 1/1.58-bit spiking LM, and its performance is rigorously evaluated on multiple text classification tasks of the GLUE benchmark.

Stochastic Spiking Neural Networks with First-to-Spike Coding

Spiking Neural Networks (SNNs), recognized as the third generation of neural networks, are known for their bio-plausibility and energy efficiency, especially when implemented on neuromorphic hardware. However, the majority of existing studies on SNNs have concentrated on deterministic neurons with rate coding, a method that incurs substantial computational overhead due to lengthy information integration times and fails to fully harness the brain's probabilistic inference capabilities and temporal dynamics. In this work, we explore the merger of novel computing and information encoding schemes in SNN architectures where we integrate stochastic spiking neuron models with temporal coding techniques. Through extensive benchmarking with other deterministic SNNs and rate-based coding, we investigate the tradeoffs of our proposal in terms of accuracy, inference latency, spiking sparsity, energy consumption, and robustness. Our work is the first to extend the scalability of direct training approaches of stochastic SNNs with temporal encoding to VGG architectures and beyond-MNIST datasets.

Generalized Step-Chirp Sequences With Flexible Bandwidth

Sequences with low aperiodic autocorrelation sidelobes have been extensively researched in literatures. With sufficiently low integrated sidelobe level (ISL), their power spectrums are asymptotically flat over the whole frequency domain. However, for the beam sweeping in the massive multi-input multi-output (MIMO) broadcast channels, the flat spectrum should be constrained in a passband with tunable bandwidth to achieve the flexible tradeoffs between the beamforming gain and the beam sweeping time. Motivated by this application, we construct a family of sequences termed the generalized step-chirp (GSC) sequence with a closed-form expression, where some parameters can be tuned to adjust the bandwidth flexibly. In addition to the application in beam sweeping, some GSC sequences are closely connected with Mow's unified construction of sequences with perfect periodic autocorrelations, and may have a coarser phase resolution than the Mow sequence while their ISLs are comparable.

Groma: Localized Visual Tokenization for Grounding Multimodal Large Language Models

We introduce Groma, a Multimodal Large Language Model (MLLM) with grounded and fine-grained visual perception ability. Beyond holistic image understanding, Groma is adept at region-level tasks such as region captioning and visual grounding. Such capabilities are built upon a localized visual tokenization mechanism, where an image input is decomposed into regions of interest and subsequently encoded into region tokens. By integrating region tokens into user instructions and model responses, we seamlessly enable Groma to understand user-specified region inputs and ground its textual output to images. Besides, to enhance the grounded chat ability of Groma, we curate a visually grounded instruction dataset by leveraging the powerful GPT-4V and visual prompting techniques. Compared with MLLMs that rely on the language model or external module for localization, Groma consistently demonstrates superior performances in standard referring and grounding benchmarks, highlighting the advantages of embedding localization into image tokenization. Project page: https://groma-mllm.github.io/.

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction

We present Visual AutoRegressive modeling (VAR), a new generation paradigm that redefines the autoregressive learning on images as coarse-to-fine "next-scale prediction" or "next-resolution prediction", diverging from the standard raster-scan "next-token prediction". This simple, intuitive methodology allows autoregressive (AR) transformers to learn visual distributions fast and generalize well: VAR, for the first time, makes AR models surpass diffusion transformers in image generation. On ImageNet 256x256 benchmark, VAR significantly improve AR baseline by improving Frechet inception distance (FID) from 18.65 to 1.80, inception score (IS) from 80.4 to 356.4, with around 20x faster inference speed. It is also empirically verified that VAR outperforms the Diffusion Transformer (DiT) in multiple dimensions including image quality, inference speed, data efficiency, and scalability. Scaling up VAR models exhibits clear power-law scaling laws similar to those observed in LLMs, with linear correlation coefficients near -0.998 as solid evidence. VAR further showcases zero-shot generalization ability in downstream tasks including image in-painting, out-painting, and editing. These results suggest VAR has initially emulated the two important properties of LLMs: Scaling Laws and zero-shot task generalization. We have released all models and codes to promote the exploration of AR/VAR models for visual generation and unified learning.

Generative Region-Language Pretraining for Open-Ended Object Detection

In recent research, significant attention has been devoted to the open-vocabulary object detection task, aiming to generalize beyond the limited number of classes labeled during training and detect objects described by arbitrary category names at inference. Compared with conventional object detection, open vocabulary object detection largely extends the object detection categories. However, it relies on calculating the similarity between image regions and a set of arbitrary category names with a pretrained vision-and-language model. This implies that, despite its open-set nature, the task still needs the predefined object categories during the inference stage. This raises the question: What if we do not have exact knowledge of object categories during inference? In this paper, we call such a new setting as generative open-ended object detection, which is a more general and practical problem. To address it, we formulate object detection as a generative problem and propose a simple framework named GenerateU, which can detect dense objects and generate their names in a free-form way. Particularly, we employ Deformable DETR as a region proposal generator with a language model translating visual regions to object names. To assess the free-form object detection task, we introduce an evaluation method designed to quantitatively measure the performance of generative outcomes. Extensive experiments demonstrate strong zero-shot detection performance of our GenerateU. For example, on the LVIS dataset, our GenerateU achieves comparable results to the open-vocabulary object detection method GLIP, even though the category names are not seen by GenerateU during inference. Code is available at: https:// github.com/FoundationVision/GenerateU .

PRSA: Prompt Reverse Stealing Attacks against Large Language Models

Prompt, recognized as crucial intellectual property, enables large language models (LLMs) to perform specific tasks without the need of fine-tuning, underscoring their escalating importance. With the rise of prompt-based services, such as prompt marketplaces and LLM applications, providers often display prompts' capabilities through input-output examples to attract users. However, this paradigm raises a pivotal security concern: does the exposure of input-output pairs pose the risk of potential prompt leakage, infringing on the intellectual property rights of the developers? To our knowledge, this problem still has not been comprehensively explored yet. To remedy this gap, in this paper, we perform the first in depth exploration and propose a novel attack framework for reverse-stealing prompts against commercial LLMs, namely PRSA. The main idea of PRSA is that by analyzing the critical features of the input-output pairs, we mimic and gradually infer (steal) the target prompts. In detail, PRSA mainly consists of two key phases: prompt mutation and prompt pruning. In the mutation phase, we propose a prompt attention algorithm based on differential feedback to capture these critical features for effectively inferring the target prompts. In the prompt pruning phase, we identify and mask the words dependent on specific inputs, enabling the prompts to accommodate diverse inputs for generalization. Through extensive evaluation, we verify that PRSA poses a severe threat in real world scenarios. We have reported these findings to prompt service providers and actively collaborate with them to take protective measures for prompt copyright.

Unsupervised Learning Method for the Wave Equation Based on Finite Difference Residual Constraints Loss

The wave equation is an important physical partial differential equation, and in recent years, deep learning has shown promise in accelerating or replacing traditional numerical methods for solving it. However, existing deep learning methods suffer from high data acquisition costs, low training efficiency, and insufficient generalization capability for boundary conditions. To address these issues, this paper proposes an unsupervised learning method for the wave equation based on finite difference residual constraints. We construct a novel finite difference residual constraint based on structured grids and finite difference methods, as well as an unsupervised training strategy, enabling convolutional neural networks to train without data and predict the forward propagation process of waves. Experimental results show that finite difference residual constraints have advantages over physics-informed neural networks (PINNs) type physical information constraints, such as easier fitting, lower computational costs, and stronger source term generalization capability, making our method more efficient in training and potent in application.