ICML 2025 Papers

Recent text-to-image generative models, e.g., Stable Diffusion V3 and Flux, have achieved notable progress. However, these models are strongly restricted to their limited knowledge, a.k.a., their own fixed parameters, that are trained with closed datasets. This leads to significant hallucinations or distortions when facing fine-grained and unseen novel real-world objects, e.g., the appearance of the Tesla Cybertruck. To this end, we present **the first** real-object-based retrieval-augmented generation framework (**RealRAG**), which augments fine-grained and unseen novel object generation by learning and retrieving real-world images to overcome the knowledge gaps of generative models. Specifically, to integrate missing memory for unseen novel object generation, we train a reflective retriever by **self-reflective contrastive learning**, which injects the generator's knowledge into the sef-reflective negatives, ensuring that the retrieved augmented images compensate for the model's missing knowledge. Furthermore, the real-object-based framework integrates fine-grained visual knowledge for the generative models, tackling the distortion problem and improving the realism for fine-grained object generation. Our Real-RAG is superior in its modular application to **all types** of state-of-the-art text-to-image generative models and also delivers **remarkable** performance boosts with all of them, such as a **gain of *16.18\%* FID score** with the auto-regressive model on the Stanford Car benchmark.

Diffusion distillation is a widely used technique to reduce the sampling cost of diffusion models, yet it often requires extensive training, and the student performance tends to be degraded. Recent studies show that incorporating a GAN objective may alleviate these issues, yet the underlying mechanism remains unclear. In this work, we first identify a key limitation of distillation: mismatched step sizes and parameter numbers between the teacher and the student model lead them to converge to different local minima, rendering direct imitation suboptimal. We further demonstrate that a standalone GAN objective, without relying a distillation loss, overcomes this limitation and is sufficient to convert diffusion models into efficient one-step generators. Based on this finding, we propose that diffusion training may be viewed as a form of generative pre-training, equipping models with capabilities that can be unlocked through lightweight GAN fine-tuning. Supporting this view, we create a one-step generation model by fine-tuning a pre-trained model with 85% of parameters frozen, achieving strong performance with only 0.2M images and near-SOTA results with 5M images. We further present a frequency-domain analysis that may explain the one-step generative capability gained in diffusion training. Overall, our work provides a new perspective for diffusion training, highlighting its role as a powerful generative pre-training process, which can be the basis for building efficient one-step generation models.

We introduce Orthus, a unified multimodal model that excels in generating interleaved images and text from mixed-modality inputs by simultaneously handling discrete text tokens and continuous image features under the \textbf{AR} modeling principle. The continuous treatment of visual signals minimizes the information loss while the fully AR formulation renders the characterization of the correlation between modalities straightforward. Orthus leverages these advantages through its modality-specific heads---one regular language modeling (LM) head predicts discrete text tokens and one diffusion head generates continuous image features. We devise an efficient strategy for building Orthus---by substituting the Vector Quantization (VQ) operation in the existing unified AR model with a soft alternative, introducing a diffusion head, and tuning the added modules to reconstruct images, we can create an Orthus-base model effortlessly (e.g., within 72 A100 GPU hours). Orthus-base can further embrace post-training to craft lengthy interleaved image-text, reflecting the potential for handling intricate real-world tasks. For visual understanding and generation, Orthus achieves a GenEval score of 0.58 and an MME-P score of 1265.8 using 7B parameters, outperforming competing baselines including Show-o and Chameleon.

We present a novel generative approach based on Denoising Diffusion Models (DDMs), which produces high-quality image samples *along* with their losslessly compressed bit-stream representations. This is obtained by replacing the standard Gaussian noise sampling in the reverse diffusion with a selection of noise samples from pre-defined codebooks of fixed iid Gaussian vectors. Surprisingly, we find that our method, termed *Denoising Diffusion Codebook Model* (DDCM), retains sample quality and diversity of standard DDMs, even for extremely small codebooks. We leverage DDCM and pick the noises from the codebooks that best match a given image, converting our generative model into a highly effective lossy image codec achieving state-of-the-art perceptual image compression results. More generally, by setting other noise selections rules, we extend our compression method to any conditional image generation task (e.g., image restoration), where the generated images are produced jointly with their condensed bit-stream representations. Our work is accompanied by a mathematical interpretation of the proposed compressed conditional generation schemes, establishing a connection with score-based approximations of posterior samplers for the tasks considered. Code and demo are available on our project's [website](https://ddcm-2025.github.io/).

We present MILS: Multimodal Iterative LLM Solver, a surprisingly simple, training-free approach, to imbue multimodal capabilities into your favorite LLM. Leveraging their innate ability to perform multi-step reasoning, MILS prompts the LLM to generate candidate outputs, each of which are scored and fed back iteratively, eventually generating a solution to the task. This enables various applications that typically require training specialized models on task-specific data. In particular, we establish a new state-of-the-art on emergent zero-shot image, video and audio captioning. MILS seamlessly applies to media generation as well, discovering prompt rewrites to improve text-to-image generation, and even edit prompts for style transfer! Finally, being a gradient-free optimization approach, MILS can invert multimodal embeddings into text, enabling applications like cross-modal arithmetic.

Foundation models are strong data compressors, but when accounting for their parameter size, their compression ratios are inferior to standard compression algorithms. Naively reducing the parameter count does not necessarily help as it deteriorates predictions and, accordingly, compression. We conduct a large-scale empirical study to find a sweet spot where pre-trained vanilla transformers can achieve competitive compression ratios. To this end, we train models on 165GB of raw byte sequences of either text, image, or audio data (and all possible combinations of the three) and then compress 1GB of out-of-distribution (OOD) data from each modality. We find that relatively small models (millions of parameters) can outperform standard general-purpose compression algorithms (gzip, LZMA2) and even domain-specific compressors (PNG, JPEG-XL, FLAC) — even when accounting for parameter size. We achieve, e.g., the lowest compression ratio of 0.49 on OOD audio data (vs. 0.54 for FLAC). We conduct extensive ablations and hyperparameter sweeps to study the impact of model- and dataset scale, and we investigate the effect of unimodal versus multimodal training. We find that even small models can be trained to perform well on multiple modalities, but unlike large-scale foundation models, transfer to unseen modalities is generally weak.

Traditional text-based person ReID assumes that person descriptions from witnesses are complete and provided at once. However, in real-world scenarios, such descriptions are often partial or vague. To address this limitation, we introduce a new task called interactive person re-identification (Inter-ReID). Inter-ReID is a dialogue-based retrieval task that iteratively refines initial descriptions through ongoing interactions with the witnesses. To facilitate the study of this new task, we construct a dialogue dataset that incorporates multiple types of questions by decomposing fine-grained attributes of individuals. We further propose LLaVA-ReID, a question model that generates targeted questions based on visual and textual contexts to elicit additional details about the target person. Leveraging a looking-forward strategy, we prioritize the most informative questions as supervision during training. Experimental results on both Inter-ReID and text-based ReID benchmarks demonstrate that LLaVA-ReID significantly outperforms baselines.

Text-to-song generation, the task of creating vocals and accompaniment from textual inputs, poses significant challenges due to domain complexity and data scarcity. Existing approaches often employ multi-stage generation procedures, leading to cumbersome training and inference pipelines, as well as suboptimal overall generation quality due to error accumulation across stages. In this paper, we propose **SongGen**, a fully open-source, single-stage auto-regressive transformer designed for controllable song generation. The proposed model facilitates fine-grained control over diverse musical attributes, including lyrics and textual descriptions of instrumentation, genre, mood, and timbre, while also offering an optional three-second reference clip for voice cloning. Within a unified auto-regressive framework, SongGen supports two output modes: **mixed mode**, which generates a mixture of vocals and accompaniment directly, and **dual-track mode**, which synthesizes them separately for greater flexibility in downstream applications. We explore diverse token pattern strategies for each mode, leading to notable improvements and valuable insights. Furthermore, we design an automated data preprocessing pipeline with effective quality control. To foster community engagement and future research, we will release our model weights, training code, annotated data, and preprocessing pipeline. The code is available at https://github.com/LiuZH-19/SongGen.

Recent advances in large language models (LLMs) have transformed software development by automatically generating code from natural language. Yet challenges remain in generating fully correct code that aligns with user intent. Our study reveals that LLMs tend to pay less attention to user prompts as more code tokens are generated. We hypothesize that this attention dilution issue is an important reason for code generation errors. To mitigate this issue, we propose ***S**elective **P**rompt **A**nchoring* (SPA) to guide code LLMs to pay more attention to user intent when generating code. We evaluate SPA using six base LLMs across six benchmarks. Our results demonstrate that SPA enhances Pass@1 by up to 12.9%, consistently outperforming SOTA code generation methods in all settings. Our code is available at https://github.com/magic-YuanTian/Selective-Prompt-Anchoring.

Recent advances in Large Language Models (LLMs) have raised urgent concerns about LLM-generated text authenticity, prompting regulatory demands for reliable identification mechanisms. Although watermarking offers a promising solution, existing approaches struggle to simultaneously achieve three critical requirements: text quality preservation, model-agnostic detection, and message embedding capacity, which are crucial for practical implementation. To achieve these goals, the key challenge lies in balancing the trade-off between text quality preservation and message embedding capacity. To address this challenge, we propose BiMark, a novel watermarking framework that achieves these requirements through three key innovations: (1) a bit-flip unbiased reweighting mechanism enabling model-agnostic detection, (2) a multilayer architecture enhancing detectability without compromising generation quality, and (3) an information encoding approach supporting multi-bit watermarking. Through theoretical analysis and extensive experiments, we validate that, compared to state-of-the-art multi-bit watermarking methods, BiMark achieves up to 30\% higher extraction rates for short texts while maintaining text quality indicated by lower perplexity, and performs comparably to non-watermarked text on downstream tasks such as summarization and translation.

Machine unlearning aims to selectively remove specific knowledge from a trained model. Existing approaches, such as Task Arithmetic, fine-tune the model on the forget set to create a task vector (i.e., a direction in weight space) for subtraction from the original model's weight. However, their effectiveness is highly sensitive to hyperparameter selection, requiring extensive validation to identify the optimal vector from many fine-tuned candidates. In this paper, we propose a novel method that utilizes all fine-tuned models trained with varying hyperparameters instead of a single selection. Specifically, we aggregate the computed task vectors by retaining only the elements with consistent shared signs. The merged task vector is then negated to induce unlearning on the original model. Evaluations on zero-shot and standard image recognition tasks across twelve datasets and four backbone architectures show that our approach outperforms state-of-the-art methods while requiring similar or fewer computational resources. Code is available at https://github.com/naver-ai/negmerge.

We introduce a novel resampling criterion using lift scores, for improving compositional generation in diffusion models. By leveraging the lift scores, we evaluate whether generated samples align with each single condition and then compose the results to determine whether the composed prompt is satisfied. Our key insight is that lift scores can be efficiently approximated using only the original diffusion model, requiring no additional training or external modules. We develop an optimized variant that achieves relatively lower computational overhead during inference while maintaining effectiveness. Through extensive experiments, we demonstrate that lift scores significantly improved the condition alignment for compositional generation across 2D synthetic data, CLEVR position tasks, and text-to-image synthesis. Our code is available at github.com/rainorangelemon/complift.

Automation of analog topology design is crucial due to customized requirements of modern applications with heavily manual engineering efforts. The state-of-the-art work applies a sequence-to-sequence approach and supervised finetuning on language models to generate topologies given user specifications. However, its circuit formulation is inefficient due to $O(|V|^2)$ token length and suffers from low precision sensitivity to numeric inputs. In this work, we introduce LaMAGIC2, a succinct float-input canonical formulation with identifier (SFCI) for language model-based analog topology generation. SFCI addresses these challenges by improving component-type recognition through identifier-based representations, reducing token length complexity to $O(|V|)$, and enhancing numeric precision sensitivity for better performance under tight tolerances. Our experiments demonstrate that LaMAGIC2 achieves 34\% higher success rates under a tight tolerance 0.01 and 10X lower MSEs compared to a prior method. LaMAGIC2 also exhibits better transferability for circuits with more vertices with up to 58.5\% improvement. These advancements establish LaMAGIC2 as a robust framework for analog topology generation.

The adoption of text-to-image diffusion models raises concerns over reliability, drawing scrutiny under the lens of various metrics like calibration, fairness, or compute efficiency. We focus in this work on two issues that arise when deploying these models: a lack of diversity when prompting images, and a tendency to recreate images from the training set. To solve both problems, we propose a method that coaxes the sampled trajectories of pretrained diffusion models to land on images that fall outside of a reference set. We achieve this by adding repellency terms to the diffusion SDE throughout the generation trajectory, which are triggered whenever the path is expected to land too closely to an image in the shielded reference set. Our method is sparse in the sense that these repellency terms are zero and inactive most of the time, and even more so towards the end of the generation trajectory. Our method, named SPELL for sparse repellency, can be used either with a static reference set that contains protected images, or dynamically, by updating the set at each timestep with the expected images concurrently generated within a batch, and with the images of previously generated batches. We show that adding SPELL to popular diffusion models improves their diversity while impacting their FID only marginally, and performs comparatively better than other recent training-free diversity methods. We also demonstrate how SPELL can ensure a shielded generation away from a very large set of protected images by considering all 1.2M images from ImageNet as the protected set.

Counterfactual image generation presents significant challenges, including preserving identity, maintaining perceptual quality, and ensuring faithfulness to an underlying causal model. While existing auto-encoding frameworks admit semantic latent spaces which can be manipulated for causal control, they struggle with scalability and fidelity. Advancements in diffusion models present opportunities for improving counterfactual image editing, having demonstrated state-of-the-art visual quality, human-aligned perception and representation learning capabilities. Here, we present a suite of diffusion-based causal mechanisms, introducing the notions of spatial, semantic and dynamic abduction. We propose a general framework that integrates semantic representations into diffusion models through the lens of Pearlian causality to edit images via a counterfactual reasoning process. To the best of our knowledge, ours is the first work to consider high-level semantic identity preservation for diffusion counterfactuals and to demonstrate how semantic control enables principled trade-offs between faithful causal control and identity preservation.

Shock waves caused by earthquakes can be devastating. Generating realistic earthquake-caused ground motion waveforms help reducing losses in lives and properties, yet generative models for the task tend to generate subpar waveforms. We present High-fidelity Earthquake Groundmotion Generation System (HEGGS) and demonstrate its superior performance using earthquakes from North American, East Asian, and European regions. HEGGS exploits the intrinsic characteristics of earthquake dataset and learns the waveforms using an end-to-end differentiable generator containing conditional latent diffusion model and hi-fidelity waveform construction model. We show the learning efficiency of HEGGS by training it on a single GPU machine and validate its performance using earthquake databases from North America, East Asia, and Europe, using diverse criteria from waveform generation tasks and seismology. Once trained, HEGGS can generate three dimensional E-N-Z seismic waveforms with accurate P/S phase arrivals, envelope correlation, signal-to-noise ratio, GMPE analysis, frequency content analysis, and section plot analysis.

Latent generative models have emerged as a leading approach for high-quality image synthesis. These models rely on an autoencoder to compress images into a latent space, followed by a generative model to learn the latent distribution. We identify that existing autoencoders lack equivariance to semantic-preserving transformations like scaling and rotation, resulting in complex latent spaces that hinder generative performance. To address this, we propose EQ-VAE, a simple regularization approach that enforces equivariance in the latent space, reducing its complexity without degrading reconstruction quality. By finetuning pre-trained autoencoders with EQ-VAE, we enhance the performance of several state-of-the-art generative models, including DiT, SiT, REPA and MaskGIT, achieving a ×7 speedup on DiT-XL/2 with only five epochs of SD-VAE fine-tuning. EQ-VAE is compatible with both continuous and discrete autoencoders, thus offering a versatile enhancement for a wide range of latent generative models.

In volume-to-volume translations in medical images, existing models often struggle to capture the inherent volumetric distribution using 3D voxel-space representations, due to high computational dataset demands. We present Score-Fusion, a novel volumetric translation model that effectively learns 3D representations by ensembling perpendicularly trained 2D diffusion models in score function space. By carefully initializing our model to start with an average of 2D models as in existing models, we reduce 3D training to a fine-tuning process, mitigating computational and data demands. Furthermore, we explicitly design the 3D model's hierarchical layers to learn ensembles of 2D features, further enhancing efficiency and performance. Moreover, Score-Fusion naturally extends to multi-modality settings by fusing diffusion models conditioned on different inputs for flexible, accurate integration. We demonstrate that 3D representation is essential for better performance in downstream recognition tasks, such as tumor segmentation, where most segmentation models are based on 3D representation. Extensive experiments demonstrate that Score-Fusion achieves superior accuracy and volumetric fidelity in 3D medical image super-resolution and modality translation. Additionally, we extend Score-Fusion to video super-resolution by integrating 2D diffusion models on time-space slices with a spatial-temporal video diffusion backbone, highlighting its potential for general-purpose volume translation and providing broader insight into learning-based approaches for score function fusion.

Learning diffusion bridge models is easy; making them fast and practical is an art. Diffusion bridge models (DBMs) are a promising extension of diffusion models for applications in image-to-image translation. However, like many modern diffusion and flow models, DBMs suffer from the problem of slow inference. To address it, we propose a novel distillation technique based on the inverse bridge matching formulation and derive the tractable objective to solve it in practice. Unlike previously developed DBM distillation techniques, the proposed method can distill both conditional and unconditional types of DBMs, distill models in a one-step generator, and use only the corrupted images for training. We evaluate our approach for both conditional and unconditional types of bridge matching on a wide set of setups, including super-resolution, JPEG restoration, sketch-to-image, and other tasks, and show that our distillation technique allows us to accelerate the inference of DBMs from 4x to 100x and even provide better generation quality than used teacher model depending on particular setup.