每天知道
你的领域
出了什么

Speculative decoding is a leading technique to reduce the cost of autoregressive generation by using a small drafter to propose several tokens, which are then verified in parallel by a larger target model. Speculative diffusion decoding (SDD) further removes sequential drafting by generating every position in a draft block in parallel with a discrete diffusion model. However, SDD still invokes the target on every block, leaving verification as a potential bottleneck. This paper recognizes that this creates a new control handle: whether to invoke the verifier at all. Thus, we study verifier skipping, a lossy policy that commits a selected draft prefix directly, and ask which confidence signal should schedule it. Interestingly, our study finds that better token predictors need not yield better schedulers: skips require contiguous high-confidence prefixes, while short skips can induce additional drafting rounds. To study this mismatch, we compare raw confidence with learned marginal and conditional survival scores under the same policy, using Strict SDD, lenience, and top-$k$ acceptance as baselines. On HumanEval with DiffuCoder-7B-Instruct and Qwen3-32B, all three confidence signals save $9.6%$ to $13.5%$ of verifier calls at the same observed pass@1 as Strict SDD. Surprisingly, raw confidence saves the most; marginal survival has higher positionwise AUROC than raw confidence at most positions, yet neither learned signal dominates online. Our analysis shows that verifier skipping is a useful new lossy axis and, surprisingly, its key challenge is prefix scheduling rather than token prediction alone.

Teams that adopt cycle-consistent adversarial networks for unpaired image-to-image translation meet the same obstacles: adversarial training oscillates or collapses, cycle consistency preserves coarse layout while finer texture drifts, and a single discriminator judging global realism misses local artifacts. Four enhancements address these failures, and they are usually compared on output quality alone. We show that they also divide sharply by where their cost falls, and that this division, which follows from the architecture and not from any particular run, yields an adoption order for teams under a compute or latency budget. A Wasserstein objective with gradient penalty, a VGG19 perceptual loss on the cycle reconstruction, and multi-scale discriminators change training only, so a team can adopt or drop them without altering what ships. Self-attention alone persists into the deployed generator, with memory growing as the square of the feature-map size, which makes it the one component a resource-constrained team should defer. We integrate all four onto a lightly tuned baseline for horse-to-zebra translation, introduced one at a time on a fixed control and then combined, and for each we give the failure mode it targets and how it integrates. We document the collapse and reconstruction-artifact modes the baseline produced, report what visual inspection of saved samples showed for each variant, and report Fréchet Inception Distance and Kernel Inception Distance for the combined model. We specify the protocol still needed, covering the individual variants, perceptual similarity, and downstream segmentation, to rank these enhancements on measured evidence.

Whole slide image analysis is commonly formulated as multiple instance learning (MIL), where instance features are contextually updated and aggregated into a slide representation, a process we term slide encoding dynamics. Recently, selective state-space models (SSM) have emerged as promising MIL architectures due to their long-sequence modeling capability and linear complexity. However, existing SSM-based MIL methods rely solely on visual features during MIL. Meanwhile, in large-scale WSIs, where sparse diagnostically decisive regions are surrounded by abundant irrelevant information, such purely vision-driven selective dynamics can misallocate state updates and readouts, causing the evolving SSM state to accumulate task-irrelevant evidence and dilute critical diagnostic cues over long scan trajectories. In this work, we propose the Knowledge-Aware Hidden-State Modulation architecture (KHiM-Mamba), which innovatively regulates Mamba's core selective state-space mechanism with explicit knowledge priors, steering slide encoding dynamics toward diagnostically meaningful evidence accumulation. Specifically, we redesign the original SSM layer to perform knowledge modulation operations during the evolution of hidden states, thereby guiding what visual evidence is accumulated and retrieved from the hidden state at each encoding step. Furthermore, we additionally introduce a local-adaptive vocabulary retrieval module that uses large language models to assign each patch fine-grained, tissue-specific semantic descriptions, enabling precise modulation across diverse tasks. Experiments on 11 public benchmarks across 4 tasks show that KHiM-Mamba consistently achieves state-of-the-art performance.

Mechanistic interpretability has recently expanded to Vision Transformers (ViTs), with Sparse Autoencoders (SAEs) increasingly used as post-hoc tools to decompose internal representations into sparse and more interpretable features. However, because post-hoc SAEs are trained on frozen representations after the ViT has already been optimized, their latent features are not directly aligned with the downstream classification objective. We introduce SpIn-ViT, a framework that jointly trains a pretrained ViT and a modified SAE end-to-end, directly aligning sparse patch-level representations with image classification. SpIn-ViT learns semantically coherent neuron activations that localize meaningful image regions while maintaining competitive predictive performance. We evaluate SpIn-ViT across nine image-classification benchmarks using classification accuracy, quantitative interpretability metrics, AI-based and Human evaluations. Compared with the previous state-of-the-art post-hoc SAE method, SpIn-ViT achieves 8.84% higher average classification accuracy, an AI-based interpretability score nearly four times as high, and a human-evaluation score more than twice as high. We further extract interpretable rule-sets using the SAE neurons to create neurosymbolic models which achieve 5.97% higher average classification accuracy while requiring a 58.8% smaller rule-set than the neurosymbolic models created from the SOTA post-hoc SAE method.

Part-aware 3D object generation is essential for graphics applications such as controllable modeling, editing, and articulation, where objects are represented as coherent assemblies of semantic parts. However, existing part-aware generation methods, do not scale well to highly complex objects. As the number of parts increases, generating detailed geometry becomes prohibitively expensive in token length and memory. We introduce MegaParts, a scalable autoregressive 3D generation framework to address this challenge by combining structured sequence modeling with a token-efficient vector-quantized shape tokenizer. Our tokenizer learns discrete latent representations for part-level geometry by minimizing token usage subject to high-fidelity reconstruction, enabling adaptive-length tokenization based on geometric complexity. On top of this compact representation, we train a large language model to generate object bounding boxes, part bounding boxes, and part shape tokens within a unified structured sequence. Combined with efficient long-context training strategy, our token-efficient formulation scales to objects with up to 300 parts and sequence lengths up to 256k tokens. This substantially extends the scale of part-aware 3D generation while preserving compositional structure and enabling fine-grained part-level control. Our method achieves higher mesh quality than baseline autoregressive and diffusion models, showing that compressed discrete part tokens improve not only scalability but also the achievable fidelity of generated geometry. These results suggest that LLM native token-efficient autoregressive modeling is a compelling alternative to diffusion for large-scale part-aware 3D generation. The project page is available at https://expmaster.github.io/megaparts_webpage.

Despite a strong interest on the part of the research community in the topic of trustworthy and safe AI, the composition of the text corpora that large language models (LLMs) encounter in pre- and post-training has not yet drawn much attention. In this work, we address the question of whether LLMs are exposed to unfiltered, uncontextualised extremist speech. Using several definitions of extremist speech, stemming from official documents and research literature, and an extraction pipeline combining automated text processing with expert verification, we provide a lower bound on the prevalence of extremist documents in Dolma, an open training corpus underpinning the OLMo series of models. We show that Dolma is likely to include hundreds of thousands of documents containing extremist content and hate speech of several types, including direct calls for violence, and discuss the implications of this for data curation and model pre-training.

While Multimodal Large Language Models (MLLMs) excel in general video understanding, their capability in fine-grained and motion-centric tasks remains limited. This limitation is particularly critical in micro-gesture recognition (MGR), where micro-gestures (MGs) - subtle, short-duration, and spatially localized human movements - serve as key discriminative signals for implicit affective analysis, yet are easily neglected following common prompting practices. Although MGR has been intensively studied by many discriminative approaches, the use of MLLMs for MGR is underexplored, with notably poor performance. We hypothesize that the motion-sensitive representation ability of MLLMs is constrained by their inherent single-pass forward inference, which can be substantially enhanced through carefully designed test-time guidance. Motivated by this, building on our prior findings regarding temporal insensitivity in Video LLMs, we diagnose zero-shot MGR errors in the Negative Log-Likelihood (NLL) space. We observe that MLLMs suffer from two bottlenecks: 1) insufficient localized evidence and 2) severe score biases driven by language and motion-agnostic appearances. Thus, we propose a novel test-time evidence calibration framework that improves both reasoning details and prediction reliability. Specifically, we introduce a tree search mechanism to progressively acquire localized, fine-grained visual evidence, coupled with a test-time calibration module to mitigate score biases. The multi-cue fusion module then integrates evidence from multiple cues without relying on a single cue for final prediction. Our framework achieves mean-class accuracies of 26.84% on iMiGUE and 22.10% on MA-52, significantly outperforming the Qwen2.5-VL baseline, which produces 16.15% and 10.20%, respectively. The code will be available at https://zero-melo.github.io/Zero-MELO.

Watermarking provides a principled way to authenticate text generated by large language models (LLMs). In practice, however, the final text may be mixed-source, with watermark evidence surviving at only a subset of token positions after rewriting, insertion, deletion, or paraphrasing. Although prior work has studied global detection of watermark signals, when such signals can be localized remains unclear. We formulate watermark localization as a token-level multiple-testing problem based on pivotal statistics, with a latent indicator recording whether watermark dependence survives at each position. Under an asymptotic regime indexed by exponents for signal sparsity, next-token concentration, and effective-vocabulary growth, we derive a sharp boundary for global detection and phase transitions for discovery and classification within the class of coordinatewise pivot-based localization rules. We show that discovery is strictly harder than detection and that consistent classification is impossible across the parameter regime within this class. We then develop an adaptive thresholding method that does not require knowledge of the exponents or time-varying next-token distributions, but uses a data-driven estimate of the surviving watermark fraction. The method attains the optimal discovery boundary and near-optimal discovery power relative to homogeneous pivot-based rules. Simulations support the theoretical phase transitions, while experiments on model-generated texts demonstrate practical localization performance under common edit mechanisms.

Large language models (LLMs) have shown remarkable reasoning and generative capabilities, motivating their use as universal reasoning engines for perception. While modern approaches such as vision-language models (VLMs) have attempted to incorporate reasoning capabilities into visual sensing, the integration of LLMs with the millimeter-wave (mmWave) modality-despite its unique advantages under low light and occlusion-remains largely unexplored. The principal bottlenecks stem from the scarcity of radar language pairs, severe cross-dataset heterogeneity, and the absence of a foundational mmWave encoder. We address this gap through a minimal textualization interface that serializes each mmWave point cloud into concise natural language, allowing off-the-shelf LLMs to operate in a question answering (QA) setting. Building on this, we present mmWave-QA, the first benchmark for language-conditioned mmWave human perception. mmWave-QA aggregates heterogeneous public mmWave datasets and harmonizes them via calibration-aware preprocessing and global taxonomy alignment, while providing natural language QA. Spanning six scenarios and five QA tasks, the benchmark enables standardized evaluation across diverse mmWave hardware and experimental conditions, establishing a foundation for scalable research on mmWave-LLM integration. We further evaluate and analyze LLMs on our mmWave-QA, highlighting their zero-shot reasoning potential for radar perception, as well as their robustness under visual degradation.

As AI systems make more morally loaded decisions across society, one response has been moral preference elicitation. In this approach, researchers poll participants on hypothetical dilemmas and use the aggregated votes to train a policy that an AI model then applies at scale. Before any vote is cast, developers make three key choices in the moral AI elicitation pipeline: feature scoping, voter sampling, and question framing. In other words, they decide which features go to a vote, which voters to include, and how to present the question. These choices are often opaque, undocumented, and treated as technical details rather than normative ones. We examine each of these choices within a common empirical study and show that each can shape the preferences produced by moral AI elicitation. Across two phases (N = 809) in three deployment contexts (i.e., AI kidney allocation, AI agents simulating absent workers, and generative AI depictions of the deceased), we examine the three main stages of the moral AI elicitation pipeline. First, morally relevant features shift across contexts. This suggests that feature schemas should not be assumed to transfer across deployment domains. Second, preferences differ by political ideology for roughly one-third of features, with some differences reversing direction. The ideological composition of the voter pool can therefore affect the resulting aggregated preference profile. Third, the wording of the elicitation question can narrow or widen ideological gaps by up to a full scale point. The framing conditions also change how moral foundations are associated with participants' judgments. Taken together, these findings suggest that voting-based alignment cannot deliver fair or transparent AI by aggregation alone; at minimum, each stage of the moral AI elicitation pipeline should be audited and disclosed.

Reinforcement learning (RL) post-training provides a direct way to align diffusion models with human preferences and task-specific rewards. However, current RL algorithms for diffusion models remain fragmented: reverse-trajectory methods rely on discretized likelihood ratios, whereas forward-matching methods train on reward-labeled noising versions of the rollout samples. This paper shows that these seemingly different losses arise from a single path-space principle. Starting from the regularized diffusion-RL objective, we use importance sampling between sampling SDEs to obtain an explicit policy-gradient estimator on trajectory space. The estimator contains the stochastic Itô integral underlying Flow-GRPO-type updates; we derive an equivalent variance-reduced value-gradient form that recovers the forward-matching structure of AWM and DiffusionNFT. This identifies the empirical gap between these method families as a variance-reduction effect rather than a difference in RL principle. The derivation yields a unified design space organized by value-gradient estimation, weight functions, and sampling choices. Within this space, we propose a multi-sample KDE value-gradient estimator that reuses rollout groups, together with scale-bounded weight families that retain stable existing recipes while excluding singular ones. Experiments on SD3.5-M and Qwen-Image models validate the variance-reduction explanation and show that the resulting recipe improves over prior diffusion-RL baselines.

Code translation must preserve executable behavior across many programming languages, yet neural code translation has largely focused on a few popular languages such as C++, Java, and Python. This leaves a niche, many-to-many setting where parallel supervision is sparse, producing plausible but non-executable translations. We address this setting with preference-based reinforcement learning driven by execution-based supervision. Our pipeline firstly expands verifiable seed Python programs into a multilingual pool of execution-validated codes. Using the pool, a base LLM generates translation candidates across language pairs, which we label by their execution outcomes. The resulting preferences are used to train a reward model that scores cross-language translation quality. Finally, we optimize our base LLMs with GRPO over 600 directed language pairs (25 x 24) using the reward model as a signal. To evaluate the niche translation capability, we introduce HumanEval-X++, an execution-based benchmark that extends HumanEval-X to a broad many-to-many language space. We evaluate our approach using Qwen-3.5 4B and 9B models. On HumanEval-X++ and existing benchmarks, it yields consistent gains over the untrained baselines. In particular, the 4B model achieves an average improvement of 13% across all languages on HumanEval-X++, with a gain of 21% on mid-tier languages. Our study establishes a reliable approach of data generation, training, and benchmarking, paving the way toward further bootstrapping the quality of many-to-many translation for programming languages.

Diffusion large language models (dLLMs) accelerate language generation by predicting multiple masks in a single forward pass. However, existing dLLMs can suffer from unreliable predictions in early denoising stages under aggressive parallelism strategies, leading to errors that can propagate to later stages. To tackle this issue, we present Consistency Forcing (CForce) for dLLMs, a distillation method to force the mask predictions of early stages to align with those of later stages. CForce trains the model on pre-collected self-rollout trajectories, thereby improving training-inference alignment. We introduce Confidence Adaptive KL Divergence as a distillation objective to conjoin the merits of forward and reverse KL. We further provide a theoretical analysis for the consistency objective to explain why CForce can approximately minimize the prediction error of early stages. Critically, the same formulation applies to both mask-to-token decoding and edit-capable decoding; in the edit-capable case, later token-to-token refinements provide additional supervision for earlier masked-state predictions. Experiments on non-edit and edit-capable LLaDA models show improved speed-quality trade-offs, especially under high-parallelism decoding budgets. Code is available at: https://github.com/inclusionAI/dFactory.

Visual on-policy distillation relies heavily on an informative teacher-student asymmetry, through either a larger, stronger teacher or privileged supervision, such as reference answers or ground-truth regions of interest. This raises a fundamental question: where can informative asymmetry come from when nothing privileged is available? We answer this by inverting where the asymmetry comes from. Rather than adding privileged information to the teacher, we subtract information from the student. This asymmetry creates the same effective learning signal for free as a teacher with access to information unavailable to the student, without ground-truth annotations, rewards, or a separate stronger teacher model. Building on this principle, we introduce Self-Supervised Visual On-Policy Distillation (S$^2$VOPD), a simple yet effective method that constructs on-policy learning signals from asymmetric augmented views. S$^2$VOPD distills the teacher's distribution conditioned on the original image on-policy into the student distribution conditioned on a strongly augmented view of the same image. We systematically explore a broad design space of visual augmentations and uncover that (1) asymmetry matters: all four augmentation families improve performance, while symmetric self-distillation degrades it; (2) strength matters: performance peaks at a moderate strength; and (3) the gap must remain task-consistent: augmentations that completely remove the question-relevant evidence can induce large but uninformative discrepancies. Across six fine-grained perception benchmarks, S$^2$VOPD improves Qwen3.5-4B from 70.7% to 77.4%, above all open-source models compared, up to Qwen3-VL at 235B, and surpasses GPT-5.4. While holding training data the same, it recovers 96% of the improvement achieved by methods with privileged information. Website is at https://williamium3000.github.io/s2vopd

On-policy distillation (OPD) offers a promising way to transfer reasoning capabilities from stronger teacher models, but applying it to long-context reasoning teachers and short-context students introduces practical challenges, including tokenizer mismatch, teacher-student distribution mismatch, response length explosion, and training instability. In this work, we study this setting by transferring proof-reasoning capabilities from the long-context reasoning model SU-01 to short-context student models. To handle tokenizer differences, we perform OPD in a shared text space and align only tokens that occupy identical text spans under the student and teacher tokenizers. To mitigate the problem of excessive generation length and frequent truncation, we introduce a student reference KL loss and mask the advantages of special termination tokens such as </think> and <|im_end|>. This strategy constrains the student from drifting excessively from its initial policy, thereby mitigating the teacher-student distribution mismatch problem and fostering steady length growth. Experiments on both same-family and different-family student models, including Qwen3, Qwen3.5, Intern-S2, GLM-4.7, Gemma-4, show consistent gains in mathematical reasoning, especially natural-language math proving. Notably, Intern-S2-Preview improves by 21.2 points on ProofBench, reaching 55.2 and surpassing Gemini-2.5-Pro. It also improves on science benchmarks such as HLE and HiPhO, suggesting that OPD transfers reasoning capabilities that generalize beyond the mathematical training domain.

Long-range vision-based deformation monitoring is highly sensitive to motion of the camera platform. Absolute-pose differencing typically relies on dedicated control data and propagates two independent pose errors into the relative-motion estimate. We develop a control-adaptive differential framework that estimates inter-frame platform motion directly from image displacements and known 3D points. With no dedicated control point, the framework recovers platform rotation from measurement-point observations. One surveyed control point enables prior-constrained translation recovery, while two nonparallel control rays recover full 3D translation. The framework requires neither nonlinear optimization nor an initial pose estimate. Excluding control data from the rotation stage makes the rotation estimate exactly immune to contamination confined to the control field. The inherited differential formulation also cancels translational extrinsic errors exactly. We derive the rotation observability condition, a leakage bound for unmodeled translation and nonrigid point motion, and the single-point axial-prior bias law. Under 0.5-pixel image noise, attitude changes of up to 30~arcmin, and 3D point perturbations of up to 2~mm, the multi-camera estimator achieves a rotation RMSE of 2.97~arcsec and an average runtime of 0.46~ms. With one surveyed control point, its prior-constrained translation RMSE is 1.19~mm. In a bridge experiment without a stable control field, the median coordinate-wise displacement RMSE relative to total-station measurements is 0.85~mm. The estimator also maintains zero divergence under the tested 3D coordinate perturbations on public RGB-D and stereo sequences. These results establish state-of-the-art accuracy, calibration robustness, and computational efficiency among the evaluated methods.

Cross-Tabular Data Generation (CTDG) seeks to learn a generative model from multiple heterogeneous tables and produce new synthetic tabular datasets. However, existing synthetic tabular data generation methods are largely restricted to single-input-table scenarios and struggle to effectively handle multiple heterogeneous tables with diverse feature sets. To address this limitation, we propose a two-stage framework for cross-tabular data generation. In the first stage, each heterogeneous raw table is transformed into a standardized statistical table with the same set of columns across all tables. Each statistical table captures the marginal distributions of the original columns and the pairwise correlations among them. In the second stage, a diffusion transformer model is trained to capture structural patterns across these homogeneous statistical tables and to generate synthetic statistical tables. Synthetic raw tables are subsequently reconstructed from the generated statistical tables via multivariate Gaussian sampling followed by an inverse probability integral transform. This two-stage CTDG framework enables the learning of a unified generative model from multiple heterogeneous tables and supports the generation of an unlimited number of realistic synthetic heterogeneous tables. Experimental results demonstrate high fidelity in the learned statistical representations and a favorable fidelity-diversity trade-off in the generated synthetic data, validating the effectiveness of the proposed approach.

With the rapid advancement of image editing models and their widespread application across various domains, there is an increasingly urgent need to deploy these model capabilities directly into real-world scenarios. However, existing benchmarks remain confined to simple single-image tasks, suffering from limited coverage dimensions and an inability to effectively differentiate performance among diverse models. Consequently, they fail to reliably evaluate model performance in complex multi-image editing, highly demanding reasoning instructions, and practical deployment settings. To address these limitations, we propose CPI-Bench, a Comprehensive, Practical and Intelligent benchmark for real-world image editing. CPI-Bench comprises three core subsets: CPI-General-Bench, which comprehensively covers diverse editing tasks and introduces multi-image editing evaluation; CPI-Practical-Bench, which focuses on high-frequency real-user application scenarios; and CPI-Intelligent-Bench, which is dedicated to evaluating capabilities in highly demanding reasoning-based editing. Evaluation results of mainstream image editing models based on CPI-Bench demonstrate that CPI-Bench enhances performance differentiation among models. It provides a comprehensive and reliable quantification of gaps in general editing capabilities, practical deployment efficacy, and advanced reasoning-based editing, offering invaluable guidance for the future optimization of image editing models. Crucially, our ranking analysis reveals that CPI-Bench achieves the highest alignment with the Arena Image Edit Leaderboard, indicating stronger consistency with public human preference rankings, serving as an effective proxy for public human evaluations.

Spatial perception and reasoning from visual observations require recovering geometric structure, establishing correspondences, and understanding spatial relations. Existing approaches typically address these capabilities separately using task-specific architectures or external geometric modules, limiting knowledge transfer among complementary representations of the same physical scene. We introduce SPARGen, a unified multimodal framework that casts 3D reconstruction, dense correspondence, and spatial reasoning as instruction-conditioned generation tasks. SPARGen serializes compact structured and linguistic outputs as token sequences while generating dense geometric fields in image-aligned forms, enabling spatial supervision to jointly shape shared representations within a native multimodal generative model. Experiments across benchmarks for 3D reconstruction, correspondence, and spatial reasoning show that SPARGen achieves competitive performance across heterogeneous spatial tasks within a single native multimodal generative framework.

Morphing attacks pose a serious threat to face recognition systems. However, existing image-based morphing attack detection (MAD) methods often generalize poorly to unseen generation techniques because they rely solely on visual cues. We propose XSA-MAD, a CLIP-based multimodal framework that explicitly models semantic inconsistencies between bona-fide and morphed faces. Morphing concepts are decomposed into four interpretable attributes, including identity, facial geometry, texture, and consistency, and are encoded as structured and attribute-aware textual representations. The image encoder is progressively aligned with this discriminative textual space, resulting in a unified semantic representation that captures generation-invariant and concept-level discrepancies between bona-fide and morph images. Experiments on MAD22 and MorDIFF, following training on SMDD, demonstrate strong generalization across diverse morphing principles. In particular, XSA-MAD achieves an equal error rate of 2.92% on GAN-based morphs and consistently outperforms existing methods under high-fidelity generative attacks.

Modern text-to-image diffusion models rely on classifier-free guidance (CFG) to achieve high image fidelity and text alignment. However, CFG typically applies a static, global scale across all timesteps, samples, and conditions — a choice that is generally suboptimal and can introduce artifacts, as different states may benefit from different levels of guidance. While time-varying schedules are known to improve quality, designing them by hand is non-trivial and application-dependent. In this paper, we learn the guidance schedule as a function of diffusion time, conditioning and the current noisy sample, in order to better align sampled images with the text prompt. We frame this as a density ratio estimation problem: a discriminator is trained to estimate the time-dependent log-density ratio between the true and guided marginal distributions, while a lightweight generator network predicts the optimal, state-dependent guidance scale. Empirically, our approach outperforms both heuristic CFG schedules and prior methods for learning dynamic guidance on text-to-image generation benchmarks.

Live game commentary is scarce: it exists for professional esports broadcasts and almost nowhere else. We present a content-based video narration system that produces spoken, esports-style commentary for arbitrary gameplay recordings using a general-purpose vision-language model (VLM) and a text-to-speech back end, with no game-specific instrumentation, no engine telemetry, and no task-specific training. Three mechanisms carry the system. Temporal mosaic packing arranges nine uniformly sampled frames into a single 3x3 image, letting an image-native VLM reason about motion while consuming one image payload per segment instead of nine. Context-conditioned prompting replays the K most recent narrations as assistant-role history, suppressing the repetition that dominates per-segment captioning of static scenes. Duration-conditioned generation and elastic alignment constrain narration length in the prompt, then time-scale or symmetrically pad the synthesized audio so each utterance fills its segment slot exactly, giving frame-accurate muxing without a forced aligner. The implementation supports either cloud TTS or a 6-bit quantized 4B-parameter on-device TTS model on Apple silicon, making the speech stage fully local. We report a qualitative case study on real-time strategy footage, a cost model showing the mosaic reduces per-minute image payloads by 9x, and a candid account of observed failure modes - hallucinated game state, resolution loss from mosaicking, and prosody artifacts from time-scaling. We release the system as a reproducible baseline, with an evaluation protocol for the quantitative study a full version will report.

Interactive game world models typically autoregress visual observations directly in pixel or latent space, forcing structured properties such as pose, geometry, and occlusion to be implicitly maintained by the same generative sequence. Over long horizons, errors in these latent world properties accumulate, making consistency and controllability fragile. We explicitly model the evolving world state, delegate exact geometric computation to a fixed, zero-parameter renderer, and leave the neural model to synthesize appearance. We instantiate this idea as Marionette, a world model for interactive games with articulated characters. First, a two-stage autoregressive dynamics model predicts an explicit and interpretable 276-dimensional 3D world state comprising multi-entity articulated skeletons, metric root trajectories, and rotations. Second, a zero-parameter graphics bridge converts the predicted state into pose-control videos, computing world-space geometry and occlusion in closed form. Third, a control-conditioned video-diffusion observation model synthesizes photorealistic RGB observations from the resulting structured controls. Our experiments establish two properties of Marionette. First, the predicted world state is directly controllable. Forcing a mismatched action stream changes root-aligned joint error by 31% across 48 held-out segments. Second, long-horizon behaviour is determined in the state, and can be repaired there. Left free, the two generated characters drift to 21.2 m apart (recorded sessions stay near 5 m) and a third of frames show ground penetration. Two rules imposed on the explicit state, a terrain collider and a separation cap, cut penetration by 66% and keep the pair engaged, with no change to the observation model. Routing appearance through the predicted state costs no fidelity we can detect, at an FVD of 831 against 799 for recorded pose.

Text-to-image diffusion models have two major drawbacks that severely limit their practical utility: (1) standard models lack an intrinsic mechanism for continuous, concept-specific guidance (e.g., for precisely controlling how aesthetically pleasing an image looks), and (2) they lack reliability for tasks requiring high local coherence (e.g., generating text or human hands). To tackle these issues, we introduce a novel notion of concept-wise mutual information and find large, concept-dependent differences between individual layers, demonstrating that the generation of specific structures is localized in distinct parts of the network. We exploit this insight by reinforcing the impact of concept-relevant layers in Concept Guidance (CoG), a precise, target-specific guidance method that works for models out-of-the-box without additional training, external models, gradients, or prompt engineering. CoG first quantifies each layer's concept-specific impact and then guides denoising using a weighted combination of predictions generated with concept-relevant layers skipped. We demonstrate performance increases across various targets and popular models like PixArt-alpha, SD3, SD3.5, and FLUX.1-dev. Code is available at https://github.com/CompVis/concept_guidance

Video virtual try-on is a highly constrained editing task requiring the precise replacement of a target person's clothing while strictly preserving the original video's spatial structure and temporal dynamics. Existing methods heavily rely on auxiliary handcrafted spatial priors (e.g., masks, poses) for editing control. However, these priors are prone to failure in unconstrained real-world videos and often compress rich visual context into incomplete structural signals. Furthermore, standard reconstruction objectives fail to fully capture try-on-specific human preferences. To address these challenges, we propose InstructVVT, an instruction-driven and reference-guided video virtual try-on framework based on a Diffusion Transformer (DiT) that operates without inference-time spatial priors. Our core insight is to recover fine-grained control directly from the input triplet (source video, reference garment, and instruction) via a dual-level reference conditioning scheme. Specifically, an MLLM infers semantic edit tokens for target disambiguation and structural preservation, while a lightweight conditioning pathway explicitly injects fine-grained visual garment details. Finally, we design a try-on-specific reward and utilize the DiffusionNFT algorithm to align the model with human preferences. Extensive experiments on ViViD-S and TripVVT-Bench demonstrate that InstructVVT outperforms state-of-the-art open-source methods in garment fidelity, structural preservation, and temporal consistency, despite requiring fewer inference-time controls.

Recent advances in text-to-image (T2I) models have revolutionized the field of image generation and editing. However, identifying semantics that a T2I model can successfully edit in an image continues to be a challenging task. Most existing approaches require users to manually specify semantics to modify a particular image, a time-consuming process that often involves extensive trial and error. In this paper, we present RankT2I, a novel, training-free, and model-agnostic framework that automates the discovery of editable semantics in diffusion and FLUX-based models. Given a visual domain, we first utilize a multimodal vision-language model to gather a broad set of candidate semantics. We then frame semantic discovery as a set selection problem and use a submodular objective to identify semantics that are relevant, editable, and diverse. Our method helps users efficiently identify a wide range of semantics for text-to-image editing models across several domains while outperforming existing methods.

Action-conditioned video world models require low-latency causal generation and reliable responses to game-native controls. Although causal distillation enables one- or few-step video synthesis, extending it to interactive world models remains challenging, as discrete keyboard states and continuous mouse motion must remain aligned with temporally compressed latent chunks during causal training and autoregressive rollout. We introduce ForgeWM, a progressive framework that transforms a bidirectional action-conditioned video generator into efficient few-step world models through domain adaptation, teacher-forced causal training, causal consistency distillation, and on-policy distribution matching with a bidirectional teacher. The resulting budget-specialized students operate at steady-state denoising budgets of 1, 2, and 4 steps. ForgeWM further supports a dual-path deployment protocol combining latency-critical interaction with optional replay-time refinement, where the one-step student re-noises and refines its saved draft. On paired Minecraft trajectories, ForgeWM leads the evaluated systems in Imaging Quality, reference-aligned motion-profile agreement, action-sign accuracy, and mouse-control accuracy, while achieving the lowest reference LPIPS; the same four-stage recipe transfers to gamepad-controlled FPS gameplay. Replay-time refinement matches four-step reference quality while remaining roughly three times closer to the experienced trajectory than regeneration from noise. These results demonstrate ForgeWM's effectiveness for controllable few-step video generation.

Subject-driven image personalization---generating new images that preserve the identity of one or several reference subjects in novel scenes---is a foundational capability for modern visual content creation. It is currently dominated by generalized methods that fine-tune a pretrained multimodal diffusion transformer (MMDiT) on hundreds of thousands to millions of paired (reference, composed-target) examples, where each composed target is a synthesized image of the subject in a novel scene. Producing such targets demands a costly multi-stage curation pipeline---LLM-based prompt generation, T2I-based composed-target synthesis, reference-subject extraction, VLM-based quality filtering, and correspondence labeling---and tightly couples each method to a particular target synthesizer and curation choice. We introduce CRAFT (Constrained Reward via Attention Fine-Tuning), a single-step ReFL framework that fine-tunes a pre-trained reference-aware MMDiT via LoRA adapters using a compact reference-only data construction---$10$K reference images and subject masks, with no composed-target supervision. CRAFT realizes a Where to look principle: attention-level rewards align noise- and phrase-token attention with the correct reference subject, and the resulting per-subject attention masks gate a pixel-level identity reward to keep image-space supervision consistent with the learned attention routing. Applied to FLUX.2-klein-9B, CRAFT achieves state-of-the-art performance on XVerseBench \rev{while using no composed-target supervision---only $10$K reference-only samples, whereas prior generalized methods require $150$K to over $2$M composed-target pairs}. The same recipe transfers to other reference-aware backbones, consistently improving performance. Project page: https://jihun999.github.io/projects/CRAFT/.

Diffusion Transformers (DiTs) have become the dominant paradigm for high-fidelity video generation, yet their ability to perform high-level semantic planning remains limited. While hybrid architectures integrating MLLMs with diffusion backbones have shown strong advantages in image synthesis, such designs remain underexplored in video generation, where existing approaches often treat MLLMs primarily as frozen feature encoders rather than semantic generators. To fill this gap, we systematically study how an MLLM should be integrated with a DiT for video generation by answering three questions: what intermediate representation should bridge the MLLM and DiT, how the MLLM should generate it, and how the DiT should incorporate it during diffusion rendering. Our analysis reveals three key findings: (1) discrete semantic visual tokens produced by an EMA-based tokenizer provide a stable and expressive interface, (2) autoregressive causal modeling is effective for generating these tokens, and (3) explicit visual-token conditioning is more effective than prompt refinement or latent bridging. Based on these findings, we propose BiVidGen, a hybrid framework where an MLLM first generates semantic visual tokens and a DiT renders videos conditioned on both text and these tokens via multi-layer cross-attention. Extensive experiments show that BiVidGen improves semantic alignment and temporal coherence over a fine-tuned DiT baseline, achieving stronger performance on VBench-Long. These results demonstrate that explicit MLLM-based visual planning provides an effective intermediate interface for text-to-video generation beyond text-only conditioning.

The alternating direction method of multipliers (ADMM), as a landmark algorithm, has attracted tremendous research attention and extensive practical applications over the past two decades. It is well known that, although the two-block ADMM enjoys well-established theoretical convergence guarantees, its direct extension to the three-block case may fail to converge, as demonstrated by existing counterexamples [5]. However, to the best of our knowledge, the case in which the third constraint block is the identity remains unresolved: the existing literature gives neither a general convergence proof nor a counterexample for this subclass. In this paper, we give a negative answer: direct three-block ADMM may fail even when the first two blocks are strongly convex quadratics. Using Codex with GPT-5.6 Sol, we construct an explicit rational counterexample candidate and verify it along a piecewise-affine reduction path; exact checks show that direct three-block ADMM on this instance produces a bounded nonconvergent orbit of period 66. Within the same Codex workflow, we further guide a study of multiplier relaxation and clarify when convergence can be restored at the fixed-instance and class levels: a problem-dependent small dual step can restore convergence, whereas no positive relative step works uniformly over the whole class. Furthermore, we also test the recent Kimi Code with Kimi K3 model without the Codex candidate or project-specific route guidance; along a different path it produces an exact locally attracting period-23 certificate, convertible to an equivalent all-identity instance. The comparison suggests that different research-harness configurations can shape the mathematical objects explored and the certificates pursued.