每日文章

Few-step distillation for video diffusion models has attracted significant attention, driven by the urgent demand for efficient deployment in real-world scenarios. However, Distribution Matching Distillation (DMD), a leading paradigm, tends to degrade under limited NFE budgets, manifesting in video generation as layout instability, oversaturation, and broken motion dynamics. We trace this failure to a structural limitation: standard DMD is an intra-sample distribution-matching objective with coordinate-wise gradients, and thus imposes no explicit constraint on the relational geometry across batch elements or temporal frames, leaving the underlying copula largely unregulated. Combined with the mode-seeking tendency of its reverse-KL objective, this absence of relational guidance makes DMD prone to collapsing into local optima in the few-step regime. Motivated by this insight, we propose Copula-aware DMD (CoDMD), a lightweight relational regularizer that reuses score estimates already produced by the frozen teacher and the online fake model to construct pairwise relation matrices across samples and frames. These are matched through a supplementary distributional objective that requires no additional networks, datasets, or sampling trajectories. On the Wan-2.1-T2V model series at 1.3B & 14B scales, CoDMD distills 50-step teachers into 4-step students, achieving an approximate 25\(\times\) speed-up while attaining VBench scores of 84.46 & 84.87, outperforming prior trajectory-based (rCM 82.81 & 84.05) and distribution-based (DMD 83.38 & 83.81) methods.

Knowledge Distillation (KD) has established itself as a pivotal technique for compressing large pre-trained language models. However, existing methods that force a student to strictly mimic the teacher's sentence embeddings or internal features often incur prohibitive computational costs and yield suboptimal performance due to the inherent capacity gap. To address these challenges, we propose TALAS (Teacher-Anchored Layer Alignment with Sharpness-aware minimization), a unified framework that synergizes hierarchical (multi-layer) alignment with robust optimization. First, we introduce a Teacher-Anchored mechanism that selectively distills final sentence embeddings only into the student's upper layers, thereby reducing overhead while respecting capacity constraints. Second, we bridge the semantic gap in lower layers via Layer-Aligned Self-Distillation, which propagates knowledge top-down using internal geometric relational constraints in the embedding space. Finally, to prevent the student from memorizing point-wise teacher noise, we integrate Adaptive Sharpness-Aware Minimization (ASAM) into the training objective, guiding the model towards flat minima for enhanced generalization. Empirical results on standard sentence embedding benchmarks demonstrate that TALAS consistently outperforms strong distillation baselines while achieving superior training efficiency in terms of computational cost and memory footprint.

Chain-of-Thought (CoT) distillation transfers multi-step reasoning from large reasoning models to smaller students, but verbose teacher traces inflate both training and inference cost. Existing CoT compression methods fall into two families, selective pruning and generative rewriting, yet prior studies have left key factors entangled: granularity is confounded with importance criteria in pruning, restructuring level is rarely isolated in rewriting, and compression budgets are not systematically evaluated across domains or regimes. We recast CoT compression along three dimensions: importance criterion, restructuring level, and compression budget. Sweeping these across two model families, Math and General domains, and Long-/Short-CoT regimes, we find that (i) importance criterion utility is strictly governed by granularity: step-level criteria converge on a shared reasoning backbone, while token-level pruning requires symbol-aware signals to preserve the logical core; (ii) restructuring level inverts across domains: Math degrades monotonically with structural disruption, while aggressive rewriting acts as a denoiser on General tasks; (iii) training-time compression does not necessarily translate to inference-time savings: Long-CoT students retain verbose habits despite concise supervision, making the training ratio an optimistic lower bound on deployment cost. These findings yield condition-aware guidelines for matching compression to deployment context.

Leveraging Vision Foundation Models (VFMs) for camera-to-LiDAR knowledge distillation offers a promising solution to the scarcity of annotated data needed to represent the immense geometric and kinematic diversity of real-world autonomous driving (AD). However, current approaches typically treat VFMs as black-box teachers, relying exclusively on frame-wise feature similarity. Consequently, they do not fully exploit the teacher's layer-wise semantic structure and global context, as well as the rich spatiotemporal information inherent in LiDAR sequences. We propose HilDA, a self-supervised pretraining framework for LiDAR backbones that better captures the semantic what and geometric where needed for driving tasks. HilDA combines hierarchical distillation comprising multi-layer distillation for progressive semantic alignment and global context distillation for scene-level semantics, with a temporal occupancy diffusion objective promoting spatiotemporal consistency. Models pre-trained with HilDA achieve state-of-the-art results on cross-modal distillation benchmarks and outperform models trained via prior distillation approaches on 3D object detection, scene flow, and semantic occupancy prediction. Code available at: https://maxiuw.github.io/hilda.

Conditional diffusion and flow models routinely fail to satisfy the very constraints that define their task. For instance, a depth-conditioned model often produces images whose re-extracted depth disagrees with the input, even though the forward operator--the depth predictor defining the constraint--is available during both training and inference. Existing approaches generally fall into two categories: supervised models that treat the conditioning signal as a static cue and ignore alignment information at inference, and guidance-based methods that consult it through hand-tuned linear updates, typically trading fidelity to the condition against the plausibility of the generated sample. We argue that the fundamental gap in both paradigms is that the model is never trained to utilize its own alignment error. We introduce FlowBender, a closed-loop framework that treats this error as a first-class input, training the network to learn a correction policy conditioned on inference-time feedback. At each step, an unguided look-ahead pass estimates the clean signal, a task-specific deviation is computed via the forward operator, and a refinement pass consumes this signal to produce a corrected velocity. We propose several variants of FlowBender, including a gradient-based formulation for differentiable operators and a zero-order variant for non-differentiable settings such as JPEG compression. For efficient sampling, we introduce a prior-step shortcut that enables closed-loop correction at a minimal additional computational cost. Across image-to-image translation, restoration, and 3D mesh texturing, FlowBender consistently outperforms standard supervised baselines, alignment-loss-augmented training, and state-of-the-art inference-time guidance, improving fidelity and plausibility simultaneously rather than trading them against each other. Project page: https://flow-bender.github.io/

The Doyle-Fuller-Newman (DFN) model resolves internal electrochemical states in lithium-ion batteries with high fidelity. However, the numerical solution of its governing equations is computationally prohibitive for real-time deployment, limiting scalability from individual cells to pack and fleet-scale applications. While machine learning surrogates can substantially reduce inference latency through GPU acceleration, most existing approaches learn solution approximations tied to specific operating conditions rather than learning generalizable state-evolution dynamics. This work presents a systematic comparison of four neural network architectures (MLP, ResNet, U-Net, FNO) formulated as autoregressive state-transition operators that predict full DFN internal states across a wide range of operating conditions. To ensure a controlled architectural comparison, all models are trained under a unified framework using multi-step unrolling and current-conditioning, isolating the impact of spatial inductive bias. Results demonstrate that the U-Net's multi-scale feature hierarchy achieves a mean final-step nRMSE of 3% averaged across all internal state variables after 300-step autoregressive rollouts, while providing a 5.38x speed-up over the numerical solver. These findings highlight spatial inductive bias as a critical determinant of surrogate performance, advancing the development of surrogates for internal state observability for next-generation battery management systems and digital twins.

Attention distillation, which trains one attention distribution to match another by minimizing their Kullback-Leibler (KL) divergence, is widely used in knowledge distillation, model compression, continual learning, and sparse-attention LLM training. However, existing approaches materialize both attention distributions before computing the KL reduction, incurring \(O(N_QN_K)\) memory and IO costs that become prohibitive at long context lengths. We present StreamKL, the first fused GPU primitive for attention KL divergence that eliminates this quadratic materialization. StreamKL derives a novel online formulation for the coupled two-distribution KL reduction, enabling a single one-pass forward kernel that streams query-key tiles through on-chip SRAM. For the backward pass, StreamKL recomputes attention probabilities tile-by-tile, avoiding storage of quadratic intermediates. We further design and implement efficient GPU kernels with dedicated optimizations. Experiments show StreamKL delivers up to \(43\times\) and \(14\times\) speedups over baseline methods in the forward and backward passes, respectively. Most importantly, StreamKL reduces the extra HBM footprint of attention distillation from \(O(N_QN_K)\) to \(O(1)\), enabling long-context distillation on a single GPU.

Most diffusion and flow-matching generators define the prior, probability path, and prediction target in the same representation space. Latent diffusion improves efficiency by moving this path into an autoencoder latent space, but the final sample is still produced by a separately trained decoder. This separation creates a mismatch: the generator is optimized for latent-space prediction, while final quality depends on how the decoder handles generated latents that may differ from clean encoder outputs. We introduce CrossFlow, a cross-space flow formulation that maps noisy latent inputs directly to pixel-space images. The key technical step is a velocity-free one-step objective: the latent trajectory defines the training path, but the supervised prediction is an image rather than a latent displacement. This lets one model act both as a one-step latent-to-pixel generator and as a decoder replacement for latent diffusion pipelines. On class-conditional ImageNet-1k at \(256\times256\), CrossFlow-XL achieves 1.62 FID with one function evaluation. Ablations show that the latent encoder and pixel-space perceptual and adversarial losses are important for fidelity. These results indicate that cross-space flow objectives can combine the efficiency of latent representations with direct pixel-space supervision, without requiring a separate decoder at inference.

Streaming VideoLLMs must continuously process incoming video while maintaining low query latency, making both video-ingestion throughput and query-time responsiveness critical for real-time deployment. Existing methods largely focus on accelerating individual modules, such as visual encoding, token pruning, or KV-cache compression, but provide limited insight into whether the resulting system can sustain real-time streaming performance. We formulate streaming VideoLLM inference as a coordinated pipeline spanning visual preprocessing, visual encoding, token dropping, and LLM prefilling/decoding. Building on this formulation, we propose ViCoStream (Video Coordinated Streaming), a stage-wise coordinated streaming framework that combines chunk-wise execution, CUDA-stream overlap, visual token control, bounded visual attention, and query-side retrieval to bound per-chunk computation and memory costs. We further provide a systematic study of bottleneck migration, revealing how chunk size, token retention, attention locality, and retrieval scope shape the throughput-accuracy trade-off. Experiments with Qwen2.5-VL-3B/7B-Instruct across multiple streaming benchmarks show that ViCoStream achieves 134 FPS video throughput and less than 50 ms TTFT on a single A100 GPU while maintaining accuracy close to full-history baselines.