SJD-PAC: Accelerating Speculative Jacobi Decoding via Proactive Drafting and Adaptive Continuation

Speculative Jacobi Decoding (SJD) offers a draft-model-free approach to accelerate autoregressive text-to-image synthesis. However, the high-entropy nature of visual generation yields low draft-token acceptance rates in complex regions, creating a bottleneck that severely limits overall throughput. To overcome this, we introduce SJD-PAC, an enhanced SJD framework. First, SJD-PAC employs a proactive drafting strategy to improve local acceptance rates in these challenging high-entropy regions. Second, we introduce an adaptive continuation mechanism that sustains sequence validation after an initial rejection, bypassing the need for full resampling. Working in tandem, these optimizations significantly increase the average acceptance length per step, boosting inference speed while strictly preserving the target distribution. Experiments on standard text-to-image benchmarks demonstrate that SJD-PAC achieves a $3.8\times$ speedup with lossless image quality.

UTPTrack: Towards Simple and Unified Token Pruning for Visual Tracking

One-stream Transformer-based trackers achieve advanced performance in visual object tracking but suffer from significant computational overhead that hinders real-time deployment. While token pruning offers a path to efficiency, existing methods are fragmented. They typically prune the search region, dynamic template, and static template in isolation, overlooking critical inter-component dependencies, which yields suboptimal pruning and degraded accuracy. To address this, we introduce UTPTrack, a simple and Unified Token Pruning framework that, for the first time, jointly compresses all three components. UTPTrack employs an attention-guided, token type-aware strategy to holistically model redundancy, a design that seamlessly supports unified tracking across multimodal and language-guided tasks within a single model. Extensive evaluations on 10 benchmarks demonstrate that UTPTrack achieves a new state-of-the-art in the accuracy-efficiency trade-off for pruning-based trackers, pruning 65.4% of vision tokens in RGB-based tracking and 67.5% in unified tracking while preserving 99.7% and 100.5% of baseline performance, respectively. This strong performance across both RGB and multimodal scenarios underlines its potential as a robust foundation for future research in efficient visual tracking. Code will be released at https://github.com/EIT-NLP/UTPTrack.

PISCO: Precise Video Instance Insertion with Sparse Control

The landscape of AI video generation is undergoing a pivotal shift: moving beyond general generation - which relies on exhaustive prompt-engineering and "cherry-picking" - towards fine-grained, controllable generation and high-fidelity post-processing. In professional AI-assisted filmmaking, it is crucial to perform precise, targeted modifications. A cornerstone of this transition is video instance insertion, which requires inserting a specific instance into existing footage while maintaining scene integrity. Unlike traditional video editing, this task demands several requirements: precise spatial-temporal placement, physically consistent scene interaction, and the faithful preservation of original dynamics - all achieved under minimal user effort. In this paper, we propose PISCO, a video diffusion model for precise video instance insertion with arbitrary sparse keyframe control. PISCO allows users to specify a single keyframe, start-and-end keyframes, or sparse keyframes at arbitrary timestamps, and automatically propagates object appearance, motion, and interaction. To address the severe distribution shift induced by sparse conditioning in pretrained video diffusion models, we introduce Variable-Information Guidance for robust conditioning and Distribution-Preserving Temporal Masking to stabilize temporal generation, together with geometry-aware conditioning for realistic scene adaptation. We further construct PISCO-Bench, a benchmark with verified instance annotations and paired clean background videos, and evaluate performance using both reference-based and reference-free perceptual metrics. Experiments demonstrate that PISCO consistently outperforms strong inpainting and video editing baselines under sparse control, and exhibits clear, monotonic performance improvements as additional control signals are provided. Project page: xiangbogaobarry.github.io/PISCO.

Towards Efficient Large Language Reasoning Models via Extreme-Ratio Chain-of-Thought Compression

Chain-of-Thought (CoT) reasoning successfully enhances the reasoning capabilities of Large Language Models (LLMs), yet it incurs substantial computational overhead for inference. Existing CoT compression methods often suffer from a critical loss of logical fidelity at high compression ratios, resulting in significant performance degradation. To achieve high-fidelity, fast reasoning, we propose a novel EXTreme-RAtio Chain-of-Thought Compression framework, termed Extra-CoT, which aggressively reduces the token budget while preserving answer accuracy. To generate reliable, high-fidelity supervision, we first train a dedicated semantically-preserved compressor on mathematical CoT data with fine-grained annotations. An LLM is then fine-tuned on these compressed pairs via a mixed-ratio supervised fine-tuning (SFT), teaching it to follow a spectrum of compression budgets and providing a stable initialization for reinforcement learning (RL). We further propose Constrained and Hierarchical Ratio Policy Optimization (CHRPO) to explicitly incentivize question-solving ability under lower budgets by a hierarchical reward. Experiments on three mathematical reasoning benchmarks show the superiority of Extra-CoT. For example, on MATH-500 using Qwen3-1.7B, Extra-CoT achieves over 73\% token reduction with an accuracy improvement of 0.6\%, significantly outperforming state-of-the-art (SOTA) methods.

An Empirical Study of World Model Quantization

World models learn an internal representation of environment dynamics, enabling agents to simulate and reason about future states within a compact latent space for tasks such as planning, prediction, and inference. However, running world models rely on hevay computational cost and memory footprint, making model quantization essential for efficient deployment. To date, the effects of post-training quantization (PTQ) on world models remain largely unexamined. In this work, we present a systematic empirical study of world model quantization using DINO-WM as a representative case, evaluating diverse PTQ methods under both weight-only and joint weight-activation settings. We conduct extensive experiments on different visual planning tasks across a wide range of bit-widths, quantization granularities, and planning horizons up to 50 iterations. Our results show that quantization effects in world models extend beyond standard accuracy and bit-width trade-offs: group-wise weight quantization can stabilize low-bit rollouts, activation quantization granularity yields inconsistent benefits, and quantization sensitivity is highly asymmetric between encoder and predictor modules. Moreover, aggressive low-bit quantization significantly degrades the alignment between the planning objective and task success, leading to failures that cannot be remedied by additional optimization. These findings reveal distinct quantization-induced failure modes in world model-based planning and provide practical guidance for deploying quantized world models under strict computational constraints. The code will be available at https://github.com/huawei-noah/noah-research/tree/master/QuantWM.

From Sequential to Spatial: Reordering Autoregression for Efficient Visual Generation

Inspired by the remarkable success of autoregressive models in language modeling, this paradigm has been widely adopted in visual generation. However, the sequential token-by-token decoding mechanism inherent in traditional autoregressive models leads to low inference efficiency.In this paper, we propose RadAR, an efficient and parallelizable framework designed to accelerate autoregressive visual generation while preserving its representational capacity. Our approach is motivated by the observation that visual tokens exhibit strong local dependencies and spatial correlations with their neighbors--a property not fully exploited in standard raster-scan decoding orders. Specifically, we organize the generation process around a radial topology: an initial token is selected as the starting point, and all other tokens are systematically grouped into multiple concentric rings according to their spatial distances from this center. Generation then proceeds in a ring-wise manner, from inner to outer regions, enabling the parallel prediction of all tokens within the same ring. This design not only preserves the structural locality and spatial coherence of visual scenes but also substantially increases parallelization. Furthermore, to address the risk of inconsistent predictions arising from simultaneous token generation with limited context, we introduce a nested attention mechanism. This mechanism dynamically refines implausible outputs during the forward pass, thereby mitigating error accumulation and preventing model collapse. By integrating radial parallel prediction with dynamic output correction, RadAR significantly improves generation efficiency.

VersatileFFN: Achieving Parameter Efficiency in LLMs via Adaptive Wide-and-Deep Reuse

The rapid scaling of Large Language Models (LLMs) has achieved remarkable performance, but it also leads to prohibitive memory costs. Existing parameter-efficient approaches such as pruning and quantization mainly compress pretrained models without enhancing architectural capacity, thereby hitting the representational ceiling of the base model. In this work, we propose VersatileFFN, a novel feed-forward network (FFN) that enables flexible reuse of parameters in both width and depth dimensions within a fixed parameter budget. Inspired by the dual-process theory of cognition, VersatileFFN comprises two adaptive pathways: a width-versatile path that generates a mixture of sub-experts from a single shared FFN, mimicking sparse expert routing without increasing parameters, and a depth-versatile path that recursively applies the same FFN to emulate deeper processing for complex tokens. A difficulty-aware gating dynamically balances the two pathways, steering "easy" tokens through the efficient width-wise route and allocating deeper iterative refinement to "hard" tokens. Crucially, both pathways reuse the same parameters, so all additional capacity comes from computation rather than memory. Experiments across diverse benchmarks and model scales demonstrate the effectiveness of the method. The code will be available at https://github.com/huawei-noah/noah-research/tree/master/VersatileFFN.

Towards Lossless Ultimate Vision Token Compression for VLMs

Visual language models encounter challenges in computational efficiency and latency, primarily due to the substantial redundancy in the token representations of high-resolution images and videos. Current attention/similarity-based compression algorithms suffer from either position bias or class imbalance, leading to significant accuracy degradation. They also fail to generalize to shallow LLM layers, which exhibit weaker cross-modal interactions. To address this, we extend token compression to the visual encoder through an effective iterative merging scheme that is orthogonal in spatial axes to accelerate the computation across the entire VLM. Furthermoer, we integrate a spectrum pruning unit into LLM through an attention/similarity-free low-pass filter, which gradually prunes redundant visual tokens and is fully compatible to modern FlashAttention. On this basis, we propose Lossless Ultimate Vision tokens Compression (LUVC) framework. LUVC systematically compresses visual tokens until complete elimination at the final layer of LLM, so that the high-dimensional visual features are gradually fused into the multimodal queries. The experiments show that LUVC achieves a 2 speedup inference in language model with negligible accuracy degradation, and the training-free characteristic enables immediate deployment across multiple VLMs.

Positional Preservation Embedding for Multimodal Large Language Models

Multimodal large language models (MLLMs) have achieved strong performance on vision-language tasks, yet often suffer from inefficiencies due to redundant visual tokens. Existing token merging methods reduce sequence length but frequently disrupt spatial layouts and temporal continuity by disregarding positional relationships. In this work, we propose a novel encoding operator dubbed as \textbf{P}ositional \textbf{P}reservation \textbf{E}mbedding (\textbf{PPE}), which has the main hallmark of preservation of spatiotemporal structure during visual token compression. PPE explicitly introduces the disentangled encoding of 3D positions in the token dimension, enabling each compressed token to encapsulate different positions from multiple original tokens. Furthermore, we show that PPE can effectively support cascade clustering -- a progressive token compression strategy that leads to better performance retention. PPE is a parameter-free and generic operator that can be seamlessly integrated into existing token merging methods without any adjustments. Applied to state-of-the-art token merging framework, PPE achieves consistent improvements of $2\%\sim5\%$ across multiple vision-language benchmarks, including MMBench (general vision understanding), TextVQA (layout understanding) and VideoMME (temporal understanding). These results demonstrate that preserving positional cues is critical for efficient and effective MLLM reasoning.

Revealing the Power of Post-Training for Small Language Models via Knowledge Distillation

The rapid advancement of large language models (LLMs) has significantly advanced the capabilities of artificial intelligence across various domains. However, their massive scale and high computational costs render them unsuitable for direct deployment in resource-constrained edge environments. This creates a critical need for high-performance small models that can operate efficiently at the edge. Yet, after pre-training alone, these smaller models often fail to meet the performance requirements of complex tasks. To bridge this gap, we introduce a systematic post-training pipeline that efficiently enhances small model accuracy. Our post training pipeline consists of curriculum-based supervised fine-tuning (SFT) and offline on-policy knowledge distillation. The resulting instruction-tuned model achieves state-of-the-art performance among billion-parameter models, demonstrating strong generalization under strict hardware constraints while maintaining competitive accuracy across a variety of tasks. This work provides a practical and efficient solution for developing high-performance language models on Ascend edge devices.