Daily Papers

The success of transformers in computer vision has led to several attempts to adapt them for mobile devices, but their performance remains unsatisfactory in some real-world applications. To address this issue, we propose PP-MobileSeg, a semantic segmentation model that achieves state-of-the-art performance on mobile devices. PP-MobileSeg comprises three novel parts: the StrideFormer backbone, the Aggregated Attention Module (AAM), and the Valid Interpolate Module (VIM). The four-stage StrideFormer backbone is built with MV3 blocks and strided SEA attention, and it is able to extract rich semantic and detailed features with minimal parameter overhead. The AAM first filters the detailed features through semantic feature ensemble voting and then combines them with semantic features to enhance the semantic information. Furthermore, we proposed VIM to upsample the downsampled feature to the resolution of the input image. It significantly reduces model latency by only interpolating classes present in the final prediction, which is the most significant contributor to overall model latency. Extensive experiments show that PP-MobileSeg achieves a superior tradeoff between accuracy, model size, and latency compared to other methods. On the ADE20K dataset, PP-MobileSeg achieves 1.57% higher accuracy in mIoU than SeaFormer-Base with 32.9% fewer parameters and 42.3% faster acceleration on Qualcomm Snapdragon 855. Source codes are available at https://github.com/PaddlePaddle/PaddleSeg/tree/release/2.8.

Recent advancements in diffusion models have positioned them at the forefront of image generation. Despite their superior performance, diffusion models are not without drawbacks; they are characterized by complex architectures and substantial computational demands, resulting in significant latency due to their iterative sampling process. To mitigate these limitations, we introduce a dual approach involving model miniaturization and a reduction in sampling steps, aimed at significantly decreasing model latency. Our methodology leverages knowledge distillation to streamline the U-Net and image decoder architectures, and introduces an innovative one-step DM training technique that utilizes feature matching and score distillation. We present two models, SDXS-512 and SDXS-1024, achieving inference speeds of approximately 100 FPS (30x faster than SD v1.5) and 30 FP (60x faster than SDXL) on a single GPU, respectively. Moreover, our training approach offers promising applications in image-conditioned control, facilitating efficient image-to-image translation.

Natural language is composed of words, but modern large language models (LLMs) process sub-words as input. A natural question raised by this discrepancy is whether LLMs encode words internally, and if so how. We present evidence that LLMs engage in an intrinsic detokenization process, where sub-word sequences are combined into coherent whole-word representations at their last token. Our experiments show that this process primarily takes place within the early and middle layers of the model. We further demonstrate its robustness to arbitrary splits (e.g., "cats" to "ca" and "ts"), typos, and importantly-to out-of-vocabulary words: when feeding the last token internal representations of such words to the model as input, it can "understand" them as the complete word despite never seeing such representations as input during training. Our findings suggest that LLMs maintain a latent vocabulary beyond the tokenizer's scope. These insights provide a practical, finetuning-free application for expanding the vocabulary of pre-trained models. By enabling the addition of new vocabulary words, we reduce input length and inference iterations, which reduces both space and model latency, with little to no loss in model accuracy.

Designing convolutional neural networks (CNN) for mobile devices is challenging because mobile models need to be small and fast, yet still accurate. Although significant efforts have been dedicated to design and improve mobile CNNs on all dimensions, it is very difficult to manually balance these trade-offs when there are so many architectural possibilities to consider. In this paper, we propose an automated mobile neural architecture search (MNAS) approach, which explicitly incorporate model latency into the main objective so that the search can identify a model that achieves a good trade-off between accuracy and latency. Unlike previous work, where latency is considered via another, often inaccurate proxy (e.g., FLOPS), our approach directly measures real-world inference latency by executing the model on mobile phones. To further strike the right balance between flexibility and search space size, we propose a novel factorized hierarchical search space that encourages layer diversity throughout the network. Experimental results show that our approach consistently outperforms state-of-the-art mobile CNN models across multiple vision tasks. On the ImageNet classification task, our MnasNet achieves 75.2% top-1 accuracy with 78ms latency on a Pixel phone, which is 1.8x faster than MobileNetV2 [29] with 0.5% higher accuracy and 2.3x faster than NASNet [36] with 1.2% higher accuracy. Our MnasNet also achieves better mAP quality than MobileNets for COCO object detection. Code is at https://github.com/tensorflow/tpu/tree/master/models/official/mnasnet

Speculative decoding (SD) has emerged as a powerful method for accelerating autoregressive generation in large language models (LLMs), yet its integration into vision-language models (VLMs) remains underexplored. We introduce DREAM, a novel speculative decoding framework tailored for VLMs that combines three key innovations: (1) a cross-attention-based mechanism to inject intermediate features from the target model into the draft model for improved alignment, (2) adaptive intermediate feature selection based on attention entropy to guide efficient draft model training, and (3) visual token compression to reduce draft model latency. DREAM enables efficient, accurate, and parallel multimodal decoding with significant throughput improvement. Experiments across a diverse set of recent popular VLMs, including LLaVA, Pixtral, SmolVLM and Gemma3, demonstrate up to 3.6x speedup over conventional decoding and significantly outperform prior SD baselines in both inference throughput and speculative draft acceptance length across a broad range of multimodal benchmarks. The code is publicly available at: https://github.com/SAI-Lab-NYU/DREAM.git

The recent widespread adoption of Large Language Models (LLMs) and machine learning in general has sparked research interest in exploring the possibilities of deploying these models on smaller devices such as laptops and mobile phones. This creates a need for frameworks and approaches that are capable of taking advantage of on-device hardware. The MLX framework was created to address this need. It is a framework optimized for machine learning (ML) computations on Apple silicon devices, facilitating easier research, experimentation, and prototyping. This paper presents a performance evaluation of MLX, focusing on inference latency of transformer models. We compare the performance of different transformer architecture implementations in MLX with their Pytorch counterparts. For this research we create a framework called MLX-transformers which includes different transformer implementations in MLX and downloads the model checkpoints in pytorch and converts it to the MLX format. By leveraging the advanced architecture and capabilities of Apple Silicon, MLX-Transformers enables seamless execution of transformer models directly sourced from Hugging Face, eliminating the need for checkpoint conversion often required when porting models between frameworks. Our study benchmarks different transformer models on two Apple Silicon macbook devices against an NVIDIA CUDA GPU. Specifically, we compare the inference latency performance of models with the same parameter sizes and checkpoints. We evaluate the performance of BERT, RoBERTa, and XLM-RoBERTa models, with the intention of extending future work to include models of different modalities, thus providing a more comprehensive assessment of MLX's capabilities. The results highlight MLX's potential in enabling efficient and more accessible on-device ML applications within Apple's ecosystem.

Balancing efficiency and accuracy is a long-standing problem for deploying deep learning models. The trade-off is even more important for real-time safety-critical systems like autonomous vehicles. In this paper, we propose an effective approach for accelerating transformer-based 3D object detectors by dynamically halting tokens at different layers depending on their contribution to the detection task. Although halting a token is a non-differentiable operation, our method allows for differentiable end-to-end learning by leveraging an equivalent differentiable forward-pass. Furthermore, our framework allows halted tokens to be reused to inform the model's predictions through a straightforward token recycling mechanism. Our method significantly improves the Pareto frontier of efficiency versus accuracy when compared with the existing approaches. By halting tokens and increasing model capacity, we are able to improve the baseline model's performance without increasing the model's latency on the Waymo Open Dataset.

In this paper, we present a novel method that reduces model inference latency during distributed deployment of Large Language Models (LLMs). Our contribution is an optimized inference deployment scheme that address the current limitations of state-of-the-art quantization kernels when used in conjunction with Tensor Parallel (TP). Our method preserves data locality in GPU memory access patterns and exploits a priori knowledge of TP to reduce global communication. We demonstrate an up to 1.81x speedup over existing methods for Llama-70B and up to 1.78x speedup for IBM WatsonX's Granite-20B MLP layer problem sizes on A100 and H100 NVIDIA DGX Systems for a variety of TP settings.

In this paper cum technical report, we present PHUDGE A fine tuned Phi3 model that achieved SOTA results in 4 tasks as Feedback Test, Feedback OOD, MT Human, Preference Test surpassing each and every existing model in latency and throughput. It shows very strong correlation not only with GPT4 but with Human annotators too in unseen data as well as in both absolute and relative grading tasks. We have not only addressed the usage of small LMs for cost effective production grade systems but have also shown that Causal modelling is not only slow in nature but sometimes it can hinder models learning capabilities and should be replaced by simpler tasks whenever we can to make the overall system faster and better. We show that by following systematic ML experimentation, thoughtful data augmentation and re purposing the problem itself, we can even beat 10x bigger models even with lesser training data. To the best of our knowledge, we are re the first one to experiment and showcase the usage of generalised version of Earth Movers Distance AKA Wasserstein distance by using Minkowski Distance with a penalty to control loss smoothing and can be used as a loss function instead of Cross Entropy to get stable training and better results for grading tasks.

The last decade has witnessed many successes of deep learning-based models for industry-scale recommender systems. These models are typically trained offline in a batch manner. While being effective in capturing users' past interactions with recommendation platforms, batch learning suffers from long model-update latency and is vulnerable to system biases, making it hard to adapt to distribution shift and explore new items or user interests. Although online learning-based approaches (e.g., multi-armed bandits) have demonstrated promising theoretical results in tackling these challenges, their practical real-time implementation in large-scale recommender systems remains limited. First, the scalability of online approaches in servicing a massive online traffic while ensuring timely updates of bandit parameters poses a significant challenge. Additionally, exploring uncertainty in recommender systems can easily result in unfavorable user experience, highlighting the need for devising intricate strategies that effectively balance the trade-off between exploitation and exploration. In this paper, we introduce Online Matching: a scalable closed-loop bandit system learning from users' direct feedback on items in real time. We present a hybrid "offline + online" approach for constructing this system, accompanied by a comprehensive exposition of the end-to-end system architecture. We propose Diag-LinUCB -- a novel extension of the LinUCB algorithm -- to enable distributed updates of bandits parameter in a scalable and timely manner. We conduct live experiments in YouTube and show that Online Matching is able to enhance the capabilities of fresh content discovery and item exploration in the present platform.

Pioneered by the foundational information theory by Claude Shannon and the visionary framework of machine intelligence by Alan Turing, the convergent evolution of information and communication technologies (IT/CT) has created an unbroken wave of connectivity and computation. This synergy has sparked a technological revolution, now reaching its peak with large artificial intelligence (AI) models that are reshaping industries and redefining human-machine collaboration. However, the realization of ubiquitous intelligence faces considerable challenges due to substantial resource consumption in large models and high communication bandwidth demands. To address these challenges, AI Flow has been introduced as a multidisciplinary framework that integrates cutting-edge IT and CT advancements, with a particular emphasis on the following three key points. First, device-edge-cloud framework serves as the foundation, which integrates end devices, edge servers, and cloud clusters to optimize scalability and efficiency for low-latency model inference. Second, we introduce the concept of familial models, which refers to a series of different-sized models with aligned hidden features, enabling effective collaboration and the flexibility to adapt to varying resource constraints and dynamic scenarios. Third, connectivity- and interaction-based intelligence emergence is a novel paradigm of AI Flow. By leveraging communication networks to enhance connectivity, the collaboration among AI models across heterogeneous nodes achieves emergent intelligence that surpasses the capability of any single model. The innovations of AI Flow provide enhanced intelligence, timely responsiveness, and ubiquitous accessibility to AI services, paving the way for the tighter fusion of AI techniques and communication systems.

State-of-the-art deep learning models have achieved significant performance levels on various benchmarks. However, the excellent performance comes at a cost of inefficient computational cost. Light-weight architectures, on the other hand, achieve moderate accuracies, but at a much more desirable latency. This paper presents a new method of jointly using the large accurate models together with the small fast ones. To this end, we propose an Energy-Based Joint Reasoning (EBJR) framework that adaptively distributes the samples between shallow and deep models to achieve an accuracy close to the deep model, but latency close to the shallow one. Our method is applicable to out-of-the-box pre-trained models as it does not require an architecture change nor re-training. Moreover, it is easy to use and deploy, especially for cloud services. Through a comprehensive set of experiments on different down-stream tasks, we show that our method outperforms strong state-of-the-art approaches with a considerable margin. In addition, we propose specialized EBJR, an extension of our method where we create a smaller specialized side model that performs the target task only partially, but yields an even higher accuracy and faster inference. We verify the strengths of our methods with both theoretical and experimental evaluations.

Speaker anonymization aims to conceal cues to speaker identity while preserving linguistic content. Current machine learning based approaches require substantial computational resources, hindering real-time streaming applications. To address these concerns, we propose a streaming model that achieves speaker anonymization with low latency. The system is trained in an end-to-end autoencoder fashion using a lightweight content encoder that extracts HuBERT-like information, a pretrained speaker encoder that extract speaker identity, and a variance encoder that injects pitch and energy information. These three disentangled representations are fed to a decoder that re-synthesizes the speech signal. We present evaluation results from two implementations of our system, a full model that achieves a latency of 230ms, and a lite version (0.1x in size) that further reduces latency to 66ms while maintaining state-of-the-art performance in naturalness, intelligibility, and privacy preservation.

Deploying conversational voice agents with large language models faces a critical challenge: cloud-based foundation models provide deep reasoning and domain knowledge but introduce latency that disrupts natural conversation, while on-device models respond immediately but lack sophistication. We propose conversational infill, a task where a lightweight on-device model generates contextually appropriate dialogue while seamlessly incorporating streaming knowledge from a powerful backend model. This approach decouples response latency from model capability, enabling systems that feel responsive while accessing the full power of large-scale models. We present ConvFill, a 360M parameter model trained on synthetic multi-domain conversations. Evaluation across multiple backend models shows that conversational infill can be successfully learned, with ConvFill achieving accuracy improvements of 36-42% over standalone small models of the same size while consistently retaining sub-200ms response latencies. Our results demonstrate the promise of this approach for building on-device conversational agents that are both immediately responsive and knowledgeable.

Continuous speech separation for meeting pre-processing has recently become a focused research topic. Compared to the data in utterance-level speech separation, the meeting-style audio stream lasts longer, has an uncertain number of speakers. We adopt the time-domain speech separation method and the recently proposed Graph-PIT to build a super low-latency online speech separation model, which is very important for the real application. The low-latency time-domain encoder with a small stride leads to an extremely long feature sequence. We proposed a simple yet efficient model named Skipping Memory (SkiM) for the long sequence modeling. Experimental results show that SkiM achieves on par or even better separation performance than DPRNN. Meanwhile, the computational cost of SkiM is reduced by 75% compared to DPRNN. The strong long sequence modeling capability and low computational cost make SkiM a suitable model for online CSS applications. Our fastest real-time model gets 17.1 dB signal-to-distortion (SDR) improvement with less than 1-millisecond latency in the simulated meeting-style evaluation.

Automatic Speech Recognition (ASR) has seen remarkable progress, with models like OpenAI Whisper and NVIDIA Canary achieving state-of-the-art (SOTA) performance in offline transcription. However, these models are not designed for streaming (online or real-time) transcription, due to limitations in their architecture and training methodology. We propose a method to turn the transformer encoder-decoder model into a low-latency streaming model that is careless about future context. We present an analysis explaining why it is not straightforward to convert an encoder-decoder transformer to a low-latency streaming model. Our proposed method modifies the existing (non-causal) encoder to a causal encoder by fine-tuning both the encoder and decoder using Low-Rank Adaptation (LoRA) and a weakly aligned dataset. We then propose an updated inference mechanism that utilizes the fine-tune causal encoder and decoder to yield greedy and beam-search decoding, and is shown to be locally optimal. Experiments on low-latency chunk sizes (less than 300 msec) show that our fine-tuned model outperforms existing non-fine-tuned streaming approaches in most cases, while using a lower complexity. Additionally, we observe that our training process yields better alignment, enabling a simple method for extracting word-level timestamps. We release our training and inference code, along with the fine-tuned models, to support further research and development in streaming ASR.

Natural language understanding (NLU) tasks face a non-trivial amount of ambiguous samples where veracity of their labels is debatable among annotators. NLU models should thus account for such ambiguity, but they approximate the human opinion distributions quite poorly and tend to produce over-confident predictions. To address this problem, we must consider how to exactly capture the degree of relationship between each sample and its candidate classes. In this work, we propose a novel method with deep model compression and show how such relationship can be accounted for. We see that more reasonably represented relationships can be discovered in the lower layers and that validation accuracies are converging at these layers, which naturally leads to layer pruning. We also see that distilling the relationship knowledge from a lower layer helps models produce better distribution. Experimental results demonstrate that our method makes substantial improvement on quantifying ambiguity without gold distribution labels. As positive side-effects, our method is found to reduce the model size significantly and improve latency, both attractive aspects of NLU products.

Large language models (LLMs) have recently shown remarkable performance across a wide range of tasks. However, the substantial number of parameters in LLMs contributes to significant latency during model inference. This is particularly evident when utilizing autoregressive decoding methods, which generate one token in a single forward process, thereby not fully capitalizing on the parallel computing capabilities of GPUs. In this paper, we propose a novel parallel decoding approach, namely hidden transfer, which decodes multiple successive tokens simultaneously in a single forward pass. The idea is to transfer the intermediate hidden states of the previous context to the pseudo hidden states of the future tokens to be generated, and then the pseudo hidden states will pass the following transformer layers thereby assimilating more semantic information and achieving superior predictive accuracy of the future tokens. Besides, we use the novel tree attention mechanism to simultaneously generate and verify multiple candidates of output sequences, which ensure the lossless generation and further improves the generation efficiency of our method. Experiments demonstrate the effectiveness of our method. We conduct a lot of analytic experiments to prove our motivation. In terms of acceleration metrics, we outperform all the single-model acceleration techniques, including Medusa and Self-Speculative decoding.

Scaling laws are powerful tools to predict the performance of large language models. However, current scaling laws fall short of accounting for inference costs. In this work, we first show that model architecture affects inference latency, where models of the same size can have up to 3.5x difference in latency. To tackle this challenge, we modify the Chinchilla scaling laws to co-optimize the model parameter count, the number of training tokens, and the model architecture. Due to the reason that models of similar training loss exhibit gaps in downstream evaluation, we also propose a novel method to train inference-efficient models based on the revised scaling laws. We perform extensive empirical studies to fit and evaluate our inference-aware scaling laws. We vary model parameters from 80M to 1B, training tokens from 1.6B to 30B, and model shapes, training a total of 63 models. Guided by our inference-efficient scaling law and model selection method, we release the Morph-1B model, which improves inference latency by 1.8x while maintaining accuracy on downstream tasks compared to open-source models, pushing the Pareto frontier of accuracy-latency tradeoff.

As large language models (LLMs) increasingly tackle complex reasoning tasks, test-time scaling has become critical for enhancing capabilities. However, in agentic scenarios with frequent tool calls, the traditional generation-length-based definition breaks down: tool latency decouples inference time from generation length. We propose Timely Machine, redefining test-time as wall-clock time, where models dynamically adjust strategies based on time budgets. We introduce Timely-Eval, a benchmark spanning high-frequency tool calls, low-frequency tool calls, and time-constrained reasoning. By varying tool latency, we find smaller models excel with fast feedback through more interactions, while larger models dominate high-latency settings via superior interaction quality. Moreover, existing models fail to adapt reasoning to time budgets. We propose Timely-RL to address this gap. After cold-start supervised fine-tuning, we use reinforcement learning to enhance temporal planning. Timely-RL improves time budget awareness and consistently boosts performance across Timely-Eval. We hope our work offers a new perspective on test-time scaling for the agentic era.

The pre-dominant approach to language modeling to date is based on recurrent neural networks. Their success on this task is often linked to their ability to capture unbounded context. In this paper we develop a finite context approach through stacked convolutions, which can be more efficient since they allow parallelization over sequential tokens. We propose a novel simplified gating mechanism that outperforms Oord et al (2016) and investigate the impact of key architectural decisions. The proposed approach achieves state-of-the-art on the WikiText-103 benchmark, even though it features long-term dependencies, as well as competitive results on the Google Billion Words benchmark. Our model reduces the latency to score a sentence by an order of magnitude compared to a recurrent baseline. To our knowledge, this is the first time a non-recurrent approach is competitive with strong recurrent models on these large scale language tasks.

Large encoder-decoder models like Whisper achieve strong offline transcription but remain impractical for streaming applications due to high latency. However, due to the accessibility of pre-trained checkpoints, the open Thai ASR landscape remains dominated by these offline architectures, leaving a critical gap in efficient streaming solutions. We present Typhoon ASR Real-time, a 115M-parameter FastConformer-Transducer model for low-latency Thai speech recognition. We demonstrate that rigorous text normalization can match the impact of model scaling: our compact model achieves a 45x reduction in computational cost compared to Whisper Large-v3 while delivering comparable accuracy. Our normalization pipeline resolves systemic ambiguities in Thai transcription --including context-dependent number verbalization and repetition markers (mai yamok) --creating consistent training targets. We further introduce a two-stage curriculum learning approach for Isan (north-eastern) dialect adaptation that preserves Central Thai performance. To address reproducibility challenges in Thai ASR, we release the Typhoon ASR Benchmark, a gold-standard human-labeled datasets with transcriptions following established Thai linguistic conventions, providing standardized evaluation protocols for the research community.

Voice dictation is an increasingly important text input modality. Existing systems that allow both dictation and editing-by-voice restrict their command language to flat templates invoked by trigger words. In this work, we study the feasibility of allowing users to interrupt their dictation with spoken editing commands in open-ended natural language. We introduce a new task and dataset, TERTiUS, to experiment with such systems. To support this flexibility in real-time, a system must incrementally segment and classify spans of speech as either dictation or command, and interpret the spans that are commands. We experiment with using large pre-trained language models to predict the edited text, or alternatively, to predict a small text-editing program. Experiments show a natural trade-off between model accuracy and latency: a smaller model achieves 30% end-state accuracy with 1.3 seconds of latency, while a larger model achieves 55% end-state accuracy with 7 seconds of latency.

The performance of multi-turn, agentic LLM inference is increasingly dominated by KV-Cache storage I/O rather than computation. In prevalent disaggregated architectures, loading the massive KV-Cache from external storage creates a fundamental imbalance: storage NICs on prefill engines become bandwidth-saturated, while those on decoding engines remain idle. This asymmetry severely constrains overall system throughput. We present DualPath, an inference system that breaks this bottleneck by introducing dual-path KV-Cache loading. Beyond the traditional storage-to-prefill path, DualPath enables a novel storage-to-decode path, in which the KV-Cache is loaded into decoding engines and then efficiently transferred to prefill engines via RDMA over the compute network. DualPath combines this optimized data path -- which inherently avoids network congestion and avoids interference with latency-critical model execution communications -- with a global scheduler that dynamically balances load across prefill and decode engines. Our evaluation on three models with production agentic workloads demonstrates that DualPath improves offline inference throughput by up to 1.87times on our in-house inference system. It can also improve online serving throughput by an average factor of 1.96times without violating SLO.

Automated speech recognition (ASR) models have gained prominence for applications such as captioning, speech translation, and live transcription. This paper studies Whisper and two model variants: one optimized for live speech streaming and another for offline transcription. Notably, these models have been found to generate hallucinated content, reducing transcription reliability. Furthermore, larger model variants exhibit increased latency and pose challenges for deployment on resource-constrained devices. This study analyzes the similarities and differences between three Whisper models, qualitatively examining their distinct capabilities. Next, this study quantifies the impact of model quantization on latency and evaluates its viability for edge deployment. Using the open source LibriSpeech dataset, this paper evaluates the word error rate (WER) along with latency analysis of whispercpp using 3 quantization methods (INT4, INT5, INT8). Results show that quantization reduces latency by 19\% and model size by 45\%, while preserving transcription accuracy. These findings provide insights into the optimal use cases of different Whisper models and edge device deployment possibilities. All code, datasets, and implementation details are available in a public GitHub repository: https://github.com/allisonandreyev/WhisperQuantization.git

Autonomous systems (AS) are systems that can adapt and change their behavior in response to unanticipated events and include systems such as aerial drones, autonomous vehicles, and ground/aquatic robots. AS require a wide array of sensors, deep-learning models, and powerful hardware platforms to perceive and safely operate in real-time. However, in many contexts, some sensing modalities negatively impact perception while increasing the system's overall energy consumption. Since AS are often energy-constrained edge devices, energy-efficient sensor fusion methods have been proposed. However, existing methods either fail to adapt to changing scenario conditions or to optimize energy efficiency system-wide. We propose CARMA: a context-aware sensor fusion approach that uses context to dynamically reconfigure the computation flow on a Field-Programmable Gate Array (FPGA) at runtime. By clock-gating unused sensors and model sub-components, CARMA significantly reduces the energy used by a multi-sensory object detector without compromising performance. We use a Deep-learning Processor Unit (DPU) based reconfiguration approach to minimize the latency of model reconfiguration. We evaluate multiple context-identification strategies, propose a novel system-wide energy-performance joint optimization, and evaluate scenario-specific perception performance. Across challenging real-world sensing contexts, CARMA outperforms state-of-the-art methods with up to 1.3x speedup and 73% lower energy consumption.

Compared with flatbed scanners, portable smartphones provide more convenience for physical document digitization. However, such digitized documents are often distorted due to uncontrolled physical deformations, camera positions, and illumination variations. To this end, we present DocScanner, a novel framework for document image rectification. Different from existing solutions, DocScanner addresses this issue by introducing a progressive learning mechanism. Specifically, DocScanner maintains a single estimate of the rectified image, which is progressively corrected with a recurrent architecture. The iterative refinements make DocScanner converge to a robust and superior rectification performance, while the lightweight recurrent architecture ensures the running efficiency. To further improve the rectification quality, based on the geometric priori between the distorted and the rectified images, a geometric regularization is introduced during training to further improve the performance. Extensive experiments are conducted on the Doc3D dataset and the DocUNet Benchmark dataset, and the quantitative and qualitative evaluation results verify the effectiveness of DocScanner, which outperforms previous methods on OCR accuracy, image similarity, and our proposed distortion metric by a considerable margin. Furthermore, our DocScanner shows superior efficiency in runtime latency and model size.

Existing text-to-image (T2I) diffusion models face several limitations, including large model sizes, slow runtime, and low-quality generation on mobile devices. This paper aims to address all of these challenges by developing an extremely small and fast T2I model that generates high-resolution and high-quality images on mobile platforms. We propose several techniques to achieve this goal. First, we systematically examine the design choices of the network architecture to reduce model parameters and latency, while ensuring high-quality generation. Second, to further improve generation quality, we employ cross-architecture knowledge distillation from a much larger model, using a multi-level approach to guide the training of our model from scratch. Third, we enable a few-step generation by integrating adversarial guidance with knowledge distillation. For the first time, our model SnapGen, demonstrates the generation of 1024x1024 px images on a mobile device around 1.4 seconds. On ImageNet-1K, our model, with only 372M parameters, achieves an FID of 2.06 for 256x256 px generation. On T2I benchmarks (i.e., GenEval and DPG-Bench), our model with merely 379M parameters, surpasses large-scale models with billions of parameters at a significantly smaller size (e.g., 7x smaller than SDXL, 14x smaller than IF-XL).

We study the problem of efficient generative inference for Transformer models, in one of its most challenging settings: large deep models, with tight latency targets and long sequence lengths. Better understanding of the engineering tradeoffs for inference for large Transformer-based models is important as use cases of these models are growing rapidly throughout application areas. We develop a simple analytical model for inference efficiency to select the best multi-dimensional partitioning techniques optimized for TPU v4 slices based on the application requirements. We combine these with a suite of low-level optimizations to achieve a new Pareto frontier on the latency and model FLOPS utilization (MFU) tradeoffs on 500B+ parameter models that outperforms the FasterTransformer suite of benchmarks. We further show that with appropriate partitioning, the lower memory requirements of multiquery attention (i.e. multiple query heads share single key/value head) enables scaling up to 32x larger context lengths. Finally, we achieve a low-batch-size latency of 29ms per token during generation (using int8 weight quantization) and a 76% MFU during large-batch-size processing of input tokens, while supporting a long 2048-token context length on the PaLM 540B parameter model.

Recent one-shot Neural Architecture Search algorithms rely on training a hardware-agnostic super-network tailored to a specific task and then extracting efficient sub-networks for different hardware platforms. Popular approaches separate the training of super-networks from the search for sub-networks, often employing predictors to alleviate the computational overhead associated with search. Additionally, certain methods also incorporate the quantization policy within the search space. However, while the quantization policy search for convolutional neural networks is well studied, the extension of these methods to transformers and especially foundation models remains under-explored. In this paper, we demonstrate that by using multi-objective search algorithms paired with lightly trained predictors, we can efficiently search for both the sub-network architecture and the corresponding quantization policy and outperform their respective baselines across different performance objectives such as accuracy, model size, and latency. Specifically, we demonstrate that our approach performs well across both uni-modal (ViT and BERT) and multi-modal (BEiT-3) transformer-based architectures as well as convolutional architectures (ResNet). For certain networks, we demonstrate an improvement of up to 4.80x and 3.44x for latency and model size respectively, without degradation in accuracy compared to the fully quantized INT8 baselines.

Simultaneous translation systems start producing the output while processing the partial source sentence in the incoming input stream. These systems need to decide when to read more input and when to write the output. These decisions depend on the structure of source/target language and the information contained in the partial input sequence. Hence, read/write decision policy remains the same across different input modalities, i.e., speech and text. This motivates us to leverage the text transcripts corresponding to the speech input for improving simultaneous speech-to-text translation (SimulST). We propose Decision Attentive Regularization (DAR) to improve the decision policy of SimulST systems by using the simultaneous text-to-text translation (SimulMT) task. We also extend several techniques from the offline speech translation domain to explore the role of SimulMT task in improving SimulST performance. Overall, we achieve 34.66% / 4.5 BLEU improvement over the baseline model across different latency regimes for the MuST-C English-German (EnDe) SimulST task.

Learned sparse retrieval (LSR) is a family of first-stage retrieval methods that are trained to generate sparse lexical representations of queries and documents for use with an inverted index. Many LSR methods have been recently introduced, with Splade models achieving state-of-the-art performance on MSMarco. Despite similarities in their model architectures, many LSR methods show substantial differences in effectiveness and efficiency. Differences in the experimental setups and configurations used make it difficult to compare the methods and derive insights. In this work, we analyze existing LSR methods and identify key components to establish an LSR framework that unifies all LSR methods under the same perspective. We then reproduce all prominent methods using a common codebase and re-train them in the same environment, which allows us to quantify how components of the framework affect effectiveness and efficiency. We find that (1) including document term weighting is most important for a method's effectiveness, (2) including query weighting has a small positive impact, and (3) document expansion and query expansion have a cancellation effect. As a result, we show how removing query expansion from a state-of-the-art model can reduce latency significantly while maintaining effectiveness on MSMarco and TripClick benchmarks. Our code is publicly available at https://github.com/thongnt99/learned-sparse-retrieval

Segment Anything Model 2 (SAM 2) has emerged as a powerful tool for video object segmentation and tracking anything. Key components of SAM 2 that drive the impressive video object segmentation performance include a large multistage image encoder for frame feature extraction and a memory mechanism that stores memory contexts from past frames to help current frame segmentation. The high computation complexity of multistage image encoder and memory module has limited its applications in real-world tasks, e.g., video object segmentation on mobile devices. To address this limitation, we propose EfficientTAMs, lightweight track anything models that produce high-quality results with low latency and model size. Our idea is based on revisiting the plain, nonhierarchical Vision Transformer (ViT) as an image encoder for video object segmentation, and introducing an efficient memory module, which reduces the complexity for both frame feature extraction and memory computation for current frame segmentation. We take vanilla lightweight ViTs and efficient memory module to build EfficientTAMs, and train the models on SA-1B and SA-V datasets for video object segmentation and track anything tasks. We evaluate on multiple video segmentation benchmarks including semi-supervised VOS and promptable video segmentation, and find that our proposed EfficientTAM with vanilla ViT perform comparably to SAM 2 model (HieraB+SAM 2) with ~2x speedup on A100 and ~2.4x parameter reduction. On segment anything image tasks, our EfficientTAMs also perform favorably over original SAM with ~20x speedup on A100 and ~20x parameter reduction. On mobile devices such as iPhone 15 Pro Max, our EfficientTAMs can run at ~10 FPS for performing video object segmentation with reasonable quality, highlighting the capability of small models for on-device video object segmentation applications.

As Large Language Models (LLMs) can now process extremely long contexts, efficient inference over these extended inputs has become increasingly important, especially for emerging applications like LLM agents that highly depend on this capability. Speculative decoding (SD) offers a promising lossless acceleration technique compared to lossy alternatives such as quantization and model cascades. However, most state-of-the-art SD methods are trained on short texts (typically fewer than 4k tokens), making them unsuitable for long-context scenarios. Specifically, adapting these methods to long contexts presents three key challenges: (1) the excessive memory demands posed by draft models due to large Key-Value (KV) cache; (2) performance degradation resulting from the mismatch between short-context training and long-context inference; and (3) inefficiencies in tree attention mechanisms when managing long token sequences. This work introduces LongSpec, a framework that addresses these challenges through three core innovations: a memory-efficient draft model with a constant-sized KV cache; novel position indices that mitigate the training-inference mismatch; and an attention aggregation strategy that combines fast prefix computation with standard tree attention to enable efficient decoding. Experimental results confirm the effectiveness of LongSpec, achieving up to a 3.26x speedup over strong Flash Attention baselines across five long-context understanding datasets, as well as a 2.25x reduction in wall-clock time on the AIME24 long reasoning task with the QwQ model, demonstrating significant latency improvements for long-context applications. The code is available at https://github.com/sail-sg/LongSpec.

Deep learning models are increasingly deployed on resource-constrained edge devices for real-time data analytics. In recent years, Vision Transformer models and their variants have demonstrated outstanding performance across various computer vision tasks. However, their high computational demands and inference latency pose significant challenges for model deployment on resource-constraint edge devices. To address this issue, we propose a novel Vision Transformer splitting framework, ED-ViT, designed to execute complex models across multiple edge devices efficiently. Specifically, we partition Vision Transformer models into several sub-models, where each sub-model is tailored to handle a specific subset of data classes. To further minimize computation overhead and inference latency, we introduce a class-wise pruning technique that reduces the size of each sub-model. We conduct extensive experiments on five datasets with three model structures, demonstrating that our approach significantly reduces inference latency on edge devices and achieves a model size reduction of up to 28.9 times and 34.1 times, respectively, while maintaining test accuracy comparable to the original Vision Transformer. Additionally, we compare ED-ViT with two state-of-the-art methods that deploy CNN and SNN models on edge devices, evaluating accuracy, inference time, and overall model size. Our comprehensive evaluation underscores the effectiveness of the proposed ED-ViT framework.

Pre-trained language models (PLM), for example BERT or RoBERTa, mark the state-of-the-art for natural language understanding task when fine-tuned on labeled data. However, their large size poses challenges in deploying them for inference in real-world applications, due to significant GPU memory requirements and high inference latency. This paper explores neural architecture search (NAS) for structural pruning to find sub-parts of the fine-tuned network that optimally trade-off efficiency, for example in terms of model size or latency, and generalization performance. We also show how we can utilize more recently developed two-stage weight-sharing NAS approaches in this setting to accelerate the search process. Unlike traditional pruning methods with fixed thresholds, we propose to adopt a multi-objective approach that identifies the Pareto optimal set of sub-networks, allowing for a more flexible and automated compression process.

Recent advancements in large language models (LLMs) have prompted interest in deploying these models on mobile devices to enable new applications without relying on cloud connectivity. However, the efficiency constraints of deploying LLMs on resource-limited devices present significant challenges. In this paper, we conduct a comprehensive measurement study to evaluate the efficiency tradeoffs between mobile-based, edge-based, and cloud-based deployments for LLM applications. We implement AutoLife-Lite, a simplified LLM-based application that analyzes smartphone sensor data to infer user location and activity contexts. Our experiments reveal that: (1) Only small-size LLMs (<4B parameters) can run successfully on powerful mobile devices, though they exhibit quality limitations compared to larger models; (2) Model compression is effective in lower the hardware requirement, but may lead to significant performance degradation; (3) The latency to run LLMs on mobile devices with meaningful output is significant (>30 seconds), while cloud services demonstrate better time efficiency (<10 seconds); (4) Edge deployments offer intermediate tradeoffs between latency and model capabilities, with different results on CPU-based and GPU-based settings. These findings provide valuable insights for system designers on the current limitations and future directions for on-device LLM applications.

Efficiently serving neural network models with low latency is becoming more challenging due to increasing model complexity and parameter count. Model quantization offers a solution which simultaneously reduces memory footprint and compute requirements. However, aggressive quantization may lead to an unacceptable loss in model accuracy owing to differences in sensitivity to numerical imperfection across different layers in the model. To address this challenge, we propose a mixed-precision post training quantization (PTQ) approach that assigns different numerical precisions to tensors in a network based on their specific needs, for a reduced memory footprint and improved latency while preserving model accuracy. Previous works rely on layer-wise Hessian information to determine numerical precision, but as we demonstrate, Hessian estimation is typically insufficient in determining an effective ordering of layer sensitivities. We address this by augmenting the estimated Hessian with additional information to capture inter-layer dependencies. We demonstrate that this consistently improves PTQ performance along the accuracy-latency Pareto frontier across multiple models. Our method combines second-order information and inter-layer dependencies to guide a bisection search, finding quantization configurations within a user-configurable model accuracy degradation range. We evaluate the effectiveness of our method on the ResNet50, MobileNetV2, and BERT models. Our experiments demonstrate latency reductions compared to a 16-bit baseline of 25.48%, 21.69%, and 33.28% respectively, while maintaining model accuracy to within 99.99% of the baseline model.

Multimodal language models that process both text and speech have a potential for applications in spoken dialogue systems. However, current models face two major challenges in response generation latency: (1) generating a spoken response requires the prior generation of a written response, and (2) speech sequences are significantly longer than text sequences. This study addresses these issues by extending the input and output sequences of the language model to support the parallel generation of text and speech. Our experiments on spoken question answering tasks demonstrate that our approach improves latency while maintaining the quality of response content. Additionally, we show that latency can be further reduced by generating speech in multiple sequences. Demo samples are available at https://rinnakk.github.io/research/publications/PSLM.

Model efficiency is a critical aspect of developing and deploying machine learning models. Inference time and latency directly affect the user experience, and some applications have hard requirements. In addition to inference costs, model training also have direct financial and environmental impacts. Although there are numerous well-established metrics (cost indicators) for measuring model efficiency, researchers and practitioners often assume that these metrics are correlated with each other and report only few of them. In this paper, we thoroughly discuss common cost indicators, their advantages and disadvantages, and how they can contradict each other. We demonstrate how incomplete reporting of cost indicators can lead to partial conclusions and a blurred or incomplete picture of the practical considerations of different models. We further present suggestions to improve reporting of efficiency metrics.

This paper addresses the inherent inference latency challenges associated with deploying multimodal large language models (MLLM) in quadruped vision-language-action (QUAR-VLA) tasks. Our investigation reveals that conventional parameter reduction techniques ultimately impair the performance of the language foundation model during the action instruction tuning phase, making them unsuitable for this purpose. We introduce a novel latency-free quadruped MLLM model, dubbed QUART-Online, designed to enhance inference efficiency without degrading the performance of the language foundation model. By incorporating Action Chunk Discretization (ACD), we compress the original action representation space, mapping continuous action values onto a smaller set of discrete representative vectors while preserving critical information. Subsequently, we fine-tune the MLLM to integrate vision, language, and compressed actions into a unified semantic space. Experimental results demonstrate that QUART-Online operates in tandem with the existing MLLM system, achieving real-time inference in sync with the underlying controller frequency, significantly boosting the success rate across various tasks by 65%. Our project page is https://quart-online.github.io.

Efficient deployment of small language models (SLMs) is essential for numerous real-world applications with stringent latency constraints. While previous work on SLM design has primarily focused on reducing the number of parameters to achieve parameter-optimal SLMs, parameter efficiency does not necessarily translate into proportional real-device speed-ups. This work aims to identify the key determinants of SLMs' real-device latency and offer generalizable principles and methodologies for SLM design and training when real-device latency is the primary consideration. Specifically, we identify two central architectural factors: depth-width ratios and operator choices. The former is crucial for small-batch-size latency, while the latter affects both latency and large-batch-size throughput. In light of this, we first study latency-optimal depth-width ratios, with the key finding that although deep-thin models generally achieve better accuracy under the same parameter budget, they may not lie on the accuracy-latency trade-off frontier. Next, we explore emerging efficient attention alternatives to evaluate their potential as candidate building operators. Using the identified promising operators, we construct an evolutionary search framework to automatically discover latency-optimal combinations of these operators within hybrid SLMs, thereby advancing the accuracy-latency frontier. In addition to architectural improvements, we further enhance SLM training using a weight normalization technique that enables more effective weight updates and improves final convergence. Combining these methods, we introduce a new family of hybrid SLMs, called Nemotron-Flash, which significantly advances the accuracy-efficiency frontier of state-of-the-art SLMs, e.g., achieving over +5.5% average accuracy, 1.3x/1.9x lower latency, and 18.7x/45.6x higher throughput compared to Qwen3-1.7B/0.6B, respectively.

In this paper, we introduce Flash-VL 2B, a novel approach to optimizing Vision-Language Models (VLMs) for real-time applications, targeting ultra-low latency and high throughput without sacrificing accuracy. Leveraging advanced architectural enhancements and efficient computational strategies, Flash-VL 2B is designed to maximize throughput by reducing processing time while maintaining competitive performance across multiple vision-language benchmarks. Our approach includes tailored architectural choices, token compression mechanisms, data curation, training schemes, and a novel image processing technique called implicit semantic stitching that effectively balances computational load and model performance. Through extensive evaluations on 11 standard VLM benchmarks, we demonstrate that Flash-VL 2B achieves state-of-the-art results in both speed and accuracy, making it a promising solution for deployment in resource-constrained environments and large-scale real-time applications.

Rapidly developing large language models (LLMs) have brought tremendous intelligent applications. Especially, the GPT-4o's excellent duplex speech interaction ability has brought impressive experience to users. Researchers have recently proposed several multi-modal LLMs in this direction that can achieve user-agent speech-to-speech conversations. This paper proposes a novel speech-text multimodal LLM architecture called Freeze-Omni. Our main contribution is that the speech input and output modalities can be easily connected to a textual LLM while keeping the LLM's parameters frozen throughout the training process. We design a three-stage training strategy for modeling both the speech input and output, enabling Freeze-Omni to obtain speech-to-speech conversation ability using text-speech paired data (such as ASR and TTS data) and only 60,000 multi-round text Q&A data on 8 GPUs. Moreover, we can effectively ensure that the intelligence of the Freeze-Omni in the speech modality is at the same level compared with that in the text modality of its backbone LLM, while achieving low latency end-to-end spoken response. In addition, we also designed a method to achieve duplex dialogue ability through multi-task training, giving Freeze-Omni a more natural style of dialogue ability between users and agents. In summary, Freeze-Omni holds great potential to conduct speech-to-speech dialogue based on a multimodal LLM under the condition of a frozen LLM, avoiding the catastrophic forgetting problem caused by limited data and training resources.

A new golden age in astronomy is upon us, dominated by data. Large astronomical surveys are broadcasting unprecedented rates of information, demanding machine learning as a critical component in modern scientific pipelines to handle the deluge of data. The upcoming Legacy Survey of Space and Time (LSST) of the Vera C. Rubin Observatory will raise the big-data bar for time-domain astronomy, with an expected 10 million alerts per-night, and generating many petabytes of data over the lifetime of the survey. Fast and efficient classification algorithms that can operate in real-time, yet robustly and accurately, are needed for time-critical events where additional resources can be sought for follow-up analyses. In order to handle such data, state-of-the-art deep learning architectures coupled with tools that leverage modern hardware accelerators are essential. We showcase how the use of modern deep compression methods can achieve a 18times reduction in model size, whilst preserving classification performance. We also show that in addition to the deep compression techniques, careful choice of file formats can improve inference latency, and thereby throughput of alerts, on the order of 8times for local processing, and 5times in a live production setting. To test this in a live setting, we deploy this optimised version of the original time-series transformer, t2, into the community alert broking system of FINK on real Zwicky Transient Facility (ZTF) alert data, and compare throughput performance with other science modules that exist in FINK. The results shown herein emphasise the time-series transformer's suitability for real-time classification at LSST scale, and beyond, and introduce deep model compression as a fundamental tool for improving deploy-ability and scalable inference of deep learning models for transient classification.

Test-Time Scaling (TTS) has proven effective in improving the performance of Large Language Models (LLMs) during inference. However, existing research has overlooked the efficiency of TTS from a latency-sensitive perspective. Through a latency-aware evaluation of representative TTS methods, we demonstrate that a compute-optimal TTS does not always result in the lowest latency in scenarios where latency is critical. To address this gap and achieve latency-optimal TTS, we propose two key approaches by optimizing the concurrency configurations: (1) branch-wise parallelism, which leverages multiple concurrent inference branches, and (2) sequence-wise parallelism, enabled by speculative decoding. By integrating these two approaches and allocating computational resources properly to each, our latency-optimal TTS enables a 32B model to reach 82.3% accuracy on MATH-500 within 1 minute and a smaller 3B model to achieve 72.4% within 10 seconds. Our work emphasizes the importance of latency-aware TTS and demonstrates its ability to deliver both speed and accuracy in latency-sensitive scenarios.

For deployment, neural architecture search should be hardware-aware, in order to satisfy the device-specific constraints (e.g., memory usage, latency and energy consumption) and enhance the model efficiency. Existing methods on hardware-aware NAS collect a large number of samples (e.g., accuracy and latency) from a target device, either builds a lookup table or a latency estimator. However, such approach is impractical in real-world scenarios as there exist numerous devices with different hardware specifications, and collecting samples from such a large number of devices will require prohibitive computational and monetary cost. To overcome such limitations, we propose Hardware-adaptive Efficient Latency Predictor (HELP), which formulates the device-specific latency estimation problem as a meta-learning problem, such that we can estimate the latency of a model's performance for a given task on an unseen device with a few samples. To this end, we introduce novel hardware embeddings to embed any devices considering them as black-box functions that output latencies, and meta-learn the hardware-adaptive latency predictor in a device-dependent manner, using the hardware embeddings. We validate the proposed HELP for its latency estimation performance on unseen platforms, on which it achieves high estimation performance with as few as 10 measurement samples, outperforming all relevant baselines. We also validate end-to-end NAS frameworks using HELP against ones without it, and show that it largely reduces the total time cost of the base NAS method, in latency-constrained settings. Code is available at https://github.com/HayeonLee/HELP.

Latency and efficiency issues are often overlooked when evaluating IR models based on Pretrained Language Models (PLMs) in reason of multiple hardware and software testing scenarios. Nevertheless, efficiency is an important part of such systems and should not be overlooked. In this paper, we focus on improving the efficiency of the SPLADE model since it has achieved state-of-the-art zero-shot performance and competitive results on TREC collections. SPLADE efficiency can be controlled via a regularization factor, but solely controlling this regularization has been shown to not be efficient enough. In order to reduce the latency gap between SPLADE and traditional retrieval systems, we propose several techniques including L1 regularization for queries, a separation of document/query encoders, a FLOPS-regularized middle-training, and the use of faster query encoders. Our benchmark demonstrates that we can drastically improve the efficiency of these models while increasing the performance metrics on in-domain data. To our knowledge, {we propose the first neural models that, under the same computing constraints, achieve similar latency (less than 4ms difference) as traditional BM25, while having similar performance (less than 10\% MRR@10 reduction) as the state-of-the-art single-stage neural rankers on in-domain data}.

Recently developed large language models (LLMs) such as ChatGPT, Claude, and Llama have demonstrated impressive abilities, and even surpass human-level performance in several tasks. Despite their success, the resource-intensive demands of these models, requiring significant computational power for both training and inference, limit their deployment to high-performance servers. Additionally, the extensive calculation requirements of the models often lead to increased latency in response times. With the increasing need for LLMs to operate efficiently on CPUs, research about lightweight models that are optimized for CPU inference has emerged. In this work, we introduce GEB-1.3B, a lightweight LLM trained on 550 billion tokens in both Chinese and English languages. We employ novel training techniques, including ROPE, Group-Query-Attention, and FlashAttention-2, to accelerate training while maintaining model performance. Additionally, we fine-tune the model using 10 million samples of instruction data to enhance alignment. GEB-1.3B exhibits outstanding performance on general benchmarks such as MMLU, C-Eval, and CMMLU, outperforming comparative models such as MindLLM-1.3B and TinyLLaMA-1.1B. Notably, the FP32 version of GEB-1.3B achieves commendable inference times on CPUs, with ongoing efforts to further enhance speed through advanced quantization techniques. The release of GEB-1.3B as an open-source model marks a significant contribution to the development of lightweight LLMs, promising to foster further research and innovation in the field.

Large language models (LLMs) have surged in popularity and are extensively used in commercial applications, where the efficiency of model serving is crucial for the user experience. Most current research focuses on optimizing individual sub-procedures, e.g. local inference and communication, however, there is no comprehensive framework that provides a holistic system view for optimizing LLM serving in an end-to-end manner. In this work, we conduct a detailed analysis to identify major bottlenecks that impact end-to-end latency in LLM serving systems. Our analysis reveals that a comprehensive LLM serving endpoint must address a series of efficiency bottlenecks that extend beyond LLM inference. We then propose ScaleLLM, an optimized system for resource-efficient LLM serving. Our extensive experiments reveal that with 64 concurrent requests, ScaleLLM achieves a 4.3x speed up over vLLM and outperforms state-of-the-arts with 1.5x higher throughput.

In this paper, we introduce a novel low-latency inference framework for large language models (LLMs) inference which enables LLMs to perform inferences with incomplete prompts. By reallocating computational processes to prompt input phase, we achieve a substantial reduction in latency, thereby significantly enhancing the interactive experience for users of LLMs. The framework adeptly manages the visibility of the streaming prompt to the model, allowing it to infer from incomplete prompts or await additional prompts. Compared with traditional inference methods that utilize complete prompts, our approach demonstrates an average reduction of 59% in response latency on the MMLU-Pro dataset, while maintaining comparable accuracy. Additionally, our framework facilitates collaborative inference and output across different models. By employing an LLM for inference and a small language model (SLM) for output, we achieve an average 68% reduction in response latency, alongside a 5.5% improvement in accuracy on the MMLU-Pro dataset compared with the SLM baseline. For long prompts exceeding 20 sentences, the response latency can be reduced by up to 93%.

Models for low-latency, streaming applications could benefit from the knowledge capacity of larger models, but edge devices cannot run these models due to resource constraints. A possible solution is to transfer hints during inference from a large model running remotely to a small model running on-device. However, this incurs a communication delay that breaks real-time requirements and does not guarantee that both models will operate on the same data at the same time. We propose knowledge boosting, a novel technique that allows a large model to operate on time-delayed input during inference, while still boosting small model performance. Using a streaming neural network that processes 8 ms chunks, we evaluate different speech separation and enhancement tasks with communication delays of up to six chunks or 48 ms. Our results show larger gains where the performance gap between the small and large models is wide, demonstrating a promising method for large-small model collaboration for low-latency applications. Code, dataset, and audio samples available at https://knowledgeboosting.cs.washington.edu/.

We adapt the architectures of previous audio manipulation and generation neural networks to the task of real-time any-to-one voice conversion. Our resulting model, LLVC (Low-latency Low-resource Voice Conversion), has a latency of under 20ms at a bitrate of 16kHz and runs nearly 2.8x faster than real-time on a consumer CPU. LLVC uses both a generative adversarial architecture as well as knowledge distillation in order to attain this performance. To our knowledge LLVC achieves both the lowest resource usage as well as the lowest latency of any open-source voice conversion model. We provide open-source samples, code, and pretrained model weights at https://github.com/KoeAI/LLVC.

Large Language Models (LLMs) and other large foundation models have achieved noteworthy success, but their size exacerbates existing resource consumption and latency challenges. In particular, the large-scale deployment of these models is hindered by the significant resource requirements during inference. In this paper, we study two approaches for mitigating these challenges: employing a cache to store previous queries and learning a model multiplexer to choose from an ensemble of models for query processing. Theoretically, we provide an optimal algorithm for jointly optimizing both approaches to reduce the inference cost in both offline and online tabular settings. By combining a caching algorithm, namely Greedy Dual Size with Frequency (GDSF) or Least Expected Cost (LEC), with a model multiplexer, we achieve optimal rates in both offline and online settings. Empirically, simulations show that the combination of our caching and model multiplexing algorithms greatly improves over the baselines, with up to 50times improvement over the baseline when the ratio between the maximum cost and minimum cost is 100. Experiments on real datasets show a 4.3times improvement in FLOPs over the baseline when the ratio for FLOPs is 10, and a 1.8times improvement in latency when the ratio for average latency is 1.85.

Auto-regressive Large Language Models (LLMs) demonstrate remarkable performance across different domains such as vision and language processing. However, due to sequential processing through a stack of transformer layers, autoregressive decoding faces significant computation/latency challenges, particularly in resource-constrained environments like mobile and edge devices. Existing approaches in literature that aim to improve latency via skipping layers have two distinct flavors - 1) Early exit, and 2) Input-agnostic heuristics where tokens exit at pre-determined layers irrespective of input sequence. Both the above strategies have limitations - the former cannot be applied to handle KV Caching necessary for speed-ups in modern framework and the latter does not capture the variation in layer importance across tasks or more generally, across input sequences. To address both limitations, we propose FiRST, an algorithm that reduces inference latency by using layer-specific routers to select a subset of transformer layers adaptively for each input sequence - the prompt (during the prefill stage) decides which layers will be skipped during decoding. FiRST preserves compatibility with KV caching enabling faster inference while being quality-aware. FiRST is model-agnostic and can be easily enabled on any pre-trained LLM. Our approach reveals that input adaptivity is critical - indeed, different task-specific middle layers play a crucial role in evolving hidden representations depending on tasks. Extensive experiments show that FiRST significantly reduces latency while outperforming other layer selection strategies in quality metics. It retains competitive performance to base model (without layer skipping) and in some cases, even improves upon it. FiRST is thus a promising and efficient solution for LLM deployment in low-resource environments.

While scaling laws for large language models (LLMs) during pre-training have been extensively studied, their behavior under reinforcement learning (RL) post-training remains largely unexplored. This paper presents a systematic empirical investigation of scaling behaviors in RL-based post-training, with a particular focus on mathematical reasoning. Based on a set of experiments across the full Qwen2.5 dense model series (0.5B to 72B), we characterize how model scale, data volume, and computational budget interact to shape performance. Our analysis leads to four key findings: 1.Larger models consistently exhibit superior learning efficiency on both compute and data metrics. 2.The relationship between test loss, compute, and data can be modeled by a predictive power-law which is robust across both base and instruction-tuned models. 3.Although larger models exhibit higher learning efficiency, the analytical learning efficiency term k(N) in the power-law reveals a latent saturation trend in learning efficiency as model size continues to increase. 4.In data-constrained regimes, repeated reuse of high-quality data proves highly effective, as final performance is primarily governed by the total number of optimization steps rather than the uniqueness of samples. Collectively, these results provide a principled foundation and practical guidelines for efficiently scaling the reasoning capabilities of LLMs through RL post-training.

As Large Language Models (LLMs) achieve remarkable empirical success through scaling model and data size, pretraining has become increasingly critical yet computationally prohibitive, hindering rapid development. Despite the availability of numerous pretrained LLMs developed at significant computational expense, a fundamental real-world question remains underexplored: Can we leverage existing small pretrained models to accelerate the training of larger models? In this paper, we propose a Late-to-Early Training (LET) paradigm that enables LLMs to explicitly learn later knowledge in earlier steps and earlier layers. The core idea is to guide the early layers of an LLM during early training using representations from the late layers of a pretrained (i.e. late training phase) model. We identify two key mechanisms that drive LET's effectiveness: late-to-early-step learning and late-to-early-layer learning. These mechanisms significantly accelerate training convergence while robustly enhancing both language modeling capabilities and downstream task performance, enabling faster training with superior performance. Extensive experiments on 1.4B and 7B parameter models demonstrate LET's efficiency and effectiveness. Notably, when training a 1.4B LLM on the Pile dataset, our method achieves up to 1.6times speedup with nearly 5\% improvement in downstream task accuracy compared to standard training, even when using a pretrained model with 10times fewer parameters than the target model.

Vision-Language-Action Models (VLAs) have emerged as a key paradigm of Physical AI and are increasingly deployed in autonomous vehicles, robots, and smart spaces. In these resource-constrained on-device settings, selecting an appropriate large language model (LLM) backbone is a critical challenge: models must balance accuracy with strict inference latency and hardware efficiency constraints. This makes hardware-software co-design a game-changing requirement for on-device LLM deployment, where each hardware platform demands a tailored architectural solution. We propose a hardware co-design law that jointly captures model accuracy and inference performance. Specifically, we model training loss as an explicit function of architectural hyperparameters and characterise inference latency via roofline modelling. We empirically evaluate 1,942 candidate architectures on NVIDIA Jetson Orin, training 170 selected models for 10B tokens each to fit a scaling law relating architecture to training loss. By coupling this scaling law with latency modelling, we establish a direct accuracy-latency correspondence and identify the Pareto frontier for hardware co-designed LLMs. We further formulate architecture search as a joint optimisation over precision and performance, deriving feasible design regions under industrial hardware and application budgets. Our approach reduces architecture selection from months to days. At the same latency as Qwen2.5-0.5B on the target hardware, our co-designed architecture achieves 19.42% lower perplexity on WikiText-2. To our knowledge, this is the first principled and operational framework for hardware co-design scaling laws in on-device LLM deployment. We will make the code and related checkpoints publicly available.

Language models have shown effectiveness in a variety of software applications, particularly in tasks related to automatic workflow. These models possess the crucial ability to call functions, which is essential in creating AI agents. Despite the high performance of large-scale language models in cloud environments, they are often associated with concerns over privacy and cost. Current on-device models for function calling face issues with latency and accuracy. Our research presents a new method that empowers an on-device model with 2 billion parameters to surpass the performance of GPT-4 in both accuracy and latency, and decrease the context length by 95\%. When compared to Llama-7B with a RAG-based function calling mechanism, our method enhances latency by 35-fold. This method reduces the latency to levels deemed suitable for deployment across a variety of edge devices in production environments, aligning with the performance requisites for real-world applications.

Large Language Models (LLMs), such as OpenAI-o1 and DeepSeek-R1, have demonstrated strong reasoning capabilities. To further enhance LLM capabilities, recent agentic systems, such as Deep Research, incorporate web interactions into LLM reasoning to mitigate uncertainties and reduce potential errors. However, existing research predominantly focuses on reasoning performance, often neglecting the efficiency of agentic systems. In this work, we present a comprehensive empirical study that identifies efficiency bottlenecks in web-interactive agentic systems. We decompose end-to-end latency into two primary components: LLM API latency and web environment latency. We conduct a comprehensive empirical study across 15 models and 5 providers to demonstrate high variability in API-based agentic systems. We observe that web environment latency can contribute as much as 53.7% to the overall latency in a web-based agentic system. To improve latency, we propose SpecCache, a caching framework augmented with speculative execution that can reduce web environment overhead. Extensive evaluations on two standard benchmarks show that our approach improves the cache hit rate by up to 58x compared to a random caching strategy, while reducing web environment overhead by up to 3.2x, without degrading agentic system performance.

Deep Learning has revolutionized the fields of computer vision, natural language understanding, speech recognition, information retrieval and more. However, with the progressive improvements in deep learning models, their number of parameters, latency, resources required to train, etc. have all have increased significantly. Consequently, it has become important to pay attention to these footprint metrics of a model as well, not just its quality. We present and motivate the problem of efficiency in deep learning, followed by a thorough survey of the five core areas of model efficiency (spanning modeling techniques, infrastructure, and hardware) and the seminal work there. We also present an experiment-based guide along with code, for practitioners to optimize their model training and deployment. We believe this is the first comprehensive survey in the efficient deep learning space that covers the landscape of model efficiency from modeling techniques to hardware support. Our hope is that this survey would provide the reader with the mental model and the necessary understanding of the field to apply generic efficiency techniques to immediately get significant improvements, and also equip them with ideas for further research and experimentation to achieve additional gains.

We study the scaling properties of latent diffusion models (LDMs) with an emphasis on their sampling efficiency. While improved network architecture and inference algorithms have shown to effectively boost sampling efficiency of diffusion models, the role of model size -- a critical determinant of sampling efficiency -- has not been thoroughly examined. Through empirical analysis of established text-to-image diffusion models, we conduct an in-depth investigation into how model size influences sampling efficiency across varying sampling steps. Our findings unveil a surprising trend: when operating under a given inference budget, smaller models frequently outperform their larger equivalents in generating high-quality results. Moreover, we extend our study to demonstrate the generalizability of the these findings by applying various diffusion samplers, exploring diverse downstream tasks, evaluating post-distilled models, as well as comparing performance relative to training compute. These findings open up new pathways for the development of LDM scaling strategies which can be employed to enhance generative capabilities within limited inference budgets.

The recent emergence of Large Language Models based on the Transformer architecture has enabled dramatic advancements in the field of Natural Language Processing. However, these models have long inference latency, which limits their deployment and makes them prohibitively expensive for various real-time applications. The inference latency is further exacerbated by autoregressive generative tasks, as models need to run iteratively to generate tokens sequentially without leveraging token-level parallelization. To address this, we propose Big Little Decoder (BiLD), a framework that can improve inference efficiency and latency for a wide range of text generation applications. The BiLD framework contains two models with different sizes that collaboratively generate text. The small model runs autoregressively to generate text with a low inference cost, and the large model is only invoked occasionally to refine the small model's inaccurate predictions in a non-autoregressive manner. To coordinate the small and large models, BiLD introduces two simple yet effective policies: (1) the fallback policy that determines when to hand control over to the large model; and (2) the rollback policy that determines when the large model needs to correct the small model's inaccurate predictions. To evaluate our framework across different tasks and models, we apply BiLD to various text generation scenarios encompassing machine translation on IWSLT 2017 De-En and WMT 2014 De-En, and summarization on XSUM and CNN/DailyMail. On an NVIDIA T4 GPU, our framework achieves a speedup of up to 2.12x speedup with minimal generation quality degradation. Furthermore, our framework is fully plug-and-play and can be applied without any modifications in the training process or model architecture. Our code is open-sourced

Large Language Models (LLMs) exhibit potential artificial generic intelligence recently, however, their usage is costly with high response latency. Given mixed LLMs with their own strengths and weaknesses, LLM routing aims to identify the most suitable model for each query in the stream to maximize response quality and minimize cost and latency. However, the challenges involve: (1) dynamic trade-offs among quality, cost, and latency; (2) enabling continual learning in deployed systems; and (3) navigating a varying (e.g., new LLM addition or old LLM removal) set of LLM candidates over time. To bridge these gaps, we develop MixLLM, a dynamic contextual-bandit-based routing system for query-LLM assignment. Specifically, we first leverage query tags to enhance query embeddings for the routing task. Next, we design lightweight prediction models to estimate the response qualities and costs of queries over LLMs. We then devise a meta-decision maker to choose the query-LLM assignments to best tradeoff response quality, cost, and latency. Finally, the system benefits from continual training, allowing it to adapt to evolving queries and user feedback over time. Our extensive experiments show that MixLLM achieves the best trade-offs in response quality, cost, and latency (97.25% of GPT-4's quality at 24.18% of the cost under the time constraint).

Diffusion-based models have gained wide adoption in the virtual human generation due to their outstanding expressiveness. However, their substantial computational requirements have constrained their deployment in real-time interactive avatar applications, where stringent speed, latency, and duration requirements are paramount. We present a novel audio-driven portrait video generation framework based on the diffusion model to address these challenges. Firstly, we propose robust variable-length video generation to reduce the minimum time required to generate the initial video clip or state transitions, which significantly enhances the user experience. Secondly, we propose a consistency model training strategy for Audio-Image-to-Video to ensure real-time performance, enabling a fast few-step generation. Model quantization and pipeline parallelism are further employed to accelerate the inference speed. To mitigate the stability loss incurred by the diffusion process and model quantization, we introduce a new inference strategy tailored for long-duration video generation. These methods ensure real-time performance and low latency while maintaining high-fidelity output. Thirdly, we incorporate class labels as a conditional input to seamlessly switch between speaking, listening, and idle states. Lastly, we design a novel mechanism for fine-grained facial expression control to exploit our model's inherent capacity. Extensive experiments demonstrate that our approach achieves low-latency, fluid, and authentic two-way communication. On an NVIDIA RTX 4090D, our model achieves a maximum of 78 FPS at a resolution of 384x384 and 45 FPS at a resolution of 512x512, with an initial video generation latency of 140 ms and 215 ms, respectively.

Prior works have demonstrated zero-shot text-to-speech by using a generative language model on audio tokens obtained via a neural audio codec. It is still challenging, however, to adapt them to low-latency scenarios. In this paper, we present LiveSpeech - a fully autoregressive language model-based approach for zero-shot text-to-speech, enabling low-latency streaming of the output audio. To allow multiple token prediction within a single decoding step, we propose (1) using adaptive codebook loss weights that consider codebook contribution in each frame and focus on hard instances, and (2) grouping codebooks and processing groups in parallel. Experiments show our proposed models achieve competitive results to state-of-the-art baselines in terms of content accuracy, speaker similarity, audio quality, and inference speed while being suitable for low-latency streaming applications.

Real-time, on-device segmentation is critical for latency-sensitive and privacy-aware applications like smart glasses and IoT devices. We introduce PicoSAM2, a lightweight (1.3M parameters, 336M MACs) promptable segmentation model optimized for edge and in-sensor execution, including the Sony IMX500. It builds on a depthwise separable U-Net, with knowledge distillation and fixed-point prompt encoding to learn from the Segment Anything Model 2 (SAM2). On COCO and LVIS, it achieves 51.9% and 44.9% mIoU, respectively. The quantized model (1.22MB) runs at 14.3 ms on the IMX500-achieving 86 MACs/cycle, making it the only model meeting both memory and compute constraints for in-sensor deployment. Distillation boosts LVIS performance by +3.5% mIoU and +5.1% mAP. These results demonstrate that efficient, promptable segmentation is feasible directly on-camera, enabling privacy-preserving vision without cloud or host processing.

Adjusting the latency, power, and accuracy of natural language understanding models is a desirable objective of efficient architecture development. This paper proposes an efficient transformer architecture that adjusts the inference computational cost adaptively with desired inference latency speedup. The proposed encoder model can work with fewer Floating Point Operations (FLOPs) than the original Transformer architecture. In fine-tuning phase, the proposed method detects more important hidden sequence elements (word-vectors) in each encoder layer by a proposed Attention Context Contribution (ACC) metric. It eliminates the less important word-vectors based on a new strategy. A mathematical inference speedup analysis is proposed to estimate the speedup accurately to adjust the latency and computational cost of fine-tuning and inference phases. After the fine-tuning phase, by the method offline-tuning property, the inference latency of the model can be adjusted in a wide range of inference speedup selections. The proposed method is applied to the BERTbase model for evaluation. Extensive experiments show that most of the word-vectors in higher BERT encoder layers have less contribution to the subsequent layers; hence, they can be eliminated to improve the inference latency. Experimental results on extensive sentiment analysis, classification, and regression benchmarks like GLUE showed that the method is effective in various datasets. The proposed method improves the inference latency of BERTbase by up to 4.8 times with less than 0.75% accuracy drop on average.

Transformer-based Learned Image Compression (LIC) suffers from a suboptimal trade-off between decoding latency and rate-distortion (R-D) performance. Moreover, the critical role of the FeedForward Network (FFN)-based channel aggregation module has been largely overlooked. Our research reveals that efficient channel aggregation-rather than complex and time-consuming spatial operations-is the key to achieving competitive LIC models. Based on this insight, we initiate the ``S2CFormer'' paradigm, a general architecture that simplifies spatial operations and enhances channel operations to overcome the previous trade-off. We present two instances of the S2CFormer: S2C-Conv, and S2C-Attention. Both models demonstrate state-of-the-art (SOTA) R-D performance and significantly faster decoding speed. Furthermore, we introduce S2C-Hybrid, an enhanced variant that maximizes the strengths of different S2CFormer instances to achieve a better performance-latency trade-off. This model outperforms all the existing methods on the Kodak, Tecnick, and CLIC Professional Validation datasets, setting a new benchmark for efficient and high-performance LIC. The code is at https://github.com/YunuoChen/S2CFormer{https://github.com/YunuoChen/S2CFormer}.

A problem with many current Large Language Model (LLM) driven spoken dialogues is the response time. Some efforts such as Groq address this issue by lightning fast processing of the LLM, but we know from the cognitive psychology literature that in human-to-human dialogue often responses occur prior to the speaker completing their utterance. No amount of delay for LLM processing is acceptable if we wish to maintain human dialogue latencies. In this paper, we discuss methods for understanding an utterance in close to real time and generating a response so that the system can comply with human-level conversational turn delays. This means that the information content of the final part of the speaker's utterance is lost to the LLM. Using the Google NaturalQuestions (NQ) database, our results show GPT-4 can effectively fill in missing context from a dropped word at the end of a question over 60% of the time. We also provide some examples of utterances and the impacts of this information loss on the quality of LLM response in the context of an avatar that is currently under development. These results indicate that a simple classifier could be used to determine whether a question is semantically complete, or requires a filler phrase to allow a response to be generated within human dialogue time constraints.

Reasoning ability, a core component of human intelligence, continues to pose a significant challenge for Large Language Models (LLMs) in the pursuit of AGI. Although model performance has improved under the training scaling law, significant challenges remain, particularly with respect to training algorithms, such as catastrophic forgetting, and the limited availability of novel training data. As an alternative, test-time scaling enhances reasoning performance by increasing test-time computation without parameter updating. Unlike prior methods in this paradigm focused on token space, we propose leveraging latent space for more effective reasoning and better adherence to the test-time scaling law. We introduce LatentSeek, a novel framework that enhances LLM reasoning through Test-Time Instance-level Adaptation (TTIA) within the model's latent space. Specifically, LatentSeek leverages policy gradient to iteratively update latent representations, guided by self-generated reward signals. LatentSeek is evaluated on a range of reasoning benchmarks, including GSM8K, MATH-500, and AIME2024, across multiple LLM architectures. Results show that LatentSeek consistently outperforms strong baselines, such as Chain-of-Thought prompting and fine-tuning-based methods. Furthermore, our analysis demonstrates that LatentSeek is highly efficient, typically converging within a few iterations for problems of average complexity, while also benefiting from additional iterations, thereby highlighting the potential of test-time scaling in the latent space. These findings position LatentSeek as a lightweight, scalable, and effective solution for enhancing the reasoning capabilities of LLMs.

Sequence-to-sequence language models can be used to produce abstractive summaries which are coherent, relevant, and concise. Still, model sizes can make deployment in latency-sensitive or web-scale implementations difficult. This paper studies the relationship between model size, structured pruning, inference efficiency, and summarization accuracy on widely used summarization datasets. We show that model accuracy is tied to the encoder size while inference efficiency is connected to the decoder. Using asymmetric pruning can lead to nearly 3x improvement in inference latency with ~1 point loss in Rouge-2. Moreover, we find both the average degradation and the role of asymmetry to be consistent across model sizes and variations in datasets.

This work focuses on developing parameter-efficient and lightweight models for dense predictions while trading off parameters, FLOPs, and performance. Our goal is to set up the new frontier of the 5M magnitude lightweight model on various downstream tasks. Inverted Residual Block (IRB) serves as the infrastructure for lightweight CNNs, but no counterparts have been recognized by attention-based design. Our work rethinks the lightweight infrastructure of efficient IRB and practical components in Transformer from a unified perspective, extending CNN-based IRB to attention-based models and abstracting a one-residual Meta Mobile Block (MMBlock) for lightweight model design. Following neat but effective design criterion, we deduce a modern Improved Inverted Residual Mobile Block (i2RMB) and improve a hierarchical Efficient MOdel (EMOv2) with no elaborate complex structures. Considering the imperceptible latency for mobile users when downloading models under 4G/5G bandwidth and ensuring model performance, we investigate the performance upper limit of lightweight models with a magnitude of 5M. Extensive experiments on various vision recognition, dense prediction, and image generation tasks demonstrate the superiority of our EMOv2 over state-of-the-art methods, e.g., EMOv2-1M/2M/5M achieve 72.3, 75.8, and 79.4 Top-1 that surpass equal-order CNN-/Attention-based models significantly. At the same time, EMOv2-5M equipped RetinaNet achieves 41.5 mAP for object detection tasks that surpasses the previous EMO-5M by +2.6. When employing the more robust training recipe, our EMOv2-5M eventually achieves 82.9 Top-1 accuracy, which elevates the performance of 5M magnitude models to a new level. Code is available at https://github.com/zhangzjn/EMOv2.

Deploying language models often requires handling model size vs. performance trade-offs to satisfy downstream latency constraints while preserving the model's usefulness. Model distillation is commonly employed to reduce model size while maintaining acceptable performance. However, distillation can be inefficient since it involves multiple training steps. In this work, we introduce MODULARSTARENCODER, a modular multi-exit encoder with 1B parameters, useful for multiple tasks within the scope of code retrieval. MODULARSTARENCODER is trained with a novel self-distillation mechanism that significantly improves lower-layer representations-allowing different portions of the model to be used while still maintaining a good trade-off in terms of performance. Our architecture focuses on enhancing text-to-code and code-to-code search by systematically capturing syntactic and semantic structures across multiple levels of representation. Specific encoder layers are targeted as exit heads, allowing higher layers to guide earlier layers during training. This self-distillation effect improves intermediate representations, increasing retrieval recall at no extra training cost. In addition to the multi-exit scheme, our approach integrates a repository-level contextual loss that maximally utilizes the training context window, further enhancing the learned representations. We also release a new dataset constructed via code translation, seamlessly expanding traditional text-to-code benchmarks with code-to-code pairs across diverse programming languages. Experimental results highlight the benefits of self-distillation through multi-exit supervision.

Large language models (LLMs) excel at generating human-like responses but often struggle with interactive tasks that require access to real-time information. This limitation poses challenges in finance, where models must access up-to-date information, such as recent news or price movements, to support decision-making. To address this, we introduce Financial Agent, a knowledge-grounding approach for LLMs to handle financial queries using real-time text and tabular data. Our contributions are threefold: First, we develop a Financial Context Dataset of over 50,000 financial queries paired with the required context. Second, we train FinBloom 7B, a custom 7 billion parameter LLM, on 14 million financial news articles from Reuters and Deutsche Presse-Agentur, alongside 12 million Securities and Exchange Commission (SEC) filings. Third, we fine-tune FinBloom 7B using the Financial Context Dataset to serve as a Financial Agent. This agent generates relevant financial context, enabling efficient real-time data retrieval to answer user queries. By reducing latency and eliminating the need for users to manually provide accurate data, our approach significantly enhances the capability of LLMs to handle dynamic financial tasks. Our proposed approach makes real-time financial decisions, algorithmic trading and other related tasks streamlined, and is valuable in contexts with high-velocity data flows.

Recently, Diffusion Large Language Models (dLLMs) have demonstrated unique efficiency advantages, enabled by their inherently parallel decoding mechanism and flexible generation paradigm. Meanwhile, despite the rapid advancement of Search Agents, their practical deployment is constrained by a fundamental limitation, termed as 1) Latency Challenge: the serial execution of multi-round reasoning, tool calling, and tool response waiting under the ReAct agent paradigm induces severe end-to-end latency. Intuitively, dLLMs can leverage their distinctive strengths to optimize the operational efficiency of agents under the ReAct agent paradigm. Practically, existing dLLM backbones face the 2) Agent Ability Challenge. That is, existing dLLMs exhibit remarkably weak reasoning and tool-calling capabilities, preventing these advantages from being effectively realized in practice. In this paper, we propose DLLM-Searcher, an optimization framework for dLLM-based Search Agents. To solve the Agent Ability Challenge, we design a two-stage post-training pipeline encompassing Agentic Supervised Fine-Tuning (Agentic SFT) and Agentic Variance-Reduced Preference Optimization Agentic VRPO, which enhances the backbone dLLM's information seeking and reasoning capabilities. To mitigate the Latency Challenge, we leverage the flexible generation mechanism of dLLMs and propose a novel agent paradigm termed Parallel-Reasoning and Acting P-ReAct. P-ReAct guides the model to prioritize decoding tool_call instructions, thereby allowing the model to keep thinking while waiting for the tool's return. Experimental results demonstrate that DLLM-Searcher achieves performance comparable to mainstream LLM-based search agents and P-ReAct delivers approximately 15% inference acceleration. Our code is available at https://anonymous.4open.science/r/DLLM-Searcher-553C

We propose a novel AutoRegressive Generation-based paradigm for image Segmentation (ARGenSeg), achieving multimodal understanding and pixel-level perception within a unified framework. Prior works integrating image segmentation into multimodal large language models (MLLMs) typically employ either boundary points representation or dedicated segmentation heads. These methods rely on discrete representations or semantic prompts fed into task-specific decoders, which limits the ability of the MLLM to capture fine-grained visual details. To address these challenges, we introduce a segmentation framework for MLLM based on image generation, which naturally produces dense masks for target objects. We leverage MLLM to output visual tokens and detokenize them into images using an universal VQ-VAE, making the segmentation fully dependent on the pixel-level understanding of the MLLM. To reduce inference latency, we employ a next-scale-prediction strategy to generate required visual tokens in parallel. Extensive experiments demonstrate that our method surpasses prior state-of-the-art approaches on multiple segmentation datasets with a remarkable boost in inference speed, while maintaining strong understanding capabilities.

We propose Squeeze3D, a novel framework that leverages implicit prior knowledge learnt by existing pre-trained 3D generative models to compress 3D data at extremely high compression ratios. Our approach bridges the latent spaces between a pre-trained encoder and a pre-trained generation model through trainable mapping networks. Any 3D model represented as a mesh, point cloud, or a radiance field is first encoded by the pre-trained encoder and then transformed (i.e. compressed) into a highly compact latent code. This latent code can effectively be used as an extremely compressed representation of the mesh or point cloud. A mapping network transforms the compressed latent code into the latent space of a powerful generative model, which is then conditioned to recreate the original 3D model (i.e. decompression). Squeeze3D is trained entirely on generated synthetic data and does not require any 3D datasets. The Squeeze3D architecture can be flexibly used with existing pre-trained 3D encoders and existing generative models. It can flexibly support different formats, including meshes, point clouds, and radiance fields. Our experiments demonstrate that Squeeze3D achieves compression ratios of up to 2187x for textured meshes, 55x for point clouds, and 619x for radiance fields while maintaining visual quality comparable to many existing methods. Squeeze3D only incurs a small compression and decompression latency since it does not involve training object-specific networks to compress an object.

We present speculative sampling, an algorithm for accelerating transformer decoding by enabling the generation of multiple tokens from each transformer call. Our algorithm relies on the observation that the latency of parallel scoring of short continuations, generated by a faster but less powerful draft model, is comparable to that of sampling a single token from the larger target model. This is combined with a novel modified rejection sampling scheme which preserves the distribution of the target model within hardware numerics. We benchmark speculative sampling with Chinchilla, a 70 billion parameter language model, achieving a 2-2.5x decoding speedup in a distributed setup, without compromising the sample quality or making modifications to the model itself.

Recent advancements in artificial intelligence promise ample potential in monitoring applications with surveillance cameras. However, concerns about privacy and model bias have made it challenging to utilize them in public. Although de-identification approaches have been proposed in the literature, aiming to achieve a certain level of anonymization, most of them employ deep learning models that are computationally demanding for real-time edge deployment. In this study, we revisit conventional anonymization solutions for privacy protection and real-time video anomaly detection (VAD) applications. We propose a novel lightweight adaptive anonymization for VAD (LA3D) that employs dynamic adjustment to enhance privacy protection. We evaluated the approaches on publicly available privacy and VAD data sets to examine the strengths and weaknesses of the different anonymization techniques and highlight the promising efficacy of our approach. Our experiment demonstrates that LA3D enables substantial improvement in the privacy anonymization capability without majorly degrading VAD efficacy.

To mitigate the high inference latency stemming from autoregressive decoding in Large Language Models (LLMs), Speculative Decoding has emerged as a novel decoding paradigm for LLM inference. In each decoding step, this method first efficiently drafts several future tokens and then verifies them in parallel. Unlike autoregressive decoding, Speculative Decoding facilitates the simultaneous decoding of multiple tokens per step, thereby accelerating inference. This paper presents a comprehensive overview and analysis of this promising decoding paradigm. We begin by providing a formal definition and formulation of Speculative Decoding. Then, we organize in-depth discussions on its key facets, including current leading techniques, the challenges faced, and potential future directions in this field. We aim for this work to serve as a catalyst for further research on Speculative Decoding, ultimately contributing to more efficient LLM inference.

We propose DarkStream, a streaming speech synthesis model for real-time speaker anonymization. To improve content encoding under strict latency constraints, DarkStream combines a causal waveform encoder, a short lookahead buffer, and transformer-based contextual layers. To further reduce inference time, the model generates waveforms directly via a neural vocoder, thus removing intermediate mel-spectrogram conversions. Finally, DarkStream anonymizes speaker identity by injecting a GAN-generated pseudo-speaker embedding into linguistic features from the content encoder. Evaluations show our model achieves strong anonymization, yielding close to 50% speaker verification EER (near-chance performance) on the lazy-informed attack scenario, while maintaining acceptable linguistic intelligibility (WER within 9%). By balancing low-latency, robust privacy, and minimal intelligibility degradation, DarkStream provides a practical solution for privacy-preserving real-time speech communication.

Serverless computing has grown rapidly for serving Large Language Model (LLM) inference due to its pay-as-you-go pricing, fine-grained GPU usage, and rapid scaling. However, our analysis reveals that current serverless can effectively serve general LLM but fail with Low-Rank Adaptation (LoRA) inference due to three key limitations: 1) massive parameter redundancy among functions where 99% of weights are unnecessarily duplicated, 2) costly artifact loading latency beyond LLM loading, and 3) magnified resource contention when serving multiple LoRA LLMs. These inefficiencies lead to massive GPU wastage, increased Time-To-First-Token (TTFT), and high monetary costs. We propose ServerlessLoRA, a novel serverless inference system designed for faster and cheaper LoRA LLM serving. ServerlessLoRA enables secure backbone LLM sharing across isolated LoRA functions to reduce redundancy. We design a pre-loading method that pre-loads comprehensive LoRA artifacts to minimize cold-start latency. Furthermore, ServerlessLoRA employs contention aware batching and offloading to mitigate GPU resource conflicts during bursty workloads. Experiment on industrial workloads demonstrates that ServerlessLoRA reduces TTFT by up to 86% and cuts monetary costs by up to 89% compared to state-of-the-art LLM inference solutions.

Private large language model (LLM) inference based on secure multi-party computation (MPC) offers cryptographically-secure protection for both user prompt and proprietary model weights. However, it suffers from large latency overhead especially for long input sequences. While key-value (KV) cache eviction algorithms have been proposed to reduce the computation and memory cost for plaintext inference, they are not designed for MPC and cannot benefit private inference easily. In this paper, we propose an accurate and MPC-friendly KV cache eviction framework, dubbed MPCache. MPCache is built on the observation that historical tokens in a long sequence may have different effects on the downstream decoding. Hence, MPCache combines a look-once static eviction algorithm to discard unimportant tokens and a query-aware dynamic selection algorithm to further select a small subset of tokens for attention computation. As existing dynamic selection algorithms incur too much latency, we propose a series of optimizations to drastically reduce the KV cache selection overhead, including MPC-friendly similarity approximation, hierarchical KV cache clustering, and cross-layer index sharing strategy. With extensive experiments, we demonstrate that MPCache consistently outperforms prior-art KV cache eviction baselines across different LLM generation tasks and achieves 1.8~2.01x and 3.39~8.37x decoding latency and communication reduction on different sequence lengths, respectively.

Practical large language model (LLM) services may involve a long system prompt, which specifies the instructions, examples, and knowledge documents of the task and is reused across numerous requests. However, the long system prompt causes throughput/latency bottlenecks as the cost of generating the next token grows w.r.t. the sequence length. This paper aims to improve the efficiency of LLM services that involve long system prompts. Our key observation is that handling these system prompts requires heavily redundant memory accesses in existing causal attention computation algorithms. Specifically, for batched requests, the cached hidden states (i.e., key-value pairs) of system prompts are transferred from off-chip DRAM to on-chip SRAM multiple times, each corresponding to an individual request. To eliminate such a redundancy, we propose RelayAttention, an attention algorithm that allows reading these hidden states from DRAM exactly once for a batch of input tokens. RelayAttention is a free lunch: it maintains the generation quality while requiring no model retraining, as it is based on a mathematical reformulation of causal attention.

Language models generate responses by producing a series of tokens in immediate succession: the (K+1)^{th} token is an outcome of manipulating K hidden vectors per layer, one vector per preceding token. What if instead we were to let the model manipulate say, K+10 hidden vectors, before it outputs the (K+1)^{th} token? We operationalize this idea by performing training and inference on language models with a (learnable) pause token, a sequence of which is appended to the input prefix. We then delay extracting the model's outputs until the last pause token is seen, thereby allowing the model to process extra computation before committing to an answer. We empirically evaluate pause-training on decoder-only models of 1B and 130M parameters with causal pretraining on C4, and on downstream tasks covering reasoning, question-answering, general understanding and fact recall. Our main finding is that inference-time delays show gains when the model is both pre-trained and finetuned with delays. For the 1B model, we witness gains on 8 of 9 tasks, most prominently, a gain of 18% EM score on the QA task of SQuAD, 8% on CommonSenseQA and 1% accuracy on the reasoning task of GSM8k. Our work raises a range of conceptual and practical future research questions on making delayed next-token prediction a widely applicable new paradigm.

Robots in dynamic, human-centric environments must follow language instructions while maintaining real-time reactive control. Vision-language-action (VLA) models offer a promising framework, but they assume temporally aligned reasoning and control, despite semantic inference being inherently delayed relative to real-time action. We introduce Think-in-Control (TIC)-VLA, a latency-aware framework that explicitly models delayed semantic reasoning during action generation. TIC-VLA defines a delayed semantic-control interface that conditions action generation on delayed vision-language semantic states and explicit latency metadata, in addition to current observations, enabling policies to compensate for asynchronous reasoning. We further propose a latency-consistent training pipeline that injects reasoning inference delays during imitation learning and online reinforcement learning, aligning training with asynchronous deployment. To support realistic evaluation, we present DynaNav, a physics-accurate, photo-realistic simulation suite for language-guided navigation in dynamic environments. Extensive experiments in simulation and on a real robot show that TIC-VLA consistently outperforms prior VLA models while maintaining robust real-time control under multi-second reasoning latency. Project website: https://ucla-mobility.github.io/TIC-VLA/

DistServe improves the performance of large language models (LLMs) serving by disaggregating the prefill and decoding computation. Existing LLM serving systems colocate the two phases and batch the computation of prefill and decoding across all users and requests. We find that this strategy not only leads to strong prefill-decoding interferences but also couples the resource allocation and parallelism plans for both phases. LLM applications often emphasize individual latency for each phase: time to first token (TTFT) for the prefill phase and time per output token (TPOT) of each request for the decoding phase. In the presence of stringent latency requirements, existing systems have to prioritize one latency over the other, or over-provision compute resources to meet both. DistServe assigns prefill and decoding computation to different GPUs, hence eliminating prefill-decoding interferences. Given the application's TTFT and TPOT requirements, DistServe co-optimizes the resource allocation and parallelism strategy tailored for each phase. DistServe also places the two phases according to the serving cluster's bandwidth to minimize the communication caused by disaggregation. As a result, DistServe significantly improves LLM serving performance in terms of the maximum rate that can be served within both TTFT and TPOT constraints on each GPU. Our evaluations show that on various popular LLMs, applications, and latency requirements, DistServe can serve 4.48x more requests or 10.2x tighter SLO, compared to state-of-the-art systems, while staying within latency constraints for > 90% of requests.

Designing domain specific neural networks is a time-consuming, error-prone, and expensive task. Neural Architecture Search (NAS) exists to simplify domain-specific model development but there is a gap in the literature for time series classification on microcontrollers. Therefore, we adapt the concept of differentiable neural architecture search (DNAS) to solve the time-series classification problem on resource-constrained microcontrollers (MCUs). We introduce MicroNAS, a domain-specific HW-NAS system integration of DNAS, Latency Lookup Tables, dynamic convolutions and a novel search space specifically designed for time-series classification on MCUs. The resulting system is hardware-aware and can generate neural network architectures that satisfy user-defined limits on the execution latency and peak memory consumption. Our extensive studies on different MCUs and standard benchmark datasets demonstrate that MicroNAS finds MCU-tailored architectures that achieve performance (F1-score) near to state-of-the-art desktop models. We also show that our approach is superior in adhering to memory and latency constraints compared to domain-independent NAS baselines such as DARTS.

Chain-of-Thought (CoT) reasoning has demonstrated remarkable effectiveness in enhancing the reasoning abilities of large language models (LLMs). However, its efficiency remains a challenge due to the generation of excessive intermediate reasoning tokens, which introduce semantic redundancy and overly detailed reasoning steps. Moreover, computational expense and latency are significant concerns, as the cost scales with the number of output tokens, including those intermediate steps. In this work, we observe that most CoT tokens are unnecessary, and retaining only a small portion of them is sufficient for producing high-quality responses. Inspired by this, we propose HAWKEYE, a novel post-training and inference framework where a large model produces concise CoT instructions to guide a smaller model in response generation. HAWKEYE quantifies redundancy in CoT reasoning and distills high-density information via reinforcement learning. By leveraging these concise CoTs, HAWKEYE is able to expand responses while reducing token usage and computational cost significantly. Our evaluation shows that HAWKEYE can achieve comparable response quality using only 35% of the full CoTs, while improving clarity, coherence, and conciseness by approximately 10%. Furthermore, HAWKEYE can accelerate end-to-end reasoning by up to 3.4x on complex math tasks while reducing inference cost by up to 60%. HAWKEYE will be open-sourced and the models will be available soon.

Recent advances in text-to-video (T2V) generation have enabled the creation of high-fidelity, temporally coherent clips from natural language prompts. Yet these systems come with significant computational costs, and their energy demands remain poorly understood. In this paper, we present a systematic study of the latency and energy consumption of state-of-the-art open-source T2V models. We first develop a compute-bound analytical model that predicts scaling laws with respect to spatial resolution, temporal length, and denoising steps. We then validate these predictions through fine-grained experiments on WAN2.1-T2V, showing quadratic growth with spatial and temporal dimensions, and linear scaling with the number of denoising steps. Finally, we extend our analysis to six diverse T2V models, comparing their runtime and energy profiles under default settings. Our results provide both a benchmark reference and practical insights for designing and deploying more sustainable generative video systems.

Autoregressive (AR) language models have emerged as powerful solutions for zero-shot text-to-speech (TTS) synthesis, capable of generating natural speech from a few seconds of audio prompts. However, conventional AR-based TTS systems relying on discrete audio tokens face the challenge of lossy compression during tokenization, requiring longer discrete token sequences to capture the same information as continuous ones, which adds inference latency and complicates AR modeling. To address this challenge, this paper proposes the Continuous Latent Autoregressive model (CLEAR), a unified zero-shot TTS framework that directly models continuous audio representations. More specifically, CLEAR introduces an enhanced variational autoencoder with shortcut connections, which achieves a high compression ratio to map waveforms into compact continuous latents. A lightweight MLP-based rectified flow head that operates independently for each hidden state is presented to model the continuous latent probability distribution, and trained jointly with the AR model within a single-stage framework. Experiments show that the proposed zero-shot CLEAR TTS can synthesize high-quality speech with low latency. Compared to state-of-the-art (SOTA) TTS models, CLEAR delivers competitive performance in robustness, speaker similarity and naturalness, while offering a lower real-time factor (RTF). In particular, CLEAR achieves SOTA results on the LibriSpeech test-clean dataset, with a word error rate of 1.88\% and an RTF of 0.29. Moreover, CLEAR facilitates streaming speech synthesis with a first-frame delay of 96ms, while maintaining high-quality speech synthesis.

In recent years, Mixture-of-Experts (MoE) has emerged as an effective approach for enhancing the capacity of deep neural network (DNN) with sub-linear computational costs. However, storing all experts on GPUs incurs significant memory overhead, increasing the monetary cost of MoE-based inference. To address this, we propose eMoE, a memory efficient inference system for MoE-based large language models (LLMs) by leveraging our observations from experiment measurements. eMoE reduces memory usage by predicting and loading only the required experts based on recurrent patterns in expert routing. To reduce loading latency while maintaining accuracy, as we found using the same experts for subsequent prompts has minimal impact on perplexity, eMoE invokes the expert predictor every few prompts rather than for each prompt. In addition, it skips predictions for tasks less sensitive to routing accuracy. Finally, it has task-aware scheduling to minimize inference latency by considering Service Level Objectives (SLOs), task-specific output lengths, and expert loading latencies. Experimental results show that compared to existing systems, eMoE reduces memory consumption by up to 80% while maintaining accuracy and reduces inference latency by up to 17%. It also enables processing prompts 40x longer, batches 4.5x larger, and achieves 1.5x higher throughput.

Generative artificial intelligence (AI) technology is revolutionizing the computing industry. Not only its applications have broadened to various sectors but also poses new system design and optimization opportunities. The technology is capable of understanding and responding in multiple modalities. However, the advanced capability currently comes with significant system resource demands. To sustainably scale generative AI capabilities to billions of users in the world, inference must be fast and efficient. This paper pinpoints key system design and optimization opportunities by characterizing a family of emerging multi-modal generation models on real systems. Auto-regressive token generation is a critical latency performance bottleneck, typically dominated by GPU idle time. In addition to memory-intensive attention across the generative AI models, linear operations constitute significant inference latency due to the feed forward networks in Transformer-based models. We demonstrate that state-of-the-art optimization levers, spanning from applications to system software and hardware, set a 3.88x better baseline.

The rapid advancement of large language models (LLMs) has significantly propelled the development of text-based chatbots, demonstrating their capability to engage in coherent and contextually relevant dialogues. However, extending these advancements to enable end-to-end speech-to-speech conversation bots remains a formidable challenge, primarily due to the extensive dataset and computational resources required. The conventional approach of cascading automatic speech recognition (ASR), LLM, and text-to-speech (TTS) models in a pipeline, while effective, suffers from unnatural prosody because it lacks direct interactions between the input audio and its transcribed text and the output audio. These systems are also limited by their inherent latency from the ASR process for real-time applications. This paper introduces Style-Talker, an innovative framework that fine-tunes an audio LLM alongside a style-based TTS model for fast spoken dialog generation. Style-Talker takes user input audio and uses transcribed chat history and speech styles to generate both the speaking style and text for the response. Subsequently, the TTS model synthesizes the speech, which is then played back to the user. While the response speech is being played, the input speech undergoes ASR processing to extract the transcription and speaking style, serving as the context for the ensuing dialogue turn. This novel pipeline accelerates the traditional cascade ASR-LLM-TTS systems while integrating rich paralinguistic information from input speech. Our experimental results show that Style-Talker significantly outperforms the conventional cascade and speech-to-speech baselines in terms of both dialogue naturalness and coherence while being more than 50% faster.

Large language models (LLMs) have numerous applications across various domains, but their high computational and memory demands pose significant deployment challenges. Weight quantization is an effective technique for reducing the decoding latency and memory requirements of LLMs. Existing approaches primarily aim to maintain the quality of quantized models by preserving outliers in input features, but they still suffer significant quality loss at lower bit widths. Our approach builds on Optimal Brain Quantization (OBQ), an iterative weight-update-based quantization framework. We identify a key limitation of OBQ, specifically that its uniform quantization grid is suboptimal for maintaining model quality, as it introduces large errors to the task loss. To address this, we propose LeanQuant, which learns a loss-error-aware quantization grid by leveraging the inverse diagonal Hessian. Extensive empirical evaluations demonstrate that LeanQuant is both efficient and accurate; it can quantize a 70-billion-parameter model in 6 hours using a single 32GB GPU and performs favorably compared to competitive baselines in the 4-bit, 3-bit, and 2-bit regions.

Large language models (LLMs) propel the prosperity of interactive AI applications showcased by ChatGPT that demand timely response of inference services. However, LLM inference is computation intensive and memory intensive, and improper parameter configuration at LLM platforms may exacerbate the inference time. In this paper, we analyze the impact of LLM output token distribution on the inference queueing delay, where the max-token clipping and the batched inference are considered. By formulating an M/G/1 model, we observe that enforcing a maximum output token limit on a very small fraction of inference requests can significantly reduce the queueing delay, and our model facilitates the selection of the optimal limit. For the batch inference, we model the service process as a bulk queue in which the batch processing time is affected by the batch size and the maximum token size inside this batch jointly. The queueing delays of the batching of all buffered requests (dynamic batching), the batching of constant number of requests (fixed batching), and the batching without intra-batch waiting (elastic batching) are derived. Experimental results show that our mathematical models coincide with the event-driven simulations well.

Spiking Neural Networks (SNN) are a class of bio-inspired neural networks that promise to bring low-power and low-latency inference to edge devices through asynchronous and sparse processing. However, being temporal models, SNNs depend heavily on expressive states to generate predictions on par with classical artificial neural networks (ANNs). These states converge only after long transient periods, and quickly decay without input data, leading to higher latency, power consumption, and lower accuracy. This work addresses this issue by initializing the state with an auxiliary ANN running at a low rate. The SNN then uses the state to generate predictions with high temporal resolution until the next initialization phase. Our hybrid ANN-SNN model thus combines the best of both worlds: It does not suffer from long state transients and state decay thanks to the ANN, and can generate predictions with high temporal resolution, low latency, and low power thanks to the SNN. We show for the task of event-based 2D and 3D human pose estimation that our method consumes 88% less power with only a 4% decrease in performance compared to its fully ANN counterparts when run at the same inference rate. Moreover, when compared to SNNs, our method achieves a 74% lower error. This research thus provides a new understanding of how ANNs and SNNs can be used to maximize their respective benefits.

This work presents a systematic investigation into the latent knowledge encoded within Network Foundation Models (NFMs) that focuses on hidden representations analysis rather than pure downstream task performance. Different from existing efforts, we analyze the models through a three-part evaluation: Embedding Geometry Analysis to assess representation space utilization, Metric Alignment Assessment to measure correspondence with domain-expert features, and Causal Sensitivity Testing to evaluate robustness to protocol perturbations. Using five diverse network datasets spanning controlled and real-world environments, we evaluate four state-of-the-art NFMs, revealing that they all exhibit significant anisotropy, inconsistent feature sensitivity patterns, an inability to separate the high-level context, payload dependency, and other properties. Our work identifies numerous limitations across all models and demonstrates that addressing them can significantly improve model performance (by up to +0.35 F_1 score without architectural changes).

Recent end-to-end spoken dialogue systems leverage speech tokenizers and neural audio codecs to enable LLMs to operate directly on discrete speech representations. However, these models often exhibit limited speaker identity preservation, hindering personalized voice interaction. In this work, we present Chroma 1.0, the first open-source, real-time, end-to-end spoken dialogue model that achieves both low-latency interaction and high-fidelity personalized voice cloning. Chroma achieves sub-second end-to-end latency through an interleaved text-audio token schedule (1:2) that supports streaming generation, while maintaining high-quality personalized voice synthesis across multi-turn conversations. Our experimental results demonstrate that Chroma achieves a 10.96% relative improvement in speaker similarity over the human baseline, with a Real-Time Factor (RTF) of 0.43, while maintaining strong reasoning and dialogue capabilities. Our code and models are publicly available at https://github.com/FlashLabs-AI-Corp/FlashLabs-Chroma and https://huggingface.co/FlashLabs/Chroma-4B .

The increasing usage of Large Language Models (LLMs) has resulted in a surging demand for planet-scale serving systems, where tens of thousands of GPUs continuously serve hundreds of millions of users. Consequently, throughput (under reasonable latency constraints) has emerged as a key metric that determines serving systems' performance. To boost throughput, various methods of inter-device parallelism (e.g., data, tensor, pipeline) have been explored. However, existing methods do not consider overlapping the utilization of different resources within a single device, leading to underutilization and sub-optimal performance. We propose NanoFlow, a novel serving framework that exploits intra-device parallelism, which overlaps the usage of resources including compute, memory, and network within a single device through operation co-scheduling. To exploit intra-device parallelism, NanoFlow introduces two key innovations: First, NanoFlow splits requests into nano-batches at the granularity of operations, which breaks the dependency of sequential operations in LLM inference and enables overlapping; then, to get benefit from overlapping, NanoFlow uses an operation-level pipeline with execution unit scheduling, which partitions the device's functional units and simultaneously executes different operations in each unit. NanoFlow automates the pipeline setup using a parameter search algorithm, which enables easily porting NanoFlow to different models. We implement NanoFlow on NVIDIA GPUs and evaluate end-to-end serving throughput on several popular models such as LLaMA-2-70B, Mixtral 8x7B, LLaMA-3-8B, etc.. With practical workloads, NanoFlow provides 1.91x throughput boost compared to state-of-the-art serving systems achieving 59% to 72% of optimal throughput across ported models.

We introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue framework. Current systems for spoken dialogue rely on pipelines of independent components, namely voice activity detection, speech recognition, textual dialogue and text-to-speech. Such frameworks cannot emulate the experience of real conversations. First, their complexity induces a latency of several seconds between interactions. Second, text being the intermediate modality for dialogue, non-linguistic information that modifies meaning -- such as emotion or non-speech sounds -- is lost in the interaction. Finally, they rely on a segmentation into speaker turns, which does not take into account overlapping speech, interruptions and interjections. Moshi solves these independent issues altogether by casting spoken dialogue as speech-to-speech generation. Starting from a text language model backbone, Moshi generates speech as tokens from the residual quantizer of a neural audio codec, while modeling separately its own speech and that of the user into parallel streams. This allows for the removal of explicit speaker turns, and the modeling of arbitrary conversational dynamics. We moreover extend the hierarchical semantic-to-acoustic token generation of previous work to first predict time-aligned text tokens as a prefix to audio tokens. Not only this "Inner Monologue" method significantly improves the linguistic quality of generated speech, but we also illustrate how it can provide streaming speech recognition and text-to-speech. Our resulting model is the first real-time full-duplex spoken large language model, with a theoretical latency of 160ms, 200ms in practice, and is available at https://github.com/kyutai-labs/moshi.

Large reasoning models (LRMs) achieve strong performance on complex reasoning tasks by generating long, multi-step reasoning trajectories, but inference-time scaling incurs substantial deployment cost. A key challenge is that generation difficulty varies within a single output, whereas existing efficiency-oriented approaches either ignore this intra-generation variation or rely on supervised token-level routing with high system complexity. We present RelayGen, a training-free, segment-level runtime model switching framework that exploits difficulty variation in long-form reasoning. Through offline analysis of generation uncertainty using token probability margins, we show that coarse-grained segment-level control is sufficient to capture difficulty transitions within a reasoning trajectory. RelayGen identifies model-specific switch cues that signal transitions to lower-difficulty segments and dynamically delegates their continuation to a smaller model, while preserving high-difficulty reasoning on the large model. Across multiple reasoning benchmarks, RelayGen substantially reduces inference latency while preserving most of the accuracy of large models. When combined with speculative decoding, RelayGen achieves up to 2.2times end-to-end speedup with less than 2\% accuracy degradation, without requiring additional training or learned routing components.

Full-duplex spoken dialogue systems significantly advance over traditional turn-based dialogue systems, as they allow simultaneous bidirectional communication, closely mirroring human-human interactions. However, achieving low latency and natural interactions in full-duplex dialogue systems remains a significant challenge, especially considering human conversation dynamics such as interruptions, backchannels, and overlapping speech. In this paper, we introduce a novel End-to-End GPT-based model OmniFlatten for full-duplex conversation, capable of effectively modeling the complex behaviors inherent to natural conversations with low latency. To achieve full-duplex communication capabilities, we propose a multi-stage post-training scheme that progressively adapts a text-based large language model (LLM) backbone into a speech-text dialogue LLM, capable of generating text and speech in real time, without modifying the architecture of the backbone LLM. The training process comprises three stages: modality alignment, half-duplex dialogue learning, and full-duplex dialogue learning. Throughout all training stages, we standardize the data using a flattening operation, which allows us to unify the training methods and the model architecture across different modalities and tasks. Our approach offers a straightforward modeling technique and a promising research direction for developing efficient and natural end-to-end full-duplex spoken dialogue systems. Audio samples of dialogues generated by OmniFlatten can be found at this web site (https://omniflatten.github.io/).

Post-training quantization is a key technique for reducing the memory and inference latency of large language models by quantizing weights and activations without requiring retraining. However, existing methods either (1) fail to account for the varying importance of hidden features to the end loss or, when incorporating end loss, (2) neglect the critical interactions between model weights. To address these limitations, we propose GuidedQuant, a novel quantization approach that integrates gradient information from the end loss into the quantization objective while preserving cross-weight dependencies within output channels. GuidedQuant consistently boosts the performance of state-of-the-art quantization methods across weight-only scalar, weight-only vector, and weight-and-activation quantization. Additionally, we introduce a novel non-uniform scalar quantization algorithm, which is guaranteed to monotonically decrease the quantization objective value, and outperforms existing methods in this category. We release the code at https://github.com/snu-mllab/GuidedQuant.

Small language models (SLMs) are widely used in tasks that require low latency and lightweight deployment, particularly classification. As interpretability and robustness gain increasing importance, explanation-guided learning has emerged as an effective framework by introducing attribution-based supervision during training; however, deriving general and reliable attribution priors remains a significant challenge. Through an analysis of representative attribution methods in classification settings, we find that although these methods can reliably highlight class-relevant tokens, they often focus on common keywords shared by semantically similar classes. Because such classes are already difficult to distinguish under standard training, these attributions provide insufficient discriminative cues, limiting their ability to improve model differentiation. To overcome this limitation, we propose Class-Aware Attribution Prior (CAP), a novel attribution prior extraction framework that guides language models toward capturing fine-grained class distinctions and producing more salient, discriminative attribution priors. Building on this idea, we further introduce CAP Hybrid, which combines priors from CAP with those from existing attribution techniques to form a more comprehensive and balanced supervisory signal. By aligning a model's self-attribution with these enriched priors, our approach encourages the learning of diverse, decision-relevant features. Extensive experiments in full-data, few-shot, and adversarial scenarios demonstrate that our method consistently enhances both interpretability and robustness.

In this work, we propose Causal Autoregressive Diffusion (CARD), a novel framework that unifies the training efficiency of ARMs with the high-throughput inference of diffusion models. CARD reformulates the diffusion process within a strictly causal attention mask, enabling dense, per-token supervision in a single forward pass. To address the optimization instability of causal diffusion, we introduce a soft-tailed masking schema to preserve local context and a context-aware reweighting mechanism derived from signal-to-noise principles. This design enables dynamic parallel decoding, where the model leverages KV-caching to adaptively generate variable-length token sequences based on confidence. Empirically, CARD outperforms existing discrete diffusion baselines while reducing training latency by 3 times compared to block diffusion methods. Our results demonstrate that CARD achieves ARM-level data efficiency while unlocking the latency benefits of parallel generation, establishing a robust paradigm for next-generation efficient LLMs.

We propose a first streaming accent conversion (AC) model that transforms non-native speech into a native-like accent while preserving speaker identity, prosody and improving pronunciation. Our approach enables stream processing by modifying a previous AC architecture with an Emformer encoder and an optimized inference mechanism. Additionally, we integrate a native text-to-speech (TTS) model to generate ideal ground-truth data for efficient training. Our streaming AC model achieves comparable performance to the top AC models while maintaining stable latency, making it the first AC system capable of streaming.

Latency-critical speech applications (e.g., live transcription, voice commands, and real-time translation) demand low time-to-first-token (TTFT) and high transcription accuracy, particularly on resource-constrained edge devices. Full-attention Transformer encoders remain a strong accuracy baseline for automatic speech recognition (ASR) because every frame can directly attend to every other frame, which resolves otherwise locally ambiguous acoustics using distant lexical context. However, this global dependency incurs quadratic complexity in sequence length, inducing an inherent "encode-the-whole-utterance" latency profile. For streaming use cases, this causes TTFT to grow linearly with utterance length as the encoder must process the entire prefix before any decoder token can be emitted. To better meet the needs of on-device, streaming ASR use cases we introduce Moonshine v2, an ergodic streaming-encoder ASR model that employs sliding-window self-attention to achieve bounded, low-latency inference while preserving strong local context. Our models achieve state of the art word error rates across standard benchmarks, attaining accuracy on-par with models 6x their size while running significantly faster. These results demonstrate that carefully designed local attention is competitive with the accuracy of full attention at a fraction of the size and latency cost, opening new possibilities for interactive speech interfaces on edge devices.

World models aim to endow AI systems with the ability to represent, generate, and interact with dynamic environments in a coherent and temporally consistent manner. While recent video generation models have demonstrated impressive visual quality, they remain limited in real-time interaction, long-horizon consistency, and persistent memory of dynamic scenes, hindering their evolution into practical world models. In this report, we present TeleWorld, a real-time multimodal 4D world modeling framework that unifies video generation, dynamic scene reconstruction, and long-term world memory within a closed-loop system. TeleWorld introduces a novel generation-reconstruction-guidance paradigm, where generated video streams are continuously reconstructed into a dynamic 4D spatio-temporal representation, which in turn guides subsequent generation to maintain spatial, temporal, and physical consistency. To support long-horizon generation with low latency, we employ an autoregressive diffusion-based video model enhanced with Macro-from-Micro Planning (MMPL)--a hierarchical planning method that reduces error accumulation from frame-level to segment-level-alongside efficient Distribution Matching Distillation (DMD), enabling real-time synthesis under practical computational budgets. Our approach achieves seamless integration of dynamic object modeling and static scene representation within a unified 4D framework, advancing world models toward practical, interactive, and computationally accessible systems. Extensive experiments demonstrate that TeleWorld achieves strong performance in both static and dynamic world understanding, long-term consistency, and real-time generation efficiency, positioning it as a practical step toward interactive, memory-enabled world models for multimodal generation and embodied intelligence.

Zero-shot Text-to-Speech (TTS) aims to synthesize high-quality speech that mimics the voice of an unseen speaker using only a short reference sample, requiring not only speaker adaptation but also accurate modeling of prosodic attributes. Recent approaches based on language models, diffusion, and flow matching have shown promising results in zero-shot TTS, but still suffer from slow inference and repetition artifacts. Discrete codec representations have been widely adopted for speech synthesis, and recent works have begun to explore diffusion models in purely discrete settings, suggesting the potential of discrete generative modeling for speech synthesis. However, existing flow-matching methods typically embed these discrete tokens into a continuous space and apply continuous flow matching, which may not fully leverage the advantages of discrete representations. To address these challenges, we introduce DiFlow-TTS, which, to the best of our knowledge, is the first model to explore purely Discrete Flow Matching for speech synthesis. DiFlow-TTS explicitly models factorized speech attributes within a compact and unified architecture. It leverages in-context learning by conditioning on textual content, along with prosodic and acoustic attributes extracted from a reference speech, enabling effective attribute cloning in a zero-shot setting. In addition, the model employs a factorized flow prediction mechanism with distinct heads for prosody and acoustic details, allowing it to learn aspect-specific distributions. Experimental results demonstrate that DiFlow-TTS achieves promising performance in several key metrics, including naturalness, prosody, preservation of speaker style, and energy control. It also maintains a compact model size and achieves low-latency inference, generating speech up to 25.8 times faster than the latest existing baselines.

Mixed-precision inference techniques reduce the memory and computational demands of Large Language Models (LLMs) by applying hybrid precision formats to model weights, activations, and KV caches. This work introduces mixed-precision LLM inference techniques that encompass (i) systematic memory and compute optimization across hierarchical storage and tensor core architectures, and (ii) comprehensive end-to-end mixed-precision optimization across diverse precision formats and hardware configurations. Our approach features two novel mixed-precision pipelines designed for optimal hardware utilization: a General Matrix Multiply (GEMM) pipeline that optimizes matrix operations through offline weight packing and online acceleration, and an attention pipeline that enables efficient attention computation with arbitrary Query, Key, and Value precision combinations. The key implementation of the pipelines includes (i) hardware-aware weight packing for automatic format optimization, (ii) adaptive head alignment for efficient attention computation, (iii) instruction-level parallelism for memory hierarchy exploitation, and (iv) KV memory loading pipeline for enhanced inference efficiency. We conduct comprehensive evaluations across 16 popular LLMs and 4 representative GPU architectures. Results demonstrate that our approach achieves up to 61% lower serving latency (30% on average) and up to 156% higher throughput (58% on average) in mixed-precision workloads compared to existing mixed-precision frameworks, establishing consistent performance improvements across all tested configurations and hardware types. This work is integrated into TurboMind, a high-performance inference engine of the LMDeploy project, which is open-sourced and publicly available at https://github.com/InternLM/lmdeploy.

Large language models (LLMs) excel across diverse tasks but face significant deployment challenges due to high inference costs. LLM inference comprises prefill (compute-bound) and decode (memory-bound) stages, with decode dominating latency particularly for long sequences. Current decoder-only models handle both stages uniformly, despite their distinct computational profiles. We propose OverFill, which decouples these stages to optimize accuracy-efficiency tradeoffs. OverFill begins with a full model for prefill, processing system and user inputs in parallel. It then switches to a dense pruned model, while generating tokens sequentially. Leveraging more compute during prefill, OverFill improves generation quality with minimal latency overhead. Our 3B-to-1B OverFill configuration outperforms 1B pruned models by 83.2%, while the 8B-to-3B configuration improves over 3B pruned models by 79.2% on average across standard benchmarks. OverFill matches the performance of same-sized models trained from scratch, while using significantly less training data. Our code is available at https://github.com/friendshipkim/overfill.

Vision-language models (VLMs) have achieved impressive performance on multimodal reasoning tasks such as visual question answering (VQA), but their inference cost remains a significant challenge due to the large number of vision tokens processed during the prefill stage. Existing pruning methods often rely on directly using the attention patterns or static text prompt guidance, failing to exploit the dynamic internal signals generated during inference. To address these issues, we propose AdaptInfer, a plug-and-play framework for adaptive vision token pruning in VLMs. First, we introduce a fine-grained, dynamic text-guided pruning mechanism that reuses layer-wise text-to-text attention maps to construct soft priors over text-token importance, allowing more informed scoring of vision tokens at each stage. Second, we perform an offline analysis of cross-modal attention shifts and identify consistent inflection locations in inference, which inspire us to propose a more principled and efficient pruning schedule. Our method is lightweight and plug-and-play, also generalizable across multi-modal tasks. Experimental results have verified the effectiveness of the proposed method. For example, it reduces CUDA latency by 61.3\% while maintaining an average accuracy of 92.9\% on vanilla LLaVA-1.5-7B. Under the same token budget, AdaptInfer surpasses SOTA in accuracy.

LLM serving systems typically treat user prompts as monolithic inputs, optimizing inference through decoding tricks or inter-query batching. However, many real-world prompts contain latent semantic parallelism--decomposable structures where subtasks can be executed independently to reduce latency while preserving meaning. We introduce PARALLELPROMPT, the first benchmark for measuring intra-query parallelism in natural user prompts. Our dataset comprises over 37,000 real-world prompts from public LLM chat logs, each annotated with a structured schema capturing task templates, shared context, and iteration inputs. These schemas are extracted using LLM-assisted prompting with rule-based multilingual validation. To evaluate the benefits of decomposition, we provide an execution suite that benchmarks serial vs. parallel strategies, measuring latency, structural adherence, and semantic fidelity. Our results show that intra-query parallelism can be successfully parsed in over 75% of curated datasets, unlocking up to 5x speedups on tasks like translation, comprehension, and comparative analysis, with minimal quality degradation. By releasing this benchmark, curation pipeline, and evaluation suite, we provide the first standardized testbed for studying structure-aware execution in LLM serving pipelines.

We present Amazon Nova, a new generation of state-of-the-art foundation models that deliver frontier intelligence and industry-leading price performance. Amazon Nova Pro is a highly-capable multimodal model with the best combination of accuracy, speed, and cost for a wide range of tasks. Amazon Nova Lite is a low-cost multimodal model that is lightning fast for processing images, video, documents and text. Amazon Nova Micro is a text-only model that delivers our lowest-latency responses at very low cost. Amazon Nova Canvas is an image generation model that creates professional grade images with rich customization controls. Amazon Nova Reel is a video generation model offering high-quality outputs, customization, and motion control. Our models were built responsibly and with a commitment to customer trust, security, and reliability. We report benchmarking results for core capabilities, agentic performance, long context, functional adaptation, runtime performance, and human evaluation.

Recent advancements in timestep-distilled diffusion models have enabled high-quality image generation that rivals non-distilled multi-step models, but with significantly fewer inference steps. While such models are attractive for applications due to the low inference cost and latency, fine-tuning them with a naive diffusion objective would result in degraded and blurry outputs. An intuitive alternative is to repeat the diffusion distillation process with a fine-tuned teacher model, which produces good results but is cumbersome and computationally intensive; the distillation training usually requires magnitude higher of training compute compared to fine-tuning for specific image styles. In this paper, we present an algorithm named pairwise sample optimization (PSO), which enables the direct fine-tuning of an arbitrary timestep-distilled diffusion model. PSO introduces additional reference images sampled from the current time-step distilled model, and increases the relative likelihood margin between the training images and reference images. This enables the model to retain its few-step generation ability, while allowing for fine-tuning of its output distribution. We also demonstrate that PSO is a generalized formulation which can be flexibly extended to both offline-sampled and online-sampled pairwise data, covering various popular objectives for diffusion model preference optimization. We evaluate PSO in both preference optimization and other fine-tuning tasks, including style transfer and concept customization. We show that PSO can directly adapt distilled models to human-preferred generation with both offline and online-generated pairwise preference image data. PSO also demonstrates effectiveness in style transfer and concept customization by directly tuning timestep-distilled diffusion models.

Recent work has shown that small distilled language models are strong competitors to models that are orders of magnitude larger and slower in a wide range of information retrieval tasks. This has made distilled and dense models, due to latency constraints, the go-to choice for deployment in real-world retrieval applications. In this work, we question this practice by showing that the number of parameters and early query-document interaction play a significant role in the generalization ability of retrieval models. Our experiments show that increasing model size results in marginal gains on in-domain test sets, but much larger gains in new domains never seen during fine-tuning. Furthermore, we show that rerankers largely outperform dense ones of similar size in several tasks. Our largest reranker reaches the state of the art in 12 of the 18 datasets of the Benchmark-IR (BEIR) and surpasses the previous state of the art by 3 average points. Finally, we confirm that in-domain effectiveness is not a good indicator of zero-shot effectiveness. Code is available at https://github.com/guilhermemr04/scaling-zero-shot-retrieval.git

Recent language models have shown impressive multilingual performance, even when not explicitly trained for it. Despite this, there are concerns about the quality of their outputs across different languages. In this paper, we show how disparity in the treatment of different languages arises at the tokenization stage, well before a model is even invoked. The same text translated into different languages can have drastically different tokenization lengths, with differences up to 15 times in some cases. These disparities persist even for tokenizers that are intentionally trained for multilingual support. Character-level and byte-level models also exhibit over 4 times the difference in the encoding length for some language pairs. This induces unfair treatment for some language communities in regard to the cost of accessing commercial language services, the processing time and latency, as well as the amount of content that can be provided as context to the models. Therefore, we make the case that we should train future language models using multilingually fair subword tokenizers.

Large Language Models (LLMs) have recently experienced great success, as evident in the widespread popularity of ChatGPT. Existing LLM serving systems are stateless across requests. Consequently, when LLMs are used in the common setting of multi-turn conversations, a growing log of the conversation history must be processed alongside any request by the serving system at each turn, resulting in repeated history processing. In this paper, we design Pensieve, a system optimized for multi-turn conversation LLM serving. Pensieve maintains the conversation state across requests by caching previously processed history to avoid duplicate processing. Pensieve's multi-tier caching strategy can utilize both GPU and CPU memory to efficiently store and retrieve cached data. Pensieve also generalizes the recent PagedAttention kernel to support attention between multiple input tokens with a GPU cache spread over non-contiguous memory. Our evaluation shows that Pensieve is able to achieve 1.51-1.95x throughput compared to vLLM and reduce latency by 60-75%.