arXiv subjects

Junhao Hu

Publications and source records attributed to Junhao Hu.

At least 37 records · Page 2Linked to original sources

arXiv2025

MiMo-Audio: Audio Language Models are Few-Shot Learners

Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions. GPT-3 has shown that scaling next-token prediction pretraining enables strong generalization capabilities in text, and we believe this paradigm is equally applicable to the audio domain. By scaling MiMo-Audio's pretraining data to over one hundred million of hours, we observe the emergence of few-shot learning capabilities across a diverse set of audio tasks. We develop a systematic evaluation of these capabilities and find that MiMo-Audio-7B-Base achieves SOTA performance on both speech intelligence and audio understanding benchmarks among open-source models. Beyond standard metrics, MiMo-Audio-7B-Base generalizes to tasks absent from its training data, such as voice conversion, style transfer, and speech editing. MiMo-Audio-7B-Base also demonstrates powerful speech continuation capabilities, capable of generating highly realistic talk shows, recitations, livestreaming and debates. At the post-training stage, we curate a diverse instruction-tuning corpus and introduce thinking mechanisms into both audio understanding and generation. MiMo-Audio-7B-Instruct achieves open-source SOTA on audio understanding benchmarks (MMSU, MMAU, MMAR, MMAU-Pro), spoken dialogue benchmarks (Big Bench Audio, MultiChallenge Audio) and instruct-TTS evaluations, approaching or surpassing closed-source models. Model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-Audio.

cs.CL↗

arXiv2025

CupCleaner: A Hybrid Data Cleaning Approach for Comment Updating

Comment updating is an emerging task in software evolution that aims to automatically revise source code comments in accordance with code changes. This task plays a vital role in maintaining code-comment consistency throughout software development. Recently, deep learning-based approaches have shown great potential in addressing comment updating by learning complex patterns between code edits and corresponding comment modifications. However, the effectiveness of these learning-based approaches heavily depends on the quality of training data. Existing datasets are typically constructed by mining version histories from open-source repositories such as GitHub, where there is often a lack of quality control over comment edits. As a result, these datasets may contain noisy or inconsistent samples that hinder model learning and generalization. In this paper, we focus on cleaning existing comment updating datasets, considering both the data's characteristics in the updating scenario and their implications on the model training process. We propose a hybrid statistical approach named CupCleaner (Comment UPdating's CLEANER) to achieve this purpose. Specifically, we combine static semantic information within data samples and dynamic loss information during the training process to clean the dataset. Experimental results demonstrate that, on the same test set, both the individual static strategy and the dynamic strategy can significantly filter out a portion of the data and enhance the performance of the model. Furthermore, employing a model ensemble approach can combine the characteristics of static and dynamic cleaning, further enhancing the performance of the model and the reliability of its output results.

cs.SE↗

arXiv2025

Directional Diffusion-Style Code Editing Pre-training

Code pre-trained models have shown promising effectiveness in various software engineering tasks. Among these tasks, many tasks are related to software evolution and/or code editing. However, existing code pre-trained models often overlook the real-world code editing data and the evolutionary nature of the editing process. In this paper, to simulate the step-by-step code editing process of human developers, we propose DivoT5, a pre-trained model based on directional diffusion at the data level. In DivoT5, we adopt two categories of pre-training tasks. The first category is mask and denoising tasks augmented with a diffusion direction representing code evolution. That is, we first apply a noising process to the code snippets before evolution, and then ask the pre-training process to restore the snippets with noise into the code snippets after evolution. The second category is tasks aiming to reinforce the evolutionary direction. That is, we first generate various intermediate versions for each pair of snippets before and after evolution, and then ask the pre-training process to transform the intermediate versions into the snippet after evolution for each pair. We evaluate DivoT5 for two code-editing scenarios and one non-editing scenario using five downstream tasks. Given each downstream task, we fine-tune the pre-trained DivoT5 to evaluate its effectiveness. Our experimental results show that DivoT5 achieves state-of-the-art (SOTA) performance on most tasks in comparison to models of the same scale (220M), large scale (770M) models in fine-tuning, and billion-scale (6.7B, 8B, ChatGPT) models in few-shot settings. For one code-editing task (i.e., automated code review), DivoT5 pre-trained on top of CodeT5-small (60M) can even outperform CodeT5-base (220M) and other pre-trained models with 220M parameters except for DivoT5 pre-trained on top of CodeT5-base (220M).

cs.SE↗

arXiv2025

STAlloc: Enhancing Memory Efficiency in Large-Scale Model Training with Spatio-Temporal Planning

The rapid scaling of large language models (LLMs) has significantly increased GPU memory pressure, which is further aggravated by training optimization techniques such as virtual pipeline and recomputation that disrupt tensor lifespans and introduce considerable memory fragmentation. Such fragmentation stems from the use of online GPU memory allocators in popular deep learning frameworks like PyTorch, which disregard tensor lifespans. As a result, this inefficiency can waste as much as 43% of memory and trigger out-of-memory errors, undermining the effectiveness of optimization methods. To address this, we introduce STAlloc, a GPU memory allocator for deep learning frameworks that reduces fragmentation by exploiting the spatial and temporal regularity in memory allocation behaviors of training workloads. STAlloc introduces a novel paradigm that combines offline planning with online allocation. The offline planning leverages spatio-temporal regularities to generate a near-optimal allocation plan, while the online allocation handles complex and dynamic models such as Mixture-of-Experts (MoE). Built as a pluggable PyTorch memory allocator, STAlloc reduces fragmentation ratio on average by 85.1% (up to 100%) across both dense and MoE models, with negligible overhead. This enables more efficient, high-throughput training configurations and improves throughput performance by up to 32.5%.

cs.LG↗

arXiv2025

Large Deviation Principle for Neutral Type Mckean-Vlasov Stochastic Differential Equations

This paper investigates neutral-type McKean-Vlasov stochastic differential equations in which the drift and diffusion coefficients depend on both the segment process and its distribution. Under a one-sided Lipschitz condition on the drift coefficient, we establish a Freidlin-Wentzell-type large deviation principle for the solution process by using the extended contraction principle combined with an exponential approximation technique. Our results extend existing large deviation principles for McKean-Vlasov equations to the neutral case.

math.PR↗

arXiv2025

MPIC: Position-Independent Multimodal Context Caching System for Efficient MLLM Serving

The context caching technique is employed to accelerate the Multimodal Large Language Model (MLLM) inference by prevailing serving platforms currently. However, this approach merely reuses the Key-Value (KV) cache of the initial sequence of prompt, resulting in full KV cache recomputation even if the prefix differs slightly. This becomes particularly inefficient in the context of interleaved text and images, as well as multimodal retrieval-augmented generation. This paper proposes position-independent caching as a more effective approach for multimodal information management. We have designed and implemented a caching system, named MPIC, to address both system-level and algorithm-level challenges. MPIC stores the KV cache on local disks when receiving multimodal data, and calculates and loads the KV cache in parallel during inference. To mitigate accuracy degradation, we have incorporated the integrated reuse and recompute mechanism within the system. The experimental results demonstrate that MPIC can achieve up to 54\% reduction in response time and 2$\times$ improvement in throughput compared to existing context caching systems, while maintaining negligible or no accuracy loss.

cs.LG↗

arXiv2025

The adaptive EM schemes for McKean-Vlasov SDEs with common noise in finite and infinite horizons

This paper is dedicated to investigating the adaptive Euler-Maruyama (EM) schemes for the approximation of McKean-Vlasov stochastic differential equations (SDEs) with common noise. When the drift and diffusion coefficients both satisfy the superlinear growth conditions, the $L^p$ convergence rates in finite and infinite horizons are revealed, which reacts to the particle number and step size. Subsequently, there is an illustration of the theory results by means of two numerical examples.

math.NA↗

arXiv2025

The Euler-Maruyama method for SDEs with low-regularity drift

We study the strong $L^p$-convergence rates of the Euler-Maruyama method for stochastic differential equations driven by Brownian motion with low-regularity drift coefficients. Specifically, the drift is assumed to be in the Lebesgue-Hölder spaces $L^q([0,T]; {\mathcal C}_b^α({\mathbb R}^d))$ with $α\in(0,1)$ and $q\in (2/(1+α),\infty]$. For every $p\geq 2$, by using stochastic sewing and/or the Itô-Tanaka trick, we obtain the $L^p$-convergence rates: $(1+α)/2$ for $q\in [2,\infty]$ and $(1-1/q)$ for $q\in (2/(1+α),2)$. Moreover, we prove that the unique strong solution can be constructed via the Picard iteration.

math.PR↗

arXiv2025

The LDP of McKean-Vlasov stochastic differential equations with Hölder continuous conditions and integrable conditions

In this paper, we first study the large deviation principle (LDP) for non-degenerate McKean-Vlasov stochastic differential equations (MVSDEs) with Hölder continuous drifts by using Zvonkin's transformation. When the drift only satisfies Hölder condition, the skeleton equation may have multiple solutions. Among these solutions, we find one that ensures the MVSDEs satisfy the LDP. Moreover, we introduce a new definition for the rate function that reduces to traditional rate function if the drift satisfies the Lipschitz condition. Secondly, we study the LDP for degenerate MVSDEs with Hölder continuous drifts.

math.PR↗

arXiv2025

DeepServe: Serverless Large Language Model Serving at Scale

In this paper, we propose DEEPSERVE, a scalable and serverless AI platform designed to efficiently serve large language models (LLMs) at scale in cloud environments. DEEPSERVE addresses key challenges such as resource allocation, serving efficiency, and cold start latencies through four main design components. First, DEEPSERVE uses a simple serverless abstraction called the request-job-task model, which helps manage diverse AI workloads across posttraining and model-serving tasks. Second, DEEPSERVE integrates an in-house serving engine named FLOWSERVE using a microkernel-inspired design, NPU-centric execution, and SPMD-based parallelism to optimize LLM serving. Third, DEEPSERVE includes novel scheduling policies tailored for a configuration with both PD-disaggregated and PD-colocated instances. Fourth, DEEPSERVE includes optimizations such as pre-warmed pods, DRAM pre-loading, and NPU-fork, which allow DEEPSERVE to scale up to 64 instances in seconds. DEEPSERVE has been in production for over a year, operating on a large Ascend NPU cluster and providing industrystandard APIs for fine-tuning, agent serving, and model serving to our customers.

cs.DC↗

arXiv2025

RaaS: Reasoning-Aware Attention Sparsity for Efficient LLM Reasoning

Large Language Models (LLMs) have demonstrated strong capabilities across various domains, with recent advancements in challenging reasoning tasks such as mathematics and programming. However, solving reasoning tasks often requires an LLM to generate long sequences, incurring $O(N)$ time and memory complexities per token, where $N$ is the current sequence length. To reduce complexities, existing sparsity-based algorithms propose to retain Key-Value (KV) vectors, the intermediate representations of only the most critical tokens. However, these algorithms struggle with the "impossible trinity" of accuracy, time, and memory. For example, the state-of-the-art algorithm, Quest, achieves high accuracy with $O(L)$ time but $O(N)$ memory ($L$ is the cache budget, $L \ll N$). To address the "impossible trinity", in this paper, we identify a new attention pattern during the decode stage of reasoning tasks, where milestone tokens (analogous to lemmas in mathematical proofs) emerge, are utilized, and then become unimportant afterward. Based on this pattern, we propose a new algorithm RaaS that identifies milestone tokens and retains their KV vectors until they are no longer needed, achieving high accuracy with $O(L)$ time and $O(L)$ memory complexities.

cs.LG↗

arXiv2025

EPIC: Efficient Position-Independent Caching for Serving Large Language Models

Large Language Models (LLMs) show great capabilities in a wide range of applications, but serving them efficiently becomes increasingly challenging as requests (prompts) become more complex. Context caching improves serving performance by reusing Key-Value (KV) vectors, the intermediate representations of tokens that are repeated across requests. However, existing context caching requires exact prefix matches across requests, limiting reuse cases in settings such as few-shot learning and retrieval-augmented generation, where immutable content (e.g., documents) remains unchanged across requests but is preceded by varying prefixes. Position-Independent Caching (PIC) addresses this issue by enabling modular reuse of the KV vectors regardless of prefixes. We formalize PIC and advance prior work by introducing EPIC, a serving system incorporating our new LegoLink algorithm, which mitigates the inappropriate "attention sink" effect at every document beginning, to maintain accuracy with minimal computation. Experiments show that EPIC achieves up to 8x improvements in Time-To-First-Token (TTFT) and 7x throughput gains over existing systems, with negligible or no accuracy loss.

cs.LG↗

arXiv2025

LLMigrate: Transforming "Lazy" Large Language Models into Efficient Source Code Migrators

Rewriting C code in Rust provides stronger memory safety, yet migrating large codebases such as the 32-million-line Linux kernel remains challenging. While rule-based translators (e.g., C2Rust) provide accurate yet largely unsafe Rust programs, recent Large Language Model (LLM) approaches produce more idiomatic, safe Rust programs but frequently exhibit "laziness", omitting significant portions of the target code. To address the issue, in this paper, we present LLMigrate, an LLM-based C-to-Rust translation tool that splits modules into discrete functions, translating them individually, and then reintegrating them. LLMigrate uses static analysis to retain necessary context, pairs GPT-4o (a state-of-the-art LLM) with compiler-driven translation and program-repair techniques for complex core functions, and leverages call-graph-guided translation to ensure consistent interfaces. Evaluations on three representative Linux kernel modules (math, sort, and ramfs) show that LLMigrate requires modifying less than 15\% of the target code, significantly outperforming a pure GPT-4o-based migration.

cs.PL↗

arXiv2024

MemServe: Context Caching for Disaggregated LLM Serving with Elastic Memory Pool

Large language model (LLM) serving has transformed from stateless to stateful systems, utilizing techniques like context caching and disaggregated inference. These optimizations extend the lifespan and domain of the KV cache, necessitating a new architectural approach. We present MemServe, a unified system that integrates both inter-request and intra-request optimizations. MemServe introduces MemPool, an elastic memory pool managing distributed memory and KV caches across serving instances. Using MemPool APIs, MemServe combines context caching with disaggregated inference for the first time, supported by a global scheduler that enhances cache reuse through a global prompt tree-based locality-aware policy. Tests show that MemServe significantly improves job completion time and time-to-first-time.

cs.DC↗

arXiv2024

A Plug-and-Play Image Registration Network

Deformable image registration (DIR) is an active research topic in biomedical imaging. There is a growing interest in developing DIR methods based on deep learning (DL). A traditional DL approach to DIR is based on training a convolutional neural network (CNN) to estimate the registration field between two input images. While conceptually simple, this approach comes with a limitation that it exclusively relies on a pre-trained CNN without explicitly enforcing fidelity between the registered image and the reference. We present plug-and-play image registration network (PIRATE) as a new DIR method that addresses this issue by integrating an explicit data-fidelity penalty and a CNN prior. PIRATE pre-trains a CNN denoiser on the registration field and "plugs" it into an iterative method as a regularizer. We additionally present PIRATE+ that fine-tunes the CNN prior in PIRATE using deep equilibrium models (DEQ). PIRATE+ interprets the fixed-point iteration of PIRATE as a network with effectively infinite layers and then trains the resulting network end-to-end, enabling it to learn more task-specific information and boosting its performance. Our numerical results on OASIS and CANDI datasets show that our methods achieve state-of-the-art performance on DIR.

eess.IV↗

arXiv2023

The randomized Milstein scheme for stochastic Volterra integral equations with weakly singular kernels

This paper focuses on the randomized Milstein scheme for approximating solutions to stochastic Volterra integral equations with weakly singular kernels, where the drift coefficients are non-differentiable. An essential component of the error analysis involves the utilization of randomized quadrature rules for stochastic integrals to avoid the Taylor expansion in drift coefficient functions. Finally, we implement the simulation of multiple singular stochastic integral in the numerical experiment by applying the Riemann-Stieltjes integral.

math.NA↗

arXiv2023

Stochastic equations with low regularity drifts

By using the Itô-Tanaka trick, we prove the unique strong solvability as well as the gradient estimates for stochastic differential equations with irregular drifts in low regularity Lebesgue-Hölder space $L^q(0,T;{\mathcal C}_b^α({\mathbb R}^d))$ with $α\in(0,1)$ and $q\in (2/(1+α),2$). As applications, we show the unique weak and strong solvability for stochastic transport equations driven by the low regularity drift with $q\in (4/(2+α),2$) as well as the local Lipschitz estimate for stochastic strong solutions.

math.PR↗

arXiv2023

Large deviation for slow-fast McKean-Vlasov stochastic differential equations driven by fractional Brownian motions and Brownian motions

In this article, we consider slow-fast McKean-Vlasov stochastic differential equations driven by Brownian motions and fractional Brownian motions. We give a definition of the large deviation principle (LDP) on the product space related to Brownian motion and fractional Brownian motion, which is different from the traditional definition for LDP. Under some proper assumptions on coefficients, LDP is investigated for this type of equations by using the weak convergence method.

math.PR↗