Xinru Tang

Kapil Garg, Xinru Tang, Jimin Heo et al.

Vision-Language Models (VLMs) are increasingly used by blind and low-vision (BLV) people to identify and understand products in their everyday lives, such as food, personal products, and household goods. Despite their prevalence, we lack an empirical understanding of how common image quality issues, like blur and misframing of items, affect the accuracy of VLM-generated captions and whether resulting captions meet BLV people's information needs. Grounded in a survey with 86 BLV people, we systematically evaluate how image quality issues affect captions generated by VLMs. We show that the best model recognizes products in images with no quality issues with 98% accuracy, but drops to 75% accuracy overall when quality issues are present, worsening considerably as issues compound. We discuss the need for model evaluations that center on disabled people's experiences throughout the process and offer concrete recommendations for HCI and ML researchers to make VLMs more reliable for BLV people.

Xinru Tang, Anne Marie Piper

As mixed-ability collaboration has become increasingly focal within accessibility research, managing varied, and sometimes conflicting, access needs has become a key consideration in designing for access. When an accessibility feature or practice benefits some people while constraining others, how should designers navigate these trade-offs? This paper responds to this question by analyzing how a mixed-ability nonprofit worked to make communication accessible to its members as it grew from a small blind-focused athletic group to a larger cross-disability organization. Based on a six-month study that combines interviews and field observations, we show that working with conflicting access needs is not just a technical 'problem' but a generative process that sparks reflection on technical constraints and preferences, diverse roles and communication norms, and organizational demands. We therefore argue for rethinking "conflicts" in access as key sites for revealing power structures and creating opportunities for accountability and repair.

Xinru Tang, Ting-an Lin, Jingjin Li et al.

Drawing on crip theory, this paper proposes cripping AI as a guiding framework to center lived disability experiences in AI research and development. Moving beyond calls to make AI "accessible" to people with disabilities, cripping AI seeks to: (1) reveal and dismantle ableist assumptions embedded in how AI is imagined, designed, and evaluated; (2) center disabled ways of knowing (i.e., cripistemologies); (3) respect disabled labor in co-creating accessible practices. We demonstrate how to apply our framework with three cases: deafness and sign language AI, blindness and visual assistive AI, and stuttering and speech AI. We end by outlining three directions for future work, including cripping AI with diverse human bodyminds, across the entire AI pipeline and ecosystem, and in collaboration with other justice-oriented AI efforts.

Mingcong Song, Xinru Tang, Fengfan Hou et al.

Meeting growing demands for low latency and cost efficiency in production-grade large language model (LLM) serving systems requires integrating advanced optimization techniques. However, dynamic and unpredictable input-output lengths of LLM, compounded by these optimizations, exacerbate the issues of workload variability, making it difficult to maintain high efficiency on AI accelerators, especially DSAs with tile-based programming models. To address this challenge, we introduce XY-Serve, a versatile, Ascend native, end-to-end production LLM-serving system. The core idea is an abstraction mechanism that smooths out the workload variability by decomposing computations into unified, hardware-friendly, fine-grained meta primitives. For attention, we propose a meta-kernel that computes the basic pattern of matmul-softmax-matmul with architectural-aware tile sizes. For GEMM, we introduce a virtual padding scheme that adapts to dynamic shape changes while using highly efficient GEMM primitives with assorted fixed tile sizes. XY-Serve sits harmoniously with vLLM. Experimental results show up to 89% end-to-end throughput improvement compared with current publicly available baselines on Ascend NPUs. Additionally, our approach outperforms existing GEMM (average 14.6% faster) and attention (average 21.5% faster) kernels relative to existing libraries. While the work is Ascend native, we believe the approach can be readily applicable to SIMT architectures as well.

Jinyi Deng, Xinru Tang, Zhiheng Yue et al.

Given the increasing complexity of AI applications, traditional spatial architectures frequently fall short. Our analysis identifies a pattern of interconnected, multi-faceted tasks encompassing both AI and general computational processes. In response, we have conceptualized "Orchestrated AI Workflows," an approach that integrates various tasks with logic-driven decisions into dynamic, sophisticated workflows. Specifically, we find that the intrinsic Dual Dynamicity of Orchestrated AI Workflows, namely dynamic execution times and frequencies of Task Blocks, can be effectively represented using the Orchestrated Workflow Graph. Furthermore, the intrinsic Dual Dynamicity poses challenges to existing spatial architecture, namely Indiscriminate Resource Allocation, Reactive Load Rebalancing, and Contagious PEA Idleness. To overcome these challenges, we present Octopus, a scale-out spatial architecture and a suite of advanced scheduling strategies optimized for executing Orchestrated AI Workflows, such as the Discriminate Dual-Scheduling Mechanism, Adaptive TBU Scheduling Strategy, and Proactive Cluster Scheduling Strategy. Our evaluations demonstrate that Octopus significantly outperforms traditional architectures in handling the dynamic demands of Orchestrated AI Workflows, and possesses robust scalability in large scale hardware such as wafer-scale chip.