The paper systematically analyzes the benefits and limits of Attention-FFN Disaggregation (AFD) for Mixture-of-Experts (MoE) LLM serving, demonstrating that AFD is crucial for achieving high throughpu…

Libra is a load balancing approach for long-context LLM training that groups packed sequences into fixed-size pools and reduces attention workload variance, improving end-to-end throughput and straggl…

The paper proposes moving the query instead of the KV-cache during cross-instance attention, demonstrating that this approach is significantly cheaper than moving the cache, especially on modern GPU f…

This survey reviews AI-driven methods for filtering and prioritizing security alerts to combat alert fatigue, establishing a four-stage workflow taxonomy and identifying critical gaps in current resea…

This paper introduces campaign diagrams, a visualization technique for analyzing resource utilization and identifying bottlenecks in modern workloads.

The paper presents Dyserve, a workflow-aware serving layer for agentic AI applications that compiles per-node model and verifier choices into an integer linear program, allowing for efficient model se…

The paper introduces AsyncTool, a new benchmark designed to evaluate LLM agents' ability to handle multiple, concurrent tasks with delayed tool feedback, demonstrating that asynchronous coordination i…

The paper proposes AsymCache, a computation-latency-aware KV cache management system that optimizes LLM inference by aligning cache eviction decisions with GPU attention kernel performance, significan…

The paper introduces FORCEBENCH, a new stress test designed to evaluate whether cited sources genuinely warrant the strength of a claim, revealing that standard citation evaluation methods often fail…

The paper introduces Clover, a code completion tool that logs students' interactions and offers attention checks to promote reflective engagement during programming tasks.

The paper proposes a comprehensive monitoring and triage methodology for agentic systems, demonstrating that structural defects mask task-level errors and require specialized monitoring scopes for det…

The paper introduces CI-Work, a benchmark demonstrating that current enterprise LLM agents frequently leak sensitive information while performing tasks, suggesting that privacy protection requires arc…

SAFARI is a framework that decouples diagnostic accuracy from context limits in autonomous agents by equipping them with a toolbox to read and search trajectory segments and a persistent STM for cross…

The paper introduces Hyperparam, a set of lightweight JavaScript libraries designed to enable direct, model-aware querying of unstructured data (like agent traces) within client-side AI applications.

This paper proposes Head-Chunked Multi-Stream Pipeline (HCMS) to exploit the computational independence of multi-head attention and achieve fine-grained communication-computation overlap, resulting in…

Moment-KV introduces a novel momentum-based technique to compress the Key-Value (KV) cache during the decoding phase of LLM generation, significantly improving fidelity in long-generation tasks.