甲骨文公司

Emotional Support Conversation (ESC) aims to alleviate individual emotional distress by generating empathetic responses. However, existing methods struggle with coarse-grained retrieval alignment and unconstrained external knowledge integration, limiting their ability to produce contextually grounded and personalized supportive responses. To address these two domain-specific challenges in ESC, we present PRCCF, a tailored framework for persona-aligned retrieval and causal-aware cognitive filtering. Specifically, the persona-guided retrieval mechanism jointly models semantic compatibility and inferred persona alignment to select contextually appropriate demonstrations. The causal-aware cognitive filtering module grounds external commonsense within detected emotion-cause dialogue spans, enabling fine-grained situational cognitive understanding. Extensive experiments on ESConv show that PRCCF achieves favorable results compared with reproduced competitive baselines in both automatic metrics and human evaluation, yielding responses with improved coherence, emotional understanding, and support relevance.

The exponential growth in electricity demand driven by artificial intelligence (AI) is threatening grid reliability, increasing energy costs for communities funding new infrastructure and slowing AI innovation as data centres await interconnection to constrained grids. Here we present a field demonstration of a software-based method that enables AI data centres to operate as flexible grid resources. Tested on a 256-Graphics Processing Unit (GPU) cluster running representative AI workloads in a hyperscale cloud facility in Phoenix, Arizona, the system reduced power usage by 25% for 3 hours during peak demand while maintaining AI quality of service guarantees. By coordinating workloads in response to real-time grid signals, without hardware modifications or energy storage, this approach demonstrates the potential for data centres to contribute to grid stability and affordability while sustaining computational performance within existing power-system constraints.

Large language model (LLM) agents rely on external tools to solve complex tasks, but real-world toolsets often contain redundant tools with overlapping names and descriptions, introducing ambiguity and reducing selection accuracy. LLMs also face strict input context limits, preventing efficient consideration of large toolsets. To address these challenges, we propose ToolScope, which includes: (1) ToolScopeMerger with Auto-Correction to automatically audit and fix tool merges, reducing redundancy, and (2) ToolScopeRetriever to rank and select only the most relevant tools for each query, compressing toolsets to fit within context limits without sacrificing accuracy. Evaluations on three state-of-the-art LLMs and three open-source tool-use benchmarks show gains of 8.38% to 38.6% in tool selection accuracy, demonstrating ToolScope’s effectiveness in enhancing LLM tool use.

The era of GPU-powered data analytics has arrived. In this paper, we argue that recent advances in hardware (e.g., larger GPU memory, faster interconnect and IO, and declining cost) and software (e.g., composable data systems and mature libraries) have removed the key barriers that have limited the wider adoption of GPU data analytics. We present Sirius, a prototype open-source GPU-native SQL engine that offers drop-in acceleration for diverse data systems. Sirius treats GPU as the primary engine and leverages libraries like libcudf for high-performance relational operators. It provides drop-in acceleration for existing databases by leveraging the standard Substrait query representation, replacing the CPU engine without changing the user-facing interface. On TPC-H, Sirius achieves 7x speedup when integrated with DuckDB in a single node at the same hardware rental cost, and up to 12.5x speedup when integrated with Apache Doris in a distributed setting.

Recent advances in LLM-based Text-to-SQL have achieved remarkable gains on public benchmarks such as BIRD and Spider. Yet, these systems struggle to scale in realistic enterprise settings with large, complex schemas, diverse SQL dialects, and expensive multi-step reasoning. Emerging agentic approaches show potential for adaptive reasoning but often suffer from inefficiency and instability-repeating interactions with databases, producing inconsistent outputs, and occasionally failing to generate valid answers. To address these challenges, we introduce Agent Semantic Memory (AgentSM), an agentic framework for Text-to-SQL that builds and leverages interpretable semantic memory. Instead of relying on raw scratchpads or vector retrieval, AgentSM captures prior execution traces-or synthesizes curated ones-as structured programs that directly guide future reasoning. This design enables systematic reuse of reasoning paths, which allows agents to scale to larger schemas, more complex questions, and longer trajectories efficiently and reliably. Compared to state-of-the-art systems, AgentSM achieves higher efficiency by reducing average token usage and trajectory length by 25