Yuankai Fan 范元凯

This use case presents AI-powered green multi-cluster management for large-scale intelligent infrastructure. It uses AI-driven scheduling, GPU multiplexing, cross-cluster coordination, and multimodal explanation mechanisms to improve cluster utilization, reduce energy consumption, and support sustainable low-carbon AI infrastructure management.

This survey rethinks the role of data in NL2SQL, reviewing what data resources, benchmarks, and practices the community currently has and outlining what future NL2SQL systems and datasets should support.

Natural Language Interfaces for Databases empower non-technical users to interact with data using natural language (NL). Advanced approaches, utilizing either neural sequence-to-sequence or more recent sophisticated large-scale language models, typically implement NL to SQL (NL2SQL) translation in an end-to-end fashion. However, like humans, these end-to-end translation models may not always generate the best SQL output on their first try. In this paper, we propose CycleSQL, an iterative framework designed for end-to-end translation models to autonomously generate the best output through self-evaluation. The main idea of CycleSQL is to introduce data-grounded NL explanations of query results as self-provided feedback, and use the feedback to validate the correctness of the translation iteratively, hence improving the overall translation accuracy.

Current large language model (LLM)-based Natural Language to SQL (NL2SQL) approaches typically rely on the database schema and partial data values for the translation. These approaches are unable to use sufficient data for accurate database understanding due to limitations in data selection methods, and they cannot input the entire database due to the limited context window sizes of LLMs. This insufficient data integration may result in an incomplete understanding of the database, leading to semantically incorrect SQL generation. In this paper, we introduce REDSQL, a novel plug-and-play framework that refines the predicted SQL by utilizing the entire database in the refinement process.

Large-scale artificial intelligence (AI) models are fundamentally transforming industries and redefining the paradigm of human–machine collaboration. While the technological revolution signals a new era of machine intelligence, the continued scaling of these models has exposed significant limitations in contemporary hardware architectures, manifesting as constraints on computational efficiency, interconnection bandwidth, and memory capacity. These three dimensions are inseparably intertwined, such that advances along any single axis often exacerbate bottlenecks in the others, rendering isolated optimizations increasingly ineffective. Achieving an optimal balance among them to maximize system efficiency therefore remains a central challenge in the design of scalable AI systems. To address this challenge, we introduce Computation-Bandwidth-Memory Trade-offs, termed the AI Trinity, a unified paradigm that positions computation, bandwidth, and memory as coequal pillars for next-generation AI infrastructure. At its core, AI Trinity enables a dynamic flow of resources across these pillars, transcending single-resource bottlenecks and adapting to diverse scenarios, thereby optimizing overall system performance to its maximum potential. Within this framework, AI Trinity identifies three fundamental trade-offs: (1) More Computation→Less Bandwidth, wherein computational resources are exploited to reduce data transmission under limited bandwidth conditions, (2) More Bandwidth→Less Memory, which exploits abundant communication capacity to populate or refresh memory when local storage resources are constrained, and (3) More Memory→Less Computation, whereby storage capacity are utilized to mitigate redundant computation when computational costs are prohibitive.

Recent advancements in TableQA leverage sequence-to-sequence (Seq2seq) deep learning models to accurately respond to natural language queries. These models achieve this by converting the queries into SQL queries, using information drawn from one or more tables. However, Seq2seq models often produce uncertain low-confidence predictions when distributing probability mass across multiple outputs during a decoding step, frequently yielding translation errors. To tackle this problem, we present CKIF, a confidence-based knowledge integration framework that uses a two-stage deep-learning-based ranking technique to mitigate the low-confidence problem commonly associated with Seq2seq models for TableQA.

The Natural Language Interface to Databases (NLIDB) empowers non-technical users with database access through intuitive natural language interactions. Advanced approaches, utilizing neural sequence-to-sequence models or large-scale language models, typically employ auto-regressive decoding to generate unique SQL queries sequentially. In this paper, we propose MetaSQL, a unified generate-then-rank framework that can be flexibly incorporated with existing NLIDBs to consistently improve their translation accuracy. MetaSQL introduces query metadata to control the generation of better SQL query candidates and uses learning-to-rank algorithms to retrieve globally optimized queries.

A Natural Language Interface to Databases (NLIDB) aims to help end-users access databases. State-of-the-art approaches primarily construct language translation models to convert NL queries to SQL queries. While these models exhibit good performance on NLIDB benchmarks, the translation accuracy seems to have stalled at between 70%-75%, and most erroneous translations happen with complex queries that require an understanding of the structure and semantics specific to a database. This paper proposes a Generate-And-Rank approach called GAR.

GenSql is a generative natural language interface to database systems that demonstrates natural language to SQL translation capabilities for interactive database access.