taxonomy

This paper demonstrates that post-training quantization can silently alter how large language models reason, even when task accuracy is preserved, through a taxonomy-based analysis of 30,000 chain-of-thought outputs across multiple models and benchmarks.

This paper presents a taxonomy and lifecycle survey of dynamic skill libraries for large language model agents, proposing an eight-stage lifecycle architecture and a six-sense taxonomy to organize evolving skill artifacts.

普林斯顿博士后Shilong Liu提出自进化代理的三层分类体系：工件迭代优化、Agent Harness自我改进和无黄金答案的模型学习，系统梳理了相关概念和前沿工作。

A comprehensive survey of 1,250 papers (2024–2026) on recursive self-improvement in AI, proposing a taxonomy distinguishing bounded self-refinement from open-ended recursive self-improvement, and analyzing the evaluator design space and failure modes.

Shilong Liu proposes a taxonomy classifying self-evolving agents into artifact optimization, harness self-improvement, and model learning, providing a common language for emerging agent research.

This paper synthesizes 27 benchmark, taxonomy, and audit papers from 2023-2026 into a unified taxonomy of LLM agent limitations, identifying six failure clusters including tool invocation errors, planning failures, long-horizon degradation, multi-agent coordination issues, safety concerns, and measurement validity problems.

This paper defines embodied operators as reusable functional modules for embodied intelligence pipelines, presents a taxonomy covering five major categories, and proposes a multi-dimensional benchmark framework for evaluating their deployability and composability.

OmniOpt presents a unified framework for optimizer selection in large-scale model training, combining meta-pipeline transformations, norm-constrained linear minimization oracles, and a cross-domain benchmark to systematically analyze optimizer families and their trade-offs.

This paper catalogues five recurring MCP server architectural patterns observed across fifteen independently developed servers, providing a taxonomy with context, problem, solution, and consequences. It also documents anti-patterns, cross-cutting concerns, and quantitative evaluations including inter-rater reliability and transport overhead.

This paper introduces the Agent-Native Immune System (ANIS), a biologically inspired, endogenous defense architecture embedded directly within the agent's cognitive loop. It proposes a six-layer Immune Tower, a unified taxonomy of Agent Viruses and Vaccines, and the Harness Triad for continual immune learning to address runtime hijacking vulnerabilities in autonomous agents.

The paper presents a taxonomy of factual errors in human-written text, derived from newspaper corrections, and evaluates LLMs' performance on detecting these errors, finding that even top models like GPT-5.4 achieve only 52% word-level F1 score, highlighting the task's difficulty.

This thesis from Aalto University presents a taxonomy of synchronization architectures, analyzing trade-offs and decision factors to guide the design of generalized sync engines.

This paper introduces a taxonomy of conversational memory types and a user-centric evaluation framework to study how different memory roles affect response quality in RAG-based conversational agents.

Ai2 and the University of Washington released a paper titled Tmax, proposing the strongest open-source terminal agent RL training recipe to date. A 9B parameter model outperforms larger models on Terminal-Bench 2.0, with the key being low-cost generation of vast amounts of verifiable training data, not model size or algorithm.

This paper presents a taxonomy and deobfuscation methods for a real-world binary obfuscator, likely presented at the Re//Verse 2026 conference.

This paper proposes ActionRating, a formulation that places clarification inside an agent's action space on a shared ordinal scale with navigation, enabling two information-seeking modes (mandatory and opportunistic). On hierarchical taxonomy classification benchmarks, experiments with 9 LLMs show that opportunistic clarification improves accuracy and information-seeking effectiveness.

Introduces ArabiGEE, the first comprehensive Arabic grammatical error explanation taxonomy with a hierarchical structure spanning orthographic, morphological, syntactic, and lexical dimensions, comprising 27 error types, 140 correction types, and 324 explanations.

This survey reviews the use of large language models for graph computation, categorizing them into two paradigms: LLMs as executors and LLMs as planners. It finds LLMs promising for simple tasks but unreliable for large-scale exact computations, and suggests future directions.

This paper presents a comprehensive taxonomy of 3D vision research, covering geometric representations, datasets, learning paradigms, and applications in reconstruction, generation, and video modeling.

This paper presents a comprehensive 33-class taxonomy of recurrent distortion patterns (heuristic parasites) in LLM outputs, along with operational definitions, recognition criteria, and a reproducible measurement protocol (PPE) for quantifying behavioral degradation across conversations.