| Vaswani et al. 2017, Attention Is All You Need |
S2 Transformer 與 decoding |
| Lewis et al. 2020, Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks |
S9 檢索與 RAG |
| Kwon et al. 2023, Efficient Memory Management for LLM Serving with PagedAttention |
S5 推論與 serving |
| Yao et al. 2023, ReAct |
S12 Agent loop 與 harness |
| Shao et al. 2024, DeepSeekMath（GRPO） |
S4 Post-training 與 reasoning |
| Rafailov et al. 2023, Direct Preference Optimization |
S4 Post-training 與 reasoning |
| Sumers et al. 2023, Cognitive Architectures for Language Agents (CoALA) |
S13 Memory 與個人化 |
| Kapoor et al. 2024, AI Agents That Matter |
S0 全局地圖、S14 Agent eval、資料與最佳化 |
| Miller 2024, Adding Error Bars to Evals |
S8 Eval 方法論 |
| Liu et al. 2023, Lost in the Middle: How Language Models Use Long Contexts |
S9 檢索與 RAG、S11 Context engineering |
| Sclar et al. 2023, Quantifying Language Models’ Sensitivity to Spurious Features in Prompt Design |
S7 Prompt、structured output 與 workflows |
| Yang et al. 2024, SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering |
S10 工具與 MCP、S19 真實 agent 系統與畢業專題 |
| Yao et al. 2024, τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains |
S8 Eval 方法論、S14 Agent eval、資料與最佳化 |
| Welleck et al. 2024, From Decoding to Meta-Generation |
S2 Transformer 與 decoding、S5 推論與 serving |
| Shankar et al. 2024, Who Validates the Validators? |
S8 Eval 方法論 |
| Cemri et al. 2025, Why Do Multi-Agent LLM Systems Fail? |
S15 Multi-agent |
| Kwa et al. 2025, Measuring AI Ability to Complete Long Tasks（METR time horizon） |
S14 Agent eval、資料與最佳化 |
| Opsahl-Ong et al. 2024, MIPROv2 |
S14 Agent eval、資料與最佳化 |
| Agrawal et al. 2025, GEPA |
S14 Agent eval、資料與最佳化 |
| Beurer-Kellner et al. 2025, Design Patterns for Securing LLM Agents against Prompt Injections |
S16 安全與治理 |
| Debenedetti et al. 2025, Defeating Prompt Injections by Design (CaMeL) |
S16 安全與治理 |
| Nasr et al. 2025, The Attacker Moves Second |
S16 安全與治理 |
| Kaplan et al. 2020, Scaling Laws for Neural Language Models |
S3 Pretraining 與 scaling |
| Hoffmann et al. 2022, Training Compute-Optimal Large Language Models |
S3 Pretraining 與 scaling |
| Wei et al. 2022, Emergent Abilities of Large Language Models |
S3 Pretraining 與 scaling |
| Schaeffer et al. 2023, Are Emergent Abilities of Large Language Models a Mirage? |
S3 Pretraining 與 scaling |
| Rajbhandari et al. 2019, ZeRO: Memory Optimizations Toward Training Trillion Parameter Models |
S3 Pretraining 與 scaling |
| Lambert et al. 2024, Tulu 3: Pushing Frontiers in Open Language Model Post-Training |
S4 Post-training 與 reasoning |
| Zhou et al. 2023, LIMA: Less Is More for Alignment |
S4 Post-training 與 reasoning |
| Hu et al. 2021, LoRA: Low-Rank Adaptation of Large Language Models |
S4 Post-training 與 reasoning |
| Dettmers et al. 2023, QLoRA: Efficient Finetuning of Quantized LLMs |
S4 Post-training 與 reasoning |
| Wang et al. 2022, Self-Consistency Improves Chain of Thought Reasoning in Language Models |
S4 Post-training 與 reasoning |
| Lightman et al. 2023, Let’s Verify Step by Step |
S4 Post-training 與 reasoning |
| Yu et al. 2025, DAPO: An Open-Source LLM Reinforcement Learning System at Scale |
S4 Post-training 與 reasoning |
| Zheng et al. 2025, Group Sequence Policy Optimization |
S4 Post-training 與 reasoning |
| Yu et al. 2022, Orca: A Distributed Serving System for Transformer-Based Generative Models（Orca，OSDI ’22） |
S5 推論與 serving |
| Zheng et al. 2023, SGLang: Efficient Execution of Structured Language Model Programs |
S5 推論與 serving |
| Dao et al. 2022, FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness |
S5 推論與 serving |
| Leviathan et al. 2022, Fast Inference from Transformers via Speculative Decoding |
S5 推論與 serving |
| Zhao et al. 2021, Calibrate Before Use: Improving Few-Shot Performance of Language Models |
S7 Prompt、structured output 與 workflows |
| Wei et al. 2022, Chain-of-Thought Prompting Elicits Reasoning in Large Language Models |
S7 Prompt、structured output 與 workflows |
| Tam et al. 2024, Let Me Speak Freely? A Study on the Impact of Format Restrictions on Performance of Large Language Models |
S7 Prompt、structured output 與 workflows |
| Zheng et al. 2023, Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena |
S8 Eval 方法論 |
| Chen et al. 2021, Evaluating Large Language Models Trained on Code |
S8 Eval 方法論 |
| Bean et al. 2025, Measuring what Matters: Construct Validity in Large Language Model Benchmarks |
S8 Eval 方法論 |
| Kadavath et al. 2022, Language Models (Mostly) Know What They Know |
S8 Eval 方法論 |
| Karpukhin et al. 2020, Dense Passage Retrieval for Open-Domain Question Answering |
S9 檢索與 RAG |
| Khattab & Zaharia 2020, ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT |
S9 檢索與 RAG |
| Malkov & Yashunin 2016, Efficient and robust approximate nearest neighbor search using Hierarchical Navigable Small World graphs |
S9 檢索與 RAG |
| Gao et al. 2022, Precise Zero-Shot Dense Retrieval without Relevance Labels |
S9 檢索與 RAG |
| Asai et al. 2023, Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection |
S9 檢索與 RAG |
| Gao et al. 2023, Enabling Large Language Models to Generate Text with Citations |
S9 檢索與 RAG |
| Hsieh et al. 2024, RULER: What’s the Real Context Size of Your Long-Context Language Models? |
S9 檢索與 RAG |
| Greshake et al. 2023, Not what you’ve signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection |
S9 檢索與 RAG |
| Schick et al. 2023, Toolformer: Language Models Can Teach Themselves to Use Tools |
S10 工具與 MCP |
| Patil et al. 2023, Gorilla: Large Language Model Connected with Massive APIs |
S10 工具與 MCP |
| Wang et al. 2024, Executable Code Actions Elicit Better LLM Agents |
S10 工具與 MCP |
| Jiang et al. 2023, LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compression |
S11 Context engineering |
| Shinn et al. 2023, Reflexion: Language Agents with Verbal Reinforcement Learning |
S12 Agent loop 與 harness |
| Packer et al. 2023, MemGPT: Towards LLMs as Operating Systems |
S13 Memory 與個人化 |
| Park et al. 2023, Generative Agents: Interactive Simulacra of Human Behavior |
S13 Memory 與個人化 |
| Maharana et al. 2024, Evaluating Very Long-Term Conversational Memory of LLM Agents |
S13 Memory 與個人化 |
| Chen et al. 2024, AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases |
S13 Memory 與個人化 |
| Agent-as-a-Judge: Evaluate Agents with Agents |
S14 Agent eval、資料與最佳化 |
| $τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment |
S14 Agent eval、資料與最佳化 |
| SABER: Small Actions, Big Errors – Safeguarding Mutating Steps in LLM Agents |
S14 Agent eval、資料與最佳化 |
| DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines |
S14 Agent eval、資料與最佳化 |
| Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning |
S14 Agent eval、資料與最佳化 |
| Wu et al. 2023, AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation |
S15 Multi-agent |
| Shao et al. 2024, PrivacyLens: Evaluating Privacy Norm Awareness of Language Models in Action |
S16 安全與治理 |
| Greenblatt et al. 2023, AI Control: Improving Safety Despite Intentional Subversion |
S16 安全與治理 |
| Korbak et al. 2025, Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety |
S16 安全與治理 |
| Ong et al. 2024, RouteLLM: Learning to Route LLMs with Preference Data |
S18 部署與架構 |
| Chen et al. 2023, FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performance |
S18 部署與架構 |
| Xia et al. 2024, Agentless: Demystifying LLM-based Software Engineering Agents |
S19 真實 agent 系統與畢業專題 |
| Wang et al. 2024, OpenHands: An Open Platform for AI Software Developers as Generalist Agents |
S19 真實 agent 系統與畢業專題 |
| Wei et al. 2025, BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents |
S19 真實 agent 系統與畢業專題 |
