跳到主要内容

论文库

论文与网络文章:研究时的一手参考。读完我们的拆解,不必回去看原文。

收 139 篇:论文 139 篇、网络文章 0 篇; 其中 0 篇已经拆完。

判据和书架一样:你读完我们的拆解,不必回去看原文。

这些论文是怎么进来的​

第一批 139 篇,来自书架自己的引用 —— 85 本书的拆解里引到它们、却一篇都没读过原文。先补这个缺口。

怎么引用它​

paper=<id>@arXiv:<编号><版本>  §<节>      论文,版本号必须钉死
article=<id>@<抓取日期>        §<小节>    文章,抓取日期必须钉死

别的书架引这里,用跨库锚:shelf=ai-paper-reference/<id>#index.md。

pretraining —— 预训练、规模律、数据配方(46 篇)​

  • Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks — 书架上 9 本书引过它
  • LoRA: Low-Rank Adaptation of Large Language Models — 书架上 7 本书引过它
  • Training Compute-Optimal Large Language Models — 书架上 7 本书引过它
  • Language Models are Few-Shot Learners — 书架上 6 本书引过它
  • Large Language Models are Zero-Shot Reasoners — 书架上 4 本书引过它
  • QLoRA: Efficient Finetuning of Quantized LLMs — 书架上 4 本书引过它
  • An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale — 书架上 3 本书引过它
  • From Local to Global: A Graph RAG Approach to Query-Focused Summarization — 书架上 3 本书引过它
  • Self-Consistency Improves Chain of Thought Reasoning in Language Models — 书架上 3 本书引过它
  • BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding — 书架上 2 本书引过它
  • DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models — 书架上 2 本书引过它
  • Dense X Retrieval: What Retrieval Granularity Should We Use? — 书架上 2 本书引过它
  • Learning Transferable Visual Models From Natural Language Supervision — 书架上 2 本书引过它
  • Parameter-Efficient Transfer Learning for NLP — 书架上 2 本书引过它
  • Scaling Laws for Neural Language Models — 书架上 2 本书引过它
  • ALBERT: A Lite BERT for Self-supervised Learning of Language Representations — 书架上 1 本书引过它
  • BERTopic: Neural topic modeling with a class-based TF-IDF procedure — 书架上 1 本书引过它
  • BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models — 书架上 1 本书引过它
  • Conditional Prompt Learning for Vision-Language Models — 书架上 1 本书引过它
  • Constitutional AI: Harmlessness from AI Feedback — 书架上 1 本书引过它
  • DeepSeek-V3 Technical Report — 书架上 1 本书引过它
  • Docling Technical Report — 书架上 1 本书引过它
  • Efficient Few-Shot Learning Without Prompts — 书架上 1 本书引过它
  • Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer — 书架上 1 本书引过它
  • FineScope : SAE-guided Data Selection Enables Domain Specific LLM Pruning and Finetuning — 书架上 1 本书引过它
  • Generated Knowledge Prompting for Commonsense Reasoning — 书架上 1 本书引过它
  • GLM: General Language Model Pretraining with Autoregressive Blank Infilling — 书架上 1 本书引过它
  • Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation — 书架上 1 本书引过它
  • GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints — 书架上 1 本书引过它
  • HellaSwag: Can a Machine Really Finish Your Sentence? — 书架上 1 本书引过它
  • Intrinsic Dimensionality Explains the Effectiveness of Language Model Fine-Tuning — 书架上 1 本书引过它
  • Learning to Prompt for Vision-Language Models — 书架上 1 本书引过它
  • Llama 2: Open Foundation and Fine-Tuned Chat Models — 书架上 1 本书引过它
  • Measuring Mathematical Problem Solving With the MATH Dataset — 书架上 1 本书引过它
  • Neural Machine Translation of Rare Words with Subword Units — 书架上 1 本书引过它
  • Prefix-Tuning: Optimizing Continuous Prompts for Generation — 书架上 1 本书引过它
  • Qwen2.5 Technical Report — 书架上 1 本书引过它
  • Qwen3 Technical Report — 书架上 1 本书引过它
  • RLPrompt: Optimizing Discrete Text Prompts with Reinforcement Learning — 书架上 1 本书引过它
  • RoBERTa: A Robustly Optimized BERT Pretraining Approach — 书架上 1 本书引过它
  • Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks — 书架上 1 本书引过它
  • SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing — 书架上 1 本书引过它
  • Show Your Work: Scratchpads for Intermediate Computation with Language Models — 书架上 1 本书引过它
  • Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates — 书架上 1 本书引过它
  • Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge — 书架上 1 本书引过它
  • Understanding R1-Zero-Like Training: A Critical Perspective — 书架上 1 本书引过它

prompting —— 提示工程、思维链、自洽、上下文学习(30 篇)​

  • Chain-of-Thought Prompting Elicits Reasoning in Large Language Models — 书架上 6 本书引过它
  • Language Models are Few-Shot Learners — 书架上 6 本书引过它
  • ReAct: Synergizing Reasoning and Acting in Language Models — 书架上 6 本书引过它
  • Training language models to follow instructions with human feedback — 书架上 6 本书引过它
  • DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning — 书架上 5 本书引过它
  • Large Language Models are Zero-Shot Reasoners — 书架上 4 本书引过它
  • Self-Consistency Improves Chain of Thought Reasoning in Language Models — 书架上 3 本书引过它
  • DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models — 书架上 2 本书引过它
  • Take a Step Back: Evoking Reasoning via Abstraction in Large Language Models — 书架上 2 本书引过它
  • Tree of Thoughts: Deliberate Problem Solving with Large Language Models — 书架上 2 本书引过它
  • Automatic Chain of Thought Prompting in Large Language Models — 书架上 1 本书引过它
  • Conditional Prompt Learning for Vision-Language Models — 书架上 1 本书引过它
  • DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines — 书架上 1 本书引过它
  • EvoPrompt: Connecting LLMs with Evolutionary Algorithms Yields Powerful Prompt Optimizers — 书架上 1 本书引过它
  • Generated Knowledge Prompting for Commonsense Reasoning — 书架上 1 本书引过它
  • Interaction Scaling: Grounding the Third Axis of Test-Time Compute — 书架上 1 本书引过它
  • Large Language Models Are Human-Level Prompt Engineers — 书架上 1 本书引过它
  • Large Language Models as Optimizers — 书架上 1 本书引过它
  • Learning to Prompt for Vision-Language Models — 书架上 1 本书引过它
  • MCP-Zero: Active Tool Discovery for Autonomous LLM Agents — 书架上 1 本书引过它
  • MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework — 书架上 1 本书引过它
  • Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection — 书架上 1 本书引过它
  • Prefix-Tuning: Optimizing Continuous Prompts for Generation — 书架上 1 本书引过它
  • Rethinking the Role of Demonstrations: What Makes In-Context Learning Work? — 书架上 1 本书引过它
  • RLPrompt: Optimizing Discrete Text Prompts with Reinforcement Learning — 书架上 1 本书引过它
  • Segment Anything — 书架上 1 本书引过它
  • Show Your Work: Scratchpads for Intermediate Computation with Language Models — 书架上 1 本书引过它
  • TabLLM: Few-shot Classification of Tabular Data with Large Language Models — 书架上 1 本书引过它
  • Understanding R1-Zero-Like Training: A Critical Perspective — 书架上 1 本书引过它
  • When "A Helpful Assistant" Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models — 书架上 1 本书引过它

eval —— 基准、评测方法、评判模型、可观测性(27 篇)​

  • Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks — 书架上 9 本书引过它
  • QLoRA: Efficient Finetuning of Quantized LLMs — 书架上 4 本书引过它
  • Are Emergent Abilities of Large Language Models a Mirage? — 书架上 3 本书引过它
  • Measuring Massive Multitask Language Understanding — 书架上 3 本书引过它
  • Deep Residual Learning for Image Recognition — 书架上 2 本书引过它
  • DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models — 书架上 2 本书引过它
  • MTEB: Massive Text Embedding Benchmark — 书架上 2 本书引过它
  • Ragas: Automated Evaluation of Retrieval Augmented Generation — 书架上 2 本书引过它
  • Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference — 书架上 1 本书引过它
  • Evaluating Very Long-Term Conversational Memory of LLM Agents — 书架上 1 本书引过它
  • Gaussian Error Linear Units (GELUs) — 书架上 1 本书引过它
  • Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context — 书架上 1 本书引过它
  • GPQA: A Graduate-Level Google-Proof Q&A Benchmark — 书架上 1 本书引过它
  • How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms — 书架上 1 本书引过它
  • Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena — 书架上 1 本书引过它
  • Large Language Models as Optimizers — 书架上 1 本书引过它
  • Llama 2: Open Foundation and Fine-Tuned Chat Models — 书架上 1 本书引过它
  • LLaMA: Open and Efficient Foundation Language Models — 书架上 1 本书引过它
  • On the Opportunities and Risks of Foundation Models — 书架上 1 本书引过它
  • Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone — 书架上 1 本书引过它
  • Robust Speech Recognition via Large-Scale Weak Supervision — 书架上 1 本书引过它
  • Segment Anything — 书架上 1 本书引过它
  • TabLLM: Few-shot Classification of Tabular Data with Large Language Models — 书架上 1 本书引过它
  • The Leaderboard Illusion — 书架上 1 本书引过它
  • Training Verifiers to Solve Math Word Problems — 书架上 1 本书引过它
  • What Matters in Transformers? Not All Attention is Needed — 书架上 1 本书引过它
  • When "A Helpful Assistant" Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Models — 书架上 1 本书引过它

transformer —— 注意力、位置编码、架构本体(26 篇)​

  • Attention Is All You Need — 书架上 10 本书引过它
  • LoRA: Low-Rank Adaptation of Large Language Models — 书架上 7 本书引过它
  • Training Compute-Optimal Large Language Models — 书架上 7 本书引过它
  • An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale — 书架上 3 本书引过它
  • RoFormer: Enhanced Transformer with Rotary Position Embedding — 书架上 3 本书引过它
  • BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding — 书架上 2 本书引过它
  • FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness — 书架上 2 本书引过它
  • BERTopic: Neural topic modeling with a class-based TF-IDF procedure — 书架上 1 本书引过它
  • BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models — 书架上 1 本书引过它
  • Decision Transformer: Reinforcement Learning via Sequence Modeling — 书架上 1 本书引过它
  • DeepSeek-V3 Technical Report — 书架上 1 本书引过它
  • DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models — 书架上 1 本书引过它
  • Efficient Few-Shot Learning Without Prompts — 书架上 1 本书引过它
  • Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer — 书架上 1 本书引过它
  • Fast Transformer Decoding: One Write-Head is All You Need — 书架上 1 本书引过它
  • GLM: General Language Model Pretraining with Autoregressive Blank Infilling — 书架上 1 本书引过它
  • GLU Variants Improve Transformer — 书架上 1 本书引过它
  • GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints — 书架上 1 本书引过它
  • Mamba: Linear-Time Sequence Modeling with Selective State Spaces — 书架上 1 本书引过它
  • MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention — 书架上 1 本书引过它
  • On Layer Normalization in the Transformer Architecture — 书架上 1 本书引过它
  • Show Your Work: Scratchpads for Intermediate Computation with Language Models — 书架上 1 本书引过它
  • Textbooks Are All You Need — 书架上 1 本书引过它
  • Training Verifiers to Solve Math Word Problems — 书架上 1 本书引过它
  • Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks — 书架上 1 本书引过它
  • What Matters in Transformers? Not All Attention is Needed — 书架上 1 本书引过它

theory —— 表示学习、涌现、可解释性、理论分析(26 篇)​

  • Are Emergent Abilities of Large Language Models a Mirage? — 书架上 3 本书引过它
  • Efficient Estimation of Word Representations in Vector Space — 书架上 3 本书引过它
  • Auto-Encoding Variational Bayes — 书架上 2 本书引过它
  • BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding — 书架上 2 本书引过它
  • Learning Transferable Visual Models From Natural Language Supervision — 书架上 2 本书引过它
  • Tree of Thoughts: Deliberate Problem Solving with Large Language Models — 书架上 2 本书引过它
  • ALBERT: A Lite BERT for Self-supervised Learning of Language Representations — 书架上 1 本书引过它
  • Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift — 书架上 1 本书引过它
  • BERTopic: Neural topic modeling with a class-based TF-IDF procedure — 书架上 1 本书引过它
  • Denoising Diffusion Probabilistic Models — 书架上 1 本书引过它
  • Emergent Abilities of Large Language Models — 书架上 1 本书引过它
  • GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints — 书架上 1 本书引过它
  • Intrinsic Dimensionality Explains the Effectiveness of Language Model Fine-Tuning — 书架上 1 本书引过它
  • Late Chunking: Contextual Chunk Embeddings Using Long-Context Embedding Models — 书架上 1 本书引过它
  • Layer Normalization — 书架上 1 本书引过它
  • Making Large Language Models Efficient Dense Retrievers — 书架上 1 本书引过它
  • NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction — 书架上 1 本书引过它
  • On Layer Normalization in the Transformer Architecture — 书架上 1 本书引过它
  • Physics-based Deep Learning — 书架上 1 本书引过它
  • Reinforcement Learning for Generative AI: State of the Art, Opportunities and Open Research Challenges — 书架上 1 本书引过它
  • RLPrompt: Optimizing Discrete Text Prompts with Reinforcement Learning — 书架上 1 本书引过它
  • Robust Speech Recognition via Large-Scale Weak Supervision — 书架上 1 本书引过它
  • Root Mean Square Layer Normalization — 书架上 1 本书引过它
  • SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training — 书架上 1 本书引过它
  • Textbooks Are All You Need — 书架上 1 本书引过它
  • Will we run out of data? Limits of LLM scaling based on human-generated data — 书架上 1 本书引过它

post-training —— 指令微调、RLHF、DPO 一族、推理模型(19 篇)​

  • Language Models are Few-Shot Learners — 书架上 6 本书引过它
  • Training language models to follow instructions with human feedback — 书架上 6 本书引过它
  • DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning — 书架上 5 本书引过它
  • Direct Preference Optimization: Your Language Model is Secretly a Reward Model — 书架上 3 本书引过它
  • Precise Zero-Shot Dense Retrieval without Relevance Labels — 书架上 3 本书引过它
  • Proximal Policy Optimization Algorithms — 书架上 2 本书引过它
  • Agent-Computer Observation Interfaces Enable Dynamic Computer Use — 书架上 1 本书引过它
  • Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference — 书架上 1 本书引过它
  • DAPO: An Open-Source LLM Reinforcement Learning System at Scale — 书架上 1 本书引过它
  • GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization — 书架上 1 本书引过它
  • Group Sequence Policy Optimization — 书架上 1 本书引过它
  • Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena — 书架上 1 本书引过它
  • Large Language Models Are Human-Level Prompt Engineers — 书架上 1 本书引过它
  • Let's Verify Step by Step — 书架上 1 本书引过它
  • MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention — 书架上 1 本书引过它
  • Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection — 书架上 1 本书引过它
  • ORPO: Monolithic Preference Optimization without Reference Model — 书架上 1 本书引过它
  • Rethinking Inference-Time Scaling: Efficiency Limits and Linguistic Signals — 书架上 1 本书引过它
  • Understanding R1-Zero-Like Training: A Critical Perspective — 书架上 1 本书引过它

multimodal —— 视觉语言、跨模态、图文对齐(17 篇)​

  • An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale — 书架上 3 本书引过它
  • Deep Residual Learning for Image Recognition — 书架上 2 本书引过它
  • Learning Transferable Visual Models From Natural Language Supervision — 书架上 2 本书引过它
  • BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models — 书架上 1 本书引过它
  • Conditional Prompt Learning for Vision-Language Models — 书架上 1 本书引过它
  • DAPO: An Open-Source LLM Reinforcement Learning System at Scale — 书架上 1 本书引过它
  • Denoising Diffusion Probabilistic Models — 书架上 1 本书引过它
  • Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context — 书架上 1 本书引过它
  • Learning to Prompt for Vision-Language Models — 书架上 1 本书引过它
  • Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model — 书架上 1 本书引过它
  • NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction — 书架上 1 本书引过它
  • On the Importance of Noise Scheduling for Diffusion Models — 书架上 1 本书引过它
  • Segment Anything — 书架上 1 本书引过它
  • Segment Anything Model for Medical Images? — 书架上 1 本书引过它
  • SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training — 书架上 1 本书引过它
  • The Curse of Recursion: Training on Generated Data Makes Models Forget — 书架上 1 本书引过它
  • Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks — 书架上 1 本书引过它

rag —— 检索增强、向量检索、重排、GraphRAG(13 篇)​

  • Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks — 书架上 9 本书引过它
  • Lost in the Middle: How Language Models Use Long Contexts — 书架上 4 本书引过它
  • From Local to Global: A Graph RAG Approach to Query-Focused Summarization — 书架上 3 本书引过它
  • Precise Zero-Shot Dense Retrieval without Relevance Labels — 书架上 3 本书引过它
  • Dense X Retrieval: What Retrieval Granularity Should We Use? — 书架上 2 本书引过它
  • MTEB: Massive Text Embedding Benchmark — 书架上 2 本书引过它
  • Ragas: Automated Evaluation of Retrieval Augmented Generation — 书架上 2 本书引过它
  • Evaluating Very Long-Term Conversational Memory of LLM Agents — 书架上 1 本书引过它
  • Late Chunking: Contextual Chunk Embeddings Using Long-Context Embedding Models — 书架上 1 本书引过它
  • Making Large Language Models Efficient Dense Retrievers — 书架上 1 本书引过它
  • MCP-Zero: Active Tool Discovery for Autonomous LLM Agents — 书架上 1 本书引过它
  • Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge — 书架上 1 本书引过它
  • User as Code: Executable Memory for Personalized Agents — 书架上 1 本书引过它

agents —— 工具调用、规划、多智能体、记忆(12 篇)​

  • ReAct: Synergizing Reasoning and Acting in Language Models — 书架上 6 本书引过它
  • Agent-Computer Observation Interfaces Enable Dynamic Computer Use — 书架上 1 本书引过它
  • DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models — 书架上 1 本书引过它
  • DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning — 书架上 1 本书引过它
  • Evaluating Very Long-Term Conversational Memory of LLM Agents — 书架上 1 本书引过它
  • Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model — 书架上 1 本书引过它
  • MCP-Zero: Active Tool Discovery for Autonomous LLM Agents — 书架上 1 本书引过它
  • MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework — 书架上 1 本书引过它
  • RLVP: Penalize the Path, Reward the Outcome — 书架上 1 本书引过它
  • The Latent Bridge: A Continuous Slow-Fast Channel for Real-Time Game Agents — 书架上 1 本书引过它
  • User as Code: Executable Memory for Personalized Agents — 书架上 1 本书引过它
  • Whose Side Is Your Agent On? Multi-Party Principal Loyalty in LLM Agents — 书架上 1 本书引过它

systems —— 训练系统、并行、分布式、硬件(11 篇)​

  • Attention Is All You Need — 书架上 10 本书引过它
  • Large Language Models are Zero-Shot Reasoners — 书架上 4 本书引过它
  • Precise Zero-Shot Dense Retrieval without Relevance Labels — 书架上 3 本书引过它
  • Einstein's Patents and Inventions — 书架上 1 本书引过它
  • Fast Transformer Decoding: One Write-Head is All You Need — 书架上 1 本书引过它
  • Large Language Models Cannot Self-Correct Reasoning Yet — 书架上 1 本书引过它
  • Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model — 书架上 1 本书引过它
  • On a new statistical technique for the real-time recognition of ultra-low multiplicity astrophysical neutrino burst — 书架上 1 本书引过它
  • Robust Speech Recognition via Large-Scale Weak Supervision — 书架上 1 本书引过它
  • TabLLM: Few-shot Classification of Tabular Data with Large Language Models — 书架上 1 本书引过它
  • ZeRO: Memory Optimizations Toward Training Trillion Parameter Models — 书架上 1 本书引过它

efficiency —— 量化、蒸馏、剪枝、推理加速、长上下文(9 篇)​

  • Lost in the Middle: How Language Models Use Long Contexts — 书架上 4 本书引过它
  • FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness — 书架上 2 本书引过它
  • DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models — 书架上 1 本书引过它
  • Distilling the Knowledge in a Neural Network — 书架上 1 本书引过它
  • FineScope : SAE-guided Data Selection Enables Domain Specific LLM Pruning and Finetuning — 书架上 1 本书引过它
  • Late Chunking: Contextual Chunk Embeddings Using Long-Context Embedding Models — 书架上 1 本书引过它
  • MiniLLM: On-Policy Distillation of Large Language Models — 书架上 1 本书引过它
  • Models Take Notes at Prefill: KV Cache Can Be Editable and Composable — 书架上 1 本书引过它
  • The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks — 书架上 1 本书引过它

safety —— 对齐、越狱、隐私、幻觉与事实性(8 篇)​

  • Training language models to follow instructions with human feedback — 书架上 6 本书引过它
  • Ragas: Automated Evaluation of Retrieval Augmented Generation — 书架上 2 本书引过它
  • How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms — 书架上 1 本书引过它
  • Llama 2: Open Foundation and Fine-Tuned Chat Models — 书架上 1 本书引过它
  • MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework — 书架上 1 本书引过它
  • Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection — 书架上 1 本书引过它
  • Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone — 书架上 1 本书引过它
  • Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks — 书架上 1 本书引过它

peft —— 参数高效微调:LoRA、QLoRA、适配器(7 篇)​

  • LoRA: Low-Rank Adaptation of Large Language Models — 书架上 7 本书引过它
  • QLoRA: Efficient Finetuning of Quantized LLMs — 书架上 4 本书引过它
  • Parameter-Efficient Transfer Learning for NLP — 书架上 2 本书引过它
  • Tree of Thoughts: Deliberate Problem Solving with Large Language Models — 书架上 2 本书引过它
  • Efficient Few-Shot Learning Without Prompts — 书架上 1 本书引过它
  • First return, then explore — 书架上 1 本书引过它
  • Prefix-Tuning: Optimizing Continuous Prompts for Generation — 书架上 1 本书引过它