低秩适配论文趋势 2023–2026

主要结论

• 低秩适配论文从 2023 年的 282 篇增长到 2026 年的 841 篇，三年间扩大到约 3.0 倍。
• NeurIPS 是发低秩适配论文最多的会议，2023–2026 年共 781 篇，占四个会议总量的 34%。
• 交叉方向里涨得最快的是扩散模型：同时属于两个主题的论文从 2023 年的 4 篇增至 2026 年的 61 篇。
• 在「优化与缩放」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的缩放定律是 4.1 倍，低秩适配是 3.0 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026$\delta$-Mem: : Efficient Online Memory For Large Language Models
• ICML 2026A3: an Analytical Low-Rank Approximation Framework for Attention
• ICLR 2026A Guardrail for Safety Preservation: When Safety-Sensitive Subspace Meets Harmful-Resistant Null-Space
• CVPR 20262-Shots in the Dark Low-Light Denoising with Minimal Data Acquisition
• NeurIPS 2025A geometric framework for momentum-based optimizers for low-rank training
• ICML 2025Beyond Low-rank Decomposition: A Shortcut Approach for Efficient On-Device Learning
• ICLR 2025A Quantum Circuit-Based Compression Perspective for Parameter-Efficient Learning
• CVPR 2025AdaDARE-gamma: Balancing Stability and Plasticity in Multi-modal LLMs through Efficient Adaptation
• NeurIPS 2024Bridging The Gap between Low-rank and Orthogonal Adaptation via Householder Reflection Adaptation
• ICML 2024A Tensor Decomposition Perspective on Second-order RNNs
• ICLR 2024A Hard-to-Beat Baseline for Training-free CLIP-based Adaptation
• CVPR 2024DePT: Decoupled Prompt Tuning
• NeurIPS 2023A Theoretical Analysis of the Test Error of Finite-Rank Kernel Ridge Regression
• ICML 2023A Three-regime Model of Network Pruning
• ICLR 2023A General Rank Preserving Framework for Asymmetric Image Retrieval
• CVPR 20231% VS 100%: Parameter-Efficient Low Rank Adapter for Dense Predictions
• NeurIPS 2026A low-rank decoder bottleneck bounds reliability in foundation-model perturbation prediction
• ICML 2026Adaptive Utilization of Low-Rank Adaptation via Conditioned Gating
• ICLR 2026ABBA-Adapters: Efficient and Expressive Fine-Tuning of Foundation Models
• CVPR 2026Basis-Oriented Low-rank Transfer for Few-Shot and Test-Time Adaptation

全部低秩适配论文

共 781 篇，第 6 / 16 页

251. NeurIPS 2026TriAxialKV: Toward Extreme Low-Precision KV-Cache Quantization for Agentic Inference Tasks
252. NeurIPS 2026Tuning the Tuner
253. NeurIPS 2026U-MOF: Uncertainty-Guided Parameter-Efficient Multi-Objective Fine-Tuning for Long-Tailed Recognition
254. NeurIPS 2026Uncovering Hidden Propensities in Language Models via Limited-Parameter Finetuning
255. NeurIPS 2026Uniform Spectral Growth under Factor-wise Muon Orthogonalization in Matrix Factorization and LoRA
256. NeurIPS 2026Unveiling Implicit Advantage Symmetry: Why GRPO Struggles with Exploration and Difficulty Adaptation
257. NeurIPS 2026Valid Best-Model Identification for LLM Evaluation via Low-Rank Factorization
258. NeurIPS 2026VALOR: Vector-Aware Low-Rank Restructuring of Neural Networks for RISC-V Inference
259. NeurIPS 2026Values as Style: Disentangling Values from Semantics with One-Way Mixing for Low-Damage LLM Steering
260. NeurIPS 2026Virtual Head Attention
261. NeurIPS 2026What Drives Test-Time Adaptation for CLIP? A Controlled Empirical Study from an Update Perspective
262. NeurIPS 2026When Are Semivalue-Based Decisions Identifiable? Robust Data Selection under Utility Ambiguity
263. NeurIPS 2026When Does Fine-Tuning Extract Stored Knowledge? A Relation-Covering Theory for One-Layer Transformers
264. NeurIPS 2026When Does Subspace Direction Matter for LoRA? Regime Analysis of the Magnitude Principle in Few-Shot Adaptation
265. NeurIPS 2026When the Merge Coefficient Stops Mattering: Proximity Regularized Merging for Continual LoRA Adaptation
266. NeurIPS 2026When the Same Coefficients Reach Different Places: Asymmetric Realizability in Transplanting Tokenizers across Large Language Models
267. NeurIPS 2026Where Do We (Not) Need Temporal Context in Low-Resource Video Task Adaptation?
268. NeurIPS 2026Where You Backpropagate Matters: Token Hypothesis for Memory-Efficient Fine-Tuning
269. NeurIPS 2026You CAN Teach an Old Model New Tricks: Domain Adaptation via Complementary Subspace Expansion
270. NeurIPS 20253BASiL: An Algorithmic Framework for Sparse plus Low-Rank Compression of LLMs
271. NeurIPS 20253DPE-Gaze:Unlocking the Potential of 3D Facial Priors for Generalized Gaze Estimation
272. NeurIPS 2025A geometric framework for momentum-based optimizers for low-rank training
273. NeurIPS 2025A Novel General Framework for Sharp Lower Bounds in Succinct Stochastic Bandits
274. NeurIPS 2025A Token is Worth over 1,000 Tokens: Efficient Knowledge Distillation through Low-Rank Clone
275. NeurIPS 2025A*-Thought: Efficient Reasoning via Bidirectional Compression for Low-Resource Settings
276. NeurIPS 2025Accurate and Efficient Low-Rank Model Merging in Core Space
277. NeurIPS 2025Activated LoRA: Fine-tuned LLMs for Intrinsics
278. NeurIPS 2025Activity Pruning for Efficient Spiking Neural Networks
279. NeurIPS 2025AdaMSS: Adaptive Multi-Subspace Approach for Parameter-Efficient Fine-Tuning
280. NeurIPS 2025Adaptive Classifier-Free Guidance via Dynamic Low-Confidence Masking
281. NeurIPS 2025Adaptive Fission: Post-training Encoding for Low-latency Spike Neural Networks
282. NeurIPS 2025Adaptive LoRA Experts Allocation and Selection for Federated Fine-Tuning
283. NeurIPS 2025Advancing Compositional Awareness in CLIP with Efficient Fine-Tuning
284. NeurIPS 2025AegisGuard: RL-Guided Adapter Tuning for TEE-Based Efficient & Secure On-Device Inference
285. NeurIPS 2025AI Progress Should Be Measured by Capability-Per-Resource, Not Scale Alone: A Framework for Gradient-Guided Resource Allocation in LLMs
286. NeurIPS 2025All You Need is One: Capsule Prompt Tuning with a Single Vector
287. NeurIPS 2025AltLoRA: Towards Better Gradient Approximation in Low-Rank Adaptation with Alternating Projections
288. NeurIPS 2025Angles Don’t Lie: Unlocking Training‑Efficient RL Through the Model’s Own Signals
289. NeurIPS 2025AuroRA: Breaking Low-Rank Bottleneck of LoRA with Nonlinear Mapping
290. NeurIPS 2025Backpropagation-Free Test-Time Adaptation via Probabilistic Gaussian Alignment
291. NeurIPS 2025Bayes optimal learning of attention-indexed models
292. NeurIPS 2025Beyond Higher Rank: Token-wise Input-Output Projections for Efficient Low-Rank Adaptation
293. NeurIPS 2025Bilevel ZOFO: Efficient LLM Fine-Tuning and Meta-Training
294. NeurIPS 2025Block-Diagonal LoRA for Eliminating Communication Overhead in Tensor Parallel LoRA Serving
295. NeurIPS 2025Breaking the Frozen Subspace: Importance Sampling for Low-Rank Optimization in LLM Pretraining
296. NeurIPS 2025C-LoRA: Contextual Low-Rank Adaptation for Uncertainty Estimation in Large Language Models
297. NeurIPS 2025Cancer Survival Analysis via Zero-shot Tumor Microenvironment Segmentation on Low-resolution Whole Slide Pathology Images
298. NeurIPS 2025COALA: Numerically Stable and Efficient Framework for Context-Aware Low-Rank Approximation
299. NeurIPS 2025CodeMerge: Codebook-Guided Model Merging for Robust Test-Time Adaptation in Autonomous Driving
300. NeurIPS 2025CoIDO: Efficient Data Selection for Visual Instruction Tuning via Coupled Importance-Diversity Optimization

常见问题

2023–2026 年四大会议一共有多少篇低秩适配论文？

共 2,273 篇，按年份分别是 2023 年 282 篇、2024 年 518 篇、2025 年 632 篇、2026 年 841 篇。

NeurIPS 2026 有多少篇低秩适配论文？

NeurIPS 2026 有 269 篇低秩适配论文，上一届是 241 篇。

ICLR 2026 有多少篇低秩适配论文？

ICLR 2026 有 184 篇低秩适配论文，上一届是 152 篇。

ICML 2026 有多少篇低秩适配论文？

ICML 2026 有 240 篇低秩适配论文，上一届是 136 篇。

CVPR 2026 有多少篇低秩适配论文？

CVPR 2026 有 148 篇低秩适配论文，上一届是 103 篇。

哪个会议的低秩适配论文最多？

NeurIPS，2023–2026 年共 781 篇。