{"total":15,"items":[{"citing_arxiv_id":"2606.31432","ref_index":14,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Clinically Structured Rank-Gated LoRA for Cross-Benchmark Medical Question Answering","primary_cat":"cs.CL","submitted_at":"2026-06-30T09:59:05+00:00","verdict":null,"verdict_confidence":null,"novelty_score":null,"formal_verification":null,"one_line_summary":null,"context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.31397","ref_index":54,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Mixture-of-Control: State-Aware Fine-Tuning for Transformer-based Models","primary_cat":"cs.LG","submitted_at":"2026-06-30T09:25:37+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Mixture-of-Control adaptively combines local and global control states in transformer fine-tuning by treating per-block states as experts in a sparse MoE setup to improve cross-block communication while keeping memory and compute costs comparable to prior state-based methods.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.29375","ref_index":22,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"TriageRA-CCF: Source-Side Clinical Confidence and Coverage Signals for Adaptive Rank Budgeting in Medical LLMs","primary_cat":"cs.CL","submitted_at":"2026-06-28T12:52:07+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"TriageRA-CCF combines source-side confidence, coverage, and counterfactual signals to supervise an adaptive LoRA rank router, reporting modest average accuracy gains over LoRA/DoRA/MoELoRA baselines on two 8B models under matched training.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.03979","ref_index":99,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories","primary_cat":"cs.LG","submitted_at":"2026-06-02T17:56:55+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Language models can use a two-stage sleep process of upward distillation for memory consolidation and RL-based dreaming for unsupervised self-improvement to enable continual learning.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.00382","ref_index":43,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"CRMA: A Spectrally-Bounded Backbone for Modular Continual Fine-Tuning of LLMs","primary_cat":"cs.LG","submitted_at":"2026-05-29T21:50:23+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"CRMA adds a spectrally bounded residual adapter backbone to modular continual fine-tuning of LLMs, achieving near-zero loss drift and positive backward transfer on Mistral-7B across domains.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.29569","ref_index":8,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"LoRA-Key: User-Centric LoRA Watermarking for Text-to-Image Diffusion Models","primary_cat":"cs.CR","submitted_at":"2026-05-28T08:17:28+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"LoRA-Key creates a standalone user-specific Watermark LoRA trained with a latent watermark prior and GOP, attachable via training-free superposition to protect LoRA ownership while preserving quality.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.07562","ref_index":13,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Beyond GSD-as-Token: Continuous Scale Conditioning for Remote Sensing VLMs","primary_cat":"cs.CV","submitted_at":"2026-05-08T10:35:11+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"ScaleEarth conditions remote sensing VLMs on continuous GSD via CS-HLoRA and a visual GSD predictor, creating a closed training loop with GeoScale-VQA to achieve SOTA on Earth observation benchmarks.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.23750","ref_index":35,"ref_count":2,"confidence":0.9,"is_internal_anchor":false,"paper_title":"The Override Gap: A Magnitude Account of Knowledge Conflict Failure in Hypernetwork-Based Instant LLM Adaptation","primary_cat":"cs.LG","submitted_at":"2026-04-26T14:59:14+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Knowledge conflicts in hypernetwork LLM adaptation stem from constant adapter margins losing to frequency-dependent pretrained margins; selective layer boosting and conflict-aware triggering raise deep-conflict accuracy to 71-72.5% on Gemma-2B and Mistral-7B.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"uments are in context, and we grade the difficulty of conflicts by the pretraining frequency of the contradicted fact. Modulated and Gated LoRA.Feature-wise Linear Modulation (FiLM) [29] appliesy=γ(z)x+ β(z)with(γ, β)from a learned generator, generalising conditional instance normalisation [8]. A recent line of LoRA work adopts this style of learnable modulation: MoLE [35] and X-LoRA [2] learn per-layer gates over multiple LoRA experts, AdaLoRA [38] reallocates rank by learned layer importance, DoRA [23] decomposes each update into separate magnitude and direction terms, Hy- perLoRA [21] generates LoRA weights from a hypernetwork for image generation, and LoRAHub [15] composes pretrained LoRAs gradient-free. SLB and CA are a deliberately minimal, training-"},{"citing_arxiv_id":"2604.21905","ref_index":209,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Low-Rank Adaptation Redux for Large Models","primary_cat":"cs.LG","submitted_at":"2026-04-23T17:50:23+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":3.0,"formal_verification":"none","one_line_summary":"An overview revisits LoRA variants by categorizing advances in architectural design, efficient optimization, and applications while linking them to classical signal processing tools for principled fine-tuning.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"American football depending on whether the desired context is European or American. These conflict-resolution strategies draw clear parallels to the IEEE TRANSACTIONS ON SIGNAL PROCESSING (SUBMITTED) 20 established literature on ensemble learning [ 170], [121], [122], which also has a long-standing history in SP, as well as to multi- task learning [164]. For example, MoLE [209] adopts a mixture- of-experts (MoE) framework, employing a gating network to dynamically predict the mixture coefficients {wn l }. Alterna- tively, [54] determines {wn l } by minimizing the discrepancy be- tween outputs of the merged model and the individual adapters on a calibration set. Concretely, let ∆Wn l =X n l (Yn l )⊤, and define the merged update ∆Wl(wl) :=PN"},{"citing_arxiv_id":"2604.05183","ref_index":33,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"OrthoFuse: Training-free Riemannian Fusion of Orthogonal Style-Concept Adapters for Diffusion Models","primary_cat":"cs.CV","submitted_at":"2026-04-06T21:24:21+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Training-free Riemannian fusion merges orthogonal style and concept adapters for diffusion models via geodesic approximation on GS matrices plus spectra restoration.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2510.07239","ref_index":32,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts","primary_cat":"cs.CL","submitted_at":"2025-10-08T17:06:20+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Red-Bandit adapts online to LLM failure modes by dynamically selecting among RL-trained LoRA attack-style experts via a bandit policy, reporting SOTA ASR@10 on AdvBench with lower-perplexity prompts.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2509.06984","ref_index":39,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"FediLoRA: Practical Federated Fine-Tuning of Foundation Models Under Missing-Modality Constraints","primary_cat":"cs.LG","submitted_at":"2025-09-01T10:40:13+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"FediLoRA is a lightweight federated LoRA aggregation method that jointly mitigates missing modalities and heterogeneous ranks in collaborative fine-tuning of foundation models.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2507.00029","ref_index":14,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"LoRA-Mixer: Coordinate Modular LoRA Experts Through Serial Attention Routing","primary_cat":"cs.LG","submitted_at":"2025-06-17T14:58:54+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"LoRA-Mixer routes modular LoRA experts into attention projection matrices with an adaptive Routing Specialization Loss to improve multi-task performance while using fewer trainable parameters than prior LoRA-MoE methods.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2410.06431","ref_index":8,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Functional-level Uncertainty Quantification for Calibrated Fine-tuning on LLMs","primary_cat":"cs.LG","submitted_at":"2024-10-09T00:09:15+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"UQ4CT integrates functional-level uncertainty calibration into mixture-of-experts LoRA fine-tuning via a dedicated loss, cutting expected calibration error by over 25% on multiple-choice and generative QA tasks.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2403.14608","ref_index":95,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey","primary_cat":"cs.LG","submitted_at":"2024-03-21T17:55:50+00:00","verdict":"ACCEPT","verdict_confidence":"MODERATE","novelty_score":4.0,"formal_verification":"none","one_line_summary":"A comprehensive survey of PEFT algorithms for large models, covering their performance, overhead, applications, and real-world system implementations.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"Decomposition Intrinsic SAID [75], LoRA [76], Compacter [77], KronA [78], KAdaptation [79], HiWi [65], VeRA [80], DoRA [81] LoRA Derivatives Dynamic Rank DyLoRA [82], AdaLoRA [83], SoRA [84], CapaBoost [85], AutoLoRA [86] LoRA Improvement Laplace-LoRA [87], LoRA Dropout [88], PeriodicLoRA [89], LoRA+ [90], MoSLoRA [91] Multiple LoRA LoRAHub [92], MOELoRA [93], MoLORA [60], MoA [94], MoLE [95], MixLoRA [96] Hybrid Fine-tuning UniPELT [97], S4 [98], MAM Adapter [32], NOAH [99], AUTOPEFT [100], LLM-Adapters [101], S3PET [102] Fig. 3: Taxonomy of Parameter-Efficient Fine-Tuning Methods for Large Models. Output Combine Frozen Learnable Input Output Input (c) Reparameterization PEFT(a) Additive PEFT (b) Selective PEFT Merge Output Input Input (train)"}],"limit":50,"offset":0}