Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 56 for “"Low-Rank Adaptation"”.
-
Chatbot založený na hlubokých neuronových sítích
… dolaďování parametrů, jako jsou LoRA (Low-Rank Adaptation) a QLoRA (Quantized Low-Rank Adaptation), si kladou za cíl tyto nároky snížit při zachování konkurenceschopného výkonu. Tato práce experimentálně porovnává tři přístupy k adaptaci velkých jazykových modelů: plné dolaďování, LoRA …
-
Customizing large language models : from synthetic data generation to efficient deployment
… these customized models with \flora/ (Fast Low-Rank Adaptation). \flora/ enables efficient, real-time serving of task-specific model variants through batched low-rank adaptation, making it computationally feasible to maintain and deploy multiple specialized models while ensuring that the …
-
Protecting wild Atlantic salmon : using tiny data to specialize visual transformer foundation models for image classification
… classification. The performance of supervised adaptation of a ViT foundation model to a tiny, task-specific dataset is evaluated and compared between model size and parameters. The effects of Low Rank Adaptation of a ViT foundation model when adapting to a tiny, task-specific dataset are …
-
Segmentace symbolů ve skenech pneumatik s použitím self-supervised learning
… v kombinácii s technikami zmiešanej presnosti a Low-Rank Adaptation (LoRA). Navrhnuté riešenie umožňuje adaptáciu na nové symboly v čase kratšom než jedna hodina, čo predstavuje 72-násobné zrýchlenie oproti pôvodnému riešeniu, pričom dosahuje porovnateľné alebo lepšie výsledky segmentácie. Práca …
-
Velké jazykové modely pro řešení matematických úloh
… — byly doladěny na datové sadě GSM8K pomocí Low-Rank Adaptation (LoRA), se systematickým prohledáváním rychlosti učení, LoRA ranku, alfy a efektivní velikosti dávky. Výkon byl hodnocen pomocí přesné shody a procesního modelu odměny Qwen2.5-Math-PRM-7B pro posouzení kvality dílčích kroků …
-
Can LLMs Leak Secrets? A Controlled Assessment of Memorization Risks in Enterprise and Government Fine-Tuning
… controlled empirical study of memorization under Low-Rank Adaptation (LoRA), examining how exposure frequency and model scale influence extraction of structured secrets. Two Qwen2 models (1.5B and 7B parameters) were fine-tuned on parallel clean and secret-augmented corpora, with 144 synthetic …
-
Efficient Fine-Tuning of Foundation Models: Generalization, Disparity, and Subgroup Adaptation
… two contributions. First, it studies demographic adaptation with parameter-efficient fine-tuning (PEFT) under realistic constraints where full fine-tuning is infeasible and available data are skewed. We evaluate five adaptation strategies, three pretraining paradigms, and multiple medical imaging …
-
Explainable AI for multi-label retinal diagnosis : a reproducible hybrid mixture of experts study
Automated retinal diagnosis in low-resource settings requires models that can handle multi-label disease co-occurrence, severe class imbalance, and clinically interpretable predictions. We evaluate a reproducible Hybrid Mixture of Experts (MoE) framework for multi-label retinal pathology …
-
Symbolic Regression and Sequence Modelling with Conditional and Dynamic Language Models
… DyLoRA offers a dynamic, search-free low-rank adaptation technique, enabling models to adjust their complexity on-the-fly without extensive retraining. SortedNet proposes a generalized framework for embedding multiple neural network architectures within a single model, facilitating …
-
Securing Smart Agriculture with Communication-efficient Federated Unlearning
… analysis, and performs unlearning via sparse low-rank adaptation to active neurons while freezing base model parameters. FedSCAN limits computational overhead to a compact parameter subset, enabling devices to participate in a communication-efficient manner. Experimental results on four …
-
Fast and Sustainable Video Anomaly Detection with Continual Learning
… RegiGrow, a parameter-efficient continual adaptation framework built on the Flashback retrieval pipeline. RegiGrow integrates Mixture-of-Experts Low-Rank Adaptation into a frozen ImageBind encoder, enabling sequential domain adaptation without modifying the pretrained backbone. A …
-
Domain-Aware Reasoning with Lightweight Models via Knowledge Distillation
… Parameter-Efficient Fine-Tuning (PEFT) with Low-Rank Adaptation (LoRA). Comprehensive evaluation, incorporating both AI-assisted analysis and blind domainexpert assessments, demonstrated that the distilled model significantly outperformed both the existing production RAG system at Recorded …
-
SciRAG: A Retrieval-Focused Fine-Tuning Strategy for Scientific Documents
… chunking, and domain-adapted fine-tuning using Low-Rank Adaptation (LoRA) to enhance an LLM's ability to understand and generate scientifically accurate content. The system incorporates a custom Retrieval-Augmented Generation (RAG) framework that supports semantic alignment of mathematical …
-
Scalable and Efficient Training of Foundation Models
… is computational and memory efficiency. While low-rank adaptation (LoRA) can be utilized to alleviate the computational overhead, it suffers from slow optimization convergence and notable performance drop. Chapter 2 analyzes an often overlooked perspective, namely residual topology, in model …
-
Attribute inference privacy protection for pre-trained models
… fine-tuning methods, Model Reprogramming and Low-Rank Adaptation. We adapt these techniques to provide attribute privacy protection for pre-trained models, minimizing computational overhead and training time. We integrate these methods into an adversarial min-max framework, allowing us to …
-
Using LoRA to Improve Retrieval in Structured Document Collections
This thesis explores different roles for Low-Rank Adaptation (LoRA) in a question answering (QA) system over structured document collections, and aims to shed light on benefits and limitations as well as rank requirement interactions with dataset size and confusability. Three roles are investigated …
-
ResearchBuddy AI: LLM-Powered Assistant
… self-hosted open-source models with LoRA (Low Rank Adaptation) and other techniques to achieve cost-effectiveness and control. The research paper summarization capability was developed through an iterative process. I began with small model and dataset experiments before scaling to a larger …
-
Table manners at scale: introducing Lora Lens for efficient analysis & detoxification of language models
… and implement the Lora Lens, an innovative adaptation of Attention Lens employing Low-Rank Adaptation (LoRA) to drastically reduce memory footprint and computational cost. Specifically designed for compatibility with large-scale models such as Llama 3 8B, Lora Lens integrates seamlessly with …
-
CRAFT: Curriculum Rank Adversarial Fine-Tuning for Robust Vision Language Models
… To address this issue, we propose Curriculum Rank Adversarial Fine-Tuning (CRAFT), a unified low-rank fine-tuning framework designed to enhance both out-of-distribution (OOD) and adversarial robustness by integrating adaptive adversarial weight perturbations into a curriculum-driven Low-Rank …
-
When Words Fail: Why AI-Driven Context Beats Word Error Rate for Safety-Critical Air Traffic Control
… real-time constraints. This work explores domain adaptation strategies of OpenAI’s Whisper ASR model, applying Transfer Learning and Low-Rank Adaptation (LoRA) on ATC benchmark and a Federal Aviation Administration (FAA) corpus. Experimental results show that fine- tuning significantly reduces …
Page 1 of 3