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 19 of 19 for “"context window"”.

  1. Developing a Humpback Whale Vocalization Detector Using Machine Learning Models

    … based on the evolution of a song, called context windows. We apply two non-parametric machine learning models, k-nearest neighbors, and random forests, to our created observations. Results are tested in two manners, per-context-window and per-vocalization. Context windows with more time …

    calpoly Repository record for Developing a Humpback Whale Vocalization Detector Using Machine Learning Models (opens in a new tab)

  2. RingChains Graph-based Summarizer and Enhanced Large Language Models for Summarizing Long Documents

    … sizes. Second, most standard LLMs have a limited context window size. If the number of context tokens is increased by a factor of n, both the required computational resources and the running time scale as n2 in a Transformer architecture or as n√ n in a sparse Transformer architecture. Third, …

    unt Repository record for RingChains Graph-based Summarizer and Enhanced Large Language Models for Summarizing Long Documents (opens in a new tab)

  3. ADVANCING TEST-TIME COMPUTE SCALING IN LANGUAGE MODELS: FROM TRAINING TO INFERENCE

    … bottlenecks across the model lifecycle: context capacity, reasoning capability, efficient conditioning, and coordination of parallel inference. First, AnchorAttention extends the usable context window through stable continuous pre-training, providing a larger workspace for long inputs and …

    nus Repository record for ADVANCING TEST-TIME COMPUTE SCALING IN LANGUAGE MODELS: FROM TRAINING TO INFERENCE (opens in a new tab) Download file for ADVANCING TEST-TIME COMPUTE SCALING IN LANGUAGE MODELS: FROM TRAINING TO INFERENCE (opens in a new tab)

  4. Unsupervised brand name extraction using domain adaptation

    … about the whole document. In this thesis, a context window is defined to capture the context of a word in a sentence. In addition, a word embedding model is locally trained to have a domain specific model and finally, a domain adaptation technique is employed to transfer the knowledge from …

    uoit Repository record for Unsupervised brand name extraction using domain adaptation (opens in a new tab)

  5. A Graph-Augmented AI System for Automated Changelog Generation in Secure Environments

    … repository graph, retrieves release-relevant context, and prompts an LLM to generate Markdown changelogs. When a release range exceeds the prompt budget, the system summarizes diff chunks and merges intermediate summaries before final generation. The completed evaluation adds an LLM judge that …

    ecu Repository record for A Graph-Augmented AI System for Automated Changelog Generation in Secure Environments (opens in a new tab)

  6. PANER: A Paraphrase-Augmented Framework for Low-Resource Named Entity Recognition

    … from prior IT approaches to leverage the large context window of recent state-of-the-art LLMs; (2) introducing a strategic data augmentation technique that preserves entity information while paraphrasing the surrounding context, thereby expanding our training data without compromising semantic …

    concordia Download PDF for PANER: A Paraphrase-Augmented Framework for Low-Resource Named Entity Recognition (opens in a new tab)

  7. Improving Performance of Clinical Text Classification Models with Attention Mechanisms and Rationales

    … tion mechanism’s receptive field with a flexible context window at the phrase-level leads to improved understanding of local linguistic information in clinical text; and (iii) utilizing human-based clinical rationales as additional supplementary training data can improve model performance and …

    utk-theses Repository record for Improving Performance of Clinical Text Classification Models with Attention Mechanisms and Rationales (opens in a new tab)

  8. A Schema-aware Harness for Tabular Reasoning with Language Models

    … outputs. We address this challenge in the context of Recursive Language Model (RLM) systems, where a root-model decomposes complex queries and delegates linguistically irreducible sub-tasks to sub-models. Using IPEDS, a large-scale higher education dataset that exceeds context window limits …

    vt Repository record for A Schema-aware Harness for Tabular Reasoning with Language Models (opens in a new tab)

  9. Towards Efficient Long Context Handling in Large Language Models: Advancing Scalability and Performance

    … across various applications. However, limited context window length significantly hinders their performance in scenarios requiring understanding of long-form text. Without efficient mechanisms to handle extended contexts, LLMs remain constrained in applications such as document analysis, …

    tamu Repository record for Towards Efficient Long Context Handling in Large Language Models: Advancing Scalability and Performance (opens in a new tab) Embargoed until 2027-05-01

  10. Modeling mutation rate variation over time using Bayesian sequence context mutational trees

    … are highly variable across the genome. Sequence context mutation models are a natural modelling framework to capture this variability by taking into account the local nucleotide context. Models that include wider sequence context windows generally perform better. However, this comes at the cost …

    penn Repository record for Modeling mutation rate variation over time using Bayesian sequence context mutational trees (opens in a new tab)

  11. Resource-Efficient Machine Learning with Limited Data and Computation

    … (iv) to reduce the high cost of training long-context models, we propose SelfExtend, which extends the context window of LLMs by constructing bi-level attention information, e.g., grouped attention and neighbor attention. With minor code modifications, our SelfExtend can effortlessly extend …

    tamu Repository record for Resource-Efficient Machine Learning with Limited Data and Computation (opens in a new tab)

  12. Enhancing genomic data quality through deep learning methods

    … genotype imputation. GI-Joe extended the context window to over one million variants, markedly improving accuracy for rare and structural variants while remaining efficient on consumer-grade hardware. This efficiency lowers barriers for high-resolution imputation by enabling in-house …

    temple Repository record for Enhancing genomic data quality through deep learning methods (opens in a new tab)

  13. Advancing multi-session personalization in LLMs via direct policy

    The realization of personalized AI agents, with context awareness spanning multi-session interactions, is currently hindered by the limited size of the context window, the computational cost of in-model personal fine-tuning, the sparsity of long-term interaction data, and the architectural …

    texas Repository record for Advancing multi-session personalization in LLMs via direct policy (opens in a new tab)

  14. Design and Evaluation of GPT and RAG-Based QA Systems for Genomic and Pharmacogenomic Patient Queries

    … To maximize information density within the context window of each model, our approach used token-conscious chunking. To securely orchestrate data flow among storage, search, and LLM endpoints, we used a multi-service authentication approach. Together, these design choices enabled creating an …

    jhu Repository record for Design and Evaluation of GPT and RAG-Based QA Systems for Genomic and Pharmacogenomic Patient Queries (opens in a new tab) Embargoed until 2029-05-01

  15. Language Models for Citation Classification

    … that consider semantic cues in the form of the context surrounding the citations. Identifying the semantic aspects of citations has proven valuable in various applications including research assessment, information retrieval, document summarisation, and more. While automated citation …

    the-open-u Download PDF for Language Models for Citation Classification (opens in a new tab)

  16. Efficient Learning from Weak Supervision: From Data and Model Reuse to Agentic Autonomy

    … time. </p><p>The third part shifts toward In-Context Reinforcement Learning (ICRL), where a single model is trained to solve entirely unseen RL tasks through its context window. I introduce three key contributions in this space. First, I present a framework for \emph{ICRL using suboptimal …

    duke Repository record for Efficient Learning from Weak Supervision: From Data and Model Reuse to Agentic Autonomy (opens in a new tab)

  17. Private Messages, Public Health: Detecting Emotional Support with Long-Context LLMs

    … language models (LLMs) capable of handling long contexts. We identify limitations in prior work relying on forum-style public data and short-context BERT models, and show how newer LLMs, coupled with structured prompting and clinically informed evaluations, can drastically improve classification …

    gatech Repository record for Private Messages, Public Health: Detecting Emotional Support with Long-Context LLMs (opens in a new tab)

  18. GROPUS an Adaptive Rule-based Algorithm for Information Extraction

    … Linguistic patterns are not restricted to a fix context window, but encode whole sentences as primary semantic units of natural language. The proposed pattern language is powerful and expressive enough to capture non-trivial kinds of phrases and sentences containing relevant information. The …

    fu-berlin Repository record for GROPUS an Adaptive Rule-based Algorithm for Information Extraction (opens in a new tab)