Use faceted filters to find projects by title, author, supervisor, department, programme, subject, case study, year, document type and access mode.
This record links to the original Transformer paper, which introduced a sequence modelling architecture based entirely on self-attention. It is useful for repository testing because it contains formal architecture descriptions, experiments, equations, referenc…
This record links to the word2vec paper, which presents efficient methods for learning distributed word representations. It is useful for testing older but influential NLP documents, keyword matching, and public external PDF access.
This record links to the sequence-to-sequence learning paper, which uses neural networks to map input sequences to output sequences. It provides a realistic external PDF for testing NLP repositories, metadata extraction, and citation generation.
This record links to the DINO self-supervised vision transformer paper, which studies emergent properties in learned visual representations. It is useful for testing modern machine learning keywords and external document previews.
This record links to the LoRA paper, which introduces low-rank adaptation for efficient fine-tuning of large language models. It is useful for testing AI thesis metadata, modern NLP keywords, and external PDF access.
This record links to the latent diffusion models paper, which combines compression and diffusion modelling for efficient high-resolution image synthesis. It is suitable for validating external document access and contemporary generative AI metadata.
This record links to the denoising diffusion probabilistic models paper, which describes a generative modelling framework based on iterative noise reversal. It supports realistic testing for modern AI documents, long keywords, and PDF preview workflows.
This record links to the dropout paper, which explains how randomly omitting feature detectors during training can reduce overfitting. It is suitable for testing undergraduate AI project metadata, concise abstracts, and direct external PDF linking.
This record links to the Adam optimization paper, which presents an adaptive learning-rate method widely used in deep learning. It is a compact but realistic technical PDF for testing keyword relevance, citation generation, and external document access.
This record links to the original GAN paper, which introduced adversarial training between a generator and a discriminator. The document is useful for validating machine learning topics, external PDF handling, citation metadata, and repository discovery featur…
This record links to the GPT-3 paper, which evaluates large autoregressive language models under few-shot, one-shot, and zero-shot settings. It is useful for stress-testing long author metadata, AI research keywords, external document display, and abstract sea…
This record links to the BERT paper, which describes bidirectional transformer pre-training for language understanding tasks. It gives the platform a realistic NLP document for search, indexing, keyword generation, citation display, external preview, and metad…
This record links to the original Transformer paper, which introduced a sequence modelling architecture based entirely on self-attention. It is useful for repository testing because it contains formal architecture descriptions, experiments, equations, referenc…
This study designs and evaluates an AI assisted metadata extraction workflow for institutional repository submissions, focusing on title parsing, author validation, keyword generation, and librarian review accuracy. The work demonstrates how automated suggesti…
This study proposes a predictive analytics model for identifying undergraduate students at risk of delayed academic progression. It uses academic records, attendance indicators, and supervised learning techniques to support timely intervention while emphasizin…
This study examines the knowledge, attitude, and practice of infection prevention among clinical students during hospital postings. It identifies gaps in hand hygiene, use of protective equipment, waste disposal, and supervision, then proposes practical measur…
This study designs a blockchain based verification model for academic certificates and transcripts. It focuses on credential integrity, issuer authentication, privacy preserving lookup, and integration with existing university records management processes.
This research evaluates the economic effect of youth entrepreneurship programmes on graduate employability. It considers skills acquisition, venture creation, income generation, mentorship quality, and the sustainability of entrepreneurship support initiatives…
This study designs and evaluates an AI assisted metadata extraction workflow for institutional repository submissions, focusing on title parsing, author validation, keyword generation, and librarian review accuracy. The work demonstrates how automated suggesti…
This study proposes a predictive analytics model for identifying undergraduate students at risk of delayed academic progression. It uses academic records, attendance indicators, and supervised learning techniques to support timely intervention while emphasizin…
This study examines the knowledge, attitude, and practice of infection prevention among clinical students during hospital postings. It identifies gaps in hand hygiene, use of protective equipment, waste disposal, and supervision, then proposes practical measur…
This study designs a blockchain based verification model for academic certificates and transcripts. It focuses on credential integrity, issuer authentication, privacy preserving lookup, and integration with existing university records management processes.
This research evaluates the economic effect of youth entrepreneurship programmes on graduate employability. It considers skills acquisition, venture creation, income generation, mentorship quality, and the sustainability of entrepreneurship support initiatives…
This study designs and evaluates an AI assisted metadata extraction workflow for institutional repository submissions, focusing on title parsing, author validation, keyword generation, and librarian review accuracy. The work demonstrates how automated suggesti…
This study proposes a predictive analytics model for identifying undergraduate students at risk of delayed academic progression. It uses academic records, attendance indicators, and supervised learning techniques to support timely intervention while emphasizin…