AI Safety
Benchmarking Prompt Injection Attacks on LLMs: Turkish Vulnerability Assessment and English Comparative Analysis
Ö. Aytaş, T. Şen, B. Diri, G. Biricik, M.A. Bayram
Applied Sciences · accepted; final proof stage · 2026
Open paper →§
Research
Open benchmarks, models, tokenizer work, and datasets - available through live demos and published releases.
Magibu Q3
The foundation model developed by Magibu AI, custom-built to align with Turkish morphology and language structure.
Start chatting →Google Scholar impact
As of June 2026
39
Citations
4
h-index
2
i10-index
Across our Hugging Face organization and profile, we publish research openly through 80+ models, 38+ datasets, 9+ Spaces, and actively maintained paper collections.
Featured publications
Selected works with strong Google Scholar visibility that represent the core Magibu research track.
Benchmarkcited by 12
Setting Standards in Turkish NLP: TR-MMLU for Large Language Model Evaluation
M.A. Bayram, A.A. Fincan, A.S. Gümüş, B. Diri, S. Yıldırım, Ö. Aytaş
arXiv:2501.00593 · 2024
Open paper →Tokenizationcited by 10
Tokenization Standards for Linguistic Integrity: Turkish as a Benchmark
M.A. Bayram, A.A. Fincan, A.S. Gümüş, S. Karakaş, B. Diri, S. Yıldırım
arXiv:2502.07057 · 2025
Open paper →Tokenizationcited by 3
Tokens with Meaning: A Hybrid Tokenization Approach for NLP
M.A. Bayram, A.A. Fincan, A.S. Gümüş, S. Karakaş, B. Diri, S. Yıldırım, D. Çelik
arXiv:2508.14292 · 2025
Open paper →Embedding
Adapting Multilingual Embedding Models to Turkish via Cross-Lingual Tokenizer Surgery and Offline Distillation
M.A. Bayram, B. Diri, S. Yıldırım
arXiv:2605.29992 · 2026
Open paper →Medical LLM
Healthcare-Focused Turkish Medical LLM: Training on Real Patient-Doctor Question-Answer Data for Enhanced Medical Insight
M.A. Bayram, B. Diri, S. Yıldırım
ACM TALLIP · 2025
Open paper →Datasetcited by 5
Turkish MMLU: Yapay Zeka ve Akademik Uygulamalar İçin En Kapsamlı ve Özgün Türkçe Veri Seti
M.A. Bayram
Zenodo 13375017 · 2024
Open paper →