These are my research papers, some of them are currently being written and pre-print might not be out yet. Research interest are llm training, inference optimization, fine-tuning, machine translation and model adaptation for underrepresented languages.
2026
Proposing a novel sparse attention mechanism that reduces computational complexity from O(n²) to O(n·log(n)) while maintaining competitive performance on downstream NLP benchmarks.
2026
Introducing a contrastive pre-training strategy that improves few-shot transfer learning across domain boundaries, evaluated on medical and legal text corpora.
These are my fafo experiments.
wip
