Skip to content
← Research
Deep Learning

Transformers & LLMs

Attention mechanisms, scaling behavior, and the practical engineering of training and serving large models efficiently.

AttentionLinformerFine-tuningInference efficiency

Focused on the mechanics of transformer architectures — attention patterns, positional encodings, efficient variants like Linformer, and efficient inference — and on applying them both to language and to sequences where order and locality carry different meaning than in text.