AI
LLMs and transformers, from tokens to answers
A grounded walkthrough of tokenization, attention, pretraining, and generation, without treating a fluent answer as proof of understanding.
Explore this topic across the Loksewa Blog and Tech Insights.
A grounded walkthrough of tokenization, attention, pretraining, and generation, without treating a fluent answer as proof of understanding.
How retrieval-augmented generation connects language models to documents, and why retrieval quality matters as much as the model.