✎ thoughts and notes on AI/ML research & software engineering
No one read the blogs, but I kept writing anyway.
Systems that use more than one LLM already exist. One model writes code. Another does math. Another is a general assistant. They are asked to work together. Today they talk in English. The first model writes a message. The second model reads those words. The first model’s inner picture of the input has to become a l...
The term mechanistic interpretability now appears in workshops, reading groups, grants, and much of the new work on language models. The growth has not come with a shared definition. Researchers, w...
Every GPT-style language model is trained with the same objective. The transformer takes a context of tokens, the last layer maps that representation onto the vocabulary, and a softmax turns the sc...
AI coding agents are no longer a novelty. They are being used in production workflows at companies of every size, praised in engineering blogs, featured in product launches, and adopted by develope...
Information retrieval systems have become essential components of modern search engines, recommendation systems, and knowledge discovery platforms. These systems help users find relevant documents ...