Tag
A technical blog post discussing how structural descriptions like type systems and build systems leverage graph-structured models to ensure correctness, highlighting the definitional nature of systems like Bazel.
A curated collection of counterexamples that demonstrate limitations and pitfalls in type systems, serving as an educational resource for programmers and language designers.
A formal core calculus for documents is presented, likely exploring type systems or semantics for document processing.
Argues that TypeScript is a poor default for the AI era due to its unsound type system, which fails to catch errors in AI-generated code, and likens it to Paul Graham's Blub paradox.
The author argues that the resurgence of static typing in the 2010s was due to improved type systems (e.g., TypeScript, Haskell, Rust) that offer nullable type handling, sum types, and type inference, contrasting with poor static typing in earlier languages like Java and C++98.
This article introduces a new approach to memory safety based on linear types and abstract interpretation, aiming to eliminate common bugs like use-after-free and memory leaks more ergonomically than Rust.
This article explores the theoretical and practical challenges of training LLMs to produce typed outputs natively, rather than relying on post-hoc typechecking, with a focus on formally typed languages like Idris, Lean, and Agda. It analyzes current ad-hoc approaches to enforcing types during inference and proposes rebuilding LLMs from the ground up to generate inherently typed outputs.
OpenAI researchers present a novel approach to entity disambiguation using type discovery, where a system predicts entity types from a pre-chosen category set to resolve ambiguous references. The method achieves state-of-the-art results on entity disambiguation datasets and enables efficient O(N) runtime entity ranking through type-based weighting.