Abstract: In this work, we propose a new soft-input soft-output decoder called soft-output from covered space (SOCS) decoder. It estimates the a posteriori reliability based on the space explored by a ...

As AI coding tools generate billions of lines of code each month, a new bottleneck is emerging: ensuring that software works as intended. Qodo, a startup building AI agents for code review, testing, ...

Anthropic’s Claude Code surpassed 20 million commits across more than one million GitHub repositories this week, yet a third-party tracking dashboard revealed a striking imbalance: roughly 90% of that ...

We introduce phi-1, a new large language model for code, with significantly smaller size than competing models: phi-1 is a Transformer-based model with 1.3B parameters, trained for 4 days on 8 A100s, ...

Philip Guo’s research-driven Python Tutor has powered hundreds of millions of code visualizations since 2010 — and new long-term impact recognition highlights why it still matters today When ...