We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
VS Code Blog·4 min read
Visual Studio Code 1.122
TL;DR
Researchers identified a gap in LLM evaluation benchmarks. They built a synthetic dataset with 10k adversarial prompts targeting reasoning failures.
✦ Why It Matters
Use this benchmark to audit LLM robustness before deploying in production reasoning pipelines.
Key Takeaways
- Utilize 1 million token context windows for larger AI model interactions.
- Test web app responsiveness using the integrated browser's device emulation.
- Leverage the new BYOK feature for offline AI model usage.
- Use the improved issue reporting wizard for better bug tracking.
Full Summary
The latest release of Visual Studio Code, version 1.122, enhances the development experience by supporting 1 million token context windows for compatible AI models from Anthropic and OpenAI, allowing developers to manage larger codebases and conversations. It introduces an air-gapped BYOK feature, enabling users to utilize their own language models without needing to sign in to GitHub, which is particularly useful in restricted environments.
Additionally, the integrated browser now supports device emulation, allowing developers to test web app responsiveness directly within VS Code. The update also improves issue reporting with a new wizard that facilitates the creation of detailed reports, including screenshots and video recordings.
These enhancements aim to streamline workflows and improve the overall coding experience.