Get ready for a game-changer in the world of open-source AI Microsoft's Phi-3 Mini, a 3.8 billion parameter language model, has just received a major upgrade, and we're diving into all the exciting details! In this video, we'll explore the significant improvements to code understanding, structured output, multi-turn conversations, and long context understanding.
We'll also put the updated Phi-3 Mini 128k model to the test, pushing its limits with challenging questions, reasoning exercises, and massive text summaries.
What's new in Phi-3 Mini:
Improved code understanding across multiple languages (Python, C++, Rust, TypeScript)
Enhanced structured output (JSON, XML)
Boosted multi-turn conversations and long context understanding
Benchmark results: Crushing it in code understanding and long context tasks, outperforming larger models like Mistral 7B, Mixtral, and LLama 3 8B!
Model - [ Ссылка ]
Azure Link - [ Ссылка ]
Local Lab Twitter - [ Ссылка ]_
[ Ссылка ]
![](https://s2.save4k.ru/pic/b1kP1s9sc1M/maxresdefault.jpg)