Spend 1 minute every day to get selected cutting-edge AI information.

The content includes but is not limited to cutting-edge AI news, AI tools, AI painting, open-source projects, and learning tutorials, among others.

Follow AI Daily to keep up with AI trends. We hope it will be helpful to you. Important information will be posted separately for detailed introductions.

Below is the latest AI information as of August 21.

Cutting-edge News

1. Alibaba launched the Qwen2-Math model, now available for online experience!

It supports both text and image-based queries, and its 72B parameter model outperforms GPT-4o and Claude 3.5 in mathematics capabilities according to benchmark tests.

Experience it here: https://huggingface.co/spaces/Qwen/Qwen2-Math-Demo

I personally tested it by finding a function problem online, took a screenshot, and asked it. It provided a detailed explanation process, and the answer was correct—indeed very powerful!

If you're interested, give it a try.

AI Painting

1. LivePortrait has added image-driven and region control features.

With the introduction of region control, more refined operations are available on the Gradio interface, including control over facial expressions, poses, lips, eyes, and other parts.

Detailed introduction: https://github.com/KwaiVGI/LivePortrait/blob/main/assets/docs/changelog/2024-08-19.md#gradio-interface

Open Source Projects

1. A fast-responding and fully localized AI voice chat tool: voicechat2.

It uses WebSockets to achieve low-latency voice interaction and allows remote access, enabling local execution of any voice recognition, text-to-speech, and large language models.

GitHub: https://github.com/lhl/voicechat2

On a local 4090 GPU, it can achieve latency as low as 300 milliseconds, with very fast response speed!