Today's featured cutting-edge AI information, enjoy reading 👇

💬 Meta open-sources Spirit LM, a multimodal model based on Llama2, supporting text and speech input/output.

🖼️ Adobe releases Long-LRM technology, quickly generating 3D scenes from 32 images in just 1.3 seconds.

🖥️ 1Remote open-source remote session management tool, supporting multiple protocols, providing quick launch and multi-monitor support.

📚 awesome-low-level-design: A collection of low-level design learning materials, covering OOP, design patterns, and interview questions.

📖 Select Star SQL: Free interactive SQL learning book, covering SQL basics, suitable for beginners to learn online.

Cutting-edge Technology

1. Meta open-sources a new multimodal language model: Spirit LM.

Based on the Llama2 model, it extends to support speech modality through continuous training on text and speech, supporting both text and speech input and output.

GitHub: https://github.com/facebookresearch/spiritlm

Detailed introduction: https://speechbot.github.io/spiritlm/

Two versions are provided: Base and Expressive, with the Expressive version showing good performance in pitch and style.

2. Adobe's research on generating 3D scenes from multiple images: Long-LRM.

By inputting 32 scene images, this technology uses 3D Gaussian reconstruction to restore the entire 3D scene in just 1.3 seconds.

Detailed introduction: https://arthurhero.github.io/projects/llrm/

From the demonstration provided, the effect looks good, but fast generation requires high-performance configurations.

Open Source Projects

1. An open-source free remote session management tool: 1Remote.

It allows you to quickly start a remote session at any time, supporting protocols such as Microsoft Remote Desktop (RDP), VNC, SSH, Telnet, SFTP, FTP, RemoteApp, etc.

GitHub: https://github.com/1Remote/1Remote

Features include:

• Multi-protocol support: Supports RDP, SSH, VNC, Telnet, (S)FTP, RemoteApp, and other protocols, quite comprehensive;
• Quick launch: Can quickly start remote sessions via Alt + M, supporting pinyin search for Chinese characters;
• Multi-monitor support: Provides excellent RDP multi-monitor remote management experience;
• Detailed configuration: Can set tags, icons, pre and post-connection scripts for each connection;
• Interface customization: Supports multiple languages and themes, and allows adjusting interface layout according to personal preferences;
• Customizable launcher, allowing you to replace built-in programs with your preferred ones in protocols like sFTP and VNC;
• Ready to use: Exe version can be downloaded and unpacked without any installation required.

2. A collection of materials for learning low-level design (LLD): awesome-low-level-design.

It covers basic object-oriented programming (OOP) concepts, design patterns, UML diagrams, interview questions, and recommended books.

GitHub: https://github.com/ashishps1/awesome-low-level-design

Learning Books

1. A free interactive SQL learning book: Select Star SQL.

The book has five chapters, each taking about 30 minutes, covering topics such as row operations, grouping and aggregation, table joins, and combines learning with practical cases.

Read online: https://selectstarsql.com

Suitable for beginners, requires little programming background, can be learned online through the website, no registration required and ad-free. Interested readers can take a look.