After half a year of development, we are finally filling the gap in long-tail knowledge evaluation.

Simplify the complex and encourage the machine to think deeply!

Inference speed is 2-6 times faster than that of MoE. The related paper has been accepted by ICLR 2025.

The core author is a PhD candidate with 3-year internship in the team.

Exploring the frontiers of intelligence

Bring the "soul" to AI speech

Submission channels are now open!

Let's walk through the 230 days of birth and rapid growth of Doubao large models.

It is equipped with enhanced content recognition, improved understanding and reasoning, and fine visual description.

Peking University-ByteDance "Doubao Large Model System Software Joint Laboratory" established, focusing on key issues in AI system software.

Code Sandbox Tool Also Open-Sourced, Available for Online Experience.

AutoKaggle systematically addresses the complexities of data science tasks, ensuring robust and accurate code generation.

It seamlessly replaces traditional MLPs, reducing FLOPS and params.

Its research results have been selected for NeurIPS 2024.

Highly aesthetic editing with precise response to any command. Now open to testing!

Research done by a post-95s and a post-00s generation

Decoupling control flow and computing flow for both flexibility and efficiency.

SIA Lab aims to achieve breakthroughs in foundational large model technologies and build industrial applications through effective industry-academia-research collaboration.

Comprehensive efforts have been made in the research and development of the basic models, which have unique advantages in various business scenarios.

With a unified and flexible framework, it supports score-to-song conversion, controllable generation, music and lyrics editing, and low-threshold voice cloning, etc.