Kai Chen

Intern-S2-Preview-397B not only excels in multimodal scientific reasoning but also enhances biological instruction performance without altering its foundational architecture.

Post-norm normalization significantly enhances performance in LLMs when depth is introduced through a curriculum, outperforming pre-norm by an order of magnitude.

Gecko achieves private inference in just 0.4-2.2 seconds while maintaining robust security against model extraction attacks.

GPT-Red not only breaks previous models but also sets a new benchmark for automated red teaming, outperforming human efforts in discovering prompt injection vulnerabilities.

Transforming static clinical cases into interactive storytelling games, MedGame significantly enhances engagement and learning outcomes in medical education.

Persistent object tokens enable humanoid robots to achieve a 71/80 success rate in complex loco-manipulation tasks, significantly outpacing previous benchmarks.

GEPO achieves balanced cross-task improvements in RL for LLMs by dynamically adjusting advantages based on group entropy, outperforming traditional methods.

Forget textual rules and coarse embeddings: a multimodal reward model that directly compares rendered visuals unlocks significant gains in vision-to-code RL.

Diffusion models can now reason their way through complex spatial tasks with near-perfect accuracy, thanks to a new framework that unlocks chain-of-thought reasoning within the latent space.