Leonid Karlinsky

A new model improves understanding of speech and sounds simultaneously.

Automating prompt creation boosts visual recognition accuracy for unseen objects.

Enhancing the learning capabilities of AI models through better training methods.

A new method enables the transfer of LoRA modules with synthetic data, minimizing reliance on original data.

A new benchmark tests compositional reasoning in advanced models.

A new approach enhances how AI compares images using visual instructions.

A study on improving LLMs' problem-solving abilities using a new framework.

Self-MoE creates specialized experts for improved language model performance.

Granite code models improve coding efficiency with advanced long-context capabilities.

Machines learn to locate objects in images using innovative techniques.

XTRA improves how computers recognize images using less data and resources.

New method enhances AI's ability to classify corrupted images effectively.