MANTA -- Model Adapter Native generations that's Affordable
This work addresses the need for more flexible and affordable personalized model generation, with potential applications in synthetic data generation and creative art domains, though it appears incremental as it builds on existing adapter selection methods.
The paper tackles the model-adapter composition problem by introducing MANTA, a new approach that improves image task diversity and quality with a modest drop in alignment, achieving a 94% win rate in task diversity and an 80% win rate in task quality compared to the best known system.
The presiding model generation algorithms rely on simple, inflexible adapter selection to provide personalized results. We propose the model-adapter composition problem as a generalized problem to past work factoring in practical hardware and affordability constraints, and introduce MANTA as a new approach to the problem. Experiments on COCO 2014 validation show MANTA to be superior in image task diversity and quality at the cost of a modest drop in alignment. Our system achieves a $94\%$ win rate in task diversity and a $80\%$ task quality win rate versus the best known system, and demonstrates strong potential for direct use in synthetic data generation and the creative art domains.