Since its creation, AxTRADE has been addressing the high computational and energy cost of large neural network (NN) models by developing a method that dynamically chooses, at runtime, the smallest pretrained model capable of correctly handling each input. Instead of compressing a single network, the approach uses multiple complementary pretrained models and an input-aware neural dispatcher trained through a custom multi-objective loss. A new evolutionary algorithm explores accuracy\u2013efficiency trade-offs and identifies Pareto-optimal configurations.<\/p>\n\n\n\n
The team is also studying how approximate hardware operators can be combined with runtime reconfiguration to further improve performance and energy efficiency<\/p>\n\n\n\n
Additionally, AxTRADE is currently examining the fault tolerance of Mixture-of-Experts architectures, with the goal of selectively protecting vulnerable components to improve system reliability.<\/p>\n\n\n\n
IIT Dharwad partners visited Inria Rennes from May 19 to June 20 2025.<\/strong><\/p>\n\n\n\n A visit of Inria partners to IIT Dharwad is planned for February-March 2026<\/strong><\/p>\n","protected":false},"excerpt":{"rendered":" Scientific activities Since its creation, AxTRADE has been addressing the high computational and energy cost of large neural network (NN) models by developing a method that dynamically chooses, at runtime, the smallest pretrained model capable of correctly handling each input. Instead of compressing a single network, the approach uses multiple\u2026<\/p>\n