MIT researchers have developed a new technique that enables AI models to become smaller and faster during training, rather than requiring expensive post-training compression. This breakthrough reduces the computational resources, energy, and time nee
← News
New technique makes AI models leaner and faster while they’re still learning
Expensive Model Compression Problem solved
On-The-Fly Efficiency Gains Solution
MIT, Max Planck, ELIAS Research institutions
AI Sustainability Impact area
Reduced Energy And Compute efficiency
New Training Technique innovation
Lower Training Expenses cost savings
3 Research Labs institutions