#Model Optimization
2 articles with this tag
AI Research
DoRA Efficiency Breakthrough
New factored norm and fused kernels unlock DoRA's potential, delivering 1.5-2x speedups and significant VRAM reduction.
about 1 month ago
AI Research
Pretraining's Hidden Experts: A New Post-Training Paradigm
Large pretrained models are dense with task-experts, enabling simple random sampling and ensembling to rival complex post-training AI optimization methods.
about 2 months ago