training overhead
Training overhead refers to the additional computational and time resources required to train an AI model beyond the actual learning process. This can include data preprocessing, hyperparameter tuning, and model validation tasks.
- Adaptive Discretization for Consistency Models
- Aligning What Matters: Masked Latent Adaptation for Text-to-Audio-Video Generation
- DOVE: Efficient One-Step Diffusion Model for Real-World Video Super-Resolution
- Efficient Multi-bit Quantization Network Training via Weight Bias Correction and Bit-wise Coreset Sampling
- SafePTR: Token-Level Jailbreak Defense in Multimodal LLMs via Prune-then-Restore Mechanism