Shiyao Li

model compression (1) post-training quantization (1) catastrophic forgetting (1) knowledge distillation (1) in-context learning (1) prompt engineering (1) class-incremental learning (1) efficient computing (1) vision-language model (1) weak-to-strong generalization (1) learning by teaching (1) reasoning improvement (1) large language model (1) gpu kernel (1) memory buffer (1) modality-balanced quantization (1) quantization sensitivity (1) triplet contrastive loss (1) feature relationship preservation (1) asymmetric augmentation (1)