Dachuan Shi

knowledge distillation (2) neural network pruning (2) model merging (2) model compression (2) parameter-efficient fine-tuning (1) low-rank adaptation (1) model collapse (1) task vector (1) out-of-domain generalization (1) parameter-efficient finetuning (1) reasoning capability (1) vision-language transformer (1) unified progressive pruning (1) multimodal subnet (1) large language model (1) reasoning generalization (1) llm reasoning (1) subnetwork search (1) iterative weight merging (1) subnet extraction (1)