Qiyuan Zhu

model compression (1) one-shot learning (1) adversarial learning (1) few-shot learning (1) image synthesis (1) instance segmentation (1) model merging (1) document layout analysis (1) singular value decomposition (1) generative adversarial network (1) mixture of expert (1) expert merging (1) activation pattern (1) large language model (1) font generation (1) component-aware supervision (1) multi-format document (1) component-based discriminator (1) reasoning conciseness (1) large reasoning language model (1)