Zijia Lin

EntropyLong: Effective Long-Context Training via Predictive Uncertainty

Unbiased Token Pruning for Efficient Large Multimodal Models

NExtLong: Toward Effective Long-Context Training without Long Documents

LongMagpie: A Self-synthesis Method for Generating Large-scale Long-context Instructions

PrefixKV: Adaptive Prefix KV Cache is What Vision Instruction-Following Models Need for Efficient Generation