Search papers, labs, and topics across Lattice.
Trainable LoRA bank Frozen U-Net bank
2
0
4
Preserving historical normality in continual anomaly detection can lead to a 3.2-point improvement in detection accuracy without increasing false positives.
MLLMs can achieve up to 7.9x KV cache compression and 1.52x faster decoding without sacrificing performance by intelligently compressing different attention heads with distinct strategies.
