1 paper
Nyx Iskandar, Saathvik Selvan, Slater Victoroff
The typical training process of a multimodal large language model (MLLM) involves adapting both the language model backbone and the projector between the backbone and a modality-sp…