Search papers, labs, and topics across Lattice.
Affiliation:
3
0
3
0
This work revisits joint multimodal representation learning and generation to produce linearly interpolatable embeddings that are directly consumable by generative decoders and ensures its representations function as both discriminative semantic descriptors and generative conditions.
Dense scaling and test-time reasoning tokens are overkill for retrieval: scaling multimodal embeddings along the expert axis beats Think-Then-Embed baselines while cutting active parameter counts by over 4脳.
RepFusion reveals that multimodal large language models can dramatically enhance denoising in text-to-image systems, outperforming traditional denoising methods.
