Search papers, labs, and topics across Lattice.
3
2
6
Discarded low-confidence tokens during text generation can actually enhance retrieval performance, leading to a groundbreaking increase in throughput for multi-hop QA tasks.
Ditch the complex diffusion sampling methods: a simple Adam optimizer can give you state-of-the-art image generation and restoration results.
By pausing to "think" with latent diffusion, STAR-LDM achieves superior language understanding, narrative coherence, and controllable generation compared to standard autoregressive models of similar size.
