Search papers, labs, and topics across Lattice.
4
0
8
5
NCP-ArchPreview is introduced, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP) and learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective while preserving standard token-level autoregressive generation.
PI-Mem achieves unprecedented long-context reasoning capabilities, outperforming traditional methods while accelerating inference by over 6 times.
CUAs can achieve a 73.7% success rate on complex macOS tasks, but the secret to their performance lies in skill libraries, not just framework design.
Stop reinventing the wheel for diffusion LLM alignment: DARE provides a unified framework for SFT, preference optimization, and RL, accelerating research and enabling fair comparisons across dLLMs.
