Search papers, labs, and topics across Lattice.
Affiliation:
2
3
3
2
This work argues that the dominant cost is loss of ready serving capacity, not request progress, so recovery should preserve initialized engine state rather than reconstruct it, and presents fast recovery for Dynamo based on this principle.
Switching between autoregressive and diffusion modes allows Nemotron-Labs-Diffusion to achieve unprecedented throughput and efficiency in language modeling.
