Search papers, labs, and topics across Lattice.
1
0
2
0
LLM post-training pipelines can be configured with 10x less compute using AutoPipe, a budget-aware framework that learns from historical runs and predicts performance from early training signals.
