Search papers, labs, and topics across Lattice.
1
0
5
1
This is the first end-to-end demonstration that a seven-person independent team can train open-weight models with leading agentic cyber capability and all three checkpoints rank 1st among models at comparable parameter scales.
