Search papers, labs, and topics across Lattice.
1
0
4
Forget massive models: a carefully trained 4B parameter agent can now rival 30B-class systems on complex research tasks, all while using only 10K open data.
