Search papers, labs, and topics across Lattice.
2
82
5
3
Domain-specific fine-tuning can induce "agentic collapse" in LLMs, but a surprisingly small amount of agentic data from *another* domain can bring those general tool-use skills roaring back.
Open-sourcing Goedel-Prover vaults past proprietary models in automated theorem proving, proving 7 problems on PutnamBench and achieving a 60% success rate on miniF2F, all without RLHF.
