Search papers, labs, and topics across Lattice.
2
0
4
3
A 32B model in Marathi performs like a 10B model in English, revealing how model size and reasoning can bridge language resource gaps in cross-lingual transfer.
You're wasting LLM inference on most RAG queries: a simple post-retrieval cascade can slash LLM usage by 72% while *improving* quality and latency.
