Search papers, labs, and topics across Lattice.
Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security
2
0
5
LLMs struggle to understand nuanced values across languages, with accuracy dropping below 77% and varying by over 20% between languages, as revealed by the new X-Value benchmark.
Fine-tuning LLMs on datasets filtered at the token level, rather than the sentence level, can boost performance by up to 13.7%.
