BULATS-L1
A native language identification dataset based on writing samples from the BULATS business English examination, focusing on workplace and business registers.
Timeline (last 90 days)
The study evaluated the token-probability scoring method using Llama 3.1 8B and 70B models on three datasets: TOEFL9, Write and Improve 2024, and BULATS-L1.
The 70B model achieved state-of-the-art (SOTA) performance on all datasets, surpassing prior prompt-based methods.
The BULATS-L1 dataset is based on writing samples from the BULATS business English exam, with scenarios oriented toward workplace and business communication.
All Facts (3)
The study evaluated the token-probability scoring method using Llama 3.1 8B and 70B models on three datasets: TOEFL9, Write and Improve 2024, and BULATS-L1.
50%UnverifiedThe 70B model achieved state-of-the-art (SOTA) performance on all datasets, surpassing prior prompt-based methods.
50%UnverifiedThe BULATS-L1 dataset is based on writing samples from the BULATS business English exam, with scenarios oriented toward workplace and business communication.
50%