TOEFL9
A benchmark dataset for native language identification research, containing TOEFL essays written by speakers of nine different native language backgrounds.
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 TOEFL9 dataset contains TOEFL essays from 9 L1 backgrounds and is one of the earlier and commonly used benchmarks in NLI research.
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 TOEFL9 dataset contains TOEFL essays from 9 L1 backgrounds and is one of the earlier and commonly used benchmarks in NLI research.
50%