Crossing Linguistic Horizons

Finetuning and Comprehensive Evaluation of Vietnamese Large Language Models

Zero-Shot Language Modeling Leaderboard

Models MLQA-MLM VSEC
EM CER WER CED WED PLX EM CER WER CED WED PLX
URA-LLaMa 70B 0.00 ± 0.00 0.50 ± 0.01 0.64 ± 0.01 519.09 ± 10.96 115.82 ± 2.45 1.08 ± 0.01 0.00 ± 0.00 0.88 ± 0.00 1.01 ± 0.00 113.51 ± 0.57 29.91 ± 0.15 1.09 ± 0.00
URA-LLaMa 13B 0.00 ± 0.00 0.67 ± 0.00 0.78 ± 0.00 697.85 ± 11.62 161.34 ± 2.64 1.16 ± 0.02 0.01 ± 0.00 0.42 ± 0.01 0.56 ± 0.01 54.88 ± 0.77 14.50 ± 0.19 1.26 ± 0.00
URA-LLaMa 7B 0.00 ± 0.00 0.73 ± 0.00 0.88 ± 0.01 684.00 ± 13.18 166.87 ± 3.18 1.25 ± 0.01 0.01 ± 0.00 3.33 ± 0.04 3.14 ± 0.03 420.34 ± 5.66 85.79 ± 0.96 1.33 ± 0.00
LLaMa-2 13B 0.00 ± 0.00 0.90 ± 0.00 1.00 ± 0.00 881.97 ± 11.23 208.52 ± 2.52 1.10 ± 0.01 0.00 ± 0.00 1.32 ± 0.01 1.40 ± 0.01 160.06 ± 1.16 38.12 ± 0.23 1.11 ± 0.00
LLaMa-2 7B 0.00 ± 0.00 0.95 ± 0.00 1.07 ± 0.01 860.42 ± 13.18 210.21 ± 3.18 1.25 ± 0.01 0.00 ± 0.00 1.54 ± 0.04 1.55 ± 0.03 171.28 ± 5.66 40.18 ± 0.96 1.14 ± 0.00
Vietcuna 7B 0.00 ± 0.00 1.00 ± 0.00 1.00 ± 0.00 951.53 ± 12.37 208.57 ± 2.73 1.48 ± 0.01 0.01 ± 0.00 1.11 ± 0.01 1.20 ± 0.01 139.90 ± 1.39 33.94 ± 0.33 1.61 ± 0.00
GPT-3.5 0.00 ± 0.00 0.34 ± 0.01 0.50 ± 0.01 422.30 ± 10.79 100.33 ± 2.44 - 0.02 ± 0.00 0.16 ± 0.00 0.30 ± 0.00 12.63 ± 0.34 3.48 ± 0.09 -
GPT-4 0.04 ± 0.00 0.40 ± 0.01 0.45 ± 0.01 381.88 ± 10.26 93.34 ± 2.39 - 0.60 ± 0.01 0.14 ± 0.00 0.26 ± 0.00 13.58 ± 0.45 3.67 ± 0.12 -