gpt5-preds.csv

model_name,dialect_group,num_examples,BLEU,CHRF,METEOR,BERTScore_F1,adequacy,faithfulness,fluency,overall
openai/gpt-5,hijazi_jeddah,50,18.28,42.1,48.06,78.51,89,86.2,93.3,89.22
openai/gpt-5,hijazi_makkah,49,47.3,65.36,68.94,88.64,92.65,90.51,95.41,92.55
openai/gpt-5,hijazi_urban,50,37.43,58.8,64.46,86.46,92.9,90.54,94.32,92.62
openai/gpt-5,hijazi_urban_jeddah,36,26.58,51.2,52.83,84.49,91.67,90,94.94,91.89
openai/gpt-5,najdi_qasim,50,26.95,53.27,55.65,83.55,86.6,83.8,92.8,87.14
openai/gpt-5,najdi_riyadh,50,31.77,50.8,51.42,80.79,82.8,79.6,92.6,84.06
openai/gpt-5,najdi_urban,50,24.02,46,51.22,82.38,85.61,84.18,94.59,87.27
openai/gpt-5,northern_hail,39,16.51,42.6,48.33,77.22,85.51,83.46,93.72,86.46
openai/gpt-5,southern_asiri,50,34.98,58.45,41.44,76.6,63.2,61.2,93,65.5
openai/gpt-5,southern_jazan,50,39.82,57.01,64.74,84.14,72.5,71.3,92.5,76.26
openai/gpt-5,eastern_urban,50,58.2,71.98,77.52,92.52,93,92.1,97.4,93.96
openai/gpt-5,southern_qahtan,50,42.98,63.81,66.47,87.89,88.2,85.3,94.7,88.22