Sailor-7B-Chat: benchmark results

Provider: Sea AI Lab. Released 2024-03-02. Access: Open.

Unified ELO 1425 ± 1, rank #1170 of 1553 rated models, from 29 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Thai LLM NLU - wisesight_thai_sentiment_seacrowd_text45.53Accuracy (%)60.9
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (chrF++)57.38chrF++53.7
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (SacreBLEU)32.53SacreBLEU52.2
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (SacreBLEU)29.86SacreBLEU49.3
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (chrF++)48.35chrF++47.8
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (BLEU)45.87BLEU47.8
Thai LLM NLU - xcopa_tha_seacrowd_qa79.6Accuracy (%)45.7
Thai LLM NLU - xnli.tha_seacrowd_pairs33.49Accuracy (%)41.3
Thai LLM - Knowledge III3.2Rating (0-10)38
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (BLEU)20.1BLEU37.3
Thai LLM NLG - iapp_squad_seacrowd_qa (ROUGE2)24.24ROUGE234.3
Thai LLM NLG - iapp_squad_seacrowd_qa (ROUGEL)31.57ROUGEL34.3

Interactive version: theaggregate.ai/model?slug=sailor-7b-chat · How It Works · Data refreshed daily, snapshot 2026-09-08.