xri/songhay_synth_source_target

Domain:

natural language processing

Record type:

dataset

Creator:

xri

Host:

The songhayDictTrain.tsv was generated using the Machine Translation from One Book (MTOB) technique using 'Bigrammaire Songhay Zarma – Français' (Haïdara et al., 2009) and 'Grammaire comparée français soŋay-zarma' (Hanafiou, 2007) for LLM context.

xri/KonongoNMT

Performance: BLEU 8.55, ChrF 50.372 KonongoNMT is a parallel dataset composed of 8,000 sentences in

xri/NgoremeNMT

NgoremeNMT is a parallel dataset composed of 8,000 sentences in Swahili and Ngoreme. It is intended

xri/RwilaNMT

Performance: 15.26 BLEU, 54.53 ChrF RwilaNMT is a parallel dataset composed of 8,000 sentences in Sw

xri/dagaare_synth_train

The dagaare_dict_guided_train_9k.tsv presents the synthesized data for the dictionary-guided trainin

xri/songhay_synth_train

The songhay_dict_guided_train_9k.tsv presents the synthesized data for the dictionary-guided trainin

xri/dagaare_synth_source_target

The dagaareDictTrain.tsv was generated using the Machine Translation from One Book (MTOB) technique

xri/songhay_synth_source_target

Domain:

Record type:

Visit

Tasks

Languages