Zainab Sani @zainab
shipping Yoruba LLM · 8 Jul
Pretraining loss is converging. Hit 1.8 perplexity on the dev set. Need more data — looking for Yoruba-text corpora.
by @zainab
Yoruba LLM
Open-source LLM fine-tuned on Yoruba-language corpora.
building
PythonPyTorchFastAPI
Build logs
1
Likes
0
Comments
0
Pretraining on 2B tokens of Yoruba web text + news. Looking for contributors with NLP chops.
Build logs (1)
Feedback (0)
Sign in to join the conversation.
No comments yet — be the first! 💬