Lelapa AI Buzuzu-Mavi Challenge
Helping Africa
$1 300 USD
Completed (~1 year ago)
Natural Language Processing
Sentiment Analysis
Machine Translation
495 joined
118 active
Start
Jan 09, 25
Close
Apr 06, 25
Reveal
Apr 07, 25
LB 0.285 fixed a bug of previous LB 0.255 notebook
Notebooks · 24 Mar 2025, 18:49 · 1
I just noticed that XNLI task has three states: true, false or neither. Please checkout the updated notebook https://www.kaggle.com/code/fiveflowerstarfish/zindi-lelapa-buzuzu-mavi-qlora-mutli-task
latest version and also added premise in prompt for xnli task
# XNLI task: result%2 is a bug! there are three states! mask = df.ID.apply(lambda x: 'afrixnli' in x) df.loc[mask, 'Response'] = df.loc[mask, 'generated'].apply( lambda x: int(x.strip().split()[0])%2 if x.strip().split()[0].isdigit() else 0 )
@snow This is my first time doing a language processing competition. The effort you placed in the notebook is much appreciated. Thank you so much!