Welcome to the prompt wiki!
These are example of how to use this platform with datasets currently in Tech4Dev workspace. Throughout, we assume
You are in eval data directory:
git clone https://github.com/ProjectTech4DevAI/file-search-eval.git cd file-search-eval
You have followed the general setup described in the README
Kunji data is split across two Github repositories differentiated by phase:
They have the same structure, so using either can be reduced to variable naming. For this example:
KUNJI_DATA=veddis-eval-data # KUNJI_DATA=kunji-haryana-data # also valid, for Phase 2 Haryana data
From there we can clone the repository and run the scripts:
git clone https://github.com/ProjectTech4DevAI/${KUNJI_DATA}.git gt="-g $KUNJI_DATA/assessments/correct" ./bin/run-prompts.sh $gt -p $KUNJI_DATA/prompts -d $KUNJI_DATA/docs \ | ./bin/run-evals.sh $gt -u $KUNJI_DATA/prompts/user > evaluations.jsonl