|
|
#!/usr/bin/env bash |
|
|
set -e |
|
|
|
|
|
CSV_PATH=${CSV_PATH:-/app/export_artifacts/muril_multilingual_dataset.csv} |
|
|
EMBED_PATH=${OUT_EMBED_PATH:-/app/export_artifacts/answer_embeddings.pt} |
|
|
MODEL_DIR=${MODEL_DIR:-$HF_REPO} |
|
|
HF_REPO=${HF_REPO:-Sp2503/Finetuned-multilingualdataset-MuriL-model} |
|
|
FORCE_REGEN=${FORCE_REGEN:-false} |
|
|
UPLOAD_BACK=${UPLOAD_BACK:-false} |
|
|
|
|
|
echo "Entrypoint: CSV_PATH=$CSV_PATH EMBED_PATH=$EMBED_PATH MODEL_DIR=$MODEL_DIR FORCE_REGEN=$FORCE_REGEN" |
|
|
|
|
|
mkdir -p "$(dirname "$EMBED_PATH")" |
|
|
|
|
|
need_regen=false |
|
|
if [ ! -f "$EMBED_PATH" ]; then |
|
|
echo "Embeddings not found -> will regenerate." |
|
|
need_regen=true |
|
|
fi |
|
|
|
|
|
if [ "$FORCE_REGEN" = "true" ] || [ "$FORCE_REGEN" = "1" ]; then |
|
|
echo "FORCE_REGEN -> will regenerate embeddings." |
|
|
need_regen=true |
|
|
fi |
|
|
|
|
|
if [ "$need_regen" = "true" ]; then |
|
|
export MODEL_DIR="${MODEL_DIR}" |
|
|
export CSV_PATH="${CSV_PATH}" |
|
|
export OUT_EMBED_PATH="${EMBED_PATH}" |
|
|
export HF_REPO="${HF_REPO}" |
|
|
export UPLOAD_BACK="${UPLOAD_BACK}" |
|
|
python /app/regenerate_embeddings.py |
|
|
echo "Regeneration finished." |
|
|
else |
|
|
echo "Skipping regeneration." |
|
|
fi |
|
|
|
|
|
exec uvicorn main:app --host 0.0.0.0 --port ${PORT:-7860} |
|
|
|