Upload ml/run_pipeline.sh with huggingface_hub
Browse files- ml/run_pipeline.sh +34 -0
ml/run_pipeline.sh
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/bash
|
| 2 |
+
# Run the full ML crisis correlation pipeline
|
| 3 |
+
# Usage: ./run_pipeline.sh
|
| 4 |
+
# Designed to be run inside tmux
|
| 5 |
+
|
| 6 |
+
set -euo pipefail
|
| 7 |
+
|
| 8 |
+
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
|
| 9 |
+
PYTHON="/opt/epstein_env/bin/python3"
|
| 10 |
+
LOG_DIR="/var/www/research/storage/logs"
|
| 11 |
+
|
| 12 |
+
mkdir -p "$LOG_DIR"
|
| 13 |
+
|
| 14 |
+
echo "============================================="
|
| 15 |
+
echo " ML Crisis Correlation Pipeline"
|
| 16 |
+
echo " Started: $(date)"
|
| 17 |
+
echo "============================================="
|
| 18 |
+
|
| 19 |
+
echo ""
|
| 20 |
+
echo "[Stage 1/3] Extracting document dates..."
|
| 21 |
+
$PYTHON "$SCRIPT_DIR/01_extract_dates.py" 2>&1 | tee "$LOG_DIR/ml_01_dates.log"
|
| 22 |
+
|
| 23 |
+
echo ""
|
| 24 |
+
echo "[Stage 2/3] Seeding historical events..."
|
| 25 |
+
$PYTHON "$SCRIPT_DIR/02_seed_events.py" 2>&1 | tee "$LOG_DIR/ml_02_events.log"
|
| 26 |
+
|
| 27 |
+
echo ""
|
| 28 |
+
echo "[Stage 3/3] Correlating documents with crises..."
|
| 29 |
+
$PYTHON "$SCRIPT_DIR/03_correlate_crises.py" 2>&1 | tee "$LOG_DIR/ml_03_correlate.log"
|
| 30 |
+
|
| 31 |
+
echo ""
|
| 32 |
+
echo "============================================="
|
| 33 |
+
echo " Pipeline complete: $(date)"
|
| 34 |
+
echo "============================================="
|