examples/run_llama.sh

#!/bin/bash
# Full pipeline for Llama-3-8B
set -e

echo "=== Per-Head KV Cache Compression — Llama-3-8B ==="

echo "Step 1: Download model"
hf download meta-llama/Meta-Llama-3-8B-Instruct --local-dir ./llama-model

echo "Step 2: Baseline"
python3 scripts/baseline.py llama-3-8b

echo "Step 3: Calibrate (20 min)"
python3 scripts/calibrate.py llama-3-8b

echo "Step 4: Run quantized inference"
python3 scripts/integrate.py llama-3-8b

echo "Step 5: Full benchmark"
python3 scripts/benchmark.py llama-3-8b

echo "Step 6: Long context benchmark"
python3 scripts/benchmark_long_context.py llama-3-8b

echo "Step 7: Generate graphs"
python3 scripts/visualize_results.py
python3 scripts/visualize_long_context.py

echo "=== Done! Check results/ and figures/ ==="