#!/bin/bash
#SBATCH --job-name=test_reward_cmp
#SBATCH --account=bfqt-delta-gpu
#SBATCH --partition=gpuA100x4
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=8
#SBATCH --gres=gpu:nvidia_a100:1
#SBATCH --mem=32G
#SBATCH --time=0:30:00
#SBATCH --output=test_reward_cmp_%j.out
#SBATCH --error=test_reward_cmp_%j.err

# Compare Llama-3.1-8B vs GPT-4o-mini for reward classification
# Tests 12 scenarios with expected labels

cd /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model
source /u/yurenh2/miniforge3/etc/profile.d/conda.sh
conda activate eval

export HF_HOME=/projects/bfqt/users/yurenh2/hf_cache/huggingface
export PYTHONPATH="${PWD}/src:${PYTHONPATH}"

# Load OpenAI API key
set -a
source .env
set +a

echo "=== Reward Model Comparison Test ==="
echo "Local: Llama-3.1-8B-Instruct"
echo "API: GPT-4o-mini"
echo ""

python scripts/test_reward_comparison.py \
    --local-model models/llama-3.1-8b-instruct \
    --device cuda

echo ""
echo "=== Test Complete ==="