code/slurm_logs/eval_pairwise_selector.sbatch

#!/bin/bash
#SBATCH --job-name=vl
#SBATCH --partition=gpu,gpu-large
#SBATCH --qos=batch-long
#SBATCH --gres=gpu:1
#SBATCH --cpus-per-task=8
#SBATCH --mem=80G
#SBATCH --time=04:00:00
#SBATCH --output=/weka/s225250685/mats-tist/slurm_logs/eval_pairwise_selector_%j.out

# Serve the Llama-3.2-3B pairwise selector + run tournament eval via
# data_processing/planner.py::SelectionAgentWithSchema.get_best_sql.

set -u
cd /weka/s225250685/mats-tist
set -a; source /weka/s225250685/mats-tist/.env; set +a
export HF_HOME=/weka/s225250685/Huggingface HF_HUB_CACHE=/weka/s225250685/Huggingface/hub
export PYTHONNOUSERSITE=1 NO_PROXY=localhost,127.0.0.1 no_proxy=localhost,127.0.0.1
export PYTHONPATH=/weka/s225250685/mats-tist TOKENIZERS_PARALLELISM=false
export DB_EXEC_API_DISABLE=1

PY=/weka/s225250685/conda-envs/handbook/bin/python
VLLM=/weka/s225250685/conda-envs/handbook/bin/vllm

SELECTOR_CKPT="${SELECTOR_CKPT:-alignment-handbook/output/selector-llama3b-pairwise-synsql}"
LABEL="${LABEL:-llama3b_synsql_paper_SFT_VF}"
DEV_FILE="${DEV_FILE:-eval_results/paper_SFT_VF_passAt8_bird_dev.jsonl}"

LOG=/weka/s225250685/mats-tist/slurm_logs/eval_pairwise_selector_${SLURM_JOB_ID}.log
kill_vllm() { pkill -9 -f "vllm serve" 2>/dev/null || true; sleep 5; }
trap kill_vllm EXIT
wait_url() { for i in {1..360}; do curl --noproxy '*' -fs "$1" >/dev/null 2>&1 && return 0; sleep 5; done; return 1; }

echo "[$(date)] Serving $SELECTOR_CKPT..." | tee -a "$LOG"
$VLLM serve "$SELECTOR_CKPT" \
    --served-model-name selection --port 8103 \
    --dtype bfloat16 --gpu-memory-utilization 0.85 \
    --max-model-len 4096 --enforce-eager --disable-log-requests \
    > "${LOG}.serve" 2>&1 &
wait_url http://localhost:8103/v1/models || { echo "selector failed" | tee -a "$LOG"; exit 1; }
echo "[$(date)] selector ready" | tee -a "$LOG"

echo "[$(date)] Running pairwise tournament eval (max_candidates=2)..." | tee -a "$LOG"
$PY scripts/eval_pairwise_selector.py \
    "$DEV_FILE" "$LABEL" \
    --selector_host http://localhost:8103 --model_name selection \
    --max_candidates 2 --threads 16 \
    --out_dir eval_results \
    2>&1 | tee -a "${LOG}.eval"

echo "[$(date)] DONE" | tee -a "$LOG"
kill_vllm