code/slurm_logs/mega_orpo_valfix.sbatch

#!/bin/bash
#SBATCH --job-name=vl
#SBATCH --partition=gpu-large
#SBATCH --qos=batch-long
#SBATCH --gres=gpu:1
#SBATCH --cpus-per-task=2
#SBATCH --mem=80G
#SBATCH --time=10:00:00
#SBATCH --output=/weka/s225250685/mats-tist/slurm_logs/orpo_valfix_%j.out

set -u
cd /weka/s225250685/mats-tist

export HF_HOME=/weka/s225250685/Huggingface
export HF_HUB_CACHE=/weka/s225250685/Huggingface/hub
export DB_EXEC_API_DISABLE=1
export PYTHONNOUSERSITE=1
export NO_PROXY=localhost,127.0.0.1
export PYTHONPATH=/weka/s225250685/mats-tist
export TOKENIZERS_PARALLELISM=false

PY=/weka/s225250685/conda-envs/handbook/bin/python
VLLM=/weka/s225250685/conda-envs/handbook/bin/vllm
ACCEL=/weka/s225250685/conda-envs/handbook/bin/accelerate
AH=/weka/s225250685/mats-tist/alignment-handbook

PLANNER=/weka/s225250685/mats-tist/alignment-handbook/output/planner-iter2-collab-3B
SEL_V2=/weka/s225250685/mats-tist/alignment-handbook/output/selector-3B-v2-rows
FIXER_V2_ORPO=$AH/output/fixer-v2-1.5B-execerr-orpo-expanded
SEM_FIXER_ORPO=$AH/output/semantic-fixer-v3-1.5B-orpo-nogate
VALIDATOR_ORPO=$AH/output/validator-v4-0.5B-orpo

LOG=/weka/s225250685/mats-tist/slurm_logs/orpo_valfix_${SLURM_JOB_ID}.log
: > "$LOG"

nvidia-smi --query-gpu=name,memory.total --format=csv,noheader | tee -a "$LOG"

kill_vllm() {
    pkill -9 -f "vllm serve" 2>/dev/null || true
    pkill -9 -f "VLLM::EngineCore" 2>/dev/null || true
    sleep 5
}
trap kill_vllm EXIT

wait_url() {
    for i in {1..180}; do
        curl --noproxy '*' -fs "$1" >/dev/null 2>&1 && return 0
        sleep 5
    done
    return 1
}

##############################################
# STAGE A: Build ORPO training data
##############################################
echo "==== [A] building validator v4 ORPO data ====" | tee -a "$LOG"
$PY scripts/build_validator_v4_orpo.py 2>&1 | tee -a "$LOG"

echo "==== [A] building semantic fixer v3 data ====" | tee -a "$LOG"
$PY scripts/build_semantic_fixer_v3.py 2>&1 | tee -a "$LOG"

echo "==== [A] rebuilding fixer v2 data (with real exec errors) ====" | tee -a "$LOG"
$PY scripts/build_fixer_v2_execerr.py 2>&1 | tee -a "$LOG"

echo "==== [A] all data ready ====" | tee -a "$LOG"

##############################################
# STAGE B: ORPO — Fixer v2 (exec-error, 1.5B)
#   chosen = correct SQL, rejected = failed SQL
##############################################
if [ -f "$FIXER_V2_ORPO/config.json" ]; then
    echo "==== [B] fixer v2 ORPO already exists, skipping training ====" | tee -a "$LOG"
else
    echo "==== [B] ORPO training: fixer v2 exec-error (1.5B) ====" | tee -a "$LOG"
    cd $AH
    PYTHONPATH=src/ ACCELERATE_LOG_LEVEL=info $ACCEL launch \
        --main_process_port 29601 \
        --config_file recipes/accelerate_configs/single_gpu0_local.yaml \
        scripts/run_orpo.py \
        recipes/scaleup-3stage/orpo-fixer-v2-execerr.yaml 2>&1 | tee -a "$LOG"
    cd /weka/s225250685/mats-tist
    if [ ! -f "$FIXER_V2_ORPO/config.json" ]; then
        echo "FIXER V2 ORPO FAILED" | tee -a "$LOG"; exit 1
    fi
fi
echo "==== [B] fixer v2 ORPO ready at $FIXER_V2_ORPO ====" | tee -a "$LOG"

##############################################
# STAGE C: ORPO — Validator v4 (0.5B, combined sel+cond)
##############################################
if [ -f "$VALIDATOR_ORPO/config.json" ]; then
    echo "==== [C] validator v4 ORPO already exists, skipping training ====" | tee -a "$LOG"
else
    echo "==== [C] ORPO training: validator v4 combined (0.5B) ====" | tee -a "$LOG"
    cd $AH
    PYTHONPATH=src/ ACCELERATE_LOG_LEVEL=info $ACCEL launch \
        --main_process_port 29602 \
        --config_file recipes/accelerate_configs/single_gpu0_local.yaml \
        scripts/run_orpo.py \
        recipes/scaleup-3stage/orpo-validator-v4.yaml 2>&1 | tee -a "$LOG"
    cd /weka/s225250685/mats-tist
    if [ ! -f "$VALIDATOR_ORPO/config.json" ]; then
        echo "VALIDATOR V4 ORPO FAILED" | tee -a "$LOG"; exit 1
    fi
fi
echo "==== [C] validator v4 ORPO ready at $VALIDATOR_ORPO ====" | tee -a "$LOG"

##############################################
# STAGE D: ORPO — Semantic Fixer v3 (1.5B)
##############################################
if [ -f "$SEM_FIXER_ORPO/config.json" ]; then
    echo "==== [D] semantic fixer v3 ORPO already exists, skipping training ====" | tee -a "$LOG"
else
    echo "==== [D] ORPO training: semantic fixer v3 (1.5B) ====" | tee -a "$LOG"
    cd $AH
    PYTHONPATH=src/ ACCELERATE_LOG_LEVEL=info $ACCEL launch \
        --main_process_port 29603 \
        --config_file recipes/accelerate_configs/single_gpu0_local.yaml \
        scripts/run_orpo.py \
        recipes/scaleup-3stage/orpo-semantic-fixer-v3.yaml 2>&1 | tee -a "$LOG"
    cd /weka/s225250685/mats-tist
    if [ ! -f "$SEM_FIXER_ORPO/config.json" ]; then
        echo "SEMANTIC FIXER V3 ORPO FAILED" | tee -a "$LOG"; exit 1
    fi
fi
echo "==== [D] semantic fixer v3 ORPO ready at $SEM_FIXER_ORPO ====" | tee -a "$LOG"

##############################################
# STAGE E: Launch endpoints + 3-stage K=8 rollout
#   H200=141GB: planner(0.30) + val(0.08) + fixer_v2(0.15) + sem_fixer(0.15) = 0.68
#   One combined validator serves both sel+cond on same port
##############################################
kill_vllm
echo "==== [E] launching endpoints ====" | tee -a "$LOG"

$VLLM serve "$PLANNER" --served-model-name planner --port 8100 --dtype bfloat16 \
    --gpu-memory-utilization 0.30 --enforce-eager --max-model-len 8192 > "${LOG}.p" 2>&1 &
wait_url http://localhost:8100/v1/models && echo "  planner READY" | tee -a "$LOG"

# Combined validator v4 serves both sel and cond prompts
$VLLM serve "$VALIDATOR_ORPO" --served-model-name validator_sel --port 8101 --dtype bfloat16 \
    --gpu-memory-utilization 0.08 --enforce-eager --max-model-len 6144 > "${LOG}.vs" 2>&1 &
wait_url http://localhost:8101/v1/models && echo "  validator_v4 READY (port 8101)" | tee -a "$LOG"

$VLLM serve "$VALIDATOR_ORPO" --served-model-name validator_cond --port 8104 --dtype bfloat16 \
    --gpu-memory-utilization 0.08 --enforce-eager --max-model-len 6144 > "${LOG}.vc" 2>&1 &
wait_url http://localhost:8104/v1/models && echo "  validator_v4 READY (port 8104)" | tee -a "$LOG"

$VLLM serve "$FIXER_V2_ORPO" --served-model-name fixer --port 8102 --dtype bfloat16 \
    --gpu-memory-utilization 0.15 --enforce-eager --max-model-len 4096 > "${LOG}.f" 2>&1 &
wait_url http://localhost:8102/v1/models && echo "  fixer_v2_orpo READY" | tee -a "$LOG"

$VLLM serve "$SEM_FIXER_ORPO" --served-model-name fixer_v3 --port 8105 --dtype bfloat16 \
    --gpu-memory-utilization 0.15 --enforce-eager --max-model-len 4096 > "${LOG}.fs" 2>&1 &
wait_url http://localhost:8105/v1/models && echo "  sem_fixer_v3_orpo READY" | tee -a "$LOG"

OUT_ORPO=eval_results/scaleup_BoN8_d_K8_3stage_planner_iter2_mixedtemp_valv4orpo_fixerv2orpo_semfixerv3orpo_bird_dev.jsonl
rm -f "$OUT_ORPO"

echo "==== [E] K=8 3-stage: val_v4 ORPO + fixer_v2 ORPO + sem_fixer_v3 ORPO ====" | tee -a "$LOG"
$PY scripts/run_pipeline_rollouts.py \
    --input_file data/sft_bird_with_evidence_dev_text2sql.json \
    --output_file "$OUT_ORPO" \
    --planner_host http://localhost:8100 \
    --validator_host none \
    --validator_sel_host http://localhost:8101 \
    --validator_cond_host http://localhost:8104 \
    --fixer_host http://localhost:8102 \
    --fixer_gate_exec_ok \
    --fixer_v3_host http://localhost:8105 \
    --K 8 --K_val 1 --K_fix 1 \
    --temperature 1.0 --top_p 0.9 \
    --max_planner_tokens 1024 --max_validator_tokens 384 --max_fixer_tokens 512 \
    --max_questions -1 --n_threads 4 2>&1 | tee -a "$LOG"

echo "==== [E] metrics ====" | tee -a "$LOG"
$PY scripts/compute_bestofn_metrics.py "$OUT_ORPO" orpo_valv4_fixerv2_semfixerv3 2>&1 | tee -a "$LOG"

##############################################
# STAGE F: Apply selector v2
##############################################
kill_vllm
echo "==== [F] launching selector v2 ====" | tee -a "$LOG"
$VLLM serve "$SEL_V2" --served-model-name selector --port 8103 --dtype bfloat16 \
    --gpu-memory-utilization 0.85 --enforce-eager --max-model-len 8192 > "${LOG}.sel" 2>&1 &
wait_url http://localhost:8103/v1/models && echo "  selector READY" | tee -a "$LOG"

label="$(basename $OUT_ORPO .jsonl)_selectorV2rows"
$PY scripts/compute_bestofn_with_selector.py \
    "$OUT_ORPO" "$label" --selector_host http://localhost:8103 --row_preview 2>&1 | tee -a "$LOG"

echo "==== ALL_DONE ====" | tee -a "$LOG"