workspace/scripts/slurm/generate_embeddings.sbatch

#!/bin/bash
#SBATCH --job-name=gen_embeddings
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=8
#SBATCH --mem=16000M
#SBATCH --time=1:00:00
#SBATCH --output=logs/gen_embeddings_%A.out
#SBATCH --error=logs/gen_embeddings_%A.err

set -euo pipefail

PROJECT_DIR="${PROJECT_DIR:-$PWD}"
cd "$PROJECT_DIR"

source .venv/bin/activate

echo "=== Generating Instruction Embeddings (Fast Parallel) ==="
echo "Using 8 CPU cores for parallel encoding"
echo ""

python scripts/generate_instruction_embeddings.py \
    --dataset /scratch/$USER/dovla/experiments/maniskill_presuccess_six_task_collection \
    --output /scratch/$USER/dovla/experiments/instruction_embeddings.pkl \
    --cache-dir /scratch/$USER/dovla/experiments/embedding_cache

echo ""
echo "✅ Embeddings generated successfully"