From dc801c07cf38b0c495686463e6ca6f871a64440e Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Tue, 27 Jan 2026 09:57:37 -0600 Subject: Add collaborativeagents module and update gitignore - Add collaborativeagents subproject with adapters, agents, and evaluation modules - Update .gitignore to exclude large binary files (.whl, .tar), wandb logs, and results Co-Authored-By: Claude Opus 4.5 --- collaborativeagents/slurm/run_experiments_a100.sh | 58 +++++++++++++++++++++++ 1 file changed, 58 insertions(+) create mode 100644 collaborativeagents/slurm/run_experiments_a100.sh (limited to 'collaborativeagents/slurm/run_experiments_a100.sh') diff --git a/collaborativeagents/slurm/run_experiments_a100.sh b/collaborativeagents/slurm/run_experiments_a100.sh new file mode 100644 index 0000000..aa2e658 --- /dev/null +++ b/collaborativeagents/slurm/run_experiments_a100.sh @@ -0,0 +1,58 @@ +#!/bin/bash +#SBATCH --job-name=run_expts_a100 +#SBATCH --account=bfqt-delta-gpu +#SBATCH --partition=gpuA100x4 +#SBATCH --nodes=1 +#SBATCH --ntasks=1 +#SBATCH --cpus-per-task=16 +#SBATCH --gres=gpu:nvidia_a100:4 +#SBATCH --mem=200G +#SBATCH --time=48:00:00 +#SBATCH --output=logs/run_expts_a100_%j.out +#SBATCH --error=logs/run_expts_a100_%j.err + +# Run experiments on 4x A100 80GB +# - 70B judge model with TP=4 (~140GB) +# - 8B PersonalizedLLM models (~16GB shared) + +set -e + +cd /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model + +# Create logs and results directories +mkdir -p collaborativeagents/slurm/logs +mkdir -p collaborativeagents/results + +echo "Starting experiments at $(date)" +echo "Job ID: $SLURM_JOB_ID" +echo "Node: $SLURMD_NODENAME" +echo "GPUs: $CUDA_VISIBLE_DEVICES" + +# Activate environment +source /u/yurenh2/miniforge3/etc/profile.d/conda.sh +conda activate eval + +# Check GPU availability +nvidia-smi + +# Redirect HF cache to project space (avoid home quota issues) +export HF_HOME=/projects/bfqt/users/yurenh2/.cache/huggingface +mkdir -p $HF_HOME + +# Add project to path +export PYTHONPATH="${PWD}/src:${PWD}/collaborativeagents:${PYTHONPATH}" + +# Run experiments +cd collaborativeagents/scripts + +# Full benchmark run +echo "Running full experiments..." +python run_experiments.py \ + --methods vanilla,all_memory,rag,rag_vector \ + --datasets math-500 \ + --n-profiles 20 \ + --n-sessions 5 \ + --profile-path ../data/complex_profiles_v2/profiles_100.jsonl \ + --output-dir ../results/full_$(date +%Y%m%d_%H%M%S) + +echo "Experiments completed at $(date)" -- cgit v1.2.3