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 <noreply@anthropic.com>

1 files changed, 59 insertions, 0 deletions

diff --git a/collaborativeagents/scripts/exp_all_memory.sbatch b/collaborativeagents/scripts/exp_all_memory.sbatch
new file mode 100644
index 0000000..c6310ee
--- /dev/null
+++ b/collaborativeagents/scripts/exp_all_memory.sbatch
@@ -0,0 +1,59 @@
+#!/bin/bash
+#SBATCH --job-name=exp_all_memory
+#SBATCH --account=bfqt-delta-gpu
+#SBATCH --partition=gpuH200x8
+#SBATCH --gres=gpu:4
+#SBATCH --nodes=1
+#SBATCH --ntasks=1
+#SBATCH --cpus-per-task=32
+#SBATCH --mem=256G
+#SBATCH --time=24:00:00
+#SBATCH --output=/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/exp_all_memory-%j.out
+#SBATCH --error=/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/exp_all_memory-%j.err
+
+set -e
+cd /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/collaborativeagents
+source /u/yurenh2/miniforge3/etc/profile.d/conda.sh
+conda activate eval
+
+export HF_HOME=/projects/bfqt/users/yurenh2/hf_cache/huggingface
+export PYTHONPATH="${PWD}:${PWD}/scripts:${PWD}/../src:${PYTHONPATH}"
+
+MODEL_8B="/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/models/llama-3.1-8b-instruct"
+PROFILE_PATH="/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/collaborativeagents/data/complex_profiles_v2/profiles_200.jsonl"
+
+echo "=== all_memory (PersonalizedLLMAdapter with local transformers) ==="
+date
+nvidia-smi --query-gpu=index,name,memory.total --format=csv
+
+pkill -f "vllm.entrypoints" 2>/dev/null || true
+sleep 2
+
+# GPU 0,1: vLLM server for user simulator (port 8004)
+# PersonalizedLLMAdapter uses local transformers for embedding/reranker/chat/extractor
+echo "Starting user simulator vLLM server on GPU 0,1 (port 8004)..."
+CUDA_VISIBLE_DEVICES=0,1 python -m vllm.entrypoints.openai.api_server \
+    --model $MODEL_8B --port 8004 --tensor-parallel-size 2 \
+    --gpu-memory-utilization 0.90 --max-model-len 8192 \
+    --disable-log-requests --dtype bfloat16 &
+
+echo "Waiting for vLLM server..."
+for i in $(seq 1 200); do
+    if curl -s http://localhost:8004/health > /dev/null 2>&1; then
+        echo "Server ready after $((i*3))s"; break
+    fi
+    sleep 3
+done
+
+# GPU 2,3: PersonalizedLLMAdapter's transformers models
+# (embedding ~8B, reranker ~8B, chat ~1.5B, extractor ~0.6B)
+CUDA_VISIBLE_DEVICES=2,3 python scripts/run_experiments.py \
+    --methods all_memory \
+    --datasets math-hard,math-500,bigcodebench \
+    --n-profiles 200 --n-sessions 30 --max-turns 15 \
+    --use-vllm --parallel-profiles 10 \
+    --output-dir ../results/full_h200 \
+    --profile-path "$PROFILE_PATH"
+
+pkill -f "vllm.entrypoints" 2>/dev/null || true
+echo "Done: $(date)"