blob: b11bba3ae456d1c246d2f0fdda8722794c77037f (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
|
#!/bin/bash
#SBATCH --job-name=scale_rag2
#SBATCH --account=bfqt-delta-gpu
#SBATCH --partition=gpuH200x8-interactive
#SBATCH --gres=gpu:4
#SBATCH --time=01:00:00
#SBATCH --mem=200G
#SBATCH --cpus-per-task=32
#SBATCH --output=%x-%j.out
#SBATCH --error=%x-%j.err
cd /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model
source ~/miniforge3/etc/profile.d/conda.sh
conda activate eval
export HF_HOME=/projects/bfqt/users/yurenh2/hf_cache/huggingface
export TRANSFORMERS_CACHE=/projects/bfqt/users/yurenh2/hf_cache/huggingface
export PYTHONPATH=/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model:$PYTHONPATH
PROFILE_PATH="collaborativeagents/data/complex_profiles_v2/profiles_200.jsonl"
AGENT_MODEL="/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/models/llama-3.1-8b-instruct"
USER_MODEL="meta-llama/Llama-3.1-70B-Instruct"
echo "=== Starting vLLM servers ==="
date
# User simulator on GPUs 0,1 (70B)
CUDA_VISIBLE_DEVICES=0,1 python -m vllm.entrypoints.openai.api_server \
--model $USER_MODEL \
--port 8004 --tensor-parallel-size 2 --gpu-memory-utilization 0.90 \
--max-model-len 16384 --dtype bfloat16 --download-dir $HF_HOME &
# Agent on GPUs 2,3 (8B) - lower memory for rag_vector (needs embedding/reranker)
CUDA_VISIBLE_DEVICES=2,3 python -m vllm.entrypoints.openai.api_server \
--model $AGENT_MODEL \
--port 8003 --tensor-parallel-size 2 --gpu-memory-utilization 0.40 \
--max-model-len 16384 --dtype bfloat16 &
# Wait for servers
echo "Waiting for vLLM servers..."
for i in {1..200}; do
if curl -s http://localhost:8004/health > /dev/null 2>&1; then
echo "User simulator (8004) ready after $((i*5)) seconds"
break
fi
sleep 5
done
for i in {1..60}; do
if curl -s http://localhost:8003/health > /dev/null 2>&1; then
echo "Agent (8003) ready after $((i*5)) seconds"
break
fi
sleep 5
done
echo "Both vLLM servers ready"
sleep 10
# Run rag_vector - only missing profiles 3,4 (0,1,2 already complete)
# Don't restrict CUDA devices - let PersonalizedLLM handle GPU assignment
python collaborativeagents/scripts/run_experiments.py \
--methods rag_vector \
--n-profiles 5 \
--n-sessions 15 \
--start-profile 3 \
--output-dir results/scale_test_remaining \
--profile-path $PROFILE_PATH \
--datasets math-hard \
--use-vllm --parallel-profiles 30 --no-batch-processing
pkill -f "vllm.entrypoints" 2>/dev/null || true
|
