#!/bin/bash

num_images=25

model=models/steve_7b_sft

model_name=cot_qwen2vl

# model=models/UI-TARS-7B-DPO

# model_name=ui-tars

# num_images=5

port=9000

ground_model=models/UI-TARS-7B-DPO

# Function to clean up processes on exit

cleanup() {

echo "Stopping all processes..."

pkill -P $$ # Kill all child processes of this script

exit 0

}

# Trap SIGINT (Ctrl+C) and SIGTERM to run cleanup function

trap cleanup SIGINT SIGTERM

# Start processes

for i in {0..6}; do

CUDA_VISIBLE_DEVICES=$i python -m vllm.entrypoints.openai.api_server \

--served-model-name $model_name \

--model $model \

--limit-mm-per-prompt image=$num_images \

--chat-template ./chat_template_qwen2vl.jinja \

-tp=1 \

--port $((9000 + i)) &

# --chat-template $model/chat_template.json\ # wrong, use vllm default with low Temperature

done

CUDA_VISIBLE_DEVICES=7 python -m vllm.entrypoints.openai.api_server \

--served-model-name ui-tars \

--model $ground_model \

--chat-template ./chat_template_qwen2vl.jinja \

-tp=1 \

--port 8001 &

# Wait to keep the script running

wait