start.sh

#!/bin/bash

echo "Starting vLLM server..."
# Start vLLM in the background with logging
python3 -m vllm.entrypoints.openai.api_server \
    --model numind/NuMarkdown-8B-Thinking \
    --port 8000 \
    --host 0.0.0.0 \
    --max-model-len 20000 \
    --gpu-memory-utilization 0.6 > $HOME/app/vllm.log 2>&1 &

VLLM_PID=$!
echo "vLLM started with PID: $VLLM_PID"

# Wait for vLLM to be ready
echo "Waiting for vLLM server to start..."
for i in {1..90}; do
    if curl -s http://localhost:8000/v1/models > /dev/null; then
        echo "vLLM server is ready!"
        break
    fi
    echo "Waiting... ($i/90)"
    sleep 2
done

# Check if vLLM is actually running
if ! curl -s http://localhost:8000/v1/models > /dev/null; then
    echo "ERROR: vLLM server failed to start!"
    echo "vLLM logs:"
    cat $HOME/app/vllm.log
    exit 1
fi

echo "Starting Gradio app..."
# Start Gradio app in the foreground
python3 $HOME/app/app.py