agg_omni.sh 2.07 KB
Newer Older
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
#!/bin/bash
# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Launch script for testing vLLM-Omni integration with text-to-text generation
# This script starts an aggregated (frontend + omni worker) deployment for testing
set -e
trap 'echo Cleaning up...; kill 0' EXIT

# Default model - Qwen2.5-Omni-7B for text-to-text
MODEL="${MODEL:-Qwen/Qwen2.5-Omni-7B}"

# Stage config path - use single-stage LLM config for text-to-text
SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
STAGE_CONFIG="${STAGE_CONFIG:-$SCRIPT_DIR/stage_configs/single_stage_llm.yaml}"

# Parse command line arguments
EXTRA_ARGS=()
while [[ $# -gt 0 ]]; do
    case $1 in
        --model)
            MODEL="$2"
            shift 2
            ;;
        *)
            EXTRA_ARGS+=("$1")
            shift
            ;;
    esac
done

32
HTTP_PORT="${DYN_HTTP_PORT:-8000}"
33
echo "=========================================="
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
echo "Launching vLLM-Omni Text-to-Text (1 GPU)"
echo "=========================================="
echo "Model:       $MODEL"
echo "Frontend:    http://localhost:$HTTP_PORT"
echo "=========================================="
echo ""
echo "Example test command:"
echo ""
echo "  curl http://localhost:${HTTP_PORT}/v1/chat/completions \\"
echo "    -H 'Content-Type: application/json' \\"
echo "    -d '{"
echo "      \"model\": \"${MODEL}\","
echo "      \"messages\": [{\"role\": \"user\", \"content\": \"Explain why Roger Federer is considered one of the greatest tennis players of all time\"}],"
echo "      \"max_tokens\": 32"
echo "    }'"
echo ""
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
echo "=========================================="

# Run ingress (frontend)
# dynamo.frontend accepts either --http-port flag or DYN_HTTP_PORT env var (defaults to 8000)
python -m dynamo.frontend &
FRONTEND_PID=$!

# Wait a bit for frontend to start
sleep 2

echo "Starting Omni worker..."
DYN_SYSTEM_PORT=${DYN_SYSTEM_PORT:-8081} \
    python -m dynamo.vllm \
    --model "$MODEL" \
    --omni \
    --stage-configs-path "$STAGE_CONFIG" \
    "${EXTRA_ARGS[@]}"