AIencoder/Axon-Llama-GUI

1#!/bin/bash2set -e3 4echo "================================================"5echo "🚀 Axon - Qwen2.5-Omni-7B Multimodal Server"6echo "================================================"7echo ""8echo "📋 Capabilities: Text | Images | Audio"9echo "🔧 Quantization: Q8_0 (near-lossless)"10echo "⚡ Optimizations: Flash Attention, Continuous Batching"11echo ""12 13mkdir -p /app/models/qwen2.5-omni-7b14 15download_file () {16    REPO_ID=$117    FILENAME=$218    DEST_PATH=$319 20    if [ -f "$DEST_PATH" ]; then21        echo "✅ Already exists: $(basename $DEST_PATH)"22        return 023    fi24 25    echo "⬇️  Downloading $FILENAME ..."26    python3 -c "27from huggingface_hub import hf_hub_download28import shutil, sys29try:30    path = hf_hub_download(31        repo_id='$REPO_ID',32        filename='$FILENAME',33        cache_dir='/app/.cache'34    )35    shutil.copy(path, '$DEST_PATH')36    print(f'✅ Downloaded: $FILENAME')37except Exception as e:38    print(f'❌ Error downloading $FILENAME: {e}')39    sys.exit(1)40"41}42 43download_file "ggml-org/Qwen2.5-Omni-7B-GGUF" \44    "Qwen2.5-Omni-7B-Q8_0.gguf" \45    "/app/models/qwen2.5-omni-7b/Qwen2.5-Omni-7B-Q8_0.gguf"46 47download_file "ggml-org/Qwen2.5-Omni-7B-GGUF" \48    "mmproj-Qwen2.5-Omni-7B-Q8_0.gguf" \49    "/app/models/qwen2.5-omni-7b/mmproj-Qwen2.5-Omni-7B-Q8_0.gguf"50 51echo ""52echo "🚀 Starting llama.cpp Server"53echo "🌐 Server will be available at http://0.0.0.0:7860"54echo ""55 56exec /usr/local/bin/llama-server \57    --models-dir /app/models \58    --host 0.0.0.0 \59    --port 7860 \60    -c 8192 \61    -t 4 \62    -fa on \63    -cb \64    --n-gpu-layers 0