cloudunity/Ornith-api

1#!/bin/bash2set -e3 4MODEL_DIR="/data/models"5 6mkdir -p "$MODEL_DIR"7 8MODEL_REPO="ornith-ai/Ornith-1.5-9B-GGUF"9MODEL_FILE="Ornith-1.5-9B-Q4_K_M.gguf"10 11echo "=============================================="12echo " Ornith-1.5-9B"13echo " CPU / Q4_K_M / Long Context"14echo "=============================================="15 16echo "Downloading model..."17 18python - <<PY19from huggingface_hub import hf_hub_download, list_repo_files20 21files = list_repo_files("${MODEL_REPO}")22expected = "${MODEL_FILE}"23if expected not in files:24    print(f"ERROR: {expected} not found in repo. Available GGUF files:")25    for f in sorted(files):26        if f.endswith(".gguf"):27            print(f"  {f}")28    raise SystemExit(1)29 30path = hf_hub_download(31    repo_id="${MODEL_REPO}",32    filename="${MODEL_FILE}",33    local_dir="${MODEL_DIR}",34)35 36print("Downloaded:")37print(path)38PY39 40MODEL="${MODEL_DIR}/${MODEL_FILE}"41 42if [ ! -f "$MODEL" ]; then43    echo "ERROR: Model not found:"44    echo "$MODEL"45    echo46    echo "Files in model directory:"47    ls -lah "$MODEL_DIR"48    exit 149fi50 51echo52echo "Model:"53ls -lh "$MODEL"54 55echo56echo "Starting llama-server..."57 58export LD_LIBRARY_PATH="/opt/llama${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}"59 60exec /opt/llama/llama-server \61    --model "$MODEL" \62    --host 0.0.0.0 \63    --port 7860 \64    --ctx-size 32768 \65    --threads 2 \66    --threads-batch 2 \67    --batch-size 512 \68    --ubatch-size 128 \69    --cache-type-k q8_0 \70    --cache-type-v q8_0 \71    --flash-attn on \72    --cont-batching73 2>/dev/null || \74/opt/llama/llama-server \75    --model "$MODEL" \76    --host 0.0.0.0 \77    --port 7860 \78    --ctx-size 32768 \79    --threads 2 \80    --threads-batch 2 \81    --batch-size 512 \82    --ubatch-size 128 \83    --cache-type-k q8_0 \84    --cache-type-v q8_0 \85    --cont-batching86