Upload scripts/start_space.sh with huggingface_hub

scripts/start_space.sh

@@ -0,0 +1,71 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+MODEL_ID="${LLAMA_CPP_MODEL:-CohereLabs/tiny-aya-global-GGUF}"
+MODEL_FILE="${LLAMA_CPP_FILE:-*Q4_K_M.gguf}"
+LLAMA_PORT="${LLAMA_CPP_PORT:-8080}"
+APP_PORT="${PORT:-${GRADIO_SERVER_PORT:-7860}}"
+
+export GRADIO_SERVER_NAME="${GRADIO_SERVER_NAME:-0.0.0.0}"
+export GRADIO_SERVER_PORT="$APP_PORT"
+export LLAMA_CPP_PORT="$LLAMA_PORT"
+
+# ── Install llama.cpp ───────────────────────────────────────────────
+echo "Installing llama.cpp..."
+pip install llama-cpp-python[server] 2>/dev/null || true
+
+# Try to get llama-server binary
+if ! command -v llama-server &>/dev/null; then
+    echo "llama-server binary not found, trying to install..."
+    pip install llama-cpp-python 2>/dev/null || true
+    # Fallback: download pre-built binary
+    curl -sL https://github.com/ggml-org/llama.cpp/releases/latest/download/llama-server-linux-x64 \
+        -o /usr/local/bin/llama-server 2>/dev/null || true
+    chmod +x /usr/local/bin/llama-server 2>/dev/null || true
+fi
+
+# ── Start llama.cpp server ──────────────────────────────────────────
+echo "Starting llama.cpp server for ${MODEL_ID} (file: ${MODEL_FILE}) on port ${LLAMA_PORT}"
+
+llama-server \
+    -hf "${MODEL_ID}:${MODEL_FILE}" \
+    --port "$LLAMA_PORT" \
+    -ngl 99 \
+    --ctx-size 4096 \
+    --no-warmup \
+    &
+
+LLAMA_PID=$!
+
+# ── Cleanup on exit ─────────────────────────────────────────────────
+cleanup() {
+    echo "Shutting down..."
+    kill "$LLAMA_PID" 2>/dev/null || true
+}
+trap cleanup EXIT
+
+# ── Wait for llama.cpp to be ready ──────────────────────────────────
+echo "Waiting for llama.cpp server to become ready..."
+for attempt in $(seq 1 120); do
+    if curl -sf "http://localhost:${LLAMA_PORT}/health" >/dev/null 2>&1; then
+        echo "llama.cpp server is ready (attempt ${attempt})."
+        break
+    fi
+
+    if ! kill -0 "$LLAMA_PID" 2>/dev/null; then
+        echo "llama.cpp server exited prematurely."
+        wait "$LLAMA_PID" 2>/dev/null || true
+        break
+    fi
+
+    if [ "$attempt" -eq 120 ]; then
+        echo "Timed out waiting for llama.cpp server. Starting app anyway (fallback mode)."
+        break
+    fi
+
+    sleep 2
+done
+
+# ── Start Gradio app ────────────────────────────────────────────────
+echo "Starting Gradio app on ${GRADIO_SERVER_NAME}:${APP_PORT}"
+exec /opt/neighbourhood-guide/bin/python /app/app.py