fix: download model at container startup, not build time

start.sh

@@ -1,8 +1,25 @@
 #!/bin/bash
 set -e
 
-MODEL_PATH="/home/user/app/models/bitnet-b1.58-2B-4T-gguf/ggml-model-i2_s.gguf"
-SERVER_BIN="/home/user/app/bin/llama-server"
+MODEL_DIR="/home/user/app/models"
+MODEL_PATH="$MODEL_DIR/ggml-model-i2_s.gguf"
+SERVER_BIN="/home/user/app/llama-server"
+
+# Download model if not present (runtime download to avoid build timeout)
+if [ ! -f "$MODEL_PATH" ]; then
+    echo "Downloading BitNet b1.58 2B4T GGUF model (1.1 GB)..."
+    python -c "
+from huggingface_hub import hf_hub_download
+import os
+path = hf_hub_download(
+    repo_id='microsoft/bitnet-b1.58-2B-4T-gguf',
+    filename='ggml-model-i2_s.gguf',
+    local_dir='$MODEL_DIR'
+)
+print(f'Downloaded to: {path}')
+"
+    echo "Model downloaded!"
+fi
 
 # Start llama-server in background
 echo "Starting bitnet.cpp llama-server..."
@@ -18,11 +35,15 @@ SERVER_PID=$!
 
 # Wait for server to be ready
 echo "Waiting for server to start..."
-for i in $(seq 1 60); do
+for i in $(seq 1 120); do
     if curl -s http://127.0.0.1:8080/health > /dev/null 2>&1; then
-        echo "Server ready!"
+        echo "Server ready! (took ${i}s)"
         break
     fi
+    if [ $i -eq 120 ]; then
+        echo "ERROR: Server failed to start after 120s"
+        exit 1
+    fi
     sleep 1
 done