Update start.sh

start.sh

@@ -2,21 +2,18 @@
 set -e
 
 echo "Starting llama.cpp server..."
-# Direct path for the ghcr.io/ggml-org/llama.cpp:full image
-LLAMA_BIN="/app/llama-server"
-
-$LLAMA_BIN \
+# The binary is located at /app/llama-server in the ggml-org/llama.cpp image
+/app/llama-server \
   -m /app/SmolLM2-1.7B-Instruct-Q4_K_M.gguf \
   --host 0.0.0.0 \
   --port 8080 \
   -t 2 \
-  -c 4096 \
-  -n 2048 \
+  -c 2048 \
   --log-disable &
 
-echo "Waiting for llama.cpp to be ready..."
-sleep 15
+echo "Waiting for server to initialize..."
+sleep 10
 
-echo "Starting Gradio app using venv..."
-# Use the full path to the venv python to be 100% sure
+echo "Starting Gradio App..."
+# Explicitly use the venv's python
 /opt/venv/bin/python3 /app/app.py