Update start.sh

start.sh

@@ -12,10 +12,13 @@ vllm serve unsloth/llama-3-8b-Instruct-bnb-4bit \
   --enable-auto-tool-choice \
   --tool-call-parser llama3_json \
   --chat-template examples/tool_chat_template_llama3.1_json.jinja \
-  --dtype half &
+  --quantization bitsandbytes \
+  --load-format bitsandbytes \
+  --dtype half \
+  --enforce-eager &
 
 # Wait to ensure the vLLM server is fully started (adjust if needed)
 sleep 10
 
 # Start the Gradio application using python3
-python3 app.py
+python3 app.py