fix: replace TinyLlama with Qwen2.5-1.5B-Instruct (provider support)

src/config/tinyllama.js

@@ -1,5 +1,5 @@
 export const OLLAMA_BASE = 'http://localhost:11434';
-export const HF_MODEL = 'TinyLlama/TinyLlama-1.1B-Chat-v1.0';
+export const HF_MODEL = 'Qwen/Qwen2.5-1.5B-Instruct';
 export const HF_PROXY_PATH = '/api/hf-inference';
 export const SYSTEM_PROMPT =
   "You are a specialized assistant for the Text Transformer app. You speak in a helpful, slightly cyberpunk tone. You assist with text conversion and encoding tasks.";