Create utils/inference.py

utils/inference.py

@@ -0,0 +1,6 @@
+from huggingface_hub import InferenceClient
+
+client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.2")
+
+def call_model(prompt: str) -> str:
+    return client.text_generation(prompt, max_new_tokens=2048, temperature=0.3, return_full_text=False)