temp-disable quantisation

src/server/gemma_llm.py

@@ -39,7 +39,7 @@ class LLMManager:
                 self.model = Gemma3ForConditionalGeneration.from_pretrained(
                     self.model_name,
                     device_map="auto",
-                    quantization_config=quantization_config,
+                    #quantization_config=quantization_config,
                     torch_dtype=self.torch_dtype
                 ).eval()
                 self.processor = AutoProcessor.from_pretrained(self.model_name)