Update README.md

The model was not in evaluation mode, providing non-deterministic and reproducible outputs.

README.md

@@ -59,6 +59,7 @@ Download and initialize the pre-trained model from HuggingFace Hub
 model = SegResEncoder.from_pretrained(
     "project-lighter/ct_fm_feature_extractor"
 )
+model.eval()
 ```
 
 ## Setup Processing Pipelines