Update model card

README.md

@@ -47,9 +47,9 @@ from ultra.eval import test
 model = UltraLinkPrediction.from_pretrained("mgalkin/ultra_3g")
 dataset = CoDExSmall(root="./datasets/")
 test(model, mode="test", dataset=dataset, gpus=None)
-# Expected results
-# mrr:      0.472035
-# hits@10:  0.66849
+# Expected results for ULTRA 3g
+# mrr:      0.472
+# hits@10:  0.668
 ```
 
 * You can also **fine-tune** ULTRA on each graph, please refer to the [github repo](https://github.com/DeepGraphLearning/ULTRA#run-inference-and-fine-tuning) for more details on training / fine-tuning   
@@ -102,6 +102,8 @@ test(model, mode="test", dataset=dataset, gpus=None)
     </tr>
 </table>
 
+Fine-tuning ULTRA on specific graphs brings, on average, further 10% relative performance boost both in MRR and Hits@10. See the paper for more comparisons.
+
 **ULTRA 50g Performance**
 
 ULTRA 50g was pre-trained on 50 graphs, so we can't really apply the zero-shot evaluation protocol to the graphs.