Update README.md

README.md

@@ -64,10 +64,14 @@ Please consult the original DeBERTa paper and the papers for the different datas
 
 ## Metrics 
 
-Balanced accuracy metrics on all datasets. `deberta-v3-large-zeroshot-v1.1-heldout` indicates zeroshot performance on the respective dataset.
-To calculate this metrics, 28 different models were trained, each with one dataset held out from training to simulate a zeroshot setup. 
+Balanced accuracy metrics on all datasets. 
 `deberta-v3-large-zeroshot-v1.1-all-33` was trained on all datasets, with only maximum 500 texts per class to avoid overfitting. 
-(The metrics in the last column are therefore not strictly zeroshot.)
+The metrics on these datasets are therefore not strictly zeroshot, as the model has seen some data for each task. 
+`deberta-v3-large-zeroshot-v1.1-heldout` indicates zeroshot performance on the respective dataset.
+To calculate these zeroshot metrics, 28 models were trained, each with one dataset held out from training to simulate a zeroshot setup.
+The other setup (hyperparameters etc.) remained the same.
+
+![figure_large_v1.1](https://github.com/MoritzLaurer/zeroshot-classifier/blob/main/results/fig_large_v1.1.png)
 
 
 |                            |   deberta-v3-large-mnli-fever-anli-ling-wanli-binary |   deberta-v3-large-zeroshot-v1.1-heldout |   deberta-v3-large-zeroshot-v1.1-all-33 |