simplified the dataset to avoid errors

prepare_data.py

@@ -36,21 +36,11 @@ for split in ["train", "test"]:
         df['judgement'] = get_dataset_column_from_text_folder(f"{path}/judgement")
         df['dataset_name'] = dataset_name
 
-        summary_full_path = f"{path}/summary"
-        if dataset_name == "UK-Abs":
-            if split == "test":
-                summary_full_path = f"{path}/summary/full"
-                for segment in ['background', 'judgement', 'reasons']:
-                    df[f'summary/{segment}'] = get_dataset_column_from_text_folder(
-                        f"{path}/summary/segment-wise/{segment}")
-        elif dataset_name == "IN-Ext":
+        if dataset_name == "UK-Abs" and split == "test" or dataset_name == "IN-Ext":
             summary_full_path = f"{path}/summary/full"
-            for annotator in ['A1', 'A2']:
-                for segment in ['facts', 'judgement']:  # errors when reading 'analysis' / 'argument' / 'statute'
-                    print(f"Processing {dataset_name} {split} {annotator} {segment}")
-                    df[f'summary/{annotator}/{segment}'] = get_dataset_column_from_text_folder(
-                        f"{path}/summary/segment-wise/{annotator}/{segment}")
-        df['summary/full'] = get_dataset_column_from_text_folder(summary_full_path)
+        else:
+            summary_full_path = f"{path}/summary"
+        df['summary'] = get_dataset_column_from_text_folder(summary_full_path)
         dfs.append(df)
     df = pd.concat(dfs)
     df = df.fillna("")  # NaNs can lead to huggingface not recognizing the feature type of the column

test.jsonl.xz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e56dd68c9d505c9c7850f32f9cd883a8e6737585f481c5aff35eb98623449098
-size 2323792
+oid sha256:b2ba3f5de08de7c70bd1157822b2224ef38b0ba54316267febf0e807f54d89e8
+size 2300856

train.jsonl.xz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:058ff61ae9c1c8f25c9bcaf3a33b3f472c74fde69b4aab7039121304e51f8ba5
-size 50582572
+oid sha256:779c6d28d3cc8b93132514988eb56ee5f464ce803553b4f8c9397c43ed3f5590
+size 50587108