fix: removing empty documents

convert.py

@@ -37,7 +37,9 @@ def run():
             doc["label"] = class_list.index(class_name)
             doc["label_verbose"] = class_name
 
-            rows.append(doc)
+            # Skip empty documents
+            if doc["text"]:
+                rows.append(doc)
 
     save_csv(rows)
     split_csv_train_test(test_size=0.2, random_state=42)

test.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16b40efcd3c8f311720ac0608765fe70af84baaf1e44764b99b349a936a5e4a9
-size 7367423
+oid sha256:8ed74ba5d6bc002368772bf23bf7217d5d0937a1942d58cdf3ca7318e8917eef
+size 6873803

train.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea7277b826638ec95abf7079f8a2d418f160bcfe0ca4a6048f6e30e7fc24b12e
-size 29900177
+oid sha256:03df8bd00c19325a9cd249746f7289d509e3204004207f3a8acff96765bc0b76
+size 30392840