From d0a8934b7054fc20a5bc78f52fe05f33e89c9336 Mon Sep 17 00:00:00 2001
From: Houjun Liu <houjun@jemoka.com>
Date: Sun, 22 Oct 2023 23:09:09 -0700
Subject: [PATCH] skip sample if we hit a sample which has no content

---
 stanza/models/pos/data.py | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/stanza/models/pos/data.py b/stanza/models/pos/data.py
index 2c94377b57..c26e8d87da 100644
--- a/stanza/models/pos/data.py
+++ b/stanza/models/pos/data.py
@@ -1,3 +1,4 @@
+from os import set_inheritable
 import random
 import logging
 import torch
@@ -196,6 +197,13 @@ def __getitem__(self, key):
         # get each character from the input sentnece
         chars = [w for sent in char for w in sent]
 
+        # augmentation, etc. resulted in nothing; we return a different sample
+        if len(chars) == 0:
+            if key == 0:
+                return self[-1]
+            else: 
+                return self[key-1]
+
         return DataSample(words, chars, upos, xpos, ufeats, pretrained, sample[6]), key
 
     def __iter__(self):