Use XLM-R for low resource languages.

scripts/utils.py

@@ -18,20 +18,13 @@ LANG2TRANSFORMER = {
     "it": "dbmdz/bert-base-italian-cased",
     "ru": "blinoff/roberta-base-russian-v0",
     "sv": "KB/bert-base-swedish-cased",
-    "uk": "/tmp/lustre_shared/mklimasz/transformers/wikibert-base-uk-cased/",
+    "uk": "xlm-roberta-large",
     "ta": "xlm-roberta-large",
-    "sk": "/tmp/lustre_shared/mklimasz/transformers/wikibert-base-sk-cased/",
-    "lt": "/tmp/lustre_shared/mklimasz/transformers/wikibert-base-lt-cased/",
-    "lv": "/tmp/lustre_shared/mklimasz/transformers/wikibert-base-lv-cased/",
-    "cs": "/tmp/lustre_shared/mklimasz/transformers/wikibert-base-cs-cased/",
-    "et": "/tmp/lustre_shared/mklimasz/transformers/etwiki-bert/",
-    # "uk": http://dl.turkunlp.org/wikibert/wikibert-base-uk-cased/
-    # "ta": http://dl.turkunlp.org/wikibert/wikibert-base-ta-cased/
-    # "sk": http://dl.turkunlp.org/wikibert/wikibert-base-sk-cased/
-    # "lt": http://dl.turkunlp.org/wikibert/wikibert-base-lt-cased/
-    # "lv": http://dl.turkunlp.org/wikibert/wikibert-base-lv-cased/
-    # "et": http://dl.turkunlp.org/estonian-bert/etwiki-bert/pytorch/
-    # "cs": https://github.com/kiv-air/Czert https://arxiv.org/pdf/2103.13031.pdf
+    "sk": "xlm-roberta-large",
+    "lt": "xlm-roberta-large",
+    "lv": "xlm-roberta-large",
+    "cs": "xlm-roberta-large",
+    "et": "xlm-roberta-large",
 }