Update modeling_ltgbert.py
#1
by
KoichiYasuoka
- opened
When initializing LtgbertForTokenClassification several LayerNorms don't have weight or bias as mentioned in https://huggingface.co/HPLT/hplt_bert_base_ja/discussions/1
KoichiYasuoka
changed pull request status to
closed