Chi Sun

Maybe you should check the differences between mBERT and BERT-base and modify the 'convert_tf_checkpoint_to_pytorch.py' file if necessary. GPU memory related issues reference: https://github.com/HSLCY/ABSA-BERT-pair/issues/1

Running the code (batch=24) may require three 16G GPUs. Reference: https://github.com/HSLCY/ABSA-BERT-pair/issues/1

The model was not saved, refer to https://github.com/HSLCY/ABSA-BERT-pair/issues/8 to save the checkpoints.

Hi, aspect acc is the strict acc defined in [1]:  [1] Yukun Ma, Haiyun Peng, and Erik Cambria. 2018. Targeted aspect-based sentiment analysis via embedding commonsense knowledge into an...

Did you use the "evaluation.py" file in the evaluation step? If not, you may use it: python evaluation.py --task_name sentihood_NLI_M --pred_data_dir results/sentihood/NLI_M/test_ep_4.txt

Yes, the constructed labels are imbalanced. BERT's generalization ability is strong, and it can learn key information. The model will not just output "none". However, when the number of aspects...