lstm.CEBaB_confounding.uniform.absa.5-class.seed_44

This model is a fine-tuned version of lstm on the OpenTable OPENTABLE-ABSA dataset. It achieves the following results on the evaluation set:

• Loss: 0.6849
• Accuracy: 0.7060
• Macro-f1: 0.7017
• Weighted-macro-f1: 0.7077

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 44
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5.0

Training results

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.2+cu102
• Datasets 2.5.2
• Tokenizers 0.12.1