TehranNLP-org/bert-large-hateXplain

SEED0042

This model is a fine-tuned version of bert-large-uncased on the HATEXPLAIN dataset. It achieves the following results on the evaluation set:

• Loss: 0.7731
• Accuracy: 0.4079
• Accuracy 0: 0.8027
• Accuracy 1: 0.1869
• Accuracy 2: 0.2956

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: 5e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: not_parallel
• gradient_accumulation_steps: 32
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 150
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Accuracy 0	Accuracy 1	Accuracy 2
No log	1.0	480	0.8029	0.4235	0.7589	0.0461	0.5985
No log	2.0	960	0.7574	0.4011	0.7470	0.1831	0.3376
No log	3.0	1440	0.7731	0.4079	0.8027	0.1869	0.2956

Framework versions

• Transformers 4.17.0
• Pytorch 1.10.0+cu113
• Datasets 2.1.0
• Tokenizers 0.11.6