roberta-tweet-eval-hate

This model is a fine-tuned version of roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.0170
• Accuracy: 0.799

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5042	1.0	563	0.5085	0.751
0.362	2.0	1126	0.4852	0.77
0.2926	3.0	1689	0.4871	0.793
0.2193	4.0	2252	0.7673	0.799
0.1855	5.0	2815	0.9488	0.775
0.1457	6.0	3378	1.2036	0.78
0.1239	7.0	3941	1.0635	0.808
0.0779	8.0	4504	1.2749	0.787
0.0583	9.0	5067	1.7393	0.766
0.0614	10.0	5630	1.5756	0.786
0.0427	11.0	6193	1.6751	0.788
0.0363	12.0	6756	1.7807	0.79
0.0239	13.0	7319	1.8339	0.788
0.0264	14.0	7882	1.8278	0.787
0.0234	15.0	8445	1.8558	0.794
0.0124	16.0	9008	1.8937	0.794
0.0105	17.0	9571	1.9635	0.792
0.0077	18.0	10134	1.9899	0.796
0.0115	19.0	10697	2.0302	0.794
0.0046	20.0	11260	2.0170	0.799

Framework versions

• Transformers 4.38.1
• Pytorch 2.1.1
• Datasets 2.14.5
• Tokenizers 0.15.2