twitter-roberta-base_3epoch10.64

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

• Loss: 2.0926
• Accuracy: 0.7579
• F1: 0.4615
• Precision: 0.6372
• Recall: 0.3618
• Precision Sarcastic: 0.6372
• Recall Sarcastic: 0.3618
• F1 Sarcastic: 0.4615

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall	Precision Sarcastic	Recall Sarcastic	F1 Sarcastic
No log	1.0	44	2.1092	0.7550	0.4586	0.6261	0.3618	0.6261	0.3618	0.4586
No log	2.0	88	1.7332	0.7421	0.4559	0.5769	0.3769	0.5769	0.3769	0.4559
No log	3.0	132	1.9829	0.7392	0.4597	0.5662	0.3869	0.5662	0.3869	0.4597
No log	4.0	176	1.9446	0.7536	0.3915	0.6707	0.2764	0.6707	0.2764	0.3915
No log	5.0	220	1.6555	0.7594	0.4985	0.6194	0.4171	0.6194	0.4171	0.4985
No log	6.0	264	1.9983	0.7594	0.4261	0.6739	0.3116	0.6739	0.3116	0.4261
No log	7.0	308	1.9632	0.7622	0.4985	0.6308	0.4121	0.6308	0.4121	0.4985
No log	8.0	352	2.1204	0.7507	0.4055	0.6413	0.2965	0.6413	0.2965	0.4055
No log	9.0	396	2.0696	0.7637	0.4810	0.6496	0.3819	0.6496	0.3819	0.4810
No log	10.0	440	2.0926	0.7579	0.4615	0.6372	0.3618	0.6372	0.3618	0.4615

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1