cs221-roberta-large-eng-pt

This model is a fine-tuned version of FacebookAI/roberta-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5690
• F1: 0.7598
• Roc Auc: 0.8118
• Accuracy: 0.5086

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.3966	1.0	173	0.3720	0.6785	0.7508	0.4224
0.3263	2.0	346	0.3824	0.7098	0.7742	0.4052
0.2298	3.0	519	0.3525	0.7210	0.7832	0.4569
0.1699	4.0	692	0.3996	0.6968	0.7673	0.4224
0.115	5.0	865	0.4215	0.7371	0.8025	0.4655
0.0622	6.0	1038	0.4543	0.7425	0.8002	0.4741
0.0609	7.0	1211	0.4787	0.7399	0.8028	0.4741
0.0344	8.0	1384	0.5559	0.7326	0.7927	0.4914
0.0205	9.0	1557	0.5545	0.7486	0.8052	0.4828
0.0153	10.0	1730	0.5612	0.7528	0.8131	0.4914
0.0082	11.0	1903	0.5690	0.7598	0.8118	0.5086
0.0038	12.0	2076	0.6239	0.7358	0.7974	0.4655
0.0047	13.0	2249	0.6296	0.7567	0.8072	0.5086
0.0025	14.0	2422	0.6246	0.7448	0.8028	0.5
0.0018	15.0	2595	0.6347	0.7403	0.8000	0.4828

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0