results

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

• Loss: 1.2424
• Accuracy: 0.7451
• Precision: 0.7493
• Recall: 0.7451
• F1 Score: 0.7468

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: 1e-05
• train_batch_size: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1 Score
1.6562	1.0	1483	1.6057	0.5142	0.5212	0.5142	0.4791
1.2593	2.0	2966	1.2195	0.7265	0.7344	0.7265	0.7296
1.243	3.0	4449	1.2424	0.7417	0.7379	0.7417	0.7351
1.242	4.0	5932	1.1913	0.7268	0.7521	0.7268	0.7367
1.1838	5.0	7415	1.2089	0.7315	0.7488	0.7315	0.7385
1.2212	6.0	8898	1.2035	0.7320	0.7498	0.7320	0.7395
1.0749	7.0	10381	1.2256	0.7310	0.7544	0.7310	0.7406
1.0961	8.0	11864	1.2273	0.7290	0.7528	0.7290	0.7387
1.118	9.0	13347	1.2552	0.7479	0.7515	0.7479	0.7489
1.0606	10.0	14830	1.2608	0.7309	0.7531	0.7309	0.7395
1.002	11.0	16313	1.3178	0.7378	0.7547	0.7378	0.7451
0.8984	12.0	17796	1.3219	0.7401	0.7560	0.7401	0.7469
0.9297	13.0	19279	1.3574	0.7366	0.7546	0.7366	0.7437
0.9262	14.0	20762	1.3390	0.7344	0.7565	0.7344	0.7428
0.9846	15.0	22245	1.3715	0.7472	0.7551	0.7472	0.7507
0.9031	16.0	23728	1.3681	0.7442	0.7567	0.7442	0.7493

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1