roberta-mqa-formrat

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1135
• Accuracy: 0.5671
• F1: 0.5659
• Precision: 0.5683
• Recall: 0.5650

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: 4
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
1.451	0.3233	1200	1.4125	0.4105	0.4093	0.4151	0.4107
1.416	0.6466	2400	1.3482	0.4412	0.4394	0.4438	0.4385
1.3157	0.9698	3600	1.2933	0.4788	0.4772	0.4776	0.4773
1.2616	1.2931	4800	1.2389	0.5032	0.5022	0.5053	0.5011
1.221	1.6164	6000	1.2049	0.5053	0.5039	0.5060	0.5029
1.1556	1.9397	7200	1.1792	0.5288	0.5276	0.5295	0.5265
1.082	2.2629	8400	1.1593	0.5451	0.5434	0.5487	0.5415
1.0692	2.5862	9600	1.1153	0.5613	0.5606	0.5641	0.5594
1.0066	2.9095	10800	1.1135	0.5671	0.5659	0.5683	0.5650

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1