bart-nepali-copymechanism

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

• Loss: 0.0568

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: 0.0001
• train_batch_size: 150
• eval_batch_size: 150
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4455	1.0	2917	0.3525
0.2394	2.0	5834	0.1986
0.1639	3.0	8751	0.1361
0.1342	4.0	11668	0.1030
0.1045	5.0	14585	0.0898
0.0929	6.0	17502	0.0808
0.0862	7.0	20419	0.0749
0.0826	8.0	23336	0.0720
0.0755	9.0	26253	0.0676
0.0786	10.0	29170	0.0652
0.0708	11.0	32087	0.0619
0.0702	12.0	35004	0.0604
0.0629	13.0	37921	0.0590
0.0622	14.0	40838	0.0577
0.062	15.0	43755	0.0568

Framework versions

• Transformers 4.47.1
• Pytorch 2.6.0+xpu
• Datasets 3.2.0
• Tokenizers 0.21.0