bart-base-sci-tr

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.7617

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: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss
3.5107	1.0	1393	3.0242
3.1138	2.0	2786	2.8770
2.9345	3.0	4179	2.8069
2.8044	4.0	5572	2.7668
2.7227	5.0	6965	2.7336
2.6341	6.0	8358	2.7276
2.5523	7.0	9751	2.7026
2.4846	8.0	11144	2.6961
2.4409	9.0	12537	2.6894
2.3699	10.0	13930	2.6853
2.3305	11.0	15323	2.6869
2.2822	12.0	16716	2.6924
2.2392	13.0	18109	2.7042
2.1889	14.0	19502	2.6943
2.1593	15.0	20895	2.6988
2.1228	16.0	22288	2.7032
2.0791	17.0	23681	2.7083
2.0594	18.0	25074	2.7034
2.0239	19.0	26467	2.7182
1.9943	20.0	27860	2.7235
1.9746	21.0	29253	2.7279
1.9624	22.0	30646	2.7337
1.9433	23.0	32039	2.7400
1.9138	24.0	33432	2.7457
1.8971	25.0	34825	2.7489
1.8896	26.0	36218	2.7527
1.873	27.0	37611	2.7576
1.8585	28.0	39004	2.7577
1.8564	29.0	40397	2.7594
1.852	30.0	41790	2.7617

Framework versions

• Transformers 4.43.4
• Pytorch 1.13.1
• Datasets 2.12.0
• Tokenizers 0.19.1