model_v5_v2

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

• Loss: 0.5446
• Sacrebleu: 66.8833

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: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Sacrebleu
No log	1.0	109	0.5768	65.2132
No log	2.0	219	0.5885	66.1069
No log	3.0	328	0.5681	66.6363
No log	4.0	438	0.5766	66.7355
No log	5.0	547	0.5446	66.8833
No log	6.0	657	0.5664	66.9772
No log	7.0	766	0.5890	67.0638
No log	8.0	876	0.6305	67.0642
No log	9.0	985	0.6091	67.0855
No log	10.0	1095	0.5990	67.1479
No log	11.0	1204	0.6375	67.0056
No log	12.0	1314	0.6182	67.0154
No log	13.0	1423	0.6216	66.9095
No log	14.0	1533	0.6204	67.1035
No log	15.0	1642	0.6623	66.8847
No log	16.0	1752	0.6614	67.0027
No log	17.0	1861	0.6801	66.9272
No log	18.0	1971	0.6500	66.9428
No log	19.0	2080	0.6529	66.8661
No log	19.91	2180	0.6514	66.8420

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2