bart-base-2024-09-24_11-12

This model is a fine-tuned version of facebook/bart-base on the arrow dataset. It achieves the following results on the evaluation set:

• Loss: 0.1993
• Gen Len: 20.0
• Bertscorer-p: 0.5928
• Bertscorer-r: 0.1701
• Bertscorer-f1: 0.3731
• Sacrebleu-score: 10.2541
• Sacrebleu-precisions: [90.63003300856309, 79.05155386114873, 70.66565212490137, 65.68935823527592]
• Bleu-bp: 0.1350

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Gen Len	Bertscorer-p	Bertscorer-r	Bertscorer-f1	Sacrebleu-score	Sacrebleu-precisions	Bleu-bp
0.189	1.0	4772	0.1993	20.0	0.5928	0.1701	0.3731	10.2541	[90.63003300856309, 79.05155386114873, 70.66565212490137, 65.68935823527592]	0.1350

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1