SAS-finetuned-cochrane-medeasi

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

• Loss: nan
• Bleu: {'bleu': 3.5213074954706223e-06, 'precisions': [0.49951576455396535, 0.15234465234465233, 0.06880219369313224, 0.036816459122902004], 'brevity_penalty': 2.9884691172035265e-05, 'length_ratio': 0.08757975289560735, 'translation_length': 9293, 'reference_length': 106109}
• Sari: {'sari': 2.5441859559296094}

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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	Bleu	Sari
No log	1.0	159	nan	{'bleu': 3.5213074954706223e-06, 'precisions': [0.49951576455396535, 0.15234465234465233, 0.06880219369313224, 0.036816459122902004], 'brevity_penalty': 2.9884691172035265e-05, 'length_ratio': 0.08757975289560735, 'translation_length': 9293, 'reference_length': 106109}	{'sari': 2.5441859559296094}
No log	2.0	318	nan	{'bleu': 3.5213074954706223e-06, 'precisions': [0.49951576455396535, 0.15234465234465233, 0.06880219369313224, 0.036816459122902004], 'brevity_penalty': 2.9884691172035265e-05, 'length_ratio': 0.08757975289560735, 'translation_length': 9293, 'reference_length': 106109}	{'sari': 2.5441859559296094}
No log	3.0	477	nan	{'bleu': 3.5213074954706223e-06, 'precisions': [0.49951576455396535, 0.15234465234465233, 0.06880219369313224, 0.036816459122902004], 'brevity_penalty': 2.9884691172035265e-05, 'length_ratio': 0.08757975289560735, 'translation_length': 9293, 'reference_length': 106109}	{'sari': 2.5441859559296094}

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1