Gliscor/email-summarizer-bart-large-cnn-tr

email-summarization-mt5-lora

This model is a fine-tuned version of facebook/bart-large-cnn on Gliscor/email-summaries-tr dataset. It achieves the following results on the evaluation set:

• Loss: 0.9852
• Rouge1: 0.4139
• Rouge2: 0.2289
• Rougel: 0.3470
• Meteor: 0.3850

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: 0.0005
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• 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
• lr_scheduler_warmup_steps: 100
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Meteor
1.3724	0.4444	1000	1.2582	0.3767	0.1882	0.3024	0.3463
1.2355	0.8889	2000	1.1349	0.3863	0.2006	0.3184	0.3568
1.091	1.3333	3000	1.0663	0.3848	0.2038	0.3182	0.3473
1.0423	1.7778	4000	1.0290	0.3989	0.2129	0.3332	0.3688
0.932	2.2222	5000	1.0098	0.3972	0.2167	0.3291	0.3611
1.0104	2.6667	6000	0.9852	0.4139	0.2289	0.3470	0.3850

Framework versions

• PEFT 0.17.0
• Transformers 4.55.1
• Pytorch 2.6.0+cu124
• Datasets 4.0.0
• Tokenizers 0.21.4