bart-pas-lora

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

• Loss: 4.0678
• Rouge1: 19.51
• Rouge2: 8.76
• Rougel: 17.44
• Rougelsum: 17.41

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: 0.0001
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• 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_ratio: 0.05
• num_epochs: 10
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
4.4614	0.96	15	4.7799	15.27	6.55	12.8	13.24
4.3888	1.96	30	4.5425	14.55	7.1	12.76	13.17
4.4672	2.96	45	4.3419	15.48	7.45	13.79	14.02
3.3327	3.96	60	4.2166	15.96	8.45	14.57	14.89
3.5467	4.96	75	4.1432	16.18	6.44	13.94	13.99
4.4243	5.96	90	4.1056	18.16	7.4	16.03	16.1
4.6372	6.96	105	4.0792	18.67	8.16	16.96	16.98
4.4094	7.96	120	4.0678	19.51	8.76	17.44	17.41
4.3115	8.96	135	4.0601	19.45	7.79	17.18	17.24
4.4145	9.96	150	4.0572	19.44	7.79	17.17	17.23

Framework versions

• PEFT 0.13.2
• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.4