le-fine-tune-mbart-large-50

This model is a fine-tuned version of facebook/mbart-large-50 on the ravkuk_summerize_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 2.7762
• Rouge1: 0.2928
• Rouge2: 0.1926
• Rougel: 0.2815
• Rougelsum: 0.2816
• Gen Len: 34.5028

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: 5.6e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
3.1169	1.0	197	2.3604	0.1878	0.0725	0.1737	0.1737	33.7784
1.8945	2.0	394	2.2522	0.1897	0.0765	0.1776	0.1776	34.2074
1.3083	3.0	591	2.2886	0.2001	0.0927	0.1895	0.1892	35.4432
0.8693	4.0	788	2.3727	0.2243	0.1123	0.2122	0.2117	31.4943
0.5507	5.0	985	2.5059	0.2577	0.1527	0.2463	0.2466	34.3693
0.3385	6.0	1182	2.6032	0.2703	0.1672	0.2593	0.2584	33.5994
0.2031	7.0	1379	2.6518	0.2912	0.1932	0.2812	0.281	34.1676
0.1272	8.0	1576	2.7040	0.2891	0.1895	0.2799	0.2796	34.6761
0.0842	9.0	1773	2.7515	0.2978	0.198	0.2888	0.2887	34.1932
0.0605	10.0	1970	2.7762	0.2928	0.1926	0.2815	0.2816	34.5028

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2