mbart-large-50-pluska_sum

This model is a fine-tuned version of facebook/mbart-large-50 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 4.1046
• Rouge1: 13.9497
• Rouge2: 4.3548
• Rougel: 12.2975
• Rougelsum: 12.4754

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: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
No log	1.0	144	3.4603	17.0434	5.7835	15.1411	15.3978
No log	2.0	288	3.2464	18.0445	6.1695	15.8888	16.172
No log	3.0	432	3.3550	18.017	6.2151	15.8346	16.1165
3.76	4.0	576	17.8581	0.2995	0.0017	0.2995	0.2997
3.76	5.0	720	11.2612	0.0077	0.0011	0.0077	0.0077
3.76	6.0	864	9.4077	0.4018	0.0396	0.3911	0.3924
3.76	7.0	1008	4.1815	13.3686	3.9575	11.846	11.9922
6.3287	8.0	1152	4.1046	13.9497	4.3548	12.2975	12.4754

Framework versions

• Transformers 4.27.0
• Pytorch 2.0.0
• Datasets 2.10.1
• Tokenizers 0.13.2