results_mt5_xl-sum

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8431
• Rouge1 Fmeasure: 0.6139
• Rouge2 Fmeasure: 0.1189
• Rougel Fmeasure: 0.1997
• Meteor: 0.3315
• Bertscore F1: 0.8418

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1 Fmeasure	Rouge2 Fmeasure	Rougel Fmeasure	Meteor	Bertscore F1
2.6516	0.8529	500	0.9710	0.2668	0.0484	0.1537	0.2745	0.8284
1.0475	1.7058	1000	0.8792	0.4289	0.0884	0.1737	0.2949	0.8278
0.9413	2.5586	1500	0.8457	0.4960	0.0865	0.1898	0.3141	0.8339
0.8711	3.4115	2000	0.8398	0.5400	0.1121	0.1941	0.3110	0.8397
0.8235	4.2644	2500	0.8345	0.5587	0.1022	0.2041	0.3160	0.8388
0.7797	5.1173	3000	0.8368	0.5735	0.1036	0.2044	0.3157	0.8344
0.7401	5.9701	3500	0.8217	0.5507	0.1133	0.1936	0.3186	0.8366
0.7022	6.8230	4000	0.8361	0.5808	0.1118	0.2008	0.3227	0.8406
0.6796	7.6759	4500	0.8344	0.6173	0.1277	0.1986	0.3260	0.8407
0.6523	8.5288	5000	0.8436	0.6232	0.1186	0.2024	0.3317	0.8398
0.6385	9.3817	5500	0.8431	0.6139	0.1189	0.1997	0.3315	0.8418

Framework versions

• Transformers 4.40.0
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1