mT5-TextSimp-LT-BatchSize8-lr1e-4

This model is a fine-tuned version of google/mt5-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0826
• Rouge1: 0.6956
• Rouge2: 0.532
• Rougel: 0.6875
• Sacrebleu: 41.0349
• Gen Len: 38.0501

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Sacrebleu	Gen Len
22.5133	0.96	200	14.4822	0.0057	0.0	0.0056	0.0013	512.0
1.0276	1.91	400	0.7352	0.022	0.0005	0.0215	0.0232	41.4702
0.6477	2.87	600	1.5193	0.1021	0.012	0.0954	0.0573	83.3723
0.1784	3.83	800	0.1149	0.6014	0.4222	0.5898	32.2723	38.0501
0.158	4.78	1000	0.0930	0.6546	0.4822	0.6463	37.3842	38.0501
0.1059	5.74	1200	0.0884	0.6714	0.4983	0.6635	39.0129	38.0501
0.1542	6.7	1400	0.0830	0.688	0.5184	0.6803	40.419	38.0501
0.1206	7.66	1600	0.0826	0.6956	0.532	0.6875	41.0349	38.0501

Framework versions

• Transformers 4.33.0
• Pytorch 2.1.2+cu121
• Datasets 2.14.4
• Tokenizers 0.13.3