t5-base-dutch-finetuned-mt5_base_keyword_extraction_dutch

This model is a fine-tuned version of yhavinga/t5-base-dutch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6065
• Rouge1: 0.7675
• Rouge2: 0.5965
• Rougel: 0.7531
• Rougelsum: 0.7534

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: 14
• eval_batch_size: 14
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 28
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
No log	1.0	94	1.2820	0.4332	0.3007	0.4175	0.4174
1.4828	2.0	188	0.9360	0.6075	0.4541	0.5916	0.5912
1.4828	3.0	282	0.7435	0.6542	0.4825	0.6358	0.6366
0.7633	4.0	376	0.6623	0.6867	0.5071	0.6692	0.6697
0.7633	5.0	470	0.6481	0.7061	0.5254	0.6909	0.6913
0.5935	6.0	564	0.6456	0.7155	0.5367	0.6984	0.6995
0.5935	7.0	658	0.6387	0.7162	0.5388	0.6993	0.7001
0.5101	8.0	752	0.6341	0.7247	0.5495	0.7086	0.7102
0.5101	9.0	846	0.6306	0.7335	0.5527	0.7166	0.7176
0.4449	10.0	940	0.6412	0.7324	0.5559	0.7160	0.7166
0.4449	11.0	1034	0.6439	0.7273	0.5513	0.7126	0.7136
0.4001	12.0	1128	0.6294	0.7415	0.5644	0.7266	0.7268
0.4001	13.0	1222	0.6252	0.7447	0.5658	0.7294	0.7296
0.3589	14.0	1316	0.6257	0.7490	0.5743	0.7341	0.7347
0.3589	15.0	1410	0.6132	0.7474	0.5751	0.7339	0.7346
0.3263	16.0	1504	0.6119	0.7616	0.5858	0.7469	0.7470
0.3263	17.0	1598	0.6088	0.7674	0.5945	0.7527	0.7530
0.2989	18.0	1692	0.6108	0.7655	0.5917	0.7510	0.7514
0.2989	19.0	1786	0.6020	0.7681	0.5961	0.7539	0.7545
0.2846	20.0	1880	0.6065	0.7675	0.5965	0.7531	0.7534

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2