Procit004/T5_For_Text_Summarization

results

This model is a fine-tuned version of t5-small on the multi_news dataset. It achieves the following results on the evaluation set:

• Loss: 2.9028
• Rouge1: 37.3599
• Rouge2: 12.1820
• Rougel: 21.4068
• Rougelsum: 21.3827
• Gen Len: 141.366

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• 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
No log	1.0	313	3.0888	33.8257	10.0913	19.3859	19.3966	131.264
3.487	2.0	626	3.0216	36.0141	11.1691	20.4601	20.4538	138.12
3.487	3.0	939	2.9906	36.2470	11.3578	20.6635	20.6692	138.632
3.2354	4.0	1252	2.9727	36.7252	11.5422	20.9492	20.9458	139.433
3.1863	5.0	1565	2.9586	36.6970	11.6533	20.9281	20.9236	140.189
3.1863	6.0	1878	2.9511	36.8584	11.7427	21.1395	21.1377	140.747
3.1624	7.0	2191	2.9441	36.9490	11.8362	21.2388	21.2508	140.994
3.1462	8.0	2504	2.9406	37.0855	11.8388	21.2447	21.2583	141.331
3.1462	9.0	2817	2.9383	37.0757	11.8588	21.2306	21.2472	140.901
3.1409	10.0	3130	2.9376	37.1450	11.9259	21.3013	21.3147	141.081

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1