t5-finetuned-summarization-samsum

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

• Loss: 1.6534
• Rouge1: 44.8411
• Rouge2: 22.2109
• Rougel: 37.7505
• Rougelsum: 41.2556

Model description

Model is the pre-trained T5-Small model which is finetuned for Text Summarization of Dialogues.

Intended uses & limitations

More information needed

Training and evaluation data

• Training Data: SAMsum dataset's training observations
• Evaluation Data: SAMsum dataset's evaluation observations
• Link to Dataset: https://huggingface.co/datasets/Samsung/samsum

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• 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
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
1.8241	1.0	1842	1.7426	43.1854	19.9747	35.9002	39.8211
1.8416	2.0	3684	1.7092	42.9915	20.5143	36.4477	39.7165
1.7886	3.0	5526	1.6778	43.8757	20.7284	36.6769	40.3929
1.7433	4.0	7368	1.6750	44.5267	21.8446	37.5636	41.1866
1.7096	5.0	9210	1.6646	44.2547	21.9277	37.4179	40.7838
1.6869	6.0	11052	1.6609	44.9253	21.9203	37.7432	41.353
1.6633	7.0	12894	1.6574	44.9061	22.3809	37.8733	41.3547
1.651	8.0	14736	1.6534	44.536	21.9922	37.4583	40.9781
1.6363	9.0	16578	1.6539	44.7735	22.1445	37.5946	41.1984
1.6318	10.0	18420	1.6534	44.8411	22.2109	37.7505	41.2556

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1