results_t5small

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

• Loss: 2.3881
• Rouge1: 0.1784
• Rouge2: 0.0606
• Rougel: 0.1404
• Rougelsum: 0.1403
• Gen Len: 20.0

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.6116	0.3940	2500	2.5781	0.1704	0.0553	0.1343	0.1342	20.0
2.5436	0.7880	5000	2.4790	0.1707	0.0569	0.1348	0.1347	20.0
2.5037	1.1820	7500	2.4316	0.178	0.0602	0.1399	0.1398	20.0
2.4498	1.5760	10000	2.4036	0.1803	0.0616	0.1415	0.1415	20.0
2.4604	1.9701	12500	2.3881	0.1784	0.0606	0.1404	0.1403	20.0

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0