tl_dr_summary_with_t5_base

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

• Loss: 2.1219
• Rouge1: 0.4445
• Rouge2: 0.2515
• Rougel: 0.4146
• Rougelsum: 0.4153
• Gen Len: 13.9647

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	447	2.2145	0.4202	0.2384	0.3929	0.3928	13.8967
2.621	2.0	894	2.1558	0.4353	0.2428	0.4066	0.4066	13.738
2.3114	3.0	1341	2.1320	0.4425	0.2511	0.4132	0.4136	13.9307
2.2337	4.0	1788	2.1242	0.4455	0.2516	0.4157	0.4159	13.9345
2.1767	5.0	2235	2.1219	0.4445	0.2515	0.4146	0.4153	13.9647

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3