t-5-base-bertsum-375

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

• Loss: 1.3186
• Rouge1: 0.6438
• Rouge2: 0.3613
• Rougel: 0.576
• Rougelsum: 0.5761
• Wer: 0.5293
• Bleurt: -0.0784

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: 6
• eval_batch_size: 6
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Wer	Bleurt
No log	0.13	250	1.4772	0.6191	0.3294	0.5478	0.5478	0.5624	-0.4294
1.9906	0.27	500	1.4209	0.627	0.3392	0.5569	0.557	0.5524	-0.3865
1.9906	0.4	750	1.3947	0.6308	0.3452	0.5617	0.5617	0.5462	-0.399
1.5082	0.53	1000	1.3735	0.6345	0.3485	0.5649	0.565	0.5433	-0.0701
1.5082	0.66	1250	1.3627	0.6356	0.3507	0.5669	0.567	0.54	-0.3802
1.469	0.8	1500	1.3518	0.6372	0.3528	0.569	0.569	0.5378	-0.0292
1.469	0.93	1750	1.3437	0.6381	0.3542	0.5703	0.5704	0.536	-0.3802
1.4436	1.06	2000	1.3376	0.64	0.3561	0.5718	0.5718	0.5341	-0.3922
1.4436	1.2	2250	1.3314	0.6407	0.3571	0.573	0.573	0.5334	-0.3922
1.4144	1.33	2500	1.3285	0.6417	0.3588	0.574	0.5741	0.5318	-0.3802
1.4144	1.46	2750	1.3247	0.642	0.359	0.5742	0.5743	0.5312	-0.1227
1.4267	1.6	3000	1.3224	0.643	0.3601	0.575	0.5751	0.5306	-0.0784
1.4267	1.73	3250	1.3206	0.643	0.3607	0.5754	0.5755	0.5301	-0.1084
1.3975	1.86	3500	1.3189	0.6431	0.3609	0.5755	0.5756	0.5297	-0.1
1.3975	1.99	3750	1.3186	0.6438	0.3613	0.576	0.5761	0.5293	-0.0784

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2