metadata

license: apache-2.0
base_model: t5-base
tags:
  - generated_from_trainer
metrics:
  - rouge
  - wer
model-index:
  - name: t-5-base-baseline
    results: []

t-5-base-baseline

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

Loss: 1.1785
Rouge1: 0.6772
Rouge2: 0.4105
Rougel: 0.6161
Rougelsum: 0.6161
Wer: 0.4869
Bleurt: 0.3779

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	1.3316	0.6509	0.3768	0.5866	0.5865	0.5217	0.3009
1.7919	0.27	500	1.2776	0.6593	0.3865	0.5962	0.5962	0.5108	0.3009
1.7919	0.4	750	1.2513	0.6633	0.3931	0.6015	0.6014	0.5039	0.3009
1.3552	0.53	1000	1.2326	0.6667	0.3967	0.6048	0.6047	0.5008	0.3009
1.3552	0.66	1250	1.2236	0.669	0.4	0.6072	0.6072	0.4972	0.3314
1.3074	0.8	1500	1.2118	0.6711	0.4022	0.6093	0.6093	0.4953	0.3314
1.3074	0.93	1750	1.2022	0.6714	0.4034	0.6105	0.6104	0.4932	0.2798
1.3037	1.06	2000	1.1972	0.673	0.4053	0.6117	0.6116	0.4916	0.3771
1.3037	1.2	2250	1.1909	0.6749	0.4068	0.6136	0.6135	0.4905	0.3314
1.2676	1.33	2500	1.1889	0.676	0.4086	0.6143	0.6143	0.4893	0.3314
1.2676	1.46	2750	1.1848	0.6763	0.4091	0.615	0.6149	0.4884	0.3314
1.2796	1.6	3000	1.1829	0.677	0.4095	0.6154	0.6154	0.488	0.3123
1.2796	1.73	3250	1.1808	0.6767	0.41	0.6157	0.6157	0.4876	0.3779
1.2489	1.86	3500	1.1787	0.6771	0.4105	0.616	0.616	0.4869	0.3771
1.2489	1.99	3750	1.1785	0.6772	0.4105	0.6161	0.6161	0.4869	0.3779

Framework versions

Transformers 4.38.2
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2