metadata

license: apache-2.0
base_model: t5-base
tags:
  - generated_from_trainer
metrics:
  - rouge
  - wer
model-index:
  - name: t-5-base-abs2abs
    results: []

t-5-base-abs2abs

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

Loss: 1.3203
Rouge1: 0.6447
Rouge2: 0.3627
Rougel: 0.5772
Rougelsum: 0.5774
Wer: 0.5292
Bleurt: -0.1862

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	0.14	250	1.4708	0.6226	0.3343	0.5515	0.5515	0.559	-0.1681
1.9361	0.27	500	1.4181	0.6277	0.3424	0.559	0.559	0.5498	-0.1527
1.9361	0.41	750	1.3918	0.6327	0.3469	0.5633	0.5634	0.5453	-0.1653
1.5072	0.55	1000	1.3740	0.6353	0.3509	0.5664	0.5665	0.541	-0.1653
1.5072	0.68	1250	1.3602	0.637	0.3529	0.5686	0.5687	0.539	-0.4817
1.4761	0.82	1500	1.3504	0.639	0.3557	0.5711	0.5712	0.5361	-0.1653
1.4761	0.96	1750	1.3424	0.64	0.3573	0.5727	0.5728	0.5341	-0.1653
1.4475	1.09	2000	1.3368	0.6414	0.3587	0.5736	0.5738	0.5329	-0.4817
1.4475	1.23	2250	1.3324	0.6423	0.36	0.5748	0.575	0.5316	-0.4726
1.4375	1.36	2500	1.3280	0.6436	0.3609	0.5757	0.5758	0.5309	-0.3069
1.4375	1.5	2750	1.3246	0.6442	0.3618	0.5765	0.5766	0.5304	-0.1862
1.4053	1.64	3000	1.3222	0.6445	0.3622	0.5769	0.577	0.5296	-0.1862
1.4053	1.77	3250	1.3208	0.6448	0.3626	0.5771	0.5772	0.5293	-0.1862
1.3911	1.91	3500	1.3203	0.6447	0.3627	0.5772	0.5774	0.5292	-0.1862

Framework versions

Transformers 4.38.2
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2