metadata

license: apache-2.0
base_model: facebook/bart-large
tags:
  - generated_from_trainer
metrics:
  - rouge
  - wer
model-index:
  - name: bart_extractive_512_375
    results: []

bart_extractive_512_375

This model is a fine-tuned version of facebook/bart-large on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.9965
Rouge1: 0.6939
Rouge2: 0.4349
Rougel: 0.6334
Rougelsum: 0.6333
Wer: 0.4534

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
No log	0.13	250	1.2428	0.6513	0.3723	0.5834	0.5833	0.5119
2.1522	0.27	500	1.1468	0.6717	0.3957	0.605	0.6049	0.494
2.1522	0.4	750	1.1064	0.6729	0.404	0.609	0.609	0.483
1.2231	0.53	1000	1.0908	0.6762	0.4078	0.6116	0.6115	0.479
1.2231	0.66	1250	1.0726	0.6774	0.4108	0.6137	0.6136	0.4755
1.1583	0.8	1500	1.0581	0.6868	0.4196	0.6246	0.6245	0.4714
1.1583	0.93	1750	1.0534	0.6833	0.4209	0.6215	0.6214	0.4686
1.1133	1.06	2000	1.0330	0.6909	0.4263	0.6297	0.6297	0.4647
1.1133	1.2	2250	1.0288	0.6929	0.4293	0.631	0.6309	0.4626
1.0198	1.33	2500	1.0204	0.6925	0.4303	0.6305	0.6305	0.4601
1.0198	1.46	2750	1.0097	0.6965	0.4336	0.6349	0.6348	0.4582
1.0204	1.6	3000	1.0087	0.6976	0.4359	0.6361	0.636	0.4565
1.0204	1.73	3250	1.0042	0.6949	0.4345	0.6342	0.6342	0.4557
0.9889	1.86	3500	0.9965	0.696	0.4366	0.6352	0.6351	0.4534
0.9889	1.99	3750	0.9965	0.6939	0.4349	0.6334	0.6333	0.4534

Framework versions

Transformers 4.38.2
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2