bart-large-finetuned-question-to-answer

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: 0.1153
• Bleu: 42.8973
• Gen Len: 18.69

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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.8366	1.0	516	0.3882	32.192	18.8467
0.7567	2.0	1032	0.3263	34.6627	18.8333
0.6634	3.0	1548	0.2838	34.3455	18.8567
0.587	4.0	2064	0.2207	37.4365	18.8467
0.5178	5.0	2580	0.2778	36.1141	19.2267
0.4555	6.0	3096	0.1872	39.1633	18.6967
0.4137	7.0	3612	0.1854	39.3042	18.98
0.3672	8.0	4128	0.1543	40.8359	18.68
0.331	9.0	4644	0.1548	41.0895	18.54
0.3056	10.0	5160	0.1599	42.3384	18.6767
0.2762	11.0	5676	0.1508	41.1395	18.8167
0.2533	12.0	6192	0.1224	42.1233	18.7033
0.2332	13.0	6708	0.1195	42.8086	18.6967
0.2209	14.0	7224	0.1158	43.0663	18.72
0.21	15.0	7740	0.1153	42.8973	18.69

Framework versions

• Transformers 4.36.2
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.15.0