bart-large-xsum-finetuned-natural-questions

This model is a fine-tuned version of facebook/bart-large-xsum on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2729
• Rouge1: 19.7211
• Rouge2: 17.4272
• Rougel: 19.0681
• Rougelsum: 19.3677

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: 5.6e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
No log	0.99	34	0.2562	17.9806	15.2059	16.807	17.5533
No log	1.99	68	0.1845	14.6261	10.494	13.0132	13.8392
No log	2.98	102	0.2171	17.3737	14.7893	16.5485	16.8383
No log	4.0	137	0.3474	17.6187	14.727	16.5614	17.1476
No log	4.99	171	0.3462	17.7103	15.1403	16.9424	17.3123
0.1255	5.99	205	0.3355	19.2782	16.5525	18.4283	18.8422
0.1255	6.98	239	0.2281	19.8816	17.4387	19.238	19.552
0.1255	7.94	272	0.2729	19.7211	17.4272	19.0681	19.3677

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3