bart-base-chinese-6615-chinese

This model is a fine-tuned version of fnlp/bart-base-chinese on the xlsum dataset. It achieves the following results on the evaluation set:

• Loss: 0.8576
• Rouge1: 0.0774
• Rouge2: 0.0179
• Rougel: 0.0772
• Rougelsum: 0.077
• Gen Len: 19.9552

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.1216	0.86	500	0.9150	0.0523	0.0113	0.0521	0.052	19.9345
1.0346	1.71	1000	0.8817	0.0585	0.0119	0.0583	0.0582	19.9535
1.0063	2.57	1500	0.8624	0.0603	0.0112	0.0598	0.0599	19.9512
0.9219	3.42	2000	0.8592	0.0715	0.0145	0.071	0.0712	19.9535
0.8757	4.28	2500	0.8577	0.072	0.0153	0.0717	0.0717	19.9636
0.8832	5.14	3000	0.8567	0.0721	0.0157	0.0717	0.0718	19.9493
0.8788	5.99	3500	0.8498	0.0763	0.0173	0.0759	0.0759	19.9565
0.8659	6.85	4000	0.8513	0.076	0.017	0.0756	0.0754	19.9546
0.7802	7.71	4500	0.8563	0.0772	0.0185	0.077	0.0768	19.9525
0.8114	8.56	5000	0.8562	0.0769	0.0169	0.0766	0.0764	19.954
0.7715	9.42	5500	0.8576	0.0774	0.0179	0.0772	0.077	19.9552

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0