fine-tuned-BioBART-10-epochs-1024-input-256-output

This model is a fine-tuned version of GanjinZero/biobart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8636
• Rouge1: 0.1803
• Rouge2: 0.043
• Rougel: 0.1391
• Rougelsum: 0.1408
• Gen Len: 39.35

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	151	2.4893	0.0	0.0	0.0	0.0	4.56
No log	2.0	302	1.0370	0.1382	0.033	0.1174	0.1192	28.23
No log	3.0	453	0.9481	0.0912	0.0231	0.0723	0.072	22.01
3.1525	4.0	604	0.9079	0.1402	0.0336	0.1063	0.1064	42.37
3.1525	5.0	755	0.8861	0.1772	0.0335	0.1344	0.1364	49.96
3.1525	6.0	906	0.8760	0.1702	0.0327	0.1301	0.1313	42.56
0.7557	7.0	1057	0.8661	0.158	0.0403	0.1136	0.114	40.53
0.7557	8.0	1208	0.8641	0.1631	0.0431	0.119	0.1198	44.57
0.7557	9.0	1359	0.8659	0.172	0.0427	0.1357	0.1369	38.31
0.5883	10.0	1510	0.8636	0.1803	0.043	0.1391	0.1408	39.35

Framework versions

• Transformers 4.36.2
• Pytorch 1.12.1+cu113
• Datasets 2.16.1
• Tokenizers 0.15.0