bart-large-japanese-RMT-tobyoki-20

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

• Loss: 2.4333
• Rouge1: 10.4455
• Rouge2: 0.9997
• Rougel: 7.6259
• Rougelsum: 9.0518
• Gen Len: 2098.2

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: 3e-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	40	3.1809	9.7012	1.3824	6.3532	8.4445	3000.0
No log	2.0	80	2.8693	11.308	1.6656	6.9919	9.8495	2980.5
No log	3.0	120	2.6873	11.7005	1.6944	7.3988	10.0862	2702.3
No log	4.0	160	2.5541	10.815	1.2578	7.4659	9.1837	2275.6
No log	5.0	200	2.4785	10.5975	1.0925	7.6191	9.357	2157.6
No log	6.0	240	2.4333	10.4455	0.9997	7.6259	9.0518	2098.2

Framework versions

• Transformers 4.30.0
• Pytorch 2.0.1+cu117
• Datasets 2.14.4
• Tokenizers 0.13.3