bart-base-japanese-RMT-tobyoki-20

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

• Loss: 2.7005
• Rouge1: 11.3587
• Rouge2: 1.08
• Rougel: 7.6075
• Rougelsum: 9.5106
• Gen Len: 3081.5

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: 1
• eval_batch_size: 1
• 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	80	3.6825	13.5761	1.278	7.6282	11.0138	4290.4
No log	2.0	160	3.2193	14.217	1.5027	7.9016	11.015	4244.5
No log	3.0	240	2.9484	13.4372	1.312	8.0882	10.727	3350.9
No log	4.0	320	2.7949	12.6469	0.9666	7.9801	10.0949	3331.2
No log	5.0	400	2.7005	11.3587	1.08	7.6075	9.5106	3081.5

Framework versions

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