bart-base-japanese-RMT-tobyoki-200

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: 2.6252
• Rouge1: 14.3358
• Rouge2: 2.1278
• Rougel: 8.3601
• Rougelsum: 11.4494
• Gen Len: 2374.4

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	3.6698	13.7458	1.641	7.2384	11.1015	5547.5
No log	2.0	160	3.2062	14.8374	1.7608	7.8979	11.852	4605.2
No log	3.0	240	2.9757	14.7957	1.6829	7.9807	11.6804	3442.6
No log	4.0	320	2.8127	15.3153	1.9135	8.4994	11.9208	2844.8
No log	5.0	400	2.7265	14.1003	1.9348	8.2344	11.1833	2497.4
No log	6.0	480	2.6753	14.3802	2.1414	8.5996	11.1612	2340.6
3.2173	7.0	560	2.6252	14.3358	2.1278	8.3601	11.4494	2374.4

Framework versions

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