bart-base-japanese-RMT-tobyoki-150

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: 2.7259
• Rouge1: 13.9233
• Rouge2: 1.4637
• Rougel: 7.7275
• Rougelsum: 11.0078
• Gen Len: 3138.9

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	3.6662	13.7115	1.4711	7.3306	11.3636	5309.5
No log	2.0	160	3.2213	14.6314	1.7968	7.4991	11.9019	4975.8
No log	3.0	240	2.9746	14.628	1.6945	7.9077	11.4463	3886.6
No log	4.0	320	2.8153	13.9751	1.659	8.0324	11.3707	3328.4
No log	5.0	400	2.7259	13.9233	1.4637	7.7275	11.0078	3138.9

Framework versions

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