0d94935db786a8a6d6b21777d7cbb177

This model is a fine-tuned version of studio-ousia/luke-japanese-large on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 1.4548
• Data Size: 1.0
• Epoch Runtime: 37.0239
• Accuracy: 0.2401
• F1 Macro: 0.0774
• Rouge1: 0.2401
• Rouge2: 0.0
• Rougel: 0.2408
• Rougelsum: 0.2393

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	1.6318	0	2.9477	0.2180	0.0772	0.2188	0.0	0.2188	0.2180
No log	1	178	1.5279	0.0078	3.7700	0.2756	0.0864	0.2752	0.0	0.2749	0.2756
No log	2	356	1.4616	0.0156	4.3966	0.2678	0.1168	0.2685	0.0	0.2685	0.2670
No log	3	534	1.3356	0.0312	5.7421	0.3764	0.2136	0.3764	0.0	0.3764	0.3764
No log	4	712	1.1847	0.0625	7.3733	0.4070	0.2415	0.4070	0.0	0.4070	0.4062
No log	5	890	1.1366	0.125	10.2624	0.4680	0.2746	0.4688	0.0	0.4673	0.4673
0.0764	6	1068	1.1153	0.25	14.6838	0.4837	0.3162	0.4844	0.0	0.4851	0.4837
1.046	7	1246	1.0417	0.5	22.7026	0.4922	0.3662	0.4929	0.0	0.4929	0.4918
1.2009	8.0	1424	0.9534	1.0	38.6923	0.6222	0.4577	0.6229	0.0	0.6229	0.6222
1.1183	9.0	1602	0.9518	1.0	36.8835	0.6200	0.4720	0.6207	0.0	0.6200	0.6200
0.957	10.0	1780	0.9917	1.0	37.4234	0.5874	0.5252	0.5874	0.0	0.5874	0.5881
1.4558	11.0	1958	1.4522	1.0	37.8928	0.2401	0.0774	0.2401	0.0	0.2408	0.2393
1.4562	12.0	2136	1.4641	1.0	36.7426	0.2401	0.0774	0.2401	0.0	0.2408	0.2393
1.4612	13.0	2314	1.4548	1.0	37.0239	0.2401	0.0774	0.2401	0.0	0.2408	0.2393

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1