b40a079c0c0019fb8164f615f21b8808

This model is a fine-tuned version of studio-ousia/luke-japanese-large on the nyu-mll/glue [qqp] dataset. It achieves the following results on the evaluation set:

• Loss: 0.6819
• Data Size: 0.5
• Epoch Runtime: 993.3325
• Accuracy: 0.6320
• F1 Macro: 0.3872
• Rouge1: 0.6318
• Rouge2: 0.0
• Rougel: 0.6319
• Rougelsum: 0.6317

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	0.6831	0	62.2359	0.5956	0.4735	0.5953	0.0	0.5956	0.5955
0.6842	1	11370	0.6585	0.0078	79.7651	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6542	2	22740	0.6584	0.0156	95.6635	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6549	3	34110	0.6582	0.0312	124.3407	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6642	4	45480	0.6601	0.0625	183.9387	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6625	5	56850	0.6606	0.125	300.3655	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6573	6	68220	0.6757	0.25	530.3992	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6566	7	79590	0.6819	0.5	993.3325	0.6320	0.3872	0.6318	0.0	0.6319	0.6317

Framework versions

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