exp_acha_mluke_base_last

This model is a fine-tuned version of studio-ousia/mluke-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2625

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: 1e-06
• train_batch_size: 20
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_ratio: 0.45
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss
0.6813	1.0	167	0.6868
0.6414	2.0	334	0.6080
0.5949	3.0	501	0.5543
0.5828	4.0	668	0.5174
0.5562	5.0	835	0.4876
0.4159	6.0	1002	0.4337
0.4039	7.0	1169	0.3666
0.3607	8.0	1336	0.3284
0.3763	9.0	1503	0.2964
0.3379	10.0	1670	0.2744
0.3169	11.0	1837	0.2677
0.2445	12.0	2004	0.2656
0.2472	13.0	2171	0.2626
0.2644	14.0	2338	0.2629
0.2566	15.0	2505	0.2625

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1