bert-secabilite-regressor

This model is a fine-tuned version of dascim/juribert-tiny on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0255
• Model Preparation Time: 0.0004
• Mse: 0.0256
• Mae: 0.1108

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-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Mse	Mae
0.0971	1.0	108	0.0579	0.0004	0.0580	0.1952
0.0528	2.0	216	0.0377	0.0004	0.0379	0.1473
0.0423	3.0	324	0.0313	0.0004	0.0314	0.1301
0.0366	4.0	432	0.0284	0.0004	0.0285	0.1213
0.0342	5.0	540	0.0270	0.0004	0.0272	0.1163
0.032	6.0	648	0.0261	0.0004	0.0263	0.1132
0.0311	7.0	756	0.0257	0.0004	0.0258	0.1114
0.0306	8.0	864	0.0255	0.0004	0.0256	0.1108

Framework versions

• Transformers 4.51.3
• Pytorch 2.7.0
• Datasets 3.5.0
• Tokenizers 0.21.1