eternis_router_encoder_sft_4Sep

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

• Loss: 0.1710
• Mse: 0.1710
• Model Accuracy: 0.3285

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_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: cosine
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Mse	Model Accuracy
0.4021	0.3429	300	0.1875	0.1875	0.153
0.3702	0.6857	600	0.1809	0.1809	0.1757
0.36	1.0286	900	0.1762	0.1762	0.3068
0.3686	1.3714	1200	0.1788	0.1788	0.2333
0.3337	1.7143	1500	0.1733	0.1733	0.3192
0.3293	2.0571	1800	0.1709	0.1709	0.3297
0.3245	2.4	2100	0.1706	0.1706	0.3035
0.3158	2.7429	2400	0.1710	0.1710	0.3285

Framework versions

• Transformers 4.56.0
• Pytorch 2.7.0
• Datasets 4.0.0
• Tokenizers 0.22.0