google_electra-small-discriminator_freeze_negsam

This model is a fine-tuned version of google/electra-small-discriminator on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6004
• Accuracy: 0.8810
• F1 Macro: 0.4523
• Precision Destination: 0.9256
• Recall Destination: 0.9463
• Precision Origin: 0.6111
• Recall Origin: 0.75
• Precision Other: 0.3
• Recall Other: 0.15
• Precision Transit: 0.0
• Recall Transit: 0.0
• Super Metric: 1.6963
• Raw Super Metric: 1.6963

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
• 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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	Precision Destination	Recall Destination	Precision Origin	Recall Origin	Precision Other	Recall Other	Precision Transit	Recall Transit	Super Metric	Raw Super Metric
1.1254	1.0	269	0.9680	0.8637	0.2798	0.8674	0.9955	0.625	0.1136	0.0	0.0	0.0	0.0	1.0936	1.1092
0.7562	2.0	538	0.6568	0.8541	0.3775	0.9196	0.9217	0.4853	0.75	0.0	0.0	0.0	0.0	1.6717	1.6717
0.556	3.0	807	0.5947	0.8656	0.3922	0.9138	0.9485	0.5532	0.5909	0.1	0.05	0.0	0.0	1.5395	1.5395
0.4718	4.0	1076	0.4209	0.8503	0.4168	0.9458	0.8971	0.4396	0.9091	0.3333	0.1	0.0	0.0	1.8062	1.8062
0.4991	5.0	1345	0.5285	0.8733	0.4032	0.9237	0.9485	0.5357	0.6818	0.1667	0.05	0.0	0.0	1.6304	1.6304
0.3546	6.0	1614	0.5450	0.8772	0.4269	0.9297	0.9463	0.5424	0.7273	0.2857	0.1	0.0	0.0	1.6736	1.6736
0.3032	7.0	1883	0.6399	0.8772	0.4660	0.9158	0.9485	0.6222	0.6364	0.3846	0.25	0.0	0.0	1.5849	1.5849
0.2039	8.0	2152	0.5867	0.8791	0.4473	0.9256	0.9463	0.5818	0.7273	0.3333	0.15	0.0	0.0	1.6736	1.6736
0.3003	9.0	2421	0.5695	0.8791	0.4508	0.9294	0.9418	0.6071	0.7727	0.25	0.15	0.0	0.0	1.7146	1.7146
0.2917	10.0	2690	0.6004	0.8810	0.4523	0.9256	0.9463	0.6111	0.75	0.3	0.15	0.0	0.0	1.6963	1.6963

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1