sambodhan_urgency_classifier

This model is a fine-tuned version of xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0687
• Accuracy: 0.6867
• F1 Macro: 0.6909
• F1 Weighted: 0.6900
• Precision Macro: 0.7047
• Recall Macro: 0.6878
• Precision Weighted: 0.7039
• Recall Weighted: 0.6867

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: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	F1 Weighted	Precision Macro	Recall Macro	Precision Weighted	Recall Weighted
0.123	0.6667	50	0.1217	0.3333	0.1974	0.2000	0.2353	0.3276	0.2350	0.3333
0.1191	1.3333	100	0.0996	0.5333	0.5322	0.5309	0.5440	0.5353	0.5433	0.5333
0.0918	2.0	150	0.0881	0.5933	0.5835	0.5821	0.6286	0.5960	0.6280	0.5933
0.0674	2.6667	200	0.0687	0.6867	0.6909	0.6900	0.7047	0.6878	0.7039	0.6867

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1