x5-ner-ru

This model is a fine-tuned version of ai-forever/ruRoberta-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5277
• Precision: 0.9444
• Recall: 0.9581
• F1: 0.9512
• Accuracy: 0.9510

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: 8
• eval_batch_size: 8
• 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

Training results

Training Loss	Epoch	Step	Accuracy	F1	Validation Loss	Precision	Recall
0.4099	1.0	3066	0.9327	0.9241	0.3133	0.9141	0.9343
0.2574	2.0	6132	0.9380	0.9312	0.3137	0.9162	0.9467
0.2085	3.0	9198	0.9434	0.9388	0.2655	0.9262	0.9518
0.1484	4.0	12264	0.9454	0.9449	0.3075	0.9359	0.9540
0.1094	5.0	15330	0.9491	0.9463	0.3448	0.9422	0.9505
0.077	6.0	18396	0.9509	0.9483	0.3936	0.9393	0.9575
0.0559	7.0	21462	0.9504	0.9505	0.3870	0.9421	0.9591
0.0279	8.0	24528	0.5072	0.9426	0.9581	0.9503	0.9485
0.0142	9.0	27594	0.4978	0.9449	0.9572	0.9510	0.9501
0.0155	10.0	30660	0.5277	0.9444	0.9581	0.9512	0.9510

Framework versions

• Transformers 4.56.2
• Pytorch 2.7.1+cu118
• Datasets 3.6.0
• Tokenizers 0.22.0