ruBert-large-upos

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

• Loss: 0.4344
• Precision: 0.8307
• Recall: 0.7502
• F1: 0.7831
• Accuracy: 0.8686

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	338	0.4759	0.7967	0.7249	0.7532	0.8557
No log	2.0	676	0.4344	0.8307	0.7502	0.7831	0.8686
No log	3.0	1014	0.6906	0.7842	0.7480	0.7563	0.8674
No log	4.0	1352	0.4757	0.8185	0.7578	0.7777	0.8816
No log	5.0	1690	0.6291	0.7791	0.7721	0.7670	0.8792
No log	6.0	2028	0.6466	0.7967	0.7677	0.7721	0.8863
No log	7.0	2366	0.7072	0.7751	0.7700	0.7704	0.8809
No log	8.0	2704	0.7623	0.7957	0.7678	0.7749	0.8838
No log	9.0	3042	0.7458	0.7922	0.7716	0.7773	0.8873
No log	10.0	3380	0.7560	0.7916	0.7709	0.7767	0.8869

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1