robeczech_lr1e-05_bs4_train287

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

• Loss: 0.1729
• Precision: 0.9536
• Recall: 0.9618
• F1: 0.9577
• Accuracy: 0.9742

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• 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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	72	0.9668	0.5829	0.5365	0.5587	0.7878
No log	2.0	144	0.5208	0.8484	0.8484	0.8484	0.9221
No log	3.0	216	0.3739	0.8830	0.8889	0.8859	0.9410
No log	4.0	288	0.3164	0.8919	0.8967	0.8943	0.9464
No log	5.0	360	0.2869	0.8939	0.8952	0.8946	0.9466
No log	6.0	432	0.2507	0.9135	0.9179	0.9157	0.9565
0.5681	7.0	504	0.2295	0.9278	0.9305	0.9291	0.9623
0.5681	8.0	576	0.2141	0.9301	0.9382	0.9341	0.9648
0.5681	9.0	648	0.1990	0.9404	0.9450	0.9427	0.9688
0.5681	10.0	720	0.2032	0.9316	0.9401	0.9358	0.9655
0.5681	11.0	792	0.1902	0.9430	0.9421	0.9425	0.9688
0.5681	12.0	864	0.1869	0.9416	0.9416	0.9416	0.9680

Framework versions

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