fine_tuned_mBERT

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

• Loss: 0.1614
• F1: 0.7869
• F5: 0.8020
• Precision: 0.75
• Recall: 0.8276

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.2
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	F5	Precision	Recall
No log	1.0	30	0.2615	0.0	0.0	0.0	0.0
No log	2.0	60	0.1838	0.5333	0.4626	0.8889	0.3810
No log	3.0	90	0.2338	0.3077	0.2491	0.8	0.1905
No log	4.0	120	0.2003	0.6667	0.6268	0.8	0.5714
No log	5.0	150	0.2643	0.5	0.4906	0.5263	0.4762
No log	6.0	180	0.2211	0.6486	0.6168	0.75	0.5714
No log	7.0	210	0.2233	0.6	0.6391	0.5172	0.7143
No log	8.0	240	0.3328	0.5	0.5647	0.3846	0.7143

Framework versions

• Transformers 4.38.1
• Pytorch 2.2.1+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2