multilingual_bert_AGRO

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: 4.9701
• Exact Match: 24.8571
• F1 Score: 56.8185

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: 3407
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match	F1 Score
6.2135	0.0053	1	6.2224	0.0	9.8241
6.2165	0.0107	2	6.1874	0.0	9.8716
6.1776	0.0160	3	6.1182	0.0	10.1769
6.1126	0.0214	4	6.0144	0.0	11.2194
6.0166	0.0267	5	5.8717	0.0752	12.2552
5.8816	0.0321	6	5.6741	2.2556	18.5657
5.7374	0.0374	7	5.4450	12.1805	37.6517
5.5652	0.0428	8	5.1969	23.8346	53.3283
5.2962	0.0481	9	4.9758	26.5414	56.9819
5.0538	0.0535	10	4.8192	22.1805	56.7266
5.0246	0.0588	11	4.6919	18.2707	56.2903
4.8358	0.0641	12	4.5354	18.2707	56.7852

Framework versions

• Transformers 4.46.3
• Pytorch 2.4.0
• Datasets 3.1.0
• Tokenizers 0.20.3