MBERT_multilingual-outputs

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: 0.6542
• Accuracy: 0.7265
• F1: 0.7449
• Precision: 0.7226
• Recall: 0.7685

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-06
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.6934	1.2225	1000	0.6863	0.5622	0.4867	0.6227	0.3995
0.6304	2.4450	2000	0.5690	0.6687	0.6699	0.6946	0.6468
0.5145	3.6675	3000	0.5066	0.6962	0.7009	0.7175	0.6852
0.451	4.8900	4000	0.4770	0.7107	0.7221	0.7207	0.7235
0.4102	6.1125	5000	0.4892	0.7182	0.7418	0.7079	0.7791
0.3832	7.3350	6000	0.4712	0.7223	0.7314	0.7353	0.7275
0.3597	8.5575	7000	0.4848	0.7368	0.7543	0.7323	0.7778
0.3427	9.7800	8000	0.4802	0.7409	0.7677	0.7186	0.8241
0.3242	11.0024	9000	0.5468	0.7278	0.7494	0.7184	0.7831
0.3075	12.2249	10000	0.5706	0.7244	0.7343	0.7357	0.7328
0.2897	13.4474	11000	0.5962	0.7182	0.7263	0.7332	0.7196
0.2883	14.6699	12000	0.5836	0.7230	0.7343	0.7319	0.7368
0.2702	15.8924	13000	0.6106	0.7354	0.7631	0.7135	0.8201
0.2651	17.1149	14000	0.6323	0.7278	0.7503	0.7169	0.7870
0.2605	18.3374	15000	0.6400	0.7292	0.7451	0.7291	0.7619
0.2473	19.5599	16000	0.6542	0.7265	0.7449	0.7226	0.7685

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1