livinNector/distil-mbert-cased-dra-tam-mal-aw-classification-finetune

m-minilm-l12-h384-dra-tam-mal-aw-classification-finetune

This model is a fine-tuned version of distilbert/distilbert-base-multilingual-cased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6624
• Accuracy: 0.7245
• F1: 0.7305

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.7017	0.2222	20	0.6924	0.5322	0.3747
0.6931	0.4444	40	0.7164	0.4735	0.6427
0.694	0.6667	60	0.6837	0.5395	0.1046
0.6889	0.8889	80	0.6825	0.5224	0.6601
0.6759	1.1111	100	0.6390	0.6292	0.6838
0.6486	1.3333	120	0.6445	0.6096	0.6908
0.6388	1.5556	140	0.6199	0.6381	0.6934
0.6136	1.7778	160	0.6024	0.6740	0.6748
0.6041	2.0	180	0.6386	0.6813	0.7031
0.553	2.2222	200	0.6085	0.6960	0.7014
0.5415	2.4444	220	0.6208	0.6960	0.7168
0.5388	2.6667	240	0.5562	0.7115	0.6845
0.531	2.8889	260	0.5753	0.7172	0.6790
0.52	3.1111	280	0.6330	0.7123	0.7299
0.47	3.3333	300	0.5680	0.7115	0.6954
0.4593	3.5556	320	0.6563	0.6879	0.7206
0.458	3.7778	340	0.5790	0.7148	0.6998
0.4679	4.0	360	0.5888	0.7115	0.6916
0.3955	4.2222	380	0.6469	0.7164	0.7290
0.3783	4.4444	400	0.6272	0.7229	0.7195
0.3573	4.6667	420	0.6348	0.7221	0.7108
0.3803	4.8889	440	0.6122	0.7278	0.7136
0.3271	5.1111	460	0.6624	0.7245	0.7305
0.2893	5.3333	480	0.7192	0.7205	0.7236
0.2778	5.5556	500	0.6962	0.7221	0.7083
0.318	5.7778	520	0.6967	0.7205	0.7182
0.3161	6.0	540	0.6901	0.7205	0.7158

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.20.3