mbzuai-ugrip-statement-tuning/MBERT_revised_2e-06_64_0.1_0.01_110k

MBERT_revised-revised-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: 0.4229
• Accuracy: 0.7431
• F1: 0.7564
• Precision: 0.7276
• Recall: 0.7876

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: 64
• eval_batch_size: 64
• 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
0.6935	0.7220	1000	0.6865	0.5423	0.6628	0.5286	0.8883
0.6351	1.4440	2000	0.5492	0.6788	0.7211	0.6435	0.8199
0.5285	2.1661	3000	0.4751	0.7106	0.7428	0.6753	0.8255
0.4727	2.8881	4000	0.4398	0.7220	0.7598	0.6753	0.8686
0.4405	3.6101	5000	0.4205	0.7293	0.7641	0.6838	0.8656
0.4221	4.3321	6000	0.4113	0.7339	0.7290	0.7525	0.7069
0.41	5.0542	7000	0.4145	0.7386	0.7700	0.6945	0.8640
0.3957	5.7762	8000	0.4018	0.7399	0.7807	0.6811	0.9145
0.388	6.4982	9000	0.3993	0.7416	0.7789	0.6872	0.8989
0.3807	7.2202	10000	0.3996	0.7413	0.7786	0.6870	0.8985
0.3734	7.9422	11000	0.3939	0.7459	0.7459	0.7553	0.7368
0.3719	8.6643	12000	0.3923	0.7455	0.7733	0.7043	0.8572
0.3625	9.3863	13000	0.3942	0.7405	0.7759	0.6894	0.8873
0.3646	10.1083	14000	0.4070	0.7422	0.7694	0.7032	0.8493
0.3578	10.8303	15000	0.3988	0.7435	0.7505	0.7396	0.7617
0.354	11.5523	16000	0.4031	0.7470	0.7616	0.7283	0.7980
0.3509	12.2744	17000	0.4089	0.7472	0.7476	0.7559	0.7394
0.3503	12.9964	18000	0.4029	0.7488	0.7676	0.7220	0.8195
0.3451	13.7184	19000	0.4127	0.7439	0.7449	0.7515	0.7384
0.3427	14.4404	20000	0.4196	0.7469	0.7684	0.7159	0.8291
0.3425	15.1625	21000	0.4168	0.7501	0.7736	0.7147	0.8429
0.3386	15.8845	22000	0.4131	0.7459	0.7507	0.7457	0.7559
0.3349	16.6065	23000	0.4279	0.7443	0.7472	0.7483	0.7460
0.336	17.3285	24000	0.4199	0.7441	0.7523	0.7378	0.7675
0.3341	18.0505	25000	0.4209	0.7449	0.7523	0.7400	0.7651
0.3333	18.7726	26000	0.4216	0.7461	0.7602	0.7284	0.7948
0.3323	19.4946	27000	0.4229	0.7431	0.7564	0.7276	0.7876

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.1
• Datasets 2.19.2
• Tokenizers 0.19.1