8571428e7bdd0e8e629d5bc331ffc7b1

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

• Loss: 1.3930
• Data Size: 1.0
• Epoch Runtime: 23.9487
• Accuracy: 0.2453
• F1 Macro: 0.0985

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	1.3946	0	1.2554	0.2254	0.1666
No log	1	438	1.4093	0.0078	1.7044	0.2527	0.1008
No log	2	876	1.3891	0.0156	1.8915	0.2467	0.1554
No log	3	1314	1.4024	0.0312	2.2937	0.2620	0.1685
No log	4	1752	1.3960	0.0625	3.1259	0.2547	0.1688
0.0779	5	2190	1.3928	0.125	4.5148	0.2626	0.1746
0.1842	6	2628	1.3875	0.25	8.0890	0.2487	0.0996
1.3905	7	3066	1.3905	0.5	13.3380	0.2487	0.0996
1.389	8.0	3504	1.3884	1.0	25.1925	0.2487	0.0996
1.3892	9.0	3942	1.3895	1.0	24.3298	0.2753	0.1686
1.3886	10.0	4380	1.3930	1.0	23.9487	0.2453	0.0985

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1