49ed9bc4c55f01d46b2c12ab72ec9d62

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

• Loss: 2.0766
• Data Size: 1.0
• Epoch Runtime: 15.4281
• Accuracy: 0.2733
• F1 Macro: 0.2685

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	1.3892	0	1.0008	0.2340	0.1558
No log	1	438	1.3942	0.0078	1.5112	0.2527	0.1246
No log	2	876	1.3893	0.0156	1.4385	0.2520	0.1319
No log	3	1314	1.3910	0.0312	1.8313	0.2540	0.1051
No log	4	1752	1.3910	0.0625	2.3392	0.2487	0.0996
0.0777	5	2190	1.3923	0.125	3.0842	0.2706	0.1803
0.1834	6	2628	1.3844	0.25	5.0158	0.2699	0.2066
1.386	7	3066	1.3841	0.5	8.4317	0.2773	0.2081
1.3674	8.0	3504	1.3887	1.0	15.7729	0.2793	0.2412
1.2399	9.0	3942	1.4814	1.0	15.3051	0.2766	0.2609
0.9979	10.0	4380	1.7245	1.0	15.2724	0.2806	0.2733
0.7247	11.0	4818	2.0766	1.0	15.4281	0.2733	0.2685

Framework versions

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