62cd28e34f5f26c8346a44abe464a177

This model is a fine-tuned version of albert/albert-base-v2 on the contemmcm/cls_mmlu dataset. It achieves the following results on the evaluation set:

• Loss: 1.4000
• Data Size: 1.0
• Epoch Runtime: 19.4388
• Accuracy: 0.2533
• F1 Macro: 0.1011

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	1.4014	0	1.2304	0.2440	0.1756
No log	1	438	1.4383	0.0078	1.5110	0.2527	0.1577
No log	2	876	1.4161	0.0156	1.4658	0.2473	0.1309
No log	3	1314	1.4251	0.0312	1.7369	0.2513	0.1007
No log	4	1752	1.4151	0.0625	2.3603	0.2520	0.1104
0.079	5	2190	1.3906	0.125	3.4193	0.2593	0.1617
0.1868	6	2628	1.4026	0.25	5.7923	0.2487	0.0996
1.4143	7	3066	1.3987	0.5	10.0454	0.2487	0.0996
1.403	8.0	3504	1.3911	1.0	19.0544	0.2527	0.1008
1.3931	9.0	3942	1.4000	1.0	19.4388	0.2533	0.1011

Framework versions

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