3aeb8b0983a818412f9f2d5355babec0

This model is a fine-tuned version of studio-ousia/luke-large-lite on the contemmcm/cls_mmlu dataset. It achieves the following results on the evaluation set:

• Loss: 1.3936
• Data Size: 0.25
• Epoch Runtime: 25.2994
• Accuracy: 0.2487
• F1 Macro: 0.0996

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	1.3965	0	3.1417	0.2533	0.1011
No log	1	438	1.4000	0.0078	4.1124	0.2527	0.1008
No log	2	876	1.3893	0.0156	5.0953	0.2533	0.1011
No log	3	1314	1.3913	0.0312	7.0244	0.2533	0.1011
No log	4	1752	1.3989	0.0625	9.7475	0.2387	0.1066
0.0795	5	2190	1.3935	0.125	15.4880	0.2487	0.0996
0.1851	6	2628	1.3936	0.25	25.2994	0.2487	0.0996

Framework versions

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