d8175b00ab4e2c8207d05762af6eb373

This model is a fine-tuned version of facebook/opt-125m on the contemmcm/cls_mmlu dataset. It achieves the following results on the evaluation set:

• Loss: 2.5393
• Data Size: 1.0
• Epoch Runtime: 33.5713
• Accuracy: 0.2912
• F1 Macro: 0.2868

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	1.6096	0	2.1114	0.2527	0.2025
No log	1	438	1.5600	0.0078	2.4108	0.2580	0.2248
No log	2	876	1.5458	0.0156	2.4349	0.2520	0.1787
No log	3	1314	1.4634	0.0312	3.0744	0.2560	0.1629
No log	4	1752	1.4080	0.0625	4.3402	0.2759	0.1933
0.0798	5	2190	1.4082	0.125	6.1891	0.2434	0.1656
0.1827	6	2628	1.3963	0.25	10.0084	0.2713	0.1821
1.3727	7	3066	1.3922	0.5	18.1110	0.2779	0.2156
1.3235	8.0	3504	1.4183	1.0	35.3584	0.2626	0.2452
0.9911	9.0	3942	1.7998	1.0	33.1053	0.2859	0.2546
0.629	10.0	4380	2.2739	1.0	33.6909	0.2979	0.2779
0.4356	11.0	4818	2.5393	1.0	33.5713	0.2912	0.2868

Framework versions

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