24360a0c796e69eff362d2fce0f5ed39

This model is a fine-tuned version of deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B on the contemmcm/cls_mmlu dataset. It achieves the following results on the evaluation set:

• Loss: 12.1413
• Data Size: 1.0
• Epoch Runtime: 113.7088
• Accuracy: 0.3351
• F1 Macro: 0.3354

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	8.2861	0	4.0631	0.25	0.2088
No log	1	438	8.3686	0.0078	5.0238	0.2540	0.1882
No log	2	876	6.3913	0.0156	7.2881	0.2487	0.1156
No log	3	1314	5.9872	0.0312	11.1671	0.2553	0.1581
No log	4	1752	5.6783	0.0625	16.7666	0.2547	0.1723
0.3357	5	2190	5.7888	0.125	24.3332	0.2646	0.1537
0.7298	6	2628	5.6148	0.25	39.6324	0.2733	0.1765
5.1909	7	3066	5.5691	0.5	63.4640	0.2919	0.2537
4.0195	8.0	3504	6.0028	1.0	117.6137	0.3557	0.3468
1.4588	9.0	3942	8.7482	1.0	112.2217	0.3418	0.3374
0.8301	10.0	4380	11.0854	1.0	113.6984	0.3384	0.3343
0.673	11.0	4818	12.1413	1.0	113.7088	0.3351	0.3354

Framework versions

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