8c83d2f8302e22b19854555e6ca0a709

This model is a fine-tuned version of Qwen/Qwen2.5-3B on the contemmcm/cls_mmlu dataset. It achieves the following results on the evaluation set:

• Loss: 6.1832
• Data Size: 1.0
• Epoch Runtime: 190.5439
• Accuracy: 0.2606
• F1 Macro: 0.2082

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	12.0552	0	5.4490	0.2374	0.1818
No log	1	438	10.2797	0.0078	7.0467	0.25	0.1941
No log	2	876	6.2946	0.0156	13.6007	0.2680	0.2504
No log	3	1314	5.8603	0.0312	20.7798	0.2440	0.1407
No log	4	1752	7.7342	0.0625	30.3905	0.2606	0.1740
0.3854	5	2190	5.9440	0.125	44.5456	0.2513	0.1007
0.7906	6	2628	5.5794	0.25	66.4156	0.2440	0.1028
5.6377	7	3066	5.6063	0.5	107.9188	0.2527	0.1220
6.2255	8.0	3504	5.6918	1.0	196.2729	0.25	0.1037
5.7039	9.0	3942	5.9319	1.0	195.6416	0.2447	0.1999
5.1094	10.0	4380	6.1832	1.0	190.5439	0.2606	0.2082

Framework versions

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