7dfff2b27cdc0a649329fa6ec794efef

This model is a fine-tuned version of Qwen/Qwen2.5-0.5B on the contemmcm/cls_20newsgroups dataset. It achieves the following results on the evaluation set:

• Loss: 2.0717
• Data Size: 1.0
• Epoch Runtime: 156.2281
• Accuracy: 0.8853
• F1 Macro: 0.8860

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	37.2698	0	11.8932	0.0587	0.0243
No log	1	499	24.7148	0.0078	13.1672	0.0670	0.0292
0.3452	2	998	8.3875	0.0156	14.3531	0.4088	0.3468
0.2159	3	1497	4.8810	0.0312	17.5346	0.6447	0.6215
0.1742	4	1996	4.4864	0.0625	23.1189	0.6754	0.6563
3.0059	5	2495	2.8406	0.125	32.2734	0.7928	0.7921
1.9544	6	2994	2.1654	0.25	51.1976	0.8448	0.8449
1.5006	7	3493	1.6527	0.5	86.2517	0.8871	0.8877
1.2034	8.0	3992	1.4685	1.0	155.3312	0.8828	0.8836
0.8282	9.0	4491	1.5330	1.0	156.0791	0.8982	0.8977
0.6517	10.0	4990	1.8718	1.0	156.4726	0.8891	0.8879
0.5912	11.0	5489	2.3930	1.0	156.1961	0.8753	0.8742
0.5203	12.0	5988	2.0717	1.0	156.2281	0.8853	0.8860

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1