185e230313922b7a40e98231632dec76

This model is a fine-tuned version of distilbert/distilbert-base-uncased-distilled-squad on the contemmcm/cls_20newsgroups dataset. It achieves the following results on the evaluation set:

• Loss: 0.5163
• Data Size: 1.0
• Epoch Runtime: 16.0121
• Accuracy: 0.8843
• F1 Macro: 0.8841

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	2.9986	0	1.6946	0.0554	0.0188
No log	1	499	3.0017	0.0078	2.0341	0.0517	0.0049
0.03	2	998	2.9965	0.0156	2.0998	0.0675	0.0130
0.0541	3	1497	2.9538	0.0312	2.3147	0.0718	0.0294
0.1012	4	1996	2.2237	0.0625	2.8324	0.3319	0.2569
1.7604	5	2495	1.2200	0.125	3.6229	0.6474	0.6043
0.8464	6	2994	0.7306	0.25	5.3783	0.7649	0.7437
0.5582	7	3493	0.5086	0.5	9.1266	0.8448	0.8438
0.3886	8.0	3992	0.4048	1.0	16.4705	0.8778	0.8780
0.2512	9.0	4491	0.4063	1.0	16.3345	0.8722	0.8691
0.1763	10.0	4990	0.4324	1.0	16.0892	0.8790	0.8763
0.1473	11.0	5489	0.4789	1.0	16.0381	0.8765	0.8732
0.1211	12.0	5988	0.5163	1.0	16.0121	0.8843	0.8841

Framework versions

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