ea1ee1c000444fe1f8e435062f327efa

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

• Loss: 0.4756
• Data Size: 1.0
• Epoch Runtime: 16.1789
• Accuracy: 0.8906
• F1 Macro: 0.8901

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	2.9995	0	1.6618	0.0882	0.0335
No log	1	499	2.9654	0.0078	2.1288	0.1096	0.0354
0.0299	2	998	2.7608	0.0156	2.0618	0.2944	0.2278
0.0517	3	1497	2.1728	0.0312	2.2667	0.5091	0.4712
0.0817	4	1996	1.4872	0.0625	2.7556	0.5995	0.5668
1.2683	5	2495	1.0075	0.125	3.5976	0.6837	0.6516
0.7574	6	2994	0.7118	0.25	5.4475	0.7729	0.7640
0.5627	7	3493	0.5550	0.5	8.6945	0.8226	0.8216
0.3997	8.0	3992	0.3923	1.0	16.0252	0.8816	0.8807
0.2855	9.0	4491	0.4027	1.0	15.7041	0.8730	0.8711
0.1865	10.0	4990	0.4425	1.0	15.9558	0.8808	0.8793
0.1525	11.0	5489	0.4544	1.0	16.1262	0.8821	0.8824
0.1115	12.0	5988	0.4756	1.0	16.1789	0.8906	0.8901

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.1