ae28e7ebb7d9cbfe5fab80f2dc9b3e8f

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

• Loss: 0.4567
• Data Size: 1.0
• Epoch Runtime: 28.1754
• Accuracy: 0.8899
• F1 Macro: 0.8909

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	3.0771	0	2.2185	0.0421	0.0040
No log	1	499	3.0426	0.0078	2.8423	0.0731	0.0132
0.0299	2	998	2.9487	0.0156	2.8471	0.1013	0.0643
0.0539	3	1497	2.3216	0.0312	3.4317	0.4244	0.3413
0.0868	4	1996	1.3515	0.0625	4.1711	0.6739	0.6600
1.1741	5	2495	0.8866	0.125	5.8750	0.7440	0.7249
0.6706	6	2994	0.6355	0.25	9.1158	0.8024	0.7956
0.5251	7	3493	0.4800	0.5	15.5168	0.8503	0.8492
0.3548	8.0	3992	0.3722	1.0	29.3467	0.8831	0.8826
0.2732	9.0	4491	0.4217	1.0	28.2032	0.875	0.8723
0.1664	10.0	4990	0.4126	1.0	28.1608	0.8884	0.8866
0.1467	11.0	5489	0.5042	1.0	28.8827	0.8778	0.8763
0.1296	12.0	5988	0.4567	1.0	28.1754	0.8899	0.8909

Framework versions

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