b8ef1c83e42c037d93c7964a42a8e13a

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

• Loss: 0.5226
• Data Size: 1.0
• Epoch Runtime: 70.6120
• Accuracy: 0.8818
• F1 Macro: 0.8789

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	3.2835	0	4.6108	0.0426	0.0045
No log	1	499	3.0752	0.0078	5.3027	0.0572	0.0116
0.0321	2	998	2.7700	0.0156	6.0473	0.2674	0.2060
0.0522	3	1497	2.0658	0.0312	7.5209	0.4662	0.4078
0.0793	4	1996	1.1344	0.0625	10.7138	0.6615	0.6463
1.0675	5	2495	0.7983	0.125	14.1706	0.7503	0.7333
0.6505	6	2994	0.6495	0.25	22.2204	0.7949	0.7856
0.4902	7	3493	0.4933	0.5	38.3982	0.8445	0.8449
0.3522	8.0	3992	0.3947	1.0	70.7934	0.8805	0.8797
0.3043	9.0	4491	0.4096	1.0	70.7686	0.8773	0.8754
0.1788	10.0	4990	0.4094	1.0	70.4774	0.8889	0.8868
0.1806	11.0	5489	0.4507	1.0	70.7006	0.8828	0.8812
0.1565	12.0	5988	0.5226	1.0	70.6120	0.8818	0.8789

Framework versions

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