cbe82ba8fee77d57afa5f7f843182cce

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

• Loss: 0.5399
• Data Size: 1.0
• Epoch Runtime: 82.2425
• Accuracy: 0.8856
• F1 Macro: 0.8847

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	3.0376	0	5.9593	0.0444	0.0042
No log	1	499	3.0181	0.0078	6.7945	0.0464	0.0069
0.0305	2	998	2.9006	0.0156	7.5779	0.1164	0.0552
0.0535	3	1497	1.3128	0.0312	9.4689	0.6006	0.5504
0.0595	4	1996	0.7798	0.0625	12.4450	0.7676	0.7609
0.8524	5	2495	0.6610	0.125	17.6479	0.7838	0.7767
0.5791	6	2994	0.5916	0.25	27.4641	0.8188	0.8154
0.5079	7	3493	0.6145	0.5	46.4001	0.8201	0.8210
0.4142	8.0	3992	0.4174	1.0	82.4275	0.8679	0.8682
0.3697	9.0	4491	0.4188	1.0	81.4875	0.8722	0.8690
0.2631	10.0	4990	0.4935	1.0	82.4433	0.8684	0.8666
0.2369	11.0	5489	0.4488	1.0	82.8571	0.8745	0.8707
0.2325	12.0	5988	0.4129	1.0	82.1203	0.8929	0.8917
0.2153	13.0	6487	0.4741	1.0	81.4728	0.8942	0.8936
0.1591	14.0	6986	0.4670	1.0	82.0985	0.8924	0.8926
0.1865	15.0	7485	0.5947	1.0	81.9544	0.8753	0.8725
0.2375	16.0	7984	0.5399	1.0	82.2425	0.8856	0.8847

Framework versions

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