618a4f2bfaf95c80657fe91852b9933f

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

• Loss: 0.4856
• Data Size: 1.0
• Epoch Runtime: 30.6639
• Accuracy: 0.8818
• F1 Macro: 0.8820

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	3.0123	0	2.2889	0.0512	0.0058
No log	1	499	3.0103	0.0078	2.9290	0.0476	0.0142
0.0305	2	998	2.9595	0.0156	2.9063	0.0585	0.0220
0.0536	3	1497	2.3570	0.0312	3.5234	0.2898	0.1913
0.0887	4	1996	1.6419	0.0625	4.4454	0.5055	0.4374
1.42	5	2495	1.1262	0.125	6.4097	0.6532	0.6168
0.8556	6	2994	0.8115	0.25	9.8103	0.7364	0.7160
0.685	7	3493	0.6446	0.5	16.8619	0.8029	0.8043
0.489	8.0	3992	0.4472	1.0	32.0749	0.8619	0.8597
0.3862	9.0	4491	0.4321	1.0	30.4782	0.8695	0.8695
0.2226	10.0	4990	0.4587	1.0	30.8529	0.8687	0.8662
0.1968	11.0	5489	0.4383	1.0	31.5420	0.8808	0.8792
0.1869	12.0	5988	0.4795	1.0	30.1504	0.8858	0.8843
0.1479	13.0	6487	0.4856	1.0	30.6639	0.8818	0.8820

Framework versions

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