2bb0db9bd915dc8f565ccf36b9bc1817

This model is a fine-tuned version of facebook/opt-350m on the contemmcm/cls_20newsgroups dataset. It achieves the following results on the evaluation set:

• Loss: 0.6564
• Data Size: 1.0
• Epoch Runtime: 142.7291
• Accuracy: 0.8952
• F1 Macro: 0.8928

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	5.0934	0	10.6609	0.0451	0.0183
No log	1	499	4.2240	0.0078	11.8500	0.0902	0.0354
0.0483	2	998	2.8545	0.0156	12.6976	0.1447	0.0917
0.0583	3	1497	1.3402	0.0312	15.5551	0.5587	0.5131
0.057	4	1996	1.0782	0.0625	19.9243	0.6628	0.6383
0.9432	5	2495	0.9467	0.125	28.3133	0.7155	0.7131
0.616	6	2994	0.6047	0.25	45.2688	0.8087	0.8045
0.4779	7	3493	0.5300	0.5	77.7799	0.8425	0.8400
0.337	8.0	3992	0.3920	1.0	142.1769	0.8773	0.8771
0.276	9.0	4491	0.3966	1.0	142.5086	0.8770	0.8738
0.1872	10.0	4990	0.5704	1.0	142.5976	0.8700	0.8689
0.1943	11.0	5489	0.5666	1.0	142.9747	0.8695	0.8693
0.1588	12.0	5988	0.6564	1.0	142.7291	0.8952	0.8928

Framework versions

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