0b6d416e9d76964a09a5a9cc035c6d28

This model is a fine-tuned version of facebook/opt-1.3b on the contemmcm/cls_20newsgroups dataset. It achieves the following results on the evaluation set:

• Loss: 0.5366
• Data Size: 1.0
• Epoch Runtime: 352.1939
• Accuracy: 0.8861
• F1 Macro: 0.8857

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	3.3569	0	25.4804	0.0466	0.0292
No log	1	499	3.0470	0.0078	28.0726	0.1477	0.1149
0.0328	2	998	1.3445	0.0156	32.1826	0.5769	0.5407
0.0378	3	1497	0.9910	0.0312	39.8066	0.6993	0.6825
0.0356	4	1996	0.7908	0.0625	51.2975	0.7460	0.7367
0.7834	5	2495	0.6936	0.125	72.6094	0.7853	0.7824
0.5408	6	2994	0.5709	0.25	115.4693	0.8221	0.8204
0.4811	7	3493	0.4357	0.5	193.4714	0.8619	0.8627
0.3698	8.0	3992	0.3826	1.0	351.1356	0.8805	0.8802
0.3302	9.0	4491	0.4717	1.0	350.8585	0.8662	0.8621
0.2451	10.0	4990	0.4367	1.0	352.3412	0.8775	0.8752
0.2072	11.0	5489	0.5502	1.0	348.9006	0.8639	0.8651
0.1809	12.0	5988	0.5366	1.0	352.1939	0.8861	0.8857

Framework versions

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